Time of Flight Mass Spectrometry-Schlag

ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands

Pages 1-414 have been published as a special issue of the International Journal of Mass Spectro-metry and Ion Processes, Volume 131 (1994)

ISBN: 0-444-81875-8

© 1994 Elsevier Science B.V. All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, P.O. Box 521, 1000 AM Amsterdam, The Netherlands.

Special regulations for readers in the USA - This publication has been registered with the Copyright Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA. All other copyright questions, including photocopying outside of the USA, should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified.

No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the materials herein.

This book is printed on acid-free paper.

Printed in The Netherlands

TIME-OF-FLIGHT MASS SPECTROMETRY AND ITS APPLICATIONS

Edited by

E.W. Schlag

Institut für Physicalische und Theoretische Chemie, Universität München, Lichtenbergstrasse 4,

D-85748, Germany

ELSEVIER Amsterdam - London - New York - Tokyo 1994

International Journal of Mass Spectrometry and Ion Processes 131 (1994) ix Elsevier Science B.V.

ix

FOREWORD

The resurgence of time-of-flight mass spectrometry (TOF-MS) has had its origin in the simplicity of con-struction and application of such instruments together with the high transmission and the great increase in resolution that has been achieved. The instrument naturally lends itself to a coupling with pulsed laser sources, though this is not a necessary requirement. It also affords a time resolution far beyond that traditionally achieved with mass spectrometric rapid scan techniques — a recent example being the real-time analysis of a multi-component mixture from an automobile exhaust. Furthermore, the mass range appears to be extremely large: mass up to 500 kDa and beyond what is being readily measured in the laboratory today.

Nevertheless, one must recognize that much yet remains to be understood, particularly in the mechanisms of large-molecule vaporization and detection, as well as the laser physics and chemistry in the ionization chamber. Using pump-probe techniques, TOF-MS can be extended into the femtosecond regime. Using ZEKE techniques, TOF-MS can be extended to laser selected state resolution of mass spectrometric break-down patterns — or state resolution in ion/molecule reactions.

The present set of contributions attempts to give a survey of current applications from many of the active groups in the field. The work starts with a contribution by Mamyrin who initially suggested the idea of building a reflecting instrument (RETOF) which in a simple way compensates for the inherent inaccuracies of an ion source so that resolutions of m/z 5000 are achieved with grid instruments; even m/z 20 000 resolution has been attained in the new grid-free RETOF instruments. In fact, today even the simplest home-made reflecting flight tubes routinely have resolutions in excess of m/z 3000. This, in combination with a simple channel plate detector and a digital oscilloscope, constitutes a working system.

The papers presented contain a variety of new applications which are no doubt just the beginning of large new areas of application. Principally, one should here mention the extreme state selection made possible by ZEKE techniques, such as demonstrated in the contributions by Johnson and Neusser — as well as the extreme mass applications for biomolecules by Hillenkamp — and for DNA analysis by Williams.

By presenting this work in journal and book form it is hoped that this will be of help to the many groups intending to initiate work in this rapidly expanding new area of mass spectrometry. It is capable of handling extremely sophisticated questions of timing and state selection, as well as questions of rapid routine analysis. In contrast to the conventional mass spectrometry of large expensive machines, TOF techniques have moved the field into the individual laboratory of home construction for many applications.

We wish to acknowledge the help in putting this volume together from Dr. Boesl, Dr. Weinkauf, Professor Neusser, Dr. Selzle, and Dr. Yeretzian who formed an important consultation base for this volume.

E.W. SCHLAG

International Journal of Mass Spectrometry and Ion Processes 131 (1994) 1-19 1 0168-1176/94/S07.00 © 1994 - Elsevier Science B.V. All rights reserved

Laser assisted reflectron time-of-flight mass spectrometry

B.A. Mamyrin A.F. Ioffe Physico-Technical Institute Russian Academy of Sciences, Polytechnicheskaya 26, 194021, St. Petersburg, Russian Federation

Pulsed-mode laser assisted ionization schemes are extensively used in connection with time-of-flight mass spectrometric techniques, particularly when large-mass, thermally labile molecules i.e. biomolecules, proteins, and DNA have to be analysed. Along with the high resolution accomplished with the introduction of the reflector fields, these techniques have received considerable attention, in particular due to their ability to record a mass spectrum over the whole mass range with each single pulse.

It is also important to note that the ionization volume (the space in which ions are created) can be considerably larger than in static mass spectrometers and with essentially unlimited mass-range. Furthermore, the sensitivity of laser-assisted reflectron time-of-flight mass spectrometry can almost reach its physical limit (a few atoms or molecules).

Many modifications of laser assisted reflectrons have been developed. The key differences reside, on the one hand in the methods employed to focus the times-of-flight of the ions on the detector, and on the other hand with the specific sources used.

One can expect to witness in the near future an ever increasing interest in these techniques with a wide range of new applications in fundamental and applied science or technologies such as materials science, analytical chemistry, pharma-cology, biochemistry and genetics, to cite only a few.

Key words: Laser assisted reflectron; Pulse-laser ion sources; Thermally-labile molecules

(Received 15 July 1993; accepted 14 October 1993)

Abstract

Introduction

A reflectron is a magnet-free time-of-flight mass spectrometer with the capacity to achieve second-order time focusing with regard to variation of ion energies and angles of divergence of their departure from the source. The concept of separating ions of differing charge-to-mass ratio via times-of-flight (TOF MS) were originally proposed in 1948 [1]. However, in its first realization, all the key elements (pulsed ion source, electronics and detector) were so imperfect that the original instrument was of no practical use. It could hardly separate ions that differed in mass by less than 30-50%.

The first TOF mass spectrometers of any practi-cal interest were proposed in the 1950s [2-4]. A schematic diagram of this type of instrument is given in Fig. 1. Their resolution m /Am was 100— 300 at a partial pressure of about 10~10 Torr in the ion source.

The main advantages of such systems in compar-ison with magnetic mass spectrometers are:

(i) A mass spectrum over the complete mass range of neutral particles within the ionization zone can be obtained in fractions of microseconds;

(ii) In principal, no upper mass limit exists for this type of mass analyzer;

(iii) The ionization volume, the volume from

SSDI0168-1176(93)03891-0

2 B.A. Mamyrinjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 1-19

Detector (MCP)

°Ooi°o" PflJ)_^to Amplifier

ΔΜ 2At 2AL

Fig. 1. Linear time-of-flight mass analyzer.

which ions can be extracted without loss of resol-ution, could exceed, by several orders of mag-nitude, the one for static magnetic devices;

(iv) The construction is rather straightforward and inexpensive.

However, resolution of simple TOF instruments is restricted due to several practical and technical reasons, a serious one being the effect of the energy spread of ions created in the pulsed source.

Since most of the questions requiring mass spec-trometric techniques were solved in the 1960s and early 1970s by well developed and industrially-produced static magnetic instruments at high resolution, together with the rather inexpensive quadrupole mass spectrometers for application at limited resolutions, a pause occurred in TOF-MS development and use. At the same time, organic chemistry, biophysics, pharmacology and the physics of clusters were in urgent need of mass-spectrometric analysis of large, thermally labile molecules whose vaporization and ionization could not be performed by traditional methods of thermal evaporation, surface or electron impact ionization. Furthermore, progress in trace analysis of molecular samples was seriously hampered by the lack of efficient ionization and volatilization techniques to obtain ions of thermally labile and non-volatile molecules.

After a long search for alternative routes, as well as improvements to existing schemes, important breakthroughs were triggered once high-mass molecular ions could be produced as very short (temporal and spacial) ion packages, a develop-

ment of fundamental importance for the future of the method (see section 4.2 below). The most convenient and "pure" technique of pulsed vapor-ization and ionization of molecules and atoms is achieved by means of pulsed lasers, a method which is still finding ever growing applications.

Compared to these modern versions of reflec-trons, the sensitivity of static instruments operat-ing with pulsed sources appears to be extremely low. Static mass spectrometers can only detect ions of one specific mass from each ion pulse. Furthermore, ions produced in pulsed sources gen-erally have considerable energy spread, reducing the sensitivity of static-field MS additionally. Moreover, pulsed methods of ion production enable ions to be obtained with masses of hundreds of thousands of daltons, however, their analysis in magnetic field MS requires extremely high energy fields.

A problem of the first TOF instruments was the large initial energy spread of the ions produced in the extraction zone, which considerably limited the theoretically-possible resolution. Important pro-gress was finally achieved as the double-gap elec-trostatic ion reflectron was introduced by us [5-8] (see Fig. 2).

The construction and operation of reflectrons is based on three fundamental concepts:

(i) The introduction of the double field reflec-trons enabled one to compensate for differ-ences of times-of-flight in the field-free drift regions (Lj,L2), effected by the energy spread in the source.

B.A. Mamyrin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 1-19 3

Pulsed Source

Detector

(MCP)

Λ Οο*

θ ο · · " o o ;

1 _ j 1

ΙΟΟ7·Ν> 1 lOo_j_·' 1 1 Mini

. u , . u R

> W. R. Ampl.

Computer

Fig. 2. Reflectron with a two-stage reflector.

(ii) With a single uniform reflecting field, only first-order time focusing of the energy spread of the ions can be accomplished. To obtain second-order focusing (necessary for high resolution at large ion energy spread) an additional indepen-dent parameter is needed, whose variation enables the second derivative of flight-times to be zero with respect to energy (see section 2). This is achieved in a double reflecting field design, in which ions are retarded in the first gap of the reflector and reflected in the second gap.

(iii) Since the region where the ions are generated in the source is not an ideal plane, but rather exhi-bit some spread along the instrument axis, one has to focus this initial spatial distribution of ions to narrow packets at the end of the extraction fields. This is accomplished by proper choice of field strengths of the two extraction fields. Such focus-ing is best at the beginning of the first field-free drift region (leaving the extraction fields), since high focusing field intensities are required. In routine linear TOF instruments weak focusing fields are used within the source so as to focus ion packets at large distances from the source, the trade-off being a more pronounced defocusing due to initial energy spread or space-charge effects.

Recently, highly sensitive reflectrons have been developed with resolutions up to 35 000 [9], and practically unlimited mass range.

Table 1 gives the basic parameters of pulsed laser ion sources as well as the properties of a reflectron

mass analyser. It follows from this table that pulsed laser ion-sources and reflectrons match as perfectly as if they were specifically created for each other. Combining pulsed lasers for ion generation with reflectrons as mass-analysers provides us with mass spectrometric techniques of high sensitivity and resolution. This explains why so many appli-cations in modern physical, chemical and bio-logical areas have been reported by now.

The renewed interest and great success of reflec-trons in a large number of modern applications led Price and Milnes to entitle their recent review on reflectrons "The renaissance of time-of-flight mass-spectrometry" [10].

1. Laser-ionization methods

7.7. Advances of laser ion generation method

The first ruby laser with chromium impurities was constructed by T. Mayman in 1960 and in 1963 the first paper appeared [11] on laser appli-cation for ion production.

Nowadays, laser-assisted methods of evapora-tion and ionization have gained wide acceptance and are used in several industrial instruments [12-14]. One can give at least five advantages of ion formation via the action of pulsed laser radia-tion over traditional ionization techniques.

(i) Resonant laser ionization methods of atoms or molecules in the gas-phase enable extremely high


Table 1 Properties of pulsed laser ion sources and reflection mass analyzers

Properties of pulsed laser ion sources

Properties of reflectron mass-analyzers

Ions are emitted with a short temporal (3-20) x 10"9 s and spatial profile along the axis of the mass analyser

Ions of all masses are concentrated in short packets; no ions are emitted between pulses

Ions of a given mass are formed with considerable energy spread

Generation of ions with masses of hundreds of thousands of atomic mass units is possible

A limit for the sensitivity (upper bond for ionization efficiency) is reached for large atomic and molecular aggregates by resonant excitation schemes with laser light

To achieve high mass resolution, ion packets are necessary with the minimum spread in their time of formation as well as spatial spread along the instrument axis

Due to the time-scan of the TOF mass-spectrometer, ions of all masses within the initial ion packet are detected; ions created between individual pulses are undesirable

Due to second order time focusing with respect to the energy and angle of the ions leaving the source, the effect of their energy spread is rather small

The mass range is essentially unlimited

Modern detection techniques enable to improve sensitivity by averaging single-shot spectra over multiple laser pulses. Transmission of ions from the source to the detector can exceed 50%

sensitivities and high selectivities to be reached (a few atoms inside ionization region can be detected) [15].

(ii) Fragmentation-free ionization with pulsed laser techniques can be realized for thermally labile, high-mass biomolecules, where traditional techniques fail (peptides, proteins and large clusters) [16,17].

(iii) For the analysis of the molecular and atomic composition of surfaces, an extremely high degree of control can be reached in the position on the surface as well as penetration depth [18].

(iv) In some cases, time for sample-analysis can be decreased by several orders of magnitude, com-pared to traditional methods [19].

In general, evaporation and ionization of atoms or molecules is required prior to mass-spectral analysis. Here, different combinations of tradi-tional techniques with laser methods are possible:

(i) thermal vaporization and subsequent laser ionization;

(ii) laser vaporization and electron impact ionization;

(iii) laser excitation of free molecules and their subsequent ionization by electric field pulse;

(iv) successive evaporation and ionization of atoms and molecules from solids by laser-light of different frequencies.

The excitation by means of laser for subsequent ionization can be either "resonant" (the frequency of the laser is tuned precisely to an atomic or mole-cular excited state) or "non-resonant" (where the radiation power density is the controlling para-meter and the frequency is of only secondary importance for the ionization efficiency). It's quite natural that, for the absorption of several photons by a molecule, sufficiently high energy density in the laser focus is required.

1.2. Non-resonant methods of ion production

The first application of pulsed lasers for ioniz-ation [20], (Fig. 3) were based on non-resonant absorption of radiation at high power densities (109—1011 Wem - 2 ) from solids, forming a dense plasma. This plasma contained singly- and multiply-charged atomic ions of elements con-tained in the sample.

Processes occurring in a non-resonant inter-action of laser radiation with solids, such as heat-ing and evaporation, vapor ionization, ejection of particles from the plasma into the vacuum, as well as recombination of multiply-charged ions, were studied in numerous works [21-26]. However, some problems concerning elemental analysis of solids still remain to be solved. In particular, systematic variations of relative sensitivity

B.A. Mamyrinjint. J. Mass Spectrom. Ion Processes 131 (1994) 1-19 5

Sample

celeration

Lens

Fig. 3. Laser ion source with non-resonance light absorption by solids.

coefficients (RSC) for different elements [27,28] still await in-depth exploration and explanation.

Numerous publications are dedicated to RSC problems of complex substances. Their aim is to develop laser-assisted methods for quantitative elemental composition analysis without the need to calibrate by means of standards. Some progress has been made in this research on the analysis of simple salts and geological samples of complex composition [29,30]. The content of 64 elements was quantitatively determined for the analysis of basalt and meteorites samples [30]. The laser mass spectrometric data fits those of the chemical analy-sis within about 20% at elemental densities exceeding 10~5%. However, at present, specific calibration samples are required in most cases.

Problems relating to the properties of the plasma created by the laser pulse, its stability and its effects on the resolution, are still poorly understood. One might assume that a flat sample surface parallel to the extraction fields from which an ion packet is ejected during a short laser pulse would be an ideal source for TOF MS. It appears, however, that the ion packet formed has considerable tem-poral and spatial spread. At the first portion of the path flight the plasma packet has such a high charge density that its conductivity is close to that of the metal. Therefore, it is impossible to control this plasma by electric fields. Only once the plasma packet has expanded and separated from the sample surface by 1-3 cm does the decrease of charge density occur simultaneously with a reduction of the charged particle density

by recombination of multiply-charged particles. As a result, given an initial spread of the ion packet along the instrument axis of 2-3 mm, and an effective drift length of about 2.5-3 m, a resol-ution of only about 1000 can be reached. This is quite sufficient to resolve ions of different elements, but is insufficient if small concentrations of elements have to be measured in the presence of cluster or multiply-charged ions.

To decrease both the space-charge effects of the plasma and the energy spread of the produced ions for elemental analysis, one can use the technique of thin film (0.1-1 ^m) laser ablation [31]. The ion energy spreads can be decreased to several tens of electronvolts instead of hundreds of electronvolts for the bulk solid sample. The initial packet width is also decreased (especially for organic matrix).

A detailed review on fundamental processes and technological realizations of laser assisted ion sources and references on elemental analysis can be found in Ref. 32.

Decreasing the laser radiation power density to 108-104W cm , complicated complexes of atoms can be vaporized and ionized (organic molecules included) from the surface of the bulk sample or from thin films deposited on metallic substrates. The ratio of molecular peak intensity to the sum of fragment peak intensities decreases with increase of radiation power density. For example, according to the description of the LAM MA-1000 instrument for a molecule of M = 292 u, variation of laser power density from 5 x l 0 8 W c m ~ 2 to 5 x 107W cm changes the ratio of fragment to parent peak from « 0.5 to 5 respectively.

Non-resonant ionization of free atoms and mol-ecules is rather difficult. Nevertheless, an efficient way of non-resonant ionization of organic mol-ecules in the gas phase has been proposed in Refs. 33 and 34 (Fig. 4).

13. Resonance methods of ion production

Resonance excitation and ionization of atoms and molecules started to be intensively studied and applied in physical experiments in the early


Film Au

— o o ; to Mass

ο·#· Reflectron

Fig. 4. Laser induced ionization source with molecular absorp-tion on the prism surface [34].

1970s after the introduction of continuously tunable dye lasers. The use of selective two-photon ionization of molecules for mass-spectrometry was proposed in Ref. 35.

The possibility of obtaining high power density in the laser beam enables the realization of resonant multiphoton ionization (MPI) of any atom or molecule by monochromatic light, if the following conditions are satisfied: nhv = W— W0

and hv is larger than the energy of the electronic transition from the W level into the continuum (here W and W0 are the high-lying discrete and ground-state levels, see Fig. 5(a)). Multifrequency versions of resonance ionization are also possible (Fig. 5(b)).

Ionization with excitation to autoionizing levels lying in the continuum (Fig. 5(c)) is a type of resonant method. In this case, the necessary radia-tion power is significantly decreased, even if hv does not correspond to the transition to any resonance level in the discrete region.

hv //////*///// /////A/////

W hv

hv

hv

0-

hvo 7T/T7X7777/

hv-,

hv,

o-

hv

hv

Fig. 5. Ionization schemes of resonance enhanced multiphoton excitation: (a), using monochromatic light; (b), using multi-frequency resonance ionization; (c), ionization with excitation to autoionization levels lying in the continuum.

The possibility of selective resonant ionization of free atoms is important for two reasons:

(i) on the one hand, the required laser radiation power is drastically decreased, which simplifies the apparatus and decreases the possible emission of background ions;

(ii) on the other hand, the selective ionization enables ions of minor components in the sample to be produced efficiently without ionization of other components.

These peculiarities of selective resonant ioniz-ation enable the best possible sensitivity to be achieved in mass spectrometry of several atoms [36,37]. More detailed information on resonant excitation and ionization of atoms and molecules is reviewed in Ref. 38.

Resonant absorption of laser light has been successfully applied in methods of matrix-assisted UV laser desorption ionization (UV-LDI) of heavy thermally labile molecules [39-43]. In this tech-nique, the molecules being analyzed are dissolved in a matrix of relatively high absorption coefficient in the UV region of the laser light (i.e. nicotine acid), which does not excite the sample molecules. After the laser pulse, a lot of energy is set free in the matrix as a result of photochemical reactions. Ionized particles appear as free electrons as well as protons. As a result of the excess energy, all species explode into vacuum including the intact sample molecules as neutral, positive or negative parent-ions, and protonated ions. The plasma which expands into vacuum is cooling down and stabilizes the sample molecule. The operation of a reflectron with such a source will be considered below (section 4).

I A. Combined methods of ion production

To produce ions, laser radiation can be used in conjunction with other methods. For example, the focused laser pulse evaporates the sample of thermally-labile substance, and the ionic com-ponent is removed by the electric field. The remain-ing neutral molecules are ionized by electron impact. This provides the possibility of investi-

B.A. Mamyrinjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 1-19 7

gating separately, ionic and neutral components and also the possibility of performing other special studies which are related to velocity distributions and densities of laser evaporated species.

Bekov et al. [44], have developed the technique of single atom detection to a high standard. Minor impurity atoms of the laser desorbed sample were excited to Rydberg levels. The ionization is then performed by pulsed field ionization.

In numerous publications the vaporization of thermally labile molecules was performed by IR light pulses with subsequent ionization of vaporized molecules by UV light of a second laser (see, for example, the description of the TOF-1 instrument of the Bruker company). Selectivity in ionization and decrease of ion energy spread are favored by cooling the vaporized molecules in a supersonic jet of buffer gas [45,46].

2. Reflectron mass analysers

2.1. Mass reflectron operation principle

Ions formed in thin packets in a plane not far from the source of Fig. 2 can be focused to similar thin packets in a plane before the detector at any spreads of their energies. For this to be performed, they have to be retarded and reflected in an electric field with the potential being adjusted according to the law Ux — ax2 where a is a constant coefficient and x is the flight path inside the retarding field. The time necessary to stop the particle and return it to the entrance of the reflecting field is

Here q and m are the charge and mass of the ion, #t/is its energy, and xmax = (U/a)1^2 is the distance to the reflection point.

It follows from Eq. (1) that the time of ion transit in a parabolic field is independent of its energy. Unfortunately, it is quite difficult to realize such a field (or its close approximation) in practice.

+u B * u R

Fig. 6. Time-of-flight focusing by reflection in a homogeneous electric field: (a), one-stage reflector; (b), two-stage reflector (deceleration and reflection).

A partial compensation of different ion times-of-flight due to energy variation can be performed during the flight-time in the field-free region as well as during the reflection in a plane electric field with a potential distribution Ux = ax (Fig. 6(a)) [47].

The total time of ion motion in such a system from plane 1 to plane 2 (Fig. 2) is

t = tL + tR (2)

where

is the time of motion in the field-free drift space;

_2dR(2qU\"2

is the time of motion in the reflecting field; L = Lx + L2; dR and UR are the distance and potential difference between the reflector electro-des respectively; qU is the ion energy, correspond-

B.A. Mamyrin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 1-19

ing to the velocity component parallel to the system axis. Assuming the extraction potential of the source to be U = KU0, where t/0 corresponds to the mean ion energy, and K is a factor close to unity, one can represent Eq. (2) in the form:

L t =

2qU0\l/2

m J where

F=K-l/2 + B(K)l/2

and

4dRU0 B =

LU*

The weak dependence of total ion time-of-flight on the energy corresponds to the condition aF/άΚ = 0 in the point K = 1 which results in the relation for the field strength calculation within the reflector:

F - U * - A U 0

«R L

As for the condition â2F/dK2 Φ 0, it is impossible to get second order time focusing in this system.

The resolution limited by ion energy variations is

M t F0 Ru = (3)

AM 2(t-t0) 2(F-F0)

where F0 = {F)K=l = 2. Assuming K — 1 + δ, where ί < 1 we got

Rv « A/62 from Eq. (3). One can see from Table 2 that a sharp decrease in

the effect of ion energy spread can be obtained as compared to conventional TOF MS.

Table 2 Resolution in reflectron mass analyzers

0.001 0.005 0.01 0.02 0.05

a U-Vp

y<>

b t0

' 2/m a x-(m i„ ·

4 x 106

1.6 x 105

4 x 104

1 x 104

1.6 x 103

The angular spread of ions leaving the source leads to a distribution of the velocity component parallel to the system axis in proportion to cos a whereas the energy component varies directly with cos2 a.

In the system with dimensions of about 1 m and maximum angles a = (1-2)° the distribution of the longitudinal energy component is less than 10~3, which is quite negligible, as can be seen in Table 2. Consequently, the system provides focusing with respect to energy and angle spread of the ions leav-ing the source.

2.2. Mass reflectron with a two-stage reflector

To obtain a higher degree of time focusing with respect to energy, a reflectron mass-analyser with a two-stage reflector has been proposed [5-8,48-50] (Fig. 6(b)).

The ion time-of-flight in this system becomes: t = iL + iB -f JR. Here tL, iB and tR are the times-of-flight in the field-free region as well as in the retarding and reflecting regions of the reflector.

Using the previous designations, we obtain

L m 4dR Un t =

ϋΛι/2 K -1/2 +

2 ^ m

-1/2

2qUo\l/2 UB

m J

Κι/Δ- - a 1/2

+ 4<4 ^('ΐ

1/2

(4)

and hence

L t = {K-"2 + AB

x [Kx/2 -{K- P)l/2} + AR(K- P)l/2}

(5)

B.A. Mamyrinllnt. J. Mass Spectrom. Ion Processes 131 (1994) 1-19 9

a i

^J

1 1 7-^\ - h -

3_j / J / " / * " N . F~ T>QX

1 ~ T~ 2 ~\~/~ m'n

1 V i y\

1 L-/ 1

i i ! -

Up un u max

Fig. 7. Ion flight time as a function of ion energy in a reflectron mass-analyzer with deceleration and reflection fields [51].

where the dimensionless parameters are

An = 4dBU0

Λρ = 4^R ί/0 Un P = —

i/o LUB ' Λ LUR

and F\ is the function representing the dependence

oît<mK=U/UQ. It follows from Eqs. (4) and (5) that the flight

time / is a function of 4 parameters

\dB dR UR C/Bl t = t

L ' L ' l/n ' £/, '0

and of the range of the permitted ion energy Umax/Umin. The graphs corresponding to Eqs. (4) and (5) are given in Fig. 7. Figure 7(a) when times-of-flight are equal at points 1 and 2 as well as at points 3 and 4 and dt/dU = 0 at points 2 and 3, corresponds to the minimum packet width. Hence there are four coupled equations whose solution for the given value Umax/Umin enables the whole-set of parameters to be determined as well as the resolution

R = U = '° u AU 2(/max - /min) Schmikk and Dubenskij [51] were the first to

publish such considerations. In more recent work [52], a similar approach was applied. Calculations show that for C/max/i/min = 1.05; 1.1; 1.2, the

resolution, without taking into account the field distortions at the reflector grids, becomes 370 x 103, 50 x 103 and 7 x 103.

2.3. Effect of the inclination angle of ion trajectory at the exit of the ion source

For the operation of a reflectron mass analyzer (Fig. 2) it is necessary that ions depart from the source at angles which will ensure their arrival at the detector after reflection. This can be achieved in several ways. The axis of the source can be rotated relative to the axis of the instrument. In this case the path length will be different from different points within the ion packet (produced in a plane) to the detector plane. In order to com-pensate for this, one should tilt the detector planes.

However, in the case of a point source, the diver-gent ion beam will not affect the resolution and compensation is not required. In fact, for an effective drift path of 2-3 m length and a detector diameter of about 6 cm, the admitted angle of the trajectory inclination is a «(0.5-1)° . The velocity component parallel to the instrument axis will vary in proportion to cos a, and hence Δ Vj V = 3 x 10~4 with respect to ion energy varia-tions of AU/U « 0.001. Such energy variations are negligible (see section 2.2).

That's why the conclusion drawn in Ref. 52, that it is necessary to turn the detector plane, is not properly justified in this case; the construction given in Fig. 8 [52] requires correction for reflec-tion point variation with the change of incidence angle (ion trajectory inclination).

The second way to induce an inclination of the ion trajectories with respect to the instrument axis is by way of a reflecting plane condenser near the source. A fairly weak field in the reflecting con-denser means that extremely small angles are required; the edge effects provide no essential influ-ence with a resolution of about (5-10) x 103

(especially in combination with a point source). The inclination of the ion beam relative to the

instrument axis can be caused by tilting the last grid

10 B.A. Mamyrin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 1-19

I I I I I I Γ I 1 I 1

•c-c—·■+■ I I

I I I I I I L

— 0 -^ ® 1

Detector

to Amplifier

Fig. 8. A linear reflectron with transparent ion source.

of the source acceleration field with respect to the plane of the ionization field grids. At resolutions up to 10 x 103 such systems ensure a trajectory inclination up to (1-2)°.

Finally, a deflection angle of the beam without noticeable sensitivity loss is reasonably provided by the natural divergence of ion trajectories at a well focused laser beam, assuming a point source.

2.4. Modifications of the reflectron mass-analyzer design

In Refs. 48-50,53-55 a new "linear" reflectron (Fig. 8) was proposed. The peculiarity of this lay-out consists in the transparent ion source which allows ions to pass to the detector once they have been reflected. In Fig. 8 the setup is shown; du d5

and d2 are the distances of the acceleration grids in the ion source; Lx and L2 are field-free drift regions; d3 and d4 are the lengths of the decelerating and reflecting fields in the reflector.

The advantages of this configuration are: Smaller diameter of the analyzer chamber is needed, distortionless ion movement for trajec-tories inclined towards the device axis, simplified ion-optical systems for decreasing beam diver-gence, and the possibility of constructing instru-ments of small size with very high sensitivity. However, the high probability of noise on the detector due to its close vicinity to the source and the pulsed extraction fields may be considered a disadvantage of the layout. Good agreement

between calculated and experimental data is shown in Ref. 54. Unfortunately, this layout has not been tested with laser ion sources.

In another modification to the reflectron, an axial-symmetric ion-beam path has been designed in which the ions pass from the source to the reflec-tor through an opening in the detector. The reflected ions hit the ring-shaped detector due to the natural angular divergence [56-58]. A certain increase of sensitivity hardly warrants the con-struction complexities.

Attempts were made to develop reflectrons with parabolic reflecting mirrors in order to focus the diverging ion beam at the detector. However, tech-nical problems, relatively low resolution and insig-nificant sensitivity gain suspended the research.

Shmikk and Dubenskii [59] designed a modifi-cation to the two-stage reflectron for the case of high ion energy spread. The multigap reflector consists of a deceleration stage (common for ions of every energy) and a number of successive stages with homogeneous fields in which ions are reflected with energies from qUm[n up to qU\, from qU\ up to qU2, from qll2 up to qU3, etc. up to #£/max. Calcu-lations for this system show that for optimum settings, ion flight times correspond to the curve 2 in Fig. 9; for comparison, curve 1 is given which corresponds to the routine double-gap reflector with the preset ratio.

Experimental results are approximately consis-tent with the calculations. The disadvantage of this layout seems to be the decrease of its "trans-parency" for high energy ions.

B.A. Mamyrinjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 1-19

' m n v \

l

1 \l

1 1 1

/ \

/ / 2 \

" " ™ ~

\ f A J\

1 I 1 1

J |

^ -^ - - ·"~

s\ J i V

1 1 1

A / 1

1 1 1

J\ -r

1 1 1—►

Umin Ul U2 U3 Um Q X

Fig. 9. Ion flight time as a function of energy: 1, two-stage reflector; 2, multi-stage reflector.

2.5. Influence of field distortions in the reflector

As the plane electrostatic field in the reflector is formed by grid electrodes with distances comparable to their diameters, field distortions appear due to edge effects and grid micro-optics. In order to decrease edge effects, ring-shaped "supporting" electrodes are used. Their potentials are set by a voltage divider and should correspond to the potential of the respective field plane. A possible reflectron structure is discussed in detail in Ref. 60.

Cellular grid structure leads to field sag in the direction of decreasing potential. This effect increases with the size of the grid-meshes and the field difference on both sides of the grid. Grid-induced field distortions lead to distortions of ion packet trajectories and therefore to a time-of-flight spread and a reduction of the instrument resol-ution. This effect is most pronounced near the sec-ond reflector grid which separates the deceleration and reflection field regions. Therefore, this grid must be chosen to be the most finely structured. Alternatively, two parallel equipotential grids may be used. Such a system (equally transparent in comparison with a single grid) produces less ion trajectory dispersion.

The problem of ion scattering during their passage through charged grids is considered in Refs. 60 and 61. In Ref. 60 an increase of the deceleration-field length is proposed in order to

minimize the effect of grid micro-optics. This reduces the differences of field intensities in the deceleration and reflection stage of the reflector. This solution was shown to provide a resolution several times higher and to achieve a value of R50% = 30 x 103.

An important increase in transmittance was achieved in gridless reflectors [62,63]. In such systems, the increase in sensitivity stems either from the absence of grids and from focusing of divergent ion beams. With such reflectrons a resolution of (10-30) x 103 has been achieved.

2.6. Peak shape in mass spectra

The important characteristic of reflectron mass spectrometers is their good peak shape. This means a drastic decrease of the signal near the peak base (the absence of "tails" caused by ion collisions with residual gas molecules and instrument walls). This results from the fact that scattered ions, even if they reach the detector, experience an important change in their time-of-flight compared to the unscattered. They end up being smeared out over a large time-of-flight. The mass spectrum in Fig. 10 [64] presents two neighboring peaks with different gain of output current. The peak base is shown not to stretch noticeably despite the gain increase by a factor of 104. Under similar con-ditions, in single-stage static magnetic devices, the peak base stretches strongly in comparison with its full width at half maximum (FWHM). Good peak shape is important if neighboring peaks with extremely different intensities have to be observed.

2.7. Influence of ion charge-exchange on reflectron operation

During ion collision with neutral species of residual gas, charge exchange is possible in which an ion turns into a neutral particle. This particle would move practically without change of velocity and direction:

X+ + Y = X + Y+ + AE

AE, which determines the change in kinetic and


T+i

"S 5CH

10H

-10 :

M M+1 Mass

Fig. 10. Signal shape in a time of flight reflectron [64].

internal energy of the particles after the reactive charge-exchange collision, is given by the differ-ence of the ionization potential of the two particles (several electron volts). At resonant charge exchange X+ + X —► X + X+, AE becomes zero. With ion energies of several kiloelectronvolts the charge exchange cross sections are sufficiently high: σ— 10~15cm2 (σ depends only slightly on the masses and velocities of interacting species).

The number of neutralized particles in this pro-cess is I0 — ÜLno, where L is the pass length and n is the numbr of gas molecules in cubic centimeters. Therefore when L = 100 cm and the background pressure « 10~6mbar, I0 « / + x 10~2. (For a gas pressure of « 10~5mbar the number of reneutra-lized molecules 70 may amount to 7+ x 10"1.)

For charge exchange processes happening between reflector and detector, the neutralized but fast particles would be detected like the ions. However, they would not be affected by any fields near the detector and may appear ahead of or behind the ion packet from which they have been produced. As a result, a base peak appears that may be recorded (it has the same shape as the ion peak). If low-intensity components (< 0.1%) have to be measured, the charge exchange phenomenon may cause considerable background noise. The performed experiments agree with the published tentative computations [65].

3. Laser-assisted ion sources

Principles of ion production using pulsed lasers

were briefly outlined in section 1. Here, we only note their general properties which affect the resolution of a reflectron. These aspects have already been discussed in detail [34,52].

3.1. The factors reducing resolution power of a mass reflectron

The mass resolution in a reflectron TOF instru-ment is primarily given by the parameters of the produced ion packet. In the case of laser evapora-tion sources, the ion energy spread is associated first of all with energy transfer processes of ions during their desorption at different moments of the laser pulse. For example, during the evaporation of solids with a laser intensity of (1010-1012)Wcm~2 the first plasma layer is ejected within the first picoseconds of the laser pulse. Afterwards it begins to screen the interior portion of the plasma as well as the surface of the sample. During this process, intense recombination of multicharged ions, an expansion of the plasma cloud and a separation of electrons and ions take place. In such cases the spread in ion energy for different parts of the ion packet can reach 103 eV or more [32]. The most unpleasant effect during quantitative measurements is the fact that the distribution of ion density over the kinetic energy may be non-uniform. This leads to important variations of relative peak intensities (ratios) in the mass spectra if a certain region of ion energies is cut off. For the ionization of free molecules, processes leading to considerable kinetic energy spread are also possible. Additional energy spreads are associated with the finite width of the produced ion packet, as its different layers are affected by different potential drops in the field of the ejecting pulse.

3.2. The influence of initial energy spread of neutral molecules and space charge of ion packets

In general, even small thermal energies of neutral molecules are of great importance because of the well-known "turn-around" effect. This turn-


around time is the time needed for an ion with an initial velocity antiparallel to the final ion flight direction to decelerate, re-accelerate and get back to its original formation coordinates. The turn-around time can be calculated as follows:

AtT = 4Δί/ τ

E; 2gUT\

m J

1/2

Ex is the extraction field strength of the source, where ions are produced and qAUT is the initial ion energy. With the initial thermal energies of qAUr « 0.03 eV, a separation of the electrodes of dx « 0.5 cm, an extraction-pulse of 500 V and the ion mass of M = 100 Da we calculate ΔίΓ to be 5 x l 0 ~ 9 s . This value can be significantly decreased by cooling the evaporated molecules in a pulsed supersonic jet of rare gas and by increasing the field strength Ex.

Furthermore, the space charge of the ion packet causes its expansion. At a charge density of « 1011 ions cm- 3 this starts to affect the mass reso-lution considerably. This is particularly undesir-able in the case where, along with high resolution, simultaneously high relative sensitivity is required.

3.3. The final resolution power

Instability in the synchronization of the laser and extraction pulses decreases the mass resolution in accumulated spectra. Instabilities of shape and amplitude of the extraction pulse and of the d.c. voltages may also contribute considerably to a broadening of peaks in mass-spectra and may cause fluctuations in the ratio of their magnitudes.

Additionally, the time constant of the signal pre-amplifier influences the finite width of the peaks, in particular in the low-mass region. All contributions to the time spread of the ion packet can easily be calculated for every particular system.

For a mass resolution of about 104 or greater the following technical specifications have to be achieved: the laser pulse width must be l-3ns; instabilities of voltage supplies should not exceed 10~6; the instability in trigger synchroniz-

ation and other electronic time parameters should not exceed about 10~6; stability and precision of geo-metric dimensions (parallelism, grid cell size etc.) must also correspond to the present resolution.

The overall influence of the above contributions on the total peak spread may be estimated assum-ing that they are statistically independent. Then, the overall dispersion of peak spread due to these factors is Δ/f = y/Âtj where t{ is the spread related to the factor /.

In order to calculate the overall peak width in the device output, ΔίΓ must be increased by the follow-ing time-independent contributions: the initial geo-metric packet width along the apparatus axes, the inaccuracy of the analyzer geometry (limitations of the geometrical precision), field distortions of grids and edge effects, "turn-around9' effect and space charge influence. Finally the mass resolution near the peak base can be calculated as

R = —F 1

ÇAfc+fe>?) 1/2

4. Laser-assisted reflectron instruments

4.1. Laser analysis of elemental composition of solids

The simplified scheme of the laser-assisted reflec-tron, constructed in our laboratory in 1975 [66] is given in Fig. 11. It enables local (high x-y resol-ution on the surface of the sample) elemental analysis of solids to be performed, laser plasma parameters at high laser power densities to be studied and energy spectra of ions and electrons to be registered. The overall effective ion drift length is 3 m, the mass resolution is about /?5o% ~ 1000, and the relative sensitivity 10~6

(accumulating about 100 pulses). Figure 12 represents the mass spectra recorded by this instru-ment during the study of HTSC ceramic samples SmBa2Cu306 and tungsten isotopes.

Figure 13 represents the scheme of the LAMMA-500 source, produced by Leybold-


Fig. 11. Laser-assisted reflectron for high space resolved element analysis of solids: 1, sample; 2, holder (3D adjustable); 3, lens; 4, laser; 5, sample illumination and observation systems; 6, reflector; 7, detector (MCP); 8, amplifier; 9, computer; 10, display.

Heraeus Co. In this device the laser beam shoots on a sample prepared as a thin film. By focusing the laser beam by means of an immersion objective, the local resolution was managed to be brought down to a diameter ^0.5 μηι. The possibility of studying sodium and potassium distributions in a single vegetal cell was demonstrated by Hillenkamp and Unsold [67].

Later on, a LAMMA-1000 device was produced in which the elemental analysis of solids and films deposited on metallic substrates could be realized. In this apparatus the laser beam is incident on the sample from the side opposite the ion reflector. For

low power densities this device allows organic film composition to be studied as well.

In the Institute of Space Research (Moscow) an unconventional laser mass reflectron layout (Fig. 14) was designed to study the surface composition of Phobos (a Martian satellite). The laser range-finder automatically adjusts the lenses to focus the laser beam on the Phobos sur-face during the flight of the space station at distances of 30-80 m over the surface. The ions produced fly up to the reflectron which is kept in a natural vacuum. The ion loss due to ion trajectory divergence is partially com-

16Λ+ Ceramic SmBa2Cu306

P. =101°W/cm2; cL = 12μιη Las ' Spot r

Cu

138^ +

/ B a

j

,5V

LJ Mass Spectra obtained in single Laser Pulse

R05 = 1000 (320-400) eV

Fig. 12. Mass spectra of HTSC ceramics and tungsten [66].


Immersion Liquid

vegetal cells

Fig. 13. Ion source with thin film sample in shoot through configuration [67]. Desorption is achieved by pulsed laser radiation.

pensated for by the large reflectron entrance area («30 cm2). The calculated resolution is about 200. Simulations under laboratory conditions have proven its efficiency.

4.2. Evaporation, desorption and ionization of thermally-labile organic molecules

Superheavy proteins differ from atoms or small organic molecules in that strong intermolecular forces are active. That is why, for example, some amino acids cannot be thermally evaporated as intact molecules.

The rate of energy migration along intramol-ecular bonds is very high (10-11 s or less). There-fore, the transfer of the externally-fed energy to the

Fig. 14. Reflectron developed applications in space; 1, pulsed laser; 2, lens system to regulate the focal point; 3, laser range-finder; 4, reflector and detector; 5, electronics.

excitation of the intramolecular bonds in molecules (to induce evaporation) can be avoided only by a sharp decrease of the laser pulse. Moreover, the energy needed for sublimation of molecules increases with increasing mass. Thus, the con-ditions for efficient and intact evaporation of thermally labile, heavy molecules can only be achieved if the proper laser power density (104-108WcnT2) can be set and with the shortest possible pulse widths 10~8 to 10"9 s).

The most progressive method of ion production of high-mass, thermally labile molecules (including protein) was developed in the works of Karas and co-workers [39-41,43] who proposed the method of matrix UV laser desorption ionization. In addition to the remarks in section 2.1 concerning this method, we present some numerical data. The mixture of protein molecules in the matrix is such that, for every protein molecule, there are « 105

matrix molecules. The laser pulse width is selected to be « 3-10 ns. The fast ejection of matrix mol-ecules into vacuum causes fast cooling and thus an effective production of intact protein molecules (the effect is similar to the one of supersonic gas jet). Intensive photochemical reactions causing free electron and proton production within the laser focus generate positively- and negatively-charged protein molecules. A number of substances (nicotinic acid, caffeic acid etc.) which resonantly absorb radiation in the wavelength range 220-370 nm are used as matrices. For effective registration of high mass ions (M > 5000 Da) further acceleration up to 20-30 kV in front of the detector was used.

16 B.A. Mamyrin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 1-19

Based on this method, the Finnigan Mat Company has developed the mass reflectron Vision-2000. By means of this instrument, a number of proteins were analysed. Mass spectra with resolved signals up to M = 300 000 Da have been produced using this mass spectrometer. Only an extremely small amount of sample is necesary (about 10"10g) for mass-spectral compositional analysis.

Nelson and Williams [68] showed DNA and RNA molecules with masses of M ^ 410000 Da, which were frozen in ice. During the sample evaporation by means of an IR light pulse, water vapor plumes were ejected into the vacuum with large velocities. The water molecules performed the cooling and transport of intact DNA mole-cules. Furthermore, the DNA molecular beams were ionized by UV laser.

Matrix assisted laser desorption-ionization coupled with reflectron mass spectrometry for the analysis of protein, peptide and biomolecular ions are of great importance for biochemistry, pharma-cology, medicine and will progress rapidly as they contribute to the elucidation of sample com-positions and the identification of mass and struc-ture of biomolecules. Nowadays, mass spectra of hundreds of proteins and polypeptides have been recorded.

An interesting device using two different lasers to desorb and ionize large molecules has been developed by the Bruker Company together with the group of Professor Schlag at the Technical University in Munich [62]. In this apparatus (Fig. 15) the molecular samples are evaporated by IR light pulses. The evaporated molecules are introduced in a synchronously switched super-sonic rare gas jet that cools them down. The pre-dominantly neutral molecules are swept away by the gas jet into the ionization region of the T O F -reflectron. There, the cooled molecules are excited by the UV laser pulse, whose wavelength may be set for resonant multiphoton ionization. The com-plete separation of benzene (l3C-14C) isotopic peaks has been demonstrated. The dye laser pulse-width was about «2 ns. An impressive mass

Sample

: . ' v V ^ : l £ 5 r r J F ^ ~ " ° * *® to Mass W ^ ~ / / ] f ® ^ Reflectron

2 / / ^ > # 3

Fig. 15. Laser source of the TOF 1 reflectron (Bruker) [62]: 1, IR laser beam; 2, supersonic pulsed jet of rare gas; 3, UV laser beam.

resolution was obtained (104) together with high sensitivity (gridless reflector).

In our laboratory a reflectron device was developed by Kozlov and Shchebelin [69] in which the application of two laser beams of an electron impact source or of a pulsed supersonic rare gas jet is possible (Fig. 16). The laser beam passes through the ion source in vertical or horizontal directions, which permits samples to evaporate and excites the evaporated particles in the gas phase. A wide ionization gap (up to 30 mm) being penetrated by an electron beam along the device axis permits the study of particle density distribution in different parts of the laser flare. Additional energy filters and controlling elec-trodes permit the separation of ionic and neutral plasma components as well as the study of the velocity-distributions of particles in the laser flare. This apparatus allows supersonic jet parameters and the state of the molecules being mixed into the buffer gas [69-71] to be studied in detail. By means of this device, investigations have been carried out on hetero-atomic and hetero-molecular cluster (including fullerenes) formation, as well film deposition on laser sputtering of HTSC ceramics etc.

The highest resolution in laser-assisted reflec-trons was reported by Geno et al. [72]. Here, the laser beam was arranged orthogonal to the ion-flight-direction-ionized Cs atoms. After ion formation in a volume of «1mm3 and pulsed field extraction, the energy spread was measured to be AE & 20 eV. The ion beam was shaped by


i Electron Becim

Cathode

\ put sed

^ E n e r g y

Supersonic Jet 4 = Η

Filter + UB *UF

Computer

W. R. Ampl.

Fig. 16. A laser-assisted reflectron to investigate vaporization, clustering and ionization processes [69].

focusing quadrupole electrostatic lenses, and after passing the field-free drift space was reflected in a double-field ion reflectron; after the second drift region the ions finally reach the detector. The effec-tive drift oath was about «6 m. An important fea-ture of the device was the long decelerating field of the reflector, which was comparable in length with the reflecting field [60]. Supersonic jet cooling of the clusters and very fast analog-to-digitial conversion (0.5 ns resolution) were used. The peak FWHM was A / « 3 n s at a flight time of /«210/ iS , R = t/2At = 35000 were used.

Conclusion

Nowadays, reflectron mass analysers constitute an indispensible analytical tool in a large variety of fundamental and applied fields: atomic and mol-ecular physics, physics of surface, atomic micro-scopy, physics and chemistry of polymers, geochemistry, physics of semiconductors, ecology (including sewage monitoring), numerous branches of technical control, production of semicon-ductors, extremely pure substances, monitoring of métallurgie processes, coal gasification, chemical technologies etc.

Up to now, many studies have been performed in

which reflectron mass analyzers are used with a mass resolution in the range 300-30000 with dif-ferent ion sources: electron impact ionization, ionization by strong pulsed electric fields, by fast atoms, by 252Cf fragments [72], molecular evapora-tion and ionization by pulsed laser radiation and by various combinations of simultaneous and subse-quent excitations of neutral particles with different methods.

Laser-assisted reflectrons are undoubtedly gain-ing ever-increasing importance due to their proper-ties which are detailed in this work. The most noticeable applications are the analysis of very large and thermally labile molecules and appli-cations in biochemistry, biophysics and medicine.

In a large number of reviews, various problems are considered in detail and numerous references provided [10,16,26,31,32,38,52,66,73,74].

Acknowledgements

The author is extremely grateful to Professor E.W. Schlag for inviting this contribution, and a number of associates of the mass spectrometry laboratory in the A.F. Ioffe Physical-Technical Institute are acknowledged for their contribution to experiments and discussions.


References

1 A.E. Camerona and D.E. Eggers, Rev. Sei. Instrum., 19 (1948) 605.

2 N.I. Ionov and B.A. Mamyrin, J. Tech. Phys., 23(11) (1953) 2101.

3 B.A. Mamyrin and N.I. Ionov, Prospectus, TOFMS for gas analysis with panoramic display, International Exhibition, Brussels, 1958.

4 W.C. Wiley and I.H. McLaren, Rev. Sei. Instrum., 26 (1955) 1150.

5 B.A. Mamyrin, Russian Patent No. 198034, (1966). 6 V.l. Karataev, B.A. Mamyrin and D.V. Schmikk, J. Tech.

Phys., 16(7) (1971) 1498. 7 B.A. Mamyrin, V.l. Karataev, D.V. Schmikk and V.A.

Zagulin, Sov. Phys. JETP, 37 (1973) 45. 8 M.A. Ivanov, B.N. Kozlov, B.A. Mamyrin, D.V. Schmikk

and V.G. Shchebelin, J. Tech. Phys., 62 (1983) 2039. 9 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei.

Instrum., 60(1989)347. 10 D. Price and G.J. Milnes, Int. J. Mass Spectrom. Ion Pro-

cesses, 99 (1990) 1. 11 R.E. Honig, Appl. Phys. Lett., 3 (1963) 8. 12 Lamma-1000, Prospectus, Leybold-Heraeus.

H.J. Heiner, S. Maier, H. Vogt and R. Wechsung, Proc. Ninth International Mass Spectrometry Conference, Vienna, Austria, 1982.

13 T. Dingle, B.W. Grifiths and J.C. Ruckman, Vacuum, 31 (1988)571. Lima, Prospectus, Cambridge Mass Spectrometry.

14 Vision-2000, Prospectus Finnigan Mat. 15 V.N. Fedoseyev, V.S. Letokhov and V.l. Mishin, Opt.

Commun., 52 (1984) 24. 16 M. Karas and U. Bahr, Trends Anal. Chem., 9(10) (1990). 17 T. Bergmann, H. Goehlich, T.P. Martin, H. Schaber

and G. Malegiannakis, Rev. Sei. Instrum., 61(10) (1990) 2585.

18 F. Hillenkamp and E. Unsold, Appl. Phys., 8 (1975) 341. 19 V.l. Belousov, J. Anal. Chem., 39(5) (1984) 909. 20 R.E. Honig and J.R. Woolston, Appl. Phys. Lett., 2

(1963)8. 21 N.G. Basov, V.A. Boiko and V.A. Grybkov, J. Exp. Theor.

Phys., 61(1) (1971) 154. 22 A.F. Haught and D.H. Polk, Phys. Fluids, 9 (1966) 2047. 23 G.G. Devjatktyh, I.D. Kovalev and G.A. Maksimov, J.

Phys. Chem., 47(1973)2917. 24 H. Zann and H.J. Dietze, Int. J. Mass Spectrom. Ion Phys.,

22(1976)623. 25 R.J. Conzemius and J.M. Cappellen, Int. J. Mass Spec-

trom. Ion Phys., 34 (1980) 171. 26 I.D. Kovalev, G.A. Maksimov, A.I. Suchkov and N.V.

Larin, Int. J. Mass Spectrom. Ion Phys., 27 (1978) 101. 27 T.A. Basova, Yu.K. Al'turov, Yu.A. Bykovskii, Yu.N.

Kolosov and V.N. Nevolin, J. Tech. Phys., 52 (1982) 1773 (in Russian).

28 G.I. Ramendik, J. Anal. Chem., 47 (1992) 1804. 29 Yu.A. Bykovskii, N.M. Vasiliev, N.N. Degtjarenko

and V.N. Nevolin, J. Tech. Phys., 42 (1972) 1749 (in Russian).

30 Yu.A. Bykovskii, G.N. Zhuralev, V.M. Gladskoi, V.G. Degtiarev and V.N. Nevolin, J. Tech. Phys., 48 (1978) 382 (in Russian).

31 F. Hillenkamp and R. Nitche, Appl. Phys., 8 (1975) 341. 32 Yu.A. Bykovskii and V.N. Nevolin, Laser Mass Spectro-

metry, Energoatomizdat, Moscow, 1985. D.S. Simons, Appl. Surf. Sei., 31 (1988) 103.

33 M. Yang, J.R. Millard and J.P. Reilly, Opt. Commun., 57 (1985) 1884.

34 M. Yang and J.P. Reilly, Int. J. Mass Spectrom. Ion Pro-cesses, 75 (1987) 209.

35 R.V. Ambarzumian and V.S. Letokhov, Appl. Opt., 11 (1972) 354.

36 V.S. Letkohov, Commun. At. Mol. Phys., 7 (1977) 93. 37 C.Th. Alkemade, J. Appl. Spectrosc., 35 (1981) 1. 38 V.S. Letokhov, Laser Photoionization Spectroscopy,

Academic Press, Orlando, FL, 1987. 39 M. Karas, D. Bachmann, U. Bahr and F. Hillenkamp, Int.

J. Mass Spectrom. Ion Processes, 78 (1987) 53. 40 M. Karas and F. Hillenkamp, Anal. Chem., 60 (1988)

2299. 41 A. Overberg, M. Karas, U. Bahr, R. Kaufmann and

F. Hillenkamp, Rapid Commun. Mass Spectrom., 4 (1990) 293.

42 R.C. Beavis and B.T. Chait, Anal. Chem., 62 (1990) 1836.

43 M. Karas, A. Ingendoh, U. Bahr and F. Hillenkamp, Biomed. Environ. Mass Spectrom., 18 (1989) 841.

44 G.I. Bekov, V.S. Letokhov, O.I. Matveiev and V.l. Mishin, J. Exp. Teor. Phys., 75 (1978) 2092 (in Russian).

45 W. Dietz, H.J. Neusser, U. Boesl and E.W. Schlag, Chem. Phys., 66 (1982) 105.

46 H. Kuhlewind, H.J. Neusser and E.W. Schlag, J. Chem. Phys., 82 (1985) 5452.

47 B.A. Mamyrin, Author's Abstract of Doctorate, A.F. Ioffe, Physical-Technical Institute, Russian Academy of Sciences, St. Petersburg, 1966.

48 TOFMS, Patent, Great Britain, No. 1474149, (1977). 49 TOFMS, Patent, France, No. 7530831, (1978).

TOFMS, Patent, USA, No. 4072862, (1978). 50 TOFMS, Patent, German Federal Republic, No. 2532552,

(1980). 51 D.V. Shmikk and B.M. Dubenskii, J. Tech. Phys., 54

(1984), 912; Sov. Phys. Tech. Phys., 29(5) (1984). 52 U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass

Spectrom. Ion Processes, 112 (1992) 121. 53 B.A. Mamyrin, V.l. Karataev and D.V. Shmikk, Russian

Patent, No. 516306, (1976). 54 B.A. Mamyrin and D.V. Shmikk, Sov. Phys., 49 (1979) 762;

J. Exp. Teor. Phys., 76 (1979) 1500. 55 D.V. Shmikk, B.M. Dubenskii and S.V. Nosal, J. Tech.

Phys., 58 (1988) 1003 (in Russian). 56 S. Delia Negra and Y. Le Beyec, Int. J. Mass Spectrom. Ion

Processes, 61 (1984) 21. 57 B. Ernstberger, H. Krause, A. Kiermeier and H.J. Neusser,

J. Chem. Phys., 92 (1990) 5285.

B.A. Mamyrin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 1-19 19

58 I.D. Kovalev, N.V. Larin, A.I. Sutchkhov, A.M. Woronov and P.A. Shmonin, Prib. Tekh. Eksp., 6 (1985) 139 (in Russian).

59 D.V. Shmikk and B.M. Dubenskii, J. Tech. Phys., 58 (1988) 1534 (in Russian).

60 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 60(1989)347.

61 O.B. Konstantinov, B.A. Mamyrin, L.E. Shchebelina and V.G. Shchebelin, J. Tech. Phys., 56 (1986) 1075 (in Russian).

62 R. Frey, G. Weiss, H. Kaminski and E.W. Schlag, Z. Naturforsch. Teil A, 40 (1985) 1349. R. Walter, U. Boesl and E.W. Schlag, Int. J. Mass Spec-trom. Ion Phys., 71 (1986) 39. Prospectus TOF-1, Bruker, Franzen Analytik, GMBH.

63 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 61 (1990) 2592.

64 B.A. Mamyrin and V.l. Karataev, in Mass Spectrometry and Chemistry Kinetics, Nauka, Moscow, 1985.

65 A.I. Gusev, V.A. Kotchiev and G.G. Managadze, J. Tech. Phys., 17 (1991) 44 (in Russian).

66 B.A. Mamyrin, Laser Mass Spectrometry, in Problems of Physical Electronics, Russian Academy of Science, St. Petersburg, 1990 (in Russian).

67 F. Hillenkamp and E. Unsold, Appl. Phys., 8 (1975) 341. 68 W. Nelson and P. Williams, Proc, Seventh International

Conference on Secondary Ions, Mass Spectrometry, Mon-terray, September 4-8, 1989, Wiley, Chichester, p. 215.

69 B.N. Kozlov and V.G. Shchebelin, Sov. Phys. Tech. Phys., 37(1) (1992) 104; J. Tech. Phys., 62 (1992) 197.

70 A.V. Bulgakov, B.N. Kozlov, A.P. Maiorov, I.I. Pilyugin, M.R. Predtechenskii and V.G. Shchebelin, Sov. Tech. Phys. Lett., 17(21) November (1991) 18.

71 A.V. Bulgakov, A.P. Maiorov, M.R. Predtechenskii, B.N. Kozlov, I.I. Pilyugin, V.G. Shchebelin, E.M. Sher and A.N. Januta, Proc. Beijing International Conference on High Tc

Superconductors, World Scientific, Singapore, 1990, pp. 109-111.

72 P.W. Geno and R.D. Macfarlane, Int. J. Mass Spectrom. Ion Processes, 77 (1987) 75.

73 D. Price and G.J. Milnes, Int. J. Mass Spectrom. Ion Pro-cesses, 60 (1984) 61.

74 S.H. Lin, Y. Fujimura, H.J. Neusser and E.W. Schlag, Multiphoton Spectroscopy of Molecules, Academic Press, Orlando, FL, 1984.

International Journal of Mass Spectrometry and Ion Processes 131 (1994) 21-41 0168-1176/94/S07.00 © 1994 - Elsevier Science B.V. All rights reserved

How to specify the ion optical system of a time-of-flight mass spectrometer

T. Bergmann21'*, T.P. Martinb

^Bergmann Messgeräte Entwicklung, Buchenweg 9a, 82441 Ohlstadt, Germany bMax-Planck-Institut für Festkörperforschung, Heisenbergstr. 1, 70569 Stuttgart, Germany

(Received 20 April 1993; accepted 11 August 1993)

Abstract

In order to make a qualified judgement of the ion optical properties of a time-of-flight system, it is necessary to have a method of presentation that does not require an extensive background in ion optical theory to understand it. The authors present such a method. This method gives a more clear understanding of state-of-the-art time-of-flight systems and gives a hint of what developments to expect in the future.

Key words: Ion optical system; Time of flight design

1. Introduction

At present, especially when it pertains to time-of-flight mass-spectrometers with gridless reflectors, quality arguments and specifications are quite nebulous. Everybody has heard or read such state-ments before: "A mass resolution of 10000 is routine" or 'The detection limit is lOOfmol".

For the resolution we usually find a plot of some mass peak, at which the quotient (total time)/(delta time) is demonstrated. For the detection limit we find a peak cropping out of the noise, perhaps with SNR = 10.

Note that the above statements are performance examples and not specifications. Performance examples will always be of importance when demonstrating the benefits of some system. What we want to be able to do is judge the value of some performance example.

* Corresponding author.

At present, the above statements are the only type that can be obtained from a time-of-flight manufacturer. These statements can mean one of the following.

(i) They refer to a complete system. The user just puts some chemical substance into a sample holder, screws it into some port, and then obtains a mass spectrum at the terminal of his computer. Even though the above statements are usually given for some specific, usually favourable, substance, putting other substances into the sample holder will give comparable results.

(ii) Somebody has purchased just the ion optics and the associated vacuum system to use it as a component in a larger system. This person may be seriously disappointed when trying to verify the above specifications. The first statement did not specify the volume of phase space used when demonstrating the resolution. The second state-ment may not tell enough about the mass reso-lution at low detection limits.

(iii) Somebody wants to build his own optics. He

SSDI0168-1176(93)03885-P

22 T. Bergmann and T.P. Martin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 21-41

then will have to find the conditions under which the above statements should hold.

In the second and third cases it is certainly desir-able to have a method of presenting the ion optical properties of the system in such a way that the user can predict the behaviour of this system under the operating conditions he has chosen.

This is not an article on how to design ion optics for a time-of-flight mass spectrometer, this article is about judging and evaluating different designs. It is the main intention of this article to introduce a method of presenting and specifying transport through the mass spectrometer. We hope that this method facilitates a clear comparison of different time-of-flight ion optics.

2. Splitting up the system

First, it is mandatory to conceptually split up the mass spectrometer system and separate out those parts that really have to do with ion optics. The main entities are:

(i) Generation of the ions, i.e. placement of these ions into the phase space of the ion optical system. This means that we have to know the coordinate and velocity distribution which the ions have at the start time of the mass analysis.

(ii) Movement through the ion optical system of the time-of-flight mass spectrometer. Specifying this movement and interpreting it in terms of sensitivity and mass resolution is the theme of this article.

(iii) Everything that comes after the ions hit the surface that defines the end of the flight path. That surface can be e.g. a microchannel plate or the ion/ electron-conversion surface of a Daly converter.

With the above separation it is possible to present the main properties of the ion optical system with a maximum of six plots, using very convenient units, on one piece of paper. This presentation will allow the prediction of the performance under most of the operating conditions encountered in practice. Above all, this presentation is so simple that performance of the complete mass spectrometer can be approxi-

mately judged by multiplying or dividing a few simple numbers.

We will start by discussing the second entity of the system. The first and third entity are equally important; however, we will postpone their dis-cussion until later because we want to discuss them with the units and definitions that we find convenient for the presentation of the second entity.

2.1. Moving through the ion optical system

Just discussing the second conceptual entity of the mass spectrometer system, we only need to use one prerequisite and only need to answer one question.

Before we look at the movement through the ion optical system we have to know from where and with what velocities the ions start. This is the pre-requisite.

Then we want to know, as a function of all starting positions and velocities, the final positions and velocities on the detector surface. This will be the answer to our question.

The answer to that question gives us information on mass resolution and sensitivity. Firstly, we want to know from what initial coordinates and velocities we have paths ending on the detector surface. Sec-ondly we want to know the associated time errors.

The coordinate and velocity distributions are certainly different for many methods of starting ions on their flight path. For example, electron bombardment of neutral gas-phase particles gives initial velocity distributions in the energy range of 1 eV, while laser-ionizing molecules out of a super-sonic expansion gives initial velocity distributions in the energy range of 1 meV. Some methods of starting ions on their flight path launch ions from very restricted coordinate regions in the ion source. Others produce ions in a very large region and thus it is very favourable if all ions from that large region can be transported to the detector.

2.1.1. Coordinate systems The latter question has to be answered numeri-

T. Bergmann and T.P. Mariin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 21-41 23

cally. In order to do that we have to define coordi-nate systems within which to take measurements.

We will define an unprimed coordinate system A somewhere in the middle of the region of the ion source where ions start on their flight path. From the center of this unprimed coordinate system with some median initial velocity (usually zero) we start a reference ion. We will orient the coordinate system A such that its z-axis points in the direction where the ion is accelerated along the symmetry axis of the ion source.

We will put the center of a second, primed coor-dinate system B on the detector surface where the reference particle hits it. The z'-axis of this coor-dinate system should point in the direction where the ion came from. To calculate the time of flight for all other paths we take the time the reference particle needed on its path.

The x- and jc'-axis of the two-coordinate systems should be in the plane defined by the line connect-ing the ion source with the detector and the z- or z'-axis.

2.1.2. Just one set of functions From classical mechanics we know the existence

of the following functions:

xfi = u{Xj,Vj) (1)

v^v'iixjtVj) (2)

These functions give us the final coordinates and velocities as a function of initial coordinates and velocities.

Of course, it is not possible to present the beha-viour of any, perhaps complicated, six-dimensional function in just a few graphs. The following dis-cussion will show, nevertheless, that it is possible to give sufficient information on the ion optical system of a time-of-flight mass spectrometer on just one sheet of paper.

In fact, for most cases of interest, the function z(xj,Vj) = Xs(Xj,Vj) has the necessary information. Plotting the six one-dimensional functions z\x, 0,0,0,0,0),z'(0, y, 0,0,0,0), . . . usually gives a sufficiently clear picture of the function z{x^ vj).

These six plots are the ones given in the z-dispersion diagrams shown in this article.

2.1.3. The equivalent length s Before discussing an example, we need to find a

convenient set of variables for our presentation. Thus, we do a "Gedankenexperiment". Take

two particles of mass m\\ we put one particle in the center of the ion source as a reference parti-cle, and put the second at some other position in the ion source. Both particles are at rest. After the start time of the mass analysis we follow their paths through the mass spectrometer. After the reference particle hits the detector surface we stop the move-ment of the second and note its distance from the detector surface. (The second particle might hit the detector before the reference particle does: we just continue the movement of the second particle until the reference particle hits the surface.)

We now start the second particle from all possible different locations in the ion source and, consequently, we get a function of final distances from the detector surface depending on the three-dimensional variables of the initial position. (This is the left-hand side of the dispersion plots given later on.)

Next, we repeat the same process that we just have done for particles of mass mu for par-ticles of mass m2. What do we get? Exactly the same function i.e. the same plots! The reason for this coincidence is that time scales with mass if ions move in time-independent electric fields.

Knowing the distance of the second particle, we can divide the distance by its velocity and find a time error. Dividing the total time needed by the reference particle by the time error of the second particle we obtain information about mass resolution.

We now assume that all other time errors in the mass spectrometer system, i.e. the time errors of the first and third entities, can be neglected. This would mean e.g. that the laser pulse for ionizing neutral particles is extremely short and that the detector and registration electronics is infinitely fast. We

24 T. Bergmann and T.P. Mariin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 21-41

then find that the quotient (time error)/(total time) is the same for all masses.

Obviously, we must find some length by which we can divide the distances given in these plots. We will multiply the mass-dependent total time-of-flight T(m) by the mass dependent velocity v0(m) of the reference particle in the drift space of the mass spectrometer:

S = v0(m)T(m) (3)

We will call S the total equivalent length of the flight path. This length is the same for all masses! We divide the length given as the result in the dispersion plots by this total equivalent length S and thus obtain information on mass resolution for all masses that does not depend on mass. If we scale that information by the ratio of the velo-city in drift space divided by the velocity of parti-cles impinging on the detector surface (again a mass independent number), we have exactly the same values obtained when dividing (time error)/ (total time).

In Eq. (3) we have defined the total equivalent length of the flight path. Likewise we can define a time variable:

s = v0{m)t (4)

Of course, the above equation should read s(m) = ... instead of s = . . . The fact that s is a coordinate-like variable allows us to completely neglect this mass dependence without losing any information: it just makes working with it easier.

The equivalent length s has more convenient properties. Appendix A will give a mathemati-cally-concise derivation of this normalization. Here we will list just a few of these properties.

(i) Velocities: if we take dx*s instead of dx /at, then the magnitude of this vector in drift space is unity for ions of all masses. In fact, we will define w0 = ax /as and, as just stated, the equality \WQ\ = 1 holds.

(ii) v0 —> w0: by definition, ν${ιή) is mass depen-dent. By virtue of the definition of s, w0 is not mass dependent any more.

(iii) Units: the unit of s is the same as the unit used for coordinates in describing the geometry of the mass-spectrometer, (mm, cm, m, etc.)

(iv) Estimating S: the total equivalent length from the ion source to the surface of the detector is approximately 20% more than the geometrical length of the ion path, or (usually) roughly twice the length of the vacuum housing.

2.1.4. Initial velocities The right-hand sides of the dispersion plots give

the final distances of particles from the detector surface as a function of initial velocity. The initial velocity is given here in units of the drift velocity v0.

These units are the same as the definition dx: /as, but to take it as a ratio to the drift velocity v0 is even simpler. Usually, the parameter that is known about initial velocities is their energy. For example, one would expect energies of about 1 eV for ion-ization by electron bombardment, and thermal energies of about lmeV (at 10K) for laser-ionizing molecules in a supersonic expansion by a one-photon process. Taking the square root of the ratio (initial energy)/(energy in drift space) gives the magnitude of the initial velocities in units of v0.

Using this formula for ions having an energy in drift space of 1 keV gives a magnitude for the initial velocities, in the case of electron bombardment, of « 0.03t;0, and for an ionization without recoil in a supersonic expansion of « 0.00lt;0. Usually, some information about the form of the distribution of initial velocities is known, so together with the absolute values and the right-hand side of the dis-persion plots, an easy judgement of mass resolution is possible.

2.1.5. Time-dependent potentials The examples in this article concern instruments

with static electric fields. If the electric potential is static, the z'-dispersion plot holds for all masses. If the electric field is time dependent then there is one z'-dispersion plot for each mass.

Usually, some typical behaviour is desired for all masses or at least for a large range of masses. One might give specifications for that situation by

T. Bergmann and T.P. Martin/Int. J. Mass Spectrom. Ion Processes 131 (1994) 21-41 25

showing one z -dispersion plot for the middle of that mass range and two more plots for the lower and higher end of the mass range.

2.1.6. Numerical computation The intention of this article is to introduce a

method of specifying time-of-flight mass-spectro-meters. In order that everyone can produce speci-fications for his instrument, the method of calculating the specifications must be simple and straightforward.

To calculate the dispersion plots shown in this article means calculating the electric potentials and ion paths for a number of different initial coor-dinates and velocities within these potentials. To underline the fact that this is the only thing done here each individual path calculated has one sym-bol entered into one of the dispersion plots. For clarity, symbols have been connected by cubic spline functions.

The techniques to calculate electric potentials and paths are standard and will not be discussed here. An excellent review of all currently employed methods can be found in ref. 1.

The method used for computing the paths of the dispersion plots shown in this article is given in ref. 2. Although this method is rather precise, such a method would not be necessary just for calculating the dispersion plots. We have used these programs simply because we had them already. The authors have not made any tests with the ion optical simulation program SIMION PC/PS2 [3], but probably this would also be sufficient.

Most of the dispersion plots in this article have been given for optimized potentials and perhaps other optimized variables. Optimizing lenses of time-of-flight mass spectrometers is a rather involved procedure and will not be discussed in this article either. The interested reader may refer to some previous publications [2,4,5].

2.2. Evaluating performance

Evaluating performance examples means uni-

fying all three conceptual entities to get an answer to the question of performance of the instrument under specific circumstances. Assume now that we want an instrument with mass resolution above 10000.

If the vacuum housing is 1 m long then the total equivalent path length will usually be around 2 m. This means that the range of path errors in the z'-dispersion plots should be less than 100 μτη. This corresponds to a scattering of ±50/xm around some median value.

If we want a mass resolution of better than 10000, then we have to stay with path errors below these ±50 μτη. To check if this is the case we have to do the following.

(i) Obtain information on initial coordinates and velocities at which ions are started in the ion source.

(iii) To use that information in the z'-dispersion plots we have to scale the initial velocities against the drift velocity v0. Section 2.1.4 shows how this is done.

(iii) For judgement, use only those initial coor-dinates and velocities that have paths actually ending on the detector. (To improve resolution, some paths might be blocked by apertures.)

(iv) Verify that the z value for all these paths stays within ±50 ^m.

Now assume that the energy in the drift space is 500 eV and that the mass of the ion is 1000 u. The speed v0 of that particle is roughly 10 000 m s"1, so the time necessary for an equivalent path of 2 m is then roughly 200 μ s. This means that start time of mass analysis should be defined better than 10 ns. (Usually this time span is the temporal length of a pulse from an ionizing laser.) Like-wise, the detector and registration electronics should also be faster than 10 ns. To be on the safe side, both times should be better than 5 ns.

Knowing the variables for ions of mass 1000 u at an energy of 500 eV it is easy to scale to different masses or energies. For example, ions of mass 40 u at 500 eV need time definitions better than (l\/25th) of the previously calculated values, which is 2 and 1 ns respectively.

26 T. Bergmann and T.P. Martinjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 21-41

2.2.1. Verifying specifications Assume now that we have purchased the ion

optical system of a time-of-flight mass spectro-meter. First, we want to verify its specifications. To do that we have to measure the arrival times of some ion as a function of initial coordinates and velocities in the ion source. From the time errors we have to calculate the z'-errors by a trivial formula and then compare the results with the data given by the manufacturer.

A first prerequisite for this measurement is that the laser pulse should be significantly shorter than the time differences that need to be determined. Likewise, the detector and registration electronics must be faster than the time differences to be measured. That means we need to pick an ion that is heavy enough or whose total time of flight is long enough so that small fractions of the total time of flight can be measured.

Directly verifying all six z'-dispersion functions precisely and completely is usually not possible. Still, in many cases we can get a good idea. Assume we are creating the ions in the ion source by a short laser pulse. We can then translate the laser beam in the x-z-plane. What we are measur-ing then, will not be the function;

z'(x,0,z, 0,0,0)

but the function

(5)

z*(x,z) = z'(x,y, z, vx, vy, vz)dydvxdvydv2

(6)

We can take care that the range of initial velocities stays much smaller than what is encountered during regular operation, e.g. by ionizing a gas out of thermal equilibrium or from a supersonic expan-sion. That simplifies our integral considerably:

X x , z ) - | z ; ( x , ^ z , 0 , 0 , 0 ) d ^ (7)

We can measure z*(x,z) as in Eq. (7) for a small range of initial velocities by a soft ionization with-out any recoil. We can measure another function Z*(JC,Z) as in Eq. (6) for a large range of initial

velocities just by ionizing molecules that fragment upon ionization. These two functions should allow enough conclusions about the behaviour of z'(x,y,z,vx,vyjvz)9 including the answer to the question as to whether the instrument fulfills the manufacturers specifications or not.

2.2.2. Sensitivity and transmitted phase space One very important thing to remember is the

Liouville theorem. This theorem states that the volume of phase space transmitted by any ion optical system stays constant on its path. It is also important to note that this statement holds, no matter how strongly the shape of the transported phase space is distorted on its path. This means that we can always distort the transported phase space volume such that it gives an optimal fit to the ions offered to the optical system. For that reason it is possible to equate the terms sensitivity and trans-mitted phase space.

We assume that it is always possible to distort the transported phase space as needed. Under this prerequisite-which has very few exceptions — an instrument designer first has to find out where exactly in phase space the ions enter the instru-ment, and then has to design the ion optical system to fit these externally imposed conditions. He has to play with that six-dimensional product of velocities and coordinates, (keeping the value of the product constant), until he has found a con-figuration that transports the maximum number of particles.

3. Demonstration: standard instrument with two-stage grid reflector

To substantiate the preceding discussion, we will now look at a well-known, time-of-flight design. Figure 1(a) shows a standard gas-phase ion source and Fig. 1(b) shows its axis potential with the lens operated in negative mode.

The lens of the ion source can be adjusted to do one of the following.

(i) The ion source can guide all ions that start on the axis at z = 0, but with different initial trans-

T. Bergmann and T.P. Martinjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 21-41 27

2.0 \—

[cm]

o.o

A

—M l· 1 '

0 [

—i · r 1 [- i p

^ *

(

—i ■ 1 —

] 1cm]

(a)

r>

-1.0 t

Fig. 1. (a) Standard gas phase ion source. The first three plates define the accelerating field, the thick fifth electrode is an Einzel lens, (b) The axis potential shows the Einzel lens operated in the negative mode.

versai velocities, to paths parallel to the axis. We will call this the velocity focussing mode.

(ii) The ion source can guide all ions that start with zero initial velocity, but different locations around the axis at z = 0, to paths parallel to the axis. We will call this the coordinate focussing or telescope mode. The difference between these two adjustments is that the first has a more negative voltage on the lens electrode.

Note that the phase space transmitted is the same for both modes. In the velocity focussing mode, paths with a larger range of initial velocities end on the detector surface, while in the coordinate focussing mode, paths with a larger range of initial coordinates end on the detector surface. All further

examples, and also the one in this section, operate the ion source in the velocity focussing mode. Section 2.2.2 discusses the possibilities of changing the shape of the transported phase space; this is what is done by changing the voltage of the lens electrode of the ion source.

Figure 2 shows the z'-dispersion curves for a time-of-flight instrument with the above stan-dard gas-phase ion source operated in the velo-city focussing mode. The ion beam entering the reflector and the axis of the reflector subtend an angle of 4°. The reflector is of the Mamyrin type [6] having two regions of homogeneous electric field separated by grids. The detector is just a flat plate of 20 mm diameter and can be rotated a few degrees around the /-axis to give


Fig. 2. Dispersion curves of a standard ion source (from Fig. 1) and two-stage grid reflector. The graphs give the error in the final z'-coordinate as a function of all six initial phase space variables. Since a grid reflector is optically active only for velocity components parallel to its axis, it only affects the function z'(0,0, z, 0,0,0). The other five dispersion functions closely reflect the behaviour of the ion source in Fig. 1. The initial velocities are given in units of v0, which is the velocity of the reference ion in the drift space of the mass

spectrometer.


an optimum z'-dispersion. The total equivalent length of the flight path is approximately 2 m.

The ionizing laser beam has a diameter of « 2.4 mm. The length of the acceleration region is about 20 mm, and thus a 2.4 mm width of the ion-izing laser beam corresponds to an energy spread of « ±6%. All free parameters are optimized using a procedure described in ref. 2, the resulting z'-dispersion being shown in Fig. 2.

All six dispersion plots in Fig. 2 show the devia-tion parallel to the incoming path of the reference particle (along the z'-axis). Since we have assumed 2 m for the total equivalent length of the flight path, a path error of Δζ' = 100 μιη corresponds to a time error of 5 x 10~5 or an error in mass determination of 10"4.

The plot in the lower left hand corner shows the dispersion as a function of the initial z-coordinate in the ion source. This dispersion corresponds to the standard solution of the Mamyrin equations [6]. The circles show the dispersion for ions initially at rest and the crosses show the dispersion for ions with an initial velocity of -0.00 li>0. (Remember that v0 is the velocity of the reference ion in the drift tube as defined in section 2.1.4.)

The plot in the lower right-hand corner shows the z'-dispersion as a function of initial velocity vz. The slope of this function z'(0,0,0,0,0, vz) is inversly proportional to the accelerating field strength in the ion source, here we have «0.5t / 0cm~l . (i/0 is the voltage corresponding to the drift velocity v0.) Note the scale: just one thousandth of the drift velocity is necessary to effect an error in path length of z' = 40 μτη\ If we assume that the energy of a particle in the drift tube is 1 keV then this initial velocity corresponds to a recoil energy of only 1 meV. This sensitivity to recoil has been pointed out before [7,8]. The usual solution for this problem is that the initial velocity vz is restricted or reduced by some means.

The upper left-hand plot shows the z'-dispersion as a function of the initial x-coordinate with the initial velocity vx as the parameter, while the upper right graph shows the z'-dispersion as a func-

tion of the initial velocity vx with the initial x-coordinate as the parameter. This is a coordinate or velocity deviation from the axis of the ion source in the plane defined by the spectrometer axis and the connecting line between the ion source and the detector. Both graphs very closely reflect the behaviour of the ion source in Fig. 1 itself.

The two graphs in the middle show the z'-disper-sion as a function of y or vy. The behaviour shown by these graphs is very similar to the z'-dispersion as a function of x and vx. Because the spectrometer is assumed to be symmetric about the x-z-plane, only two instead of three parameters are plotted.

The initial x- and ^-coordinates are limited to ±0.5 mm by the magnification of the ion source optics and the diameter of the detector.

The performance of such an instrument can be verified with the methods described in section 2.2.1. Figure 6 of ref. 2 shows the results of a measure-ment that can be used to determine

z*(0,z) = Jz,(05>;5z,050,0)d>; (8)

Scanning the laser in the x-z-plane would have given the function z*(x, z).

4. Examples: ions from the gas phase

This section addresses instruments that start the ions in the time-of-flight instrument out of the gas phase.

Instruments of this category are used for all kinds of photoionization experiments. When elec-tron impact is used, the interest is usually to have a mass-selective detector as in gas chromatography-mass spectrometry.

In secondary neutral mass spectrometry (SNMS) either a laser or electron pulse is used to ionize sputtered neutrals. This ionization event is then the start event for time-of-flight analysis.

4.1. State-of-the-art gridless reflectors

The purpose of this section is to warn the reader about believing statements like "As it seems


logical, this reflector can do everything at the same time! It cannot only perform energy focussing but can also do space focussing. It can effect a signifi-cant enhancement of sensitivity at a superior mass resolution".

What is logical? The Liouville theorem always holds, even for gridless reflectors! State-of-the-art instruments with gridless reflectors presently sell for twice the price of regular instruments on the basis of the above arguments. Yet, their per-formance is much worse, as will be shown in this section.

We will just pick out two examples, not calling them by name. The authors have looked at more gridless reflector designs to be found in the literature or in patents, all of which have a similar design and probably a comparable performance. Some newer designs have been announced, but no information about their detailed construction is given. So, if you think about buying an instrument with a gridless reflector, be sure to ask for the specifications!

The reflector of patent A has four adjustable voltages, the ion paths entering the reflector sub-tend an angle of 4° with its axis. The total length of the drift path is again « 2 m. The ions are started in the standard ion source of Fig. 1. If we optimize the four adjustable voltages in a range of ±2% of the values prescribed in the patent we get the disper-sion curves shown in Fig. 3.

Naturally, the function z'(0,0,0,0,0,uz) shown in the lower right-hand corner is identical to the corresponding function in Fig. 2, since we use the same ion source with the same accelerating field strength. The z'-dispersion as a function of z is also comparable in quality.

The z'-dispersions of initial variables perpen-dicular to the direction of acceleration are much worse for this gridless reflector than for a regular two-grid reflector as can be seen by comparing Fig. 2 and Fig. 3. We can set apertures behind the ion source such that paths with an error of z > 100/xm are blocked. Looking at the trans-mitted phase space in the x- ivplane we find that the two-grid reflector of Fig. 2 transmits a

trapezoid of area « 1.0 mm x 0.06^0 and the grid-less reflector of Fig. 3 transmits a diamond of area ^0.5(0.5 mm x 0.06u0).

This makes a factor of 4 in the x-i^-plane. The same situation is found in the ^-^-plane, such that, disregarding transmission losses by the meshes in a two-grid reflector, a gridless reflector after patent A transmits a phase space 16 times smaller than a two-grid reflector. If we take trans-mission losses of about 50% passing four times through a mesh into account, the transmission of this gridless reflector is still eight times smaller.

Figure 4 shows the z'-dispersion curves of a reflector after patent A, if not the prescribed vol-tages are used, but an optimized, completely differ-ent set of values. The phase space trans-mission is close to a factor of 2 better than in Fig. 3, still it is a factor of 4 worse than a standard two-grid reflector.

The reflector after patent B has a total of 17 independent voltage adjustments. The angle that ion paths entering the reflector subtend with the reflector axis is not given, so that it was freely varied for optimum performance. The total length of the drift path is not given so that it was used also as a free variable in the optimization. The detector and ion source are assumed to be 3 cm off the spectrometer axis. The voltages are optimized in a ±2% range around the prescribed values.

One should think that this reflector gives a much better performance than the reflector after patent A. Figure 5 gives the resulting performance. It closely corresponds to Fig. 3 and Fig. 4. An instru-ment with this reflector does not have a higher phase space transmission than instruments using reflector A.

4.2, Can reflectors without grids have a superior transmission?

The authors do not want to convey the impression that gridless reflectors are worthless. The purpose of this section is to show that it is theoretically possible to construct gridless reflectors that can transmit a phase space volume a factor of 20


-0.2 -0.1

x=z'(x.O. 0.-0.02. 0,0)

o-zMx. 0.0.0.0.0)

□ =z'(x. 0,0.0.02.0.0)

-50

-0.03 -0.01

χ=ζΊ-0.2.0.0,νχ .0.0)

ο=ζ'(0.0.0.νχ.0.0)

□ =z'(0.2.0.0.vx.0.0)

•50 f-

[vj

-0.2 -0.1

o * z ' (0. y. 0.0.0. 0)

x - z ' ( 0 . y . 0.0.0.02.0)

-0.03 -0.01

o-zMO. 0.0.0. vy.0)

x-z ' (0 .0 .2 ,0 .0 .v r 0) •50 I-

[v0l

-1.0

x«z'(0 .0 .z . 0.0.-0.001)

o = z ' (0. 0. z. 0. 0. 0) 50

50 h

1 , 1 i

-0.0010

o-zMO.O.O.O.O.vJ

-50

Fig. 3. Dispersion curves of a standard ion source and gridless reflector after patent A. The reflector voltages are restricted to ±2% of the values prescribed in the patent. The z'-dispersion as a function of z and vz are comparable to the dispersions of a two-grid reflector shown in Fig. 2. The z'-dispersion as a function of x,y, vx and vy show that the phase space transmission of this gridless reflector is a factor of 8

smaller than a two-grid reflector.


-0.5

x-zMx.O. 0,-0.02.0.0)

o-z'(x. 0.0. 0.0.0)

o-zMx. 0.0. 0.02.0.0)

-50

-0.03 -0.01

x=z'(-0.2.0.0.vx.0.0)

o = z * (0. 0.0. v „. 0.0)

G = Z * ( 0 . 2 . 0 . 0 . V X . 0 . 0 )

-50 I-

-0.5

o-z'(0.y.0.0.0.0)

x=z'(0.y. 0.0. 0.02.0) -50 I-

-0.03 -0.01

o=z'(0.0.0.0.vy.0)

χ-ζΊθ.Ο^.Ο.Ο.ν,,Ο) -50 h

-1.0

χ - ζ ' (0 .0 .2 . 0.0. -0.001)

ο=ζ'(0.0.Ζ. 0.0.0) -50

50

■'■Γ""2 I ' I ' I ' I ' I ' I ' I ' I ' I

-o.ooio

o-z,(0.0.0.0.0.vj

-50

i " " I " " ' v 0.0010 z

Fig. 4. Dispersion curves of a standard ion source and gridless reflector after patent A. The reflector voltages have been optimized to give the lowest possible timing or z'-errors. The z'-dispersion as a function of z is somewhat smoother than in Fig. 3. The z'-dispersions as a function of intial x and y are a bit wider. The transmitted phase space of this adjustment is close to a factor of 2 better than the

adjustment of Fig. 3.


-0.4 -0.2

x=z'(x.O. 0.-0.02, 0,0)

o=z'(x. 0,0,0.0.0)

n*z'(x. 0,0.0.02.0.0)

-50

-0.01

χ=ζΊ-0.2.0.0.ν„0.0)

o- z* (0,0.0, ν,,.0,0)

π=ζ·(0.2.0.0.νχ.0.0)

-50 [Vo

T V 0.03 V *

-0.4 -0.2

o-z,(0.y10.010>0)

x=z'(0.y. 0.0. 0.02.0)

o=z'(0.0.0.0.vv.0)

x=z'(0.0.2.0.0.v,0) -50 (-

z 1 0 0 '

[pin)

Φ 50

z ' (0, 0. z. 0. 0. 0) - s o

k

-

■ I 1.0

': (mm]

50

" " I i - 0 . 0 0 1 0

o=z'(0.0.0.0.0.v.)

-50

0.0010 z

[vj

Fig. 5. Dispersion curves of a standard ion source and gridless reflector after patent B. The reflector voltages are restricted to ±2% of the values prescribed in the patent. The z'-dispersion functions show that this reflector gives a phase space transmission somewhere between

Fig. 3 and Fig. 4


larger than a standard two-grid reflector. The following section will then show the perform-ance of an instrument presently under construc-tion that is expected to perform a factor of 7 better.

This section shows one instrument which has a low accelerating field of « 0.5 t/0cm~l in the ion source and another instrument with a high acceler-ating field of « 14 i/0cm_1 in the ion source. U0 is the voltage corresponding to the drift velocity v0. Both of these examples will not be actual construc-tions with electrode shapes etc. but will just be defined through their axis potentials which are given by spline functions (see ref. 2). Whether an actual construction will achieve these limits depends very much on geometric limitations of electrode design.

Fig. 6. shows the optimum performance of an instrument with a low accelerating field of « 0.5 i/o cm"1. By comparing Fig. 6 with Fig. 2 we see that this instrument transmits a phase space volume at least a factor of 20 larger than a standard instrument with two-grid reflector. (A factor of 10 for larger phase space and another factor of 2 because it uses no grids.)

The z'-dispersion as a function of z is almost completely flat. That is not too surprising since the reflector potential is controlled by a multi-variable spline function. The phase space volume in the x- and ^-direction is so large that the question might arise as to whether it is possible at all to admit ions to fill this whole volume. The z'-disper-sion as a function of vz is still as sensitive as in all the previous figures. For that reason one might try to change the shape of the transported phase space volume to allow for a larger variation of vz, perhaps reducing the phase space in x- and ^-directions.

This is what has been done in Fig. 7. The length of the transported z-coordinate is much shorter than in Fig. 6, but the accelerating field is also much higher: « 14t/0cm~l. Thus, the range of z is a factor of 5 less but the allowable range of vz is a factor of 28 higher. The transported phase in the x- and y- directions is also less, so that the total

transmitted phase space volume is again the same as in Fig. 6.

Note that the allowable recoil upon ioniz-ation for this instrument is 0.03^0. If the energy of ions on the drift tube is 1 keV, this recoil corresponds to 0.9 eV. This instrument should have a high mass resolution even for "hard" ionization.

43. SNMS-time-of-flight instrument

Figure 8 shows the performance data of a completely designed SNMS time-of-flight instru-ment. All potentials, also that of the detector, are given by explicitly defined electrodes. The post-acceleration in the detector is assumed to be 6K0. (For ion energies of 1 keV this corresponds to 6keV.) The accelerating field is « 1 0 i/o cm"1. The z'-dispersion plots in Fig. 8 have been scaled by the ratio of the velocity in drift space divided by the velocity of particles impinging on the detector surface, in this case lyVl + 6 (see section 2.1.3). Thus, the data in Fig. 8 should be compared, as in all previous cases, to a total equivalent length of S = 2m.

The origin of the unprimed coordinate system xt

has been fixed to a repeller plate 0.5 mm behind the starting point of the reference path; the reference path is also assumed to start with a velocity of 0.05^0. This corresponds to what is expected for SNMS operation. Before the laser ionizes the sputtered neutrals they are desorbed from the analysed surface at z = 0mm. When they pass through the volume around z = 0.5mm, they have a velocity of « 0.05t>0.

The angle that the ion source electrodes above the sample may subtend is limited to 20° and thus it is not possible to freely model the field as in the theoretical calculations for Fig. 7. This partially explains why the total transmitted phase space volume, as shown in Fig. 8, is only a factor of 7 better than a standard two-grid reflector can handle and not a factor of 20 as for Fig. 7. That the instrument shown in Fig. 8 uses a detector while Fig. 7 just uses an imaginary


x = z' (0, y, 0, 0, 0.03, 0) x = z' (0, 0.8, 0, 0, vy, 0)

Fig. 6. Dispersion curves showing the optimum performance of a time-of-flight instrument with a low accelerating field of « 0.5 [/0cm_1. The transmitted phase space volume for this instrument is at least 20 times as large as for an instrument with a two-

grid reflector shown in Fig. 2. Note that the behaviour in the x-z-plane for this instrument is different than in the j-z-plane.


-0.5

x = z'(x.0, 0,-0.02.0.0)

o = z*(x. 0.0. 0.0.0)

Ü = Z ' ( X . 0.0. 0.02. 0.0)

-50 f-

-0.04 -0.02

x = Z'(-0.4.0.0.vx.0.0)

o-z ' (0.0.0.vx .0.0)

o -z ' (0 .4 ,0 .0 .v l ,0 . 0)

-50 I-

0.02 0.04 VX (vj

-0.5

ο = ζ·(0.γ.0,0.0.0)

x = z'(0, y. 0.0,0.02.0) -50 I-

i i i i

-0.04 -0.02

ο-ζΜΟ,Ο.Ο,Ο,ν,.0)

x = z'(0.0.4.0.0.vv.O) -50

I I I ! *■

0.02 IvJ

0.04 y

z 5 0 '

(|im]

of® °~t*-©-e-<

-0.3 -0.1

0-zMO.O.z. 0,0.0)

-50

-

. " ■ • ' ■ • ■ ■ i ' · " '

0.1 . . . . !. n ^ . ... ! 3

0 . 3 [mm]

-0.03 -0.01

o=z'(0,0,0,0,0.vz)

-50 I-

Fig. 7. Dispersion curves showing the optimum performance of a time-of-flight instrument with a high accelerating field of « 14 U0 cm"1. Because the accelerating field is higher by a factor of 28, the range of transported vz is also higher by a factor of 28. The ranges of the remaining five phase space variables multiply to about 1/30, so that the total transmitted phase space volume for this

instrument is the same as in Fig. 6.


-0.4

x»z'(x. 0.0.5.-0.02.0.0.05)

o=z'(x. 0.0.5.0.0.0.05) "50

D -Z ' (X .0 .0 .5 .0 .02 .0 .0 .05 )

-0.04

ü=z'(0.2.0.0.5.vx.0.0.05)

o = z'(0.0.0.5. v„ 0.0.05)

x = z,(-0.2.0.0.5.vx.0.0.05)

-50 I-

-0.4

o-z'IO.y. 0.5.0.0. 0.05) _50

x=z'(0.y.0.5.0,0.02.0.05)

-0.04

o = z'(0.0.0.5.0.v,.0.05) _5Q

x = l' (0.0.2.0.5.0.vy. 0.05)

Z 100

[fim]

- H - H - X - ^

0.2 075

x= 2'{0,0,1. 0.0. 0.025)

o*z'(0.0.z. 0.0. 0.05)

>#*%£-

[mm]

(Jim) t

Fig. 8. Dispersion curves showing the performance of a fully designed SNMS-time-of-flight mass-spectrometer. The origin of the coordinate system x, has been fixed to a repeller plate 0.5 mm behind the starting point of the reference path; the reference path is also assumed to start with a velocity of 0.05υ0. This instrument has a high accelerating field of « 10 U0 cm-1 to achieve a high mass resolution in spite of the expected recoil upon ionization. The total transmitted phase space volume is higher by a factor of 7 compared to

the standard instrument with two-grid reflector of Fig. 2.


plate might account for another part of the difference.

This instrument is presently under construction.

5. Examples: ions from surfaces

This section addresses those instruments that create ions at surfaces from where they are drawn into the time-of-flight instrument. Typical examples of this type are secondary ion (SIMS) and laser desorption (LD) mass spectrometers.

In SIMS instruments a short pulse of primary ions sputters off material from the analysed surface. The ionized part of this material is drawn into the time-of-flight mass spectrometer.

In LD instruments the ions are desorbed from a surface by a short and intense laser pulse. In the case of matrix assisted laser desorption the mole-cules of interest are imbedded in a matrix of light-absorbing material which is heated by the laser pulse and which then evaporates and carries the analyte along into the time-of-flight mass-spectrometer.

In both cases the z-coordinate of the initial phase space is fixed to the analysed surface (z = 0). Since it is desired to obtain information from different locations of the surface of the sample, the x- and ^-coordinates may vary. Likewise, the ions have a finite spread of initial velocities in all three directions. Thus we have to consider five and

not six phase space variables in the following discussion.

Figure 9 shows the design of a typical SIMS ion source as can be seen for example in Fig. 1 of ref. 9. The ions start at z = 0 and around x, y — 0 and are accelerated in the positive z-direction. The thick ring at z « 5.8 cm is the active electrode of a focuss-ing lens. The ion source taken from ref. 9 is just meant as an example and the measures in Fig. 9 are only approximate!, similar designs can be found in any publication on SIMS-time-of-flight mass-spectrometers.

Figure 10 gives the performance of a system comprising the standard SIMS ion source of Fig. 9, a single-grid-reflector starting 70 cm above the sample surface, and a standard detector design (not shown). The distance of the extraction lens to the sample surface has been adjusted, such that the accelerating field is « 10 i/o cm- 1. This has been done to provide a better comparison with Fig. 11.

Figure 11 gives the performance data of the instrument shown in Fig. 8 operated in SIMS mode. Because the accelerating field above the sample surface is also « 10 U0cm~l the z'-dis-persion as a function of vz is roughly the same as in Fig. 10. It is important to note that the z'-dispersions in the plane perpendicular to the z-axis are significantly better than for the standard instrument of Fig. 10. This is due to the gridless reflector.

2.0

[en]

1.0

0.0

lern] 10

Fig. 9. Standard SIMS ion source [9]. The ions start at z = 0 and around x,y = 0 and are accelerated in the positive z-direction. The thick ring at z « 5.8 cm is the active electrode of a focussing lens.


x = Z*(x.O, 0.-0.02.0.0.05)

o«z*(x. 0.0. 0.0.0.05)

n-zMx. 0.0. 0.02. 0.0.05)

I " " ' " " ! " " ' " " ! " " ' " "

0.03 -0.01

χ = ζ · ( - 0 . 1 . 0 . 0 . ν χ . 0 . (0.05-v*) Ui)

o = z * (0.0. 0. v„ 0. (0.05-v^) Ui) " 5 0 μ

□ - ζ ' ( 0 . 1 . 0 . 0 . ν „ . 0 . (0.05-vi;)1/a)

0.01 Γ V

0.03 Υ χ

o« z* (0. y. 0.0. 0.0.05) _SQ

x-z'IO.y. 0.0. 0.02. 0.05)

-0.03

o = z ' (0. 0. 0. 0. v y. (0.05-vJ)1/2) . ^

x « z ' ( 0 . 0 . 1 . 0 . 0 . v v . 10.05-v')l/2)

0.05

o = z ' (0. 0.0.0.0. v z)

Fig. 10. Dispersion curves showing the performance of a standard SIMS instrument with a single-grid reflector. This instrument uses the ion source of Fig. 9, a single-grid reflector and a standard detector design.

References

1 P.W. Hawkes and E. Kasper, Principles of Electron Optics, Academic Press, London, 1989.

2 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 61(10), (1990) 2592.

3 D.A. Dahl and J.E. Delmore, SIMION PC/PS2, Version 4.0,

EGG-CS-7233 Rev. 2, April, 1988. 4 M. Szilagyi, Proc. IEEE, 73(3) (1985) 412. 5 J.P. Adriaanse, H.W.G. van der Steen and J.E. Barth, J. Vac

Sei. Technol. B, 7(4) (1989) 651. 6 B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A.

Zagulin, Sov. Phys. JETP, 37 (1973) 45.


-0.3 -0.1

x=Z*(x. 0.0, -0.02,0,0.05)

o=z'(x. 0.0. 0.0. 0.05)

□ =z'(x. 0.0. 0.02, 0,0.05)

50

-0.05

x - z' (-0.15,0,0. v K. 0. (0.05-v^l/2)

o=z'(0.0.0.vx,0. (0.05-vJ)1/a)

a = z ' (0.15. 0. 0. vx. 0, (0.05-v^)1/2)

-50 f-

o=z*(0.y. 0.0. 0.0.05) _50 μ

x=z'(0.y.0.0.0.02.0.05)

-0.05

ο = ζ·(0.0.0,0.νν. (Ο.Οδ-ν^)"2)^ I

χ = ζ'(0.0.15.0.0.νν. (0.05-v?)1/2)

»-©-e-e-w 0.05

0-z'IO.O.O.O.O.vJ

Fig. 11. Dispersion curves showing the performance of the SNMS instrument of Fig. 8 operated in SIMS mode. The exceptional z'-dispersion as a function of x,y, vx and vy is the result of the combined ion optical properties of the ion source and the gridless reflector.

7 U. Boesl, J. Grotemeyer, K. Walter and E.W. Schlag, Anal. Instrum., 16(1) (1987) 151.

8 U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spec-trom. Ion Processes, 112 (1992) 121.

9 E. Niehuis, T. Heller, H. Feld and A. Benninghoven, J. Vac. Sei. Technol. A, 5(4) (1987) 1243.

Appendix A: Notation

To simplify the notation we redefine the variable for energy, potential, time and velocity. If K is the total energy of an ion and K0 is the average energy


of all ions, then let e be defined by

K=(\+e)K0 (Al)

If U is the electric potential, then u is given by

U = uK0 (A2)

The location of ionization for ions with e = 0 and with vx = 0 will be at z = 0. The average velocity of ions at ground potential U = 0 is given by VQ = s/ΪΚφη. Let s be defined by

s = v0t (A3)

We give s the name "equivalent length". In solving Newton's equations we use s instead of / as the independent variable. As can easily be seen, s has

a very close relationship to the length dimensions of the problem. The velocities are always given as the s derivative of coordinates. Newton's equations now become

Defining axf/ds = wh it follows that

(w2x + v?y + w2

z) = 1 + 6 (A5)

In the dispersion curves we always define K0 and VQ as the energy and velocity of a particle on the reference path at ground potential. In the examples discussing time-of-flight instruments, ground poten-tial is defined as the potential on the drift paths.

International Journal of Mass Spectrometry and Ion Processes 131 (1994) 43-65 43 0168-1176/94/507.00 © 1994 - Elsevier Science B.V. All rights reserved

The application of ion optics in time-of-flight mass spectrometry

D. Ioanoviciu Institute of Isotopic and Molecular Technology, P.O. Box 700, R - 3400 Cluj-Napoca, Romania

(Received 26 February 1993; accepted 24 June 1993)

Abstract

The ion time-of-flight (TOF) through accelerating, decelerating and reflecting fields is the key to determining ion optical properties and the resolution of most TOF mass spectrometers. The initial velocity- and position-induced energy spread aberrations, as well as the role of the time lag, are considered for linear drift TOF mass spectrometers. Peak shapes for stable and metastable ions are presented for these instruments. On a specific geometry the resolution is estimated for realistic assumptions. The post source focusing method ion-optical capabilities are discussed.

The ion-optical properties of the single and double stage mirrors with first and second order energy focusing in time respectively are detailed for linear and packet oblique incidence geometries. Again, illustrative peak shapes for stable and metastable ions are given. The resolution of typical mirror TOF mass spectrometers was calculated and compared with those of linear drift instruments.

The TOF ion optical properties of the parabolic potential with plane and axial symmetry are described for mirrors and accelerating systems. Matrix elements are given for cylindrical electrostatic mirrors, applicable to electrostatic particle guides. Means to focus transversally and to tune ion packets to the detector i.e. electrostatic lenses and plane deflectors are briefly mentioned. The TOF basic parameters of multisector electrostatic systems are exemplified on a three deflector geometry. The main TOF ion optical parameters of four magnetic-sector analyzers are also specified.

Key words: Ion optics; Ion mirror; Reflectron

Introduction

The outstanding features of time-of-flight (TOF) mass spectrometry became obvious immediately at the time when Stephens demonstrated his instru-ment [1]. The capability of delivering complete mass spectra at matchless speeds, while extending the mass range at will and keeping sensitivity high, was a promise fully accomplished only after dec-ades. This accomplishment came as a corollary of a long process of technical refinement, especially in electronics, and by creatively introducing new ion-optical solutions. Among them, "space" and "time lag" focusing in time [2], and the use of mainly electrostatic deflectors [3] cannot be omitted. How-

ever, to date most successful have been electrostatic mirror TOF mass spectrometers. These have pushed the resolution of mass spectrometers in this category into the range of tens of thousands [4-6].

This overview concentrates on ion-optical solu-tions incorporated in TOF mass spectrometers and their refinements during recent years, as well as on the understanding of the time focusing properties of instruments in widespread use.

Accelerating and decelerating electric fields

TOF mass spectrometers use ion sources, electrostatic reflectrons and plane deflector plates

SSDI0168-1176(93)03879-Q

44 D. Ioanoviciujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 43-65

whose fields are, or may be approximated by homogeneous electrostatic fields. A brief descrip-tion of the ion movement inside such a field is worthwhile.

The movement of an ion of mass mx and electronic charge e inside a constant homogeneous electric field Ex = E (Fig. 1) is governed by the equations .. eE x — —

mi (1)

Integration accounting for initial coordinates and velocity components (subscript "i") gives

x = eEf 2mx + v

ax eEt — = h vx. t at m,

(2)

(3)

y = yi + Vy.t (4)

z = Zi + vz.t (5)

as the ion leaves the x — 0 plane when time begins. The ion will reach the plane x = df at the time /f. From Eqs. (2)-(5) we obtain the final quantities (subscript "f ")

(6) m, 'f = ^ K - v * i )

vX{ = I vX: +

v7r = v7.

2eEdf\l/2

m; ) (V)

(8)

Fig. 1. Ion movement in homogeneous electrostatic fields: coordinate system and reference planes.

yf = yi + vyitr

Zf = Zi+Vz.t{

(9)

For a decelerating field only the first two equa-tions are affected by the sign change of E:

m, >f = ^ K - v * f )

vXf = \vx. -2eEd(\

l/2

mx )

(10)

(11)

The ion penetrates the decelerating field barrier only if its initial kinetic energy is higher than eEdf. Otherwise the ion stops at the depth h, after a flight lasting tr:

h =

tx =

mivx,

m-,Vr.

eE

(12)

(13)

Equation (13) represents half of the so called "turn around" time [7].

Resolution

Resolution may be defined for peak pairs or for individual peaks. It is the ratio of the mean mass to the mass difference of the two peaks, separated by a valley of a specified depth. It also can be expressed as mass-to-peak width ratio, the width measured at a given level. For the idealized case of sharply-cut-off, constant-charge-density ion packets, only two particular cases are relevant: (1) the ion packets coming to the detector just one after another, with-out mixing: (2) the packets separated in time by an interval at least equal to a registration channel width. The first case leads to resolution at peak half height, the second at its basis. In the latter case the distance between the centers of the two packets ν{ί/η)η, each of length At and the channel width /el must satisfy (Fig. 2) the following relation:

7 > Δ ί + id (14)

Here (t/j) is the mass dispersion in time coefficient, numerically equal to half the total flight time t of

D. Ioanoviciu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 43-65 45

J

v{t/f)t*m/m)

Fig. 2. Resolved packets of m and m + Am mass ions and their picture in the spectrum.

the mass m ion through the mass spectrometer, Am is the mass difference between ions belonging to the two packets, 7 = Am/m. So the resolution R results from Eq. (14):

C/7) R = m (15) Am At + id

or neglecting iel for the half height resolution Rh:

Ä h = 2At

(16)

The packet length at the detector results from its initial longitudinal extent (eventually compressed by a post source focusing procedure) and from components due to velocity, energy and spatial spread. These components may be simply added together or the packet length obtained as the square root of the sum of component squares (or their variances) [8].

Linear free drift-space TOF mass spectrometers

A linear free drift space TOF mass spectrometer is the simplest mass analyzer structure: ions from a

detector

Fig. 3. Linear drift, two field source TOF mass spectrometer: geometric configuration and electric fields.

two field ion source are left to fly freely in a space without field until they fall on the detector. Single field ion sources followed by a short drift space ensure "space focusing" in time as a preliminary stage in reflectron TOF mass spectrometers [9]. The ion-optical study of linear TOF mass spectro-meters is useful by itself due to their widespread use and also to better understand why reflectrons became so successful.

Let us consider a two-field ion source (Fig. 3) £a

being the field intensity in the extraction region, Eb

that on the accelerating interval. The ion path depends on the formation site, defined by the x-coordinate difference s measured from the ionization region middle (the latter at a distance da from the first grid). The single field source is the particular case when £ a = £b , the first grid being absent.

From Eqs. (6) and (7), the flight time ia and the velocity vx at the first grid are

m, / a= -^ (^ - v J

vY = vl+2eE» * 'XQ

ÏYi\

1/2 (17)

where vXo is the ion velocity component at its crea-tion, directed along the field.

The time spent in the accelerating interval th and the ion velocity at the source exit vXb are

m. 'b = —^Γ (VJC —Vx)

eEbK b Λ)

Jx* 2 2eEhdh

1/2 (18)

46 D. loanoviciujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 43-65

W e will often relate an individual ion movement to tha t of a witness (or s tandard) particle formed at rest, at a distance da from the first grid. Its energy at the first and second grids will be t/a = eE^ and Us = eEada + eEbdb respectively, its velocity at the final grid vs = (2Us/m)^2 where m is its mass . We denote the combined initial posi t ion <5C, and veloc-ity generated energy spread and the rat io v, defined by:

δΓ =

and

eEas + mv2xJ2

v2 = υΛ

(19)

T h r o u g h o u t this paper δ means "relative energy spread" , except when dealing with metas table peak shapes, where it is defined otherwise.

In this case

£ a db/d&

v2-\ (20)

Here mx is replaced by m because we pursue only the first order effect of relative mass difference 7 , that of mass dispersion. In the following para -graphs the subscript " i " will often be omit ted . By using this no ta t ion , vXb = vs(l + 6c/u

2)1^2. Sum-ming flight time contr ibut ions from the source and from the field free space of length D:

'D = M\+ll^\ (21)

we obtain the total flight t ime /d. W e perform series developments in 6C to second order. By cancelling the global coefficient of <5C, energy focusing in time is obtained for differences in initial posi t ion and initial kinetic energy (without having an effect on the first order initial velocity " t u r n a r o u n d " term). This happens when

dh D = 2v2

vd~ — l + i / (22)

Besides the initial velocity induced ion packet lengthening, the second order energy spread aber-ration is also to be accounted for.

The initial velocities contribute to the ion packet length by at least Δίυ = -2ι/άάνχ0/νΙ for ions pro-duced from a surface, or by twice this value if ions are produced from molecules in the gas phase. The second order energy spread contribution is

[4,(3 - v2) + dh(v - ! ) ( ! / + 2)/i/(l 4- u)]6\ ^h\ 1

4vsv

(23)

The resolution can be obta ined from the formula

W2 h i0 + Atv + Δ/„

For an ionization event of t0, replacing t by

vK l + i /

(24)

(25)

Time lag effect

A time lag between ion produc t ion and extrac-t ion allows ions to migrate towards new posi t ions in space, commensura t e with their initial velocities. The dis t r ibut ion of the flight times to the detector thus resulting could be nar rower than tha t of the ions star t ing from their format ion points (if the ion generat ion region is no t too wide).

The velocity at the first grid of an ion submit ted to the extract ing field T\ after its format ion is obtained by replacing df = da + s ± vx T\ in Eq. (7). The positive sign means "initial velocity directed against the packet general movement sense". The total flight time is described again by expressions similar to Eqs. (17)—(21) but now the relative energy spread is

δι = d.A

The flight t ime difference related to the witness ion Δ/ι is expressed by the relation

Atl = Fs±vx(FTl + 2v2dJvs

where

F=v 4 ( 1 ! v) D d~ 2v2d»

(26)

(27)

D. loanoviciu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 43-65 47

While "space focusing" happens when F = 0, velocity focusing is possible if F— -2v2d&/(vsT\), only if F < 0. For a determined configuration of the source fields the vsT\ product must be kept constant. In other words time lag focusing is mass dependent through vs. In practice, to mini-mize the detected packet length, a compromise is established between initial velocity and position components (less negative F values than for cor-rect velocity focusing).

Stable ion peak shapes in linear drift TOF mass spectrometers

To derive stable ion peak shapes in linear TOF mass spectrometers the following preliminary observations and assumptions must be made: (a) only first order packet length contributions are relevant, those of higher order influencing peak details being difficult to notice; (b) the ionizing particle or photon flux density in the ion source is constant in space across the ionization region as well as in time from the beginning to the end of the ionizing pulse; (c) the condition of space focus-ing in time is satisfied. We set the time origin at half the ionizing pulse. An ion created at t0 from this origin with the initial velocity component v directed along the flight direction will be collected at the moment ta„:

mv = /w + / o ± iZ

(28)

iw being the flight time of an ion created at rest at the time origin, plus the sign to be used for ions turning around.

We account for the ions of a selected mass received by the detector after the beginning of the cycle, after the ions belonging to the preceding packet are already detected. The number N of the ions detected is obtained by summing over the three distributions initial velocity f(v), time g(t0) and site h(s) of formation, respectively [10]:

N= lllf(v)dvg{to)dt0h(s)as (29)

space and time intervals, which vanish outside. To obtain analytical expressions for the detected ion current distributions, we approximate the Max-well-Boltzmann initial velocity distribution of the ions created in the gas phase by the following cosinusoidal function:

f(v) = cos2 πηιν

leEJ) (30)

Thus, the highest initial velocity accounted for in the peak formation will be i>max = (eEJ>)/m. The arbitrary coefficient b may be so selected that the resulting Foster distribution covers more than 90% of the molecules, with less than 10% error on the original Maxwell-Boltzmann distribution [11].

We introduce a new variable: y = mv/(eEa) and a local detection time: r = /arr - /w. The integra-tion domain will be defined by the following con-ditions: — T < t0 < T for the ionization process lasting 2T: -d < s < d for the ionization process happening inside a space slice 2d thick, 0 < y < b for the range of the initial velocities: t0 = τ ± y9

resulting from Eq. (28). A typical configuration of the integration domain defined by the yOt0

plane is depicted in Fig. 4 for ions starting with a velocity component directed towards the detector. The detected electric charge g , counted from some very negative (after the detection of the preceding ion packet), constant r value (trace not shown in

- y

Fig. 4. Calculation of stable ion carried charge: Maxwell-Boltz-mann initial velocity distribution approximated by a cosinusoidal

g(t0) and h (s) are constants over the ionization one.


Fig. 4, will be

■Γ-τ Ô = C , U o ^ ^ J T d'°

(31)

with C\ — 2de ghE^/m. The detected ion current distribution / results as

a derivative of Q with respect to the arrival time r. As peaks are often given in arbitrary units, for clarity we will disregard the constant in front of the detected current expression. Accounting for both signs in Eq. (28), summing these contribu-tions the following distributions, all symmetric with respect to r = 0, results in:

(i) If b < T the peak is flat topped because for 0 < r < T - b

i = b (32)

when T-b<r <T+b

h = b+T· T b . (T— τ)π

(ii) If b > T then for 0 < r < b - T

h = T+T* ' . (τ+Τ)π . (Τ-τ)π sin -—-—— + sin -—-——

(33)

(34)

while for b-T<r <b+T

ι — i\

The peak shape depends in fact only on a single parameter: the T/b ratio. For an ion source with a 320V-cm_1 extracting field, for ions of 130u, from a gas at 510 K only 1.5% of the molecules are left out by choosing b = 1.877 x 10~8. Depend-ing on the packet length, the effect of the initial velocities is more (Fig. 5, T = 4 ns) or less (Fig. 6 T — 15 ns) obvious.

Numerical integrations are needed when initial velocity distributions, such as those of Maxwell-Boltzmann are used to describe peak shapes. Per-forming the same integration steps as in the preced-ing procedure, but accounting for b values pushed to infinity, differentiating again with respect to r, we obtain the following detected current expres-

Fig. 5. Velocity spread dominated, stable ion peak in a linear drift TOF mass spectrometer, Maxwell-Boltzmann distribution, approximated: T = 4 n s .

sions which are easy to calculate numerically: for 0 < r < T

{p(\-r/T) rp(l+r/T)

i=\ f(v)dv+\ f(v)dv (35) Jo Jo

if T < T < oo

|7>(1+τ/Γ)

/ = f(v)dv Jp(r/T-\)

(36) h(r/T-\)

with p = eEa T/m and r = 0 the peak symmetry axis.

For a Maxwell-Boltzmann initial velocity dis-

Fig. 6. Stable ion peak shape dominated by ionization process length, approximated Maxwell-Boltzmann distribution: T — 15 ns.

D. Ioanoviciujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 43-65 49

=1 ρ(1-τ/Τ)

■ρ(1+τ/Τ) f(v) dv (39)

while for - T < τ < T in the above integral the lower integration limit must be replaced by zero. The peak cannot be symmetric anymore.

Fig. 7. Stable ion peak, Maxwell-Boltzmann initial velocity distribution.

tribution function, in gas phase, the currents are when 0 < r < T

-l ra(\-r/T) ra(\+T/T) 2

i = { e"* djc + e"* ax (37)

while for the branch T < τ < oo

( a(l ·α(1+τ/Γ) _ 2

i=l e x ax » ( T / 7 - 1 )

(38)

where a = eEaT/(2mkTk)l/2

9 k being the Boltz-mann constant, Tk the temperature in Kelvin. The peak shape in Fig. 7 gives the result for m = 300 u, £a = 800 V cm- 1, T= 2.5 ns, Tk = 500 K.

For ions originating from solid surfaces only the negative branch of Eq. (28) must be used while the integration with respect to the variable s is replaced by a "delta" function. The resulted detected cur-rent distributions, for an arbitrary f(v) function gives, for ion packets originating from solid sur-faces, if -oo < r < —T then

Metastable peaks in linear drift decelerating ramp TOF mass spectrometers

To detect daughter ions from metastable frag-mentations in a linear TOF mass spectrometer, a decelerating-reaccelerating potential ramp must be placed on the ion path. Otherwise, having almost the same velocity, daughter and parent ions would mix together at the detector. The calcu-lations refer to stick-shaped primary ion packets of negligible cross section, the ions disintegrating with the constant λ, releasing in this process the energy 6. The ramp is assumed to be located in front of the detector, being of negligible length (compared to the whole flight path). Let us set the time origin at the moment when half of the primary ion packet has already left the ion source. Some metastable ion, located at a distance s from the middle of the packet (Fig. 8) disintegrates when the whole packet has already travelled a distance X. In the center of mass the daughter ion movement makes the angle Θ with the packet axis. The longitudinal velocity of the daughter ion becomes [11]

V[ = Vs(l +^COSÖ)

μ = ( i / p - i K

υΛ

1/2

ion ψ packet

ion source grid

v> ΖΖΔ

2dn

I detector

Fig. 8. Primary ion packet and daughter ion flight parameters.


where p = md/mp is the daughter ion md to the parent ion mp mass ratio.

The daughter ion spends a first time interval as a parent flying with the longitudinal velocity vs, then travels the remaining field-free distance with the new velocity v\.

(40) X D - + — V\

Or keeping only first order quantities, as μ is also small

Varr = D - S + ßCOS0(X - D) (41)

To obtain the number N of the daughter ions detected after some reference time, selected after the passage of the ions in the preceding cycle, we have to integrate the daughters produced during a short time interval at = dX/vs in the small solid angle element sin0d0d<^, from a part ds of the parent ion packet:

N -^-sintfdfld^dXd.s (42)

where N0 is the linear density of the primary ions in the packet.

z 1

χ=χ/(<Γ-Τ)

— y

(a)

Some substitutions simplify further calculations:

X=D(\ -x)

tanvs - D

cos Θ = -y

μϋ

s = -ζμϋ (43)

(44)

After integration over φ we obtain

^ ^ O J d x J d z (45) with C2 = 2πμϋ2Ν0\/ν5

The integration volume is defined by the follow-ing conditions:

- O O < J > < - 1 \<y<oo (46)

as it results from

0<6><7Γ 0 < χ < 1 -δ<ζ<δ (47)

with δ = ά%/(μϋ), primary ion length = 2dg and

(48) x

Z = - + T y

from Eq. (41). A surface for very negative, constant τ is taken

as a reference to understand the time evolution of

^ x

Fig. 9. (a) Detected daughter ion carried charge calculation, yOz cross section of the integration domain, (b) Detected daughter ion carried charge calculation, projection onto the xOy plane of the integration domain.

D. loanoviciujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 43-65 51

the integration volume. The resulting peak is sym-metric with respect to the instant r = 0, because the substitution of r by —r, associated with the changes y to —y and z to - z does not change functions or integration intervals.

The integration over z gives two function types (Fig. 9(a)) presents a typical case of yOz section through the integration domain):

dz = - + r + δ or 26 Jo y

(49)

The projection of the integration domain onto the xOy plane allows us to set the final integration limits (Fig. 9(b) gives a relevant configuration). The detected charge Q is then

( r + <5 r - l

τ-δ J-x/Ιτ

r-x/(T-s) 26 ay

-X/(T-S) \y ) y

+ âxl - + T + 6 }τ+δ 3-χ/{τ-6) \y

dy (50)

The detected ion current distribution results as derivative of Q with respect to r. Depending on the packet length and the released energy, the following possible distributions result (symmetric peaks) for ή: if δ > 1 for 0 < r < δ - 1, ή = 2 the multiplicative constant properly chosen, while for <5- 1 < r <8+ 1

Fig. 10. Metastable ion peak shape in a linear drift TOF mass spectrometer, δ — 0.2734.

Fig. 11. Daughter ion peak in a linear TOF mass spectrometer, δ= 1.282.

i]t = l + (6-T)[l-\n\6-T\]

When δ < l , f o r O < T < 1 -δ

ή. =2δ-(δ-τ) In \δ-τ\-(δ + r) \η(δ + r)

(51)

(52)

while for 1 — δ < τ < δ + 1, i\ = ήβ. Aniline ions of 93 u disintegrate to 66 u daugh-

ters releasing energies of 42.4 meV per process [12]. In a linear TOF mass spectrometer of 1 m length, packets of primary ions accelerated to 5 keV, last-ing 10 ns would produce the peak shape in Fig. 10, for the mentioned transition (6 = 0.2734). In simi-lar conditions the transition 88+ —> 70+ of the ethyl acetate ions, generates, for 3.24meV released energy, the peak in Fig. 11.

Calculated free drift TOF mass spectrometer performance

The following data are illustrative of the resolving power capability of the Wiley-McLaren [2] TOF mass spectrometer configuration. We consider a structure where da = 1.27 mm, dh = 10 mm, D = 1 m, s = 0.2 mm, the accelerating voltage is 2.8 kV, and v = 7.65. Ions of 200 u result from a sample in the gas phase at 525 K (VXQ = 147.7ms-1), in 50 ns packets. The contribution of the initial kinetic ener-gies to the value of 6C is entirely negligible; the reso-lution is then estimated to be 343. If the packet length is reduced to 1 ns, the resolution should increase to 440. If, instead of thermal molecules, a 10 K supersonic jet is used for ion production, we arrive at a resolution jump to 1582, again by ioniza-


detector

Fig. 12. Linear drift TOF mass spectrometer with additional grid for post source focusing.

tion phenomena lasting 1 ns. To feel the effect of the residual initial position energy spread generated aberration we neglected all the other contributions and thus Rh = 3200.

Post source focusing

Post source focusing is obtained by applying time dependent electric fields somewhere on the ion path between source and detector. Muga [13] used the so called velocity compaction method. Kissel et al. [14] applied pulses of rectangular shape between properly located grids.

The analysis provided below refers to rectangular applied pulses. The time origin is taken at the moment when ionization begins. An ion leaves the ion source with the velocity vXb = vs(l + <5C/V2)ly/2

at the time /s — t\ + /a + /b: t-x being the time when the ion was created. The ion flies through the field-free space D (Fig. 12) until the potential on the ion source exit grid Gs is suddenly increased. This hap-pens at the time Tp following the convened time origin. The ion was allowed to fly freely during t¥ = Tp — /s. The accelerating field Ep, applied after / = Tp between the grids Gs and Gp, will act upon the ion along the distance dç — D — D0— vXbtF. The ion velocity at the grid Gp, vx and the time after the field Ep was established from Eqs. (7) and (6). After successive substitutions of vXb, d^ and tF we obtain

ôceEp

(2f/f) Tv*

Vf z c f

2>

(2^2)

4»(i + i/) 4 ( ^ - 1 ) V u(\+u)

(53)

where the new symbols are defined by the relations

2 Us σ =Vf

x

Uf=Us + eEp

eEpdz

D-D0 ' s * p \+V

(54)

To reach the detector the ion has to fly the always-field-free space of length D0. It arrives at the detector at the time ta:

, (2<4/χ)(1/σ-1) + £>0σ d^v2

M = 'pi VsX

+ -(DoXa2-2da) \4dy \+u2_(db/d!l)(u+\)

+ u2+^-v.

2v

2da 2σ2χ (55)

The inspection of this result reveals that this post source focusing procedure can eliminate simul-taneously both the effects of initial velocity and ionization process length when da = v2D0/2. The correction is mass dependent as σ depends on vs. To obtain the expected focusing effect, the whole packet must be admitted to the Gs - Gp space before t = Tp.

"Space focusing" can be obtained if:

^(DoXa2-2da)

x [vsTp/(2d,) - 1/(1 + i?) + (4/rfa)(i/ - l)(i/ + 1)

- 1 / ( 2 χ ) ] - 4 / χ = 0 (56)

It is obvious, that Eq. (56) is conflicting with the condition for velocity focusing. However, as usual, a compromise can be found for the best resolution.

D. loanoviciu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 43-65 53

Fig. 13. Ion-trajectory-defining parameters in linear mirror TOF mass spectrometers.

Linear reflectron TOF mass spectrometers

The ion-optical properties of the TOF mass spec-trometers designed to direct the ion packet move-ment normal to the electrostatic mirror boundary approximate fairly well also to those of the oblique entry instruments in use, because, usually, inci-dence angles do not exceed 2°. Calculations for normal entry reflectrons were performed by Mamyrin et al. [15]. Gohl et al. [16] and Brunelle et al. [17], the latter group also trying to account for the source path section.

A single field source (Fig. 13) achieves space focusing at 2ds beside its grid, as Eq. (22) shows after the substitutions: v = 1: db = 0: */a = ds. The time that ions spend inside the source and in cover-ing the 2i/a distance is tx\

U = (57)

To fly over the remaining field-free spaces L — 2ds

the ions need to travel a distance t2:

h = L-2ds (58)

Inside the mirror's first stage, ions are decelerated from their original velocity vXb to vXr, only the x component being affected (Eq. (11)):

V .2x1/2 (59)

where rf = Us/Ur, Ur = Us- eExdx, Ex being the field inside the decelerating stage, d\ its depth. The time during which the ion remains inside the mirror is the sum of twice the deceleration time (the reacceleration time after reflection is identical) and the "turn around" time in the second, reflecting stage where it enters with νΧχ

normal velocity component. So the time spent inside the mirror f3 is

2m h=7E[ vXb + vx I- (60)

E2 being the reflecting field intensity. After the usual series development we cancel the

coefficient of 6 in the total flight time expression td = tx + t2 + Î3- The condition for first order energy focusing in time gives the total field-free path length L:

L=(Vrf,)(l-^ + ) / ( ^ - l ) + 2^ (61)

Then the second order energy-spread-dependent contribution to the packet length becomes

ΔΓ„ = K + η2άχ[η{Εχ/Ε2){Ζ - η2) + η3

-3η + 2]/(η2-\)}δ2/2ν8 (62)

The total flight time for the witness ion is

Αηαχ{Εχ/Ε1){\+η1)-{η-\)1 . 4ds

vs rf - 1 s

These formulae are valid for any two stage mir-ror with first order energy focusing in a time TOF mass spectrometer.

The single stage mirror is the particular case for when the two fields are identical (E2/Ex = 1) and the ion has lost all its energy penetrating the first stage t/r, thus vanishing and η becoming infinite. Taking the limit for η —> oo in Eqs. (61)—(63)


we have, for the single stage reflectron TOF spectrometer

(64)

where d\ is now the mirror depth. The two stage mirror structure allows the simul-

taneous cancellation of the first and second order energy dependent contribution to the packet length. Then, the ratio of the electric field intensity in the two mirror stages must be

T.. -Ml

= 2(4 + 24)

rfW/2 + rfi) — 0

2L

Ex = [rf -3 r / + 2 + ( 4 M ) ( r ? 2 -1)/η2} E2 η(ηΐ - 3)

(65)

The total field-free space length and flight time through the mass spectrometer are

L = 2 η2-3

2„2i

ίΣ = 4[2rf1<^-l)+4(rr-l)V] vs η2 - 3

or connecting the two together:

fc-iLÎÎ+ΰ

Now the second order energy aberration is absent and the third order term is to be accounted for:

(66)

(67)

(68)

ΔΓΙ Π = 8vs

(69)

where the small contribution due to ds was neglected.

The allowed range for the η values results from Eq. (65). The smallest allowed value for η making the field ratio positive, must be somewhat greater than 31/2, depending also on the source correction term. This limiting case corresponds to a vanishing field in the second stage, the ion remaining in the reflecting region for an indefinitely long time (Eq. (67)). For η = 31/2, 66% of the ion energy is lost in penetrating the first mirror stage.

The other limiting case, when η —► oo, makes the Ex/E2 ratio approach unity (Eq. (65)).

The total flight time defines the TOF mass spectrometer mass dispersive power and conse-quently the resolution if t0 is the main contribution to the detected ion packet. For a specified drift-space length, the single stage mirror is the most dispersive.

The depth h of the two stage mirror, calculated for the witness ion is

A _ L ( t y + l ) ( ^ - 2 ) / 4 - ( r ? + l ) 2 ( 4 / 2 ) η2 - 3

Again the two limiting cases are: (1) for infinite r/, when the mirror becomes single staged, the mirror depth is roughly h = dx or h = L/4; (2) for η approaching 31/2, when the second stage field becomes extremely weak, the mirror depth as well as the field-free spaces grow indefinitely.

The ion packet transverse size at the detector level results from the maximal y and z coordinates for t — /Σ, namely

and

Zd =Z0 + VZQtx

(71)

because along these directions, the movement is free of any force in the electric field assumed to be homogeneous.

Transfer matrix elements and resolution for oblique incidence mirror TOF mass spectrometers

Oblique ion incidence on the homogeneous elec-tric field mirror boundary (Fig. 14) causes addi-tional aberrations as it introduces differences in the flight path across the packet. However, for incidence angles of a few degrees, the supplemen-tary aberrations are kept small and the often long field-free spaces are moved away from the source and detector assemblies, ensuring more freedom in their design. All of the oblique incidence mirror transfer matrix elements, oblique incidence effect included, are given. The notation for these elements is: column index/line index. A second order approx-


imation was adopted for time matrix elements, only first order terms being kept for transverse elements. The relevant elements [18] are

(x/x) = (a/a) = -\

(x/a) = Asin2e-Ccos29

sin(20) (x/6) = (A + C)

(t/a) = (A + C)

{t/l) =

(t/δ) =

4

sin(26>) 2Ve

(2vs)

Acos2e-Csin2d

2v*

. , x {3A + B+C)ûn2e-A

(t/αδ) = sin(20) 2A + B

, ,ccx 3Csin2 e-(A-B)cos2 Θ {t/66) = -

8vc

(t/ßß) = -(t/S)

(72)

(73)

(74)

(75)

(76)

(77)

(78)

where JC is the transverse distance from the packet axis in the plane where the witness ion trajectory lies, a the angle made by an individual ion trajec-tory with the same axis in that plane, ß the angle of the ion trajectory with the deflection plane, and Θ the incidence angle:

A = AdytfcosO

1 + (E1/E2 - l)r7COsfl/(l - ??2sin2fl)1/2

rf-\

B = 4dlV2 cos3 Θ l-E*/E>

( l-772sin20)3 / 2

C = 4 77 ηοοϊθ+(Ει/Ε2 - 1)(1 -7/2sin2fl)1/2

η2-\

(79)

By using the above matrix elements and those of the field-free space of length L [19], all of the

Fig. 14. Oblique ion packet incidence double stage mirror TOF mass spectrometer: geometry defining parameters.

instrument matrix elements were obtained. The single stage, first order energy focusing in time mirror TOF mass spectrometer properties are summarized in the simple relations A = C = L and B = 0, when the resolution results from the formula

Rh L Lvs L V 2 /

+ (a tan ΘΫ + 2a6 tan Θ + — 4

(80)

xp being the packet geometric transverse extent. In the formula of ref. 18, the "turn around" correc-tion has been included.

For double stage, second order energy focusing mirror TOF mass spectrometers, the conditions become: A = L; B = - 2L , then

Rh = (1 - l/v2)/{[vst0/L + 4(ds/L)(vJvs)

+ (xp/L) tan 0( 1 + a tan θ + δ/2)

+ (r/2-l)<53/8]cos20} (81)

56 D. loanoviciujlnt, J. Mass Spectrom. Ion Processes 131 (1994) 43-65

In both resolution Eqs. (80) and (81) there are terms of second order having a tan Θ coefficient, also, there are third or higher order quantities.

Inhomogeneous electric field gridless mirrors

The grids in the mirror structure are at the origin of some unpleasant effects on instrument performance. In fact, as systems of miniature diverging lenses, they induce angular deviations on the passing ions; how-ever, calculations show small ion deviations [20,21]. Grids cause ion losses: even with 90% transparency grids, a two stage mirror will waste 36% of the oncoming ions, for this reason alone. Grids are a strong background source in experiments with clus-ters because they cause fragmentation. The reasons already mentioned are enough to justify a major study of gridless mirrors. Gridless mirror TOF mass spectrometer designs were reported by Frey et al. [22], Wollnik et al. [23] and Bergmann et al. [24].

The design procedure is entirely numerical and as well as the elimination of the enumerated unwanted side effects, it is possible to obtain simul-taneous energy focusing in time, and transverse packet concentration on the detector. Some poten-tial distributions [24] suggest, as a rough approx-imation, two, almost homogeneous field regions separated by a short lens-like potential kink. Such an approximation, of little interest for final mirror design, could be of some use for consideration in early design stages.

Stable and metastable ion peak shapes in reflectron

TOF mass spectrometers

The formulas describing stable peak shapes in linear TOF mass spectrometers may be used also for reflectrons, as long as only first order packet components are accounted for. However, with all the other conditions and suppositions unchanged, in Eqs. (35) and (36) we may use, for single field ion sources, a different form of p: p = 2Tds/vl.

The peak shape in Fig. 7 could correspond to

ds = 5 mm, m = 300 u, Us = 2 keV, T = 0.5 ns and TK = 500 K in a reflectron TOF mass spectrometer.

A detailed study of the metastable peak shapes in single stage reflectron energy focusing TOF mass spectrometers tuned to the parent ions was per-formed in ref. 25. In this work the peak shapes are built from fewer functions, with the additional need to combine two distributions together accounting for their individual variation intervals.

As for linear TOF mass spectrometers, we assume that the probability of primary ion disintegration is λ, an energy ε then being released; the packet cross section is negligible. ν·χ and μ are also defined by identical relations. The daughter ion flight time to the detector /arr includes the "turn around" time inside the mirror for the daughter ion:

X L — X - s 2maV[ t =—i 1 — *arr » ' 1 7

v« v\ eE

(82)

where L is the sum of the field-free spaces. Now, first order energy focusing condition having been satisfied, we have

Van- = L 0 +p)-s + ßCOse[X-L(l - p)] (83)

The calculations follow the same pattern as in the linear TOF mass spectrometer case. To bring the expression of N from the initial form (Eq. (42)) to the more convenient one (Eq. (45)) we perform the set of substitutions as follows:

X=L(l - p ) ( l -x)

1 ^ " c o s f l

s = — z/iL(l — p)

and

(84)

r = tMVv% - L(\ + p)

/ i L ( l - p ) (85)

Besides the conditions in Eq. (46) for y (unchanged), we have

1 -q<x<1

and

-6<z<6

(86)


Fig. 17. Flat topped metastable peak generated for 6 = 0.016 and q = 0.638.

Fig. 15. Metastable peak shape in a reflectron TOF mass spectrometer for δ = 0.029 and q — 1.5.

where the newly introduced quantity q is given by q = LQ/[L(\ — p)]9 L0 being the first field-free space length, and δ changes to 6 = dg/[pL(\ - p)].

The consideration of the various parameter ranges shows that the detected daughter ion distri-butions im may be built up as the sum of two terms, the first i\ being given by Eqs. (51) and (52):

*m = h ± h (87)

where the positive sign is to be used when q > 1. The following is functions must be used depending on the values of the relevant parameters F o r C < £ i f 0 < r <<5-Cthen

h = 2C

ΐονδ-ζ<τ<δ + ζ

is=C + (f>-r) 1 - l n l*-r|" (88)

C where ζ = \q - 1|. When ζ > δ and 0 < r < ζ - δ

Fig. 16. Metastable peak in reflectron TOF mass spectrometers when δ = 0.237 and q = 2.459.

4 b = 2 6 - ( 6 - r ) l n J ^ - ( ( 5 + r ) l n ^ (89)

while ΐοτζ-δ<τ<ζ + δ

In a single stage symmetric reflectron, 2L0 = L = lm, C 6 Hj primary ion packets accelerated to 0.7 keV, lasting 4 ns could produce the C4H4 daughter ion peak with long tails as in Fig. 15, an energy of 0.10 eV being released (the parameters of interest are δ = 0.029 and q = 1.5).

Consider another example: that of the cationized leucine-enkephalin ion (M + H) + (€28Η38Η507)+

disintegrations [26] in a symmetric reflectron with L= lm. Assume that ions are emitted in 4ns bursts, accelerated to 8keV. For an energy of 0.15eV released in the process, the transition 556.3+ -► 443.2+ (δ = 0.237 and q = 2.459) gives the peak in Fig. 16, the daughters of mass 120.1 u, and the flat topped peak of Fig. 17. The relevant parameters for this last case are δ = 0.016 and q = 0.638.

An oversimplified case is that of a very short parent ion packet [25] allowing us to find a simple expression for the metastable peak width at half height. The detected current distributions take the form

i ' = - l n ( l - t f )

for 0 < r < 1 - q

and (90)

i = — In r

for 1 - q < T < 1

The peak temporal half height width Atm is

58 D. Ioanovkiu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 43-65

Atm = 2-^(\-p) 1 - (1 -P)A) 1/2

(91)

The parent ion disintegrations inside the mirror spread daughter ions all over the mass range between the metastable mass position ra* = mp(\ + p)2/4 and that of the parent, while those fragmenting on the second field-free space contrib-ute to the parent ion peak intensity.

Performance of mirror TOF mass spectrometers

We estimate the resolution of a single stage mir-ror TOF mass spectrometer of usual size L= \m ("effective" ion path 2 m). Assume that 372 u mass ions are accelerated to 3 keV by a 2 mm gap. If the mean velocity of the ions leaving the surface is 509.3 ms"1 , the resolution of a linear reflectron TOF mass spectrometer delivering 1 ns packets should be 10970. An aperture angle of 1° and an incidence angle of 1.5° leaves the above value prac-tically unchanged (a decrease of a few tens only). However, a detector placed parallel to the mirror face could cause a resolution loss to 8530 for a 1 mm-wide packet.

For ions resulting from thermal molecules, a long path TOF mirror spectrometer is able to reach very high resolutions. For t0 = 1 ns, Γ = 5 2 5 Κ , Us= IkeV, m = 1 3 2 u , L = 4M, accelerating gap of 2 mm, already a single stage mirror TOF spectrometer could reach a resolution over 52 000 (oblique incidence induced aberrations neglected).

Comparison of linear drift and mirror TOF mass spectrometers

There are no means to eliminate completely initial velocity spread effects by static electric fields, neither in simple linear nor in mirror TOF mass spectrometers. Two ions formed simulta-neously at the same point with the same velocity but with oppositely directed longitudinal compo-nents will be separated all along their flight path by the turn-around time. The turn-around time, for

a specified initial velocity representative of an ion group (as most probable velocity, for instance), may be used to calculate packet length. Half of this value should be accounted for when dealing with a velocity distribution originating from a sur-face. As turn-around time is inversely proportional to the extracting field intensity, for a given ion energy, single field sources are advantageous. Long flying times favor turn-around-time-effect reduction.

For Wiley-McLaren instruments [2], the resolu-tion calculated only from the second order aberra-tion term increases with drift length. However, the drift space increase is connected with the increase of the parameter v which reduces the extracting field. For a single stage mirror the similar expres-sion is independent of the instrument size, it being possible to extend it to reduce other aberrations. In double stage mirrors the resolution including only

Fig. 18. Parabolic potential mirror created between a hyperbolic rod structure.


the third order component contains a factor \/(η2δ3), again independent of the instrument size.

Summarizing, the advantages of the mirror over free drift TOF mass spectrometers are: (i) they obtain shorter turn-around times at the source exit for a given nominal ion energy, reducing their influence on the resolution to the desired level by increasing the flight path properly; (ii) double stage mirrors also cancel the second order energy spread aberrations.

Parabolic potential mirrors

There are two ways to take advantage of the outstanding, theoretically perfect, time focusing properties of the parabolic electrostatic potential distributions: (1) as a mirror performing simulta-neous collection of ions leaving its boundary together, no matter what their energy or entry angle are; (2) as a means to simultaneously detect accelerated ions created in a deep space region.

It is possible to generate a parabolic potential distribution with plane symmetry as in quadrupoles fed by static potentials, or with axial symmetry as in ion traps.

In a plane symmetric parabolic potential the ion motion (Fig. 18) is governed by Eq. (92):

·· eG

x-\ x = 0

mi

.. eG (92) y y = 0 v )

m/ z = 0 where the potential φ = G(x2 - y2)/2 was substi-tuted, the constant G = 2φ0/α

2 resulting for a potential φ0 applied on electrodes located at a dis-tance a from the axis.

Integration gives the detection coordinates, sub-script "d" and the function of the initial coordi-nates and velocity components, subscript 0:

vx

*d = *o c o s ωί H s i n ωί

ω yd = y0 cosh ωί + — sinh ωί

ω

(93)

where ω2 = eG/m. As the x equation states, all the ions leaving the mirror boundary simultaneously i.e., the plane x = 0 will focus in time on the same plane, the arrival time depending only on the ion mass:

'd=z,="fe) (94)

The ions penetrate inside the mirror at a depth h depending on their velocity, the fast ions enter deeper:

x = h = vju (95)

The transversal coordinates at the detector are:

yd =z yQ cosh π + — sinh π ω

ζά=ζ0 + ω

(96)

(97)

The y coordinate dilates more due to the transverse defocusing action of the field in this direction.

Similar properties are displayed by the axially symmetric static ion trap field [27], having the potential φ = g(z2 - r2/2), the constant being g = φ0/α

2 for the electrode at z = a from the origin of a cylindrical coordinate system.

The movement equations can be written:

r - ^ = 0

z + 2egz

= 0 (98) rrt\

mr ψ = constant

The coordinates at the detector result from the initial conditions through the relations

rd = ro c o s h ωΑ H—" s i nh ωΑ

ζά — ζ0 cos ωζί H—- sin ωζί ω7

(99)

(100)

with uç — u?zj2 — eg j m. Again mirror face-to-face time focusing is only mass dependent:

π k= —

ω7

m %eÛ

1/2 (101)

zd=z0 + vZnî The mirror depth must be at least h deep:

60 D. loanoviciu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 43-65

ω7 (102)

As the field defocusing action on r is distributed over all the φ directions, it is less severe than along the y coordinate in the quadrupole field.

The resolution of such mirror TOF mass spectrometers depends only on the time interval /0 during which ions are admitted inside the field:

and

n(m/eG)l/2

2h

(103)

w(m/2eg)l/2

2h

respectively. The same movement equations describe the ion

behavior inside accelerating parabolic potentials. Now the ions begin their paths at L + s from the origin with small initial velocity components vXo

and Vj,0.

In the constant quadrupole field the ion coordi-nates at the detector are

xd — x0 c o s ωίά + ~ sin ωίά ω vv

ya = y0 cosh ωίά + — sinh ωίά ω

(104)

Neglecting initial velocities, ions formed simulta-neously arrive from different distances to the x = 0 plane when /d = π/(2ω). The initial veloc-ities create a flight time difference tv = vXo/(u

2L). Similarly, for the axially symmetric static field

quadrupole trap [28], we obtain the coordinates

zd = z0 cos ωζίά sin uztd

ω7

rd = r0 cosho;rid ^sinhu;r/d

(105)

Collection from different starting points happens after ta = π/(2ωζ) and the initial velocity-induced arrival time spread will be Atv = vzJ{u?zt). There are no essential differences in the structure of the resolution formulas for the two kinds of accelerat-ing parabolic potential distributions:

Fig. 19. Cylindrical mirror geometry and reflected ion trajectories.

/?h = (ΤΓ/4)

Mo+ 2^/(0^)]

for plane symmetry.

(7Γ/4) R* =

[u)zt0 + 2vzJ(ojzL)]

(106)

(107)

for axial symmetry. It is assumed that ions are produced during an

ionization event of duration /0, in gas phase, the factor of two being introduced to account for opposite senses of initial velocity components.

The time aberrations arising from packet-oblique incidence at the limit of a six-rod generated 1400 mm long, double deflecting quadrupole field mirror were calculated in ref. 29.

Cylindrical mirror TOF properties

Cylindrical mirror TOF properties seem to be of more interest for auxiliary transverse focusing in TOF mass spectrometric devices than for main mirror analyzers. This is because a big flight gap, with a small extent along the cylinder generatrices,


deteriorates the ideal potential distribution. Its merit can not be disregarded in primary ion guns or as an electrostatic particle guide [30].

The potential distribution created between two concentric cylindrical electrodes, the inner one of radius R0, the radial field intensity being EQ at its surface, produces the radial field Er = -E0R0/r (cylindrical coordinates). The movement in an r—z plane (Fig. 19) is described by the equation: mr = -eE0R0/r that integrated gives [31]:

z — ZQ = IviR^sinOi

x[m/(eE0Ro)Y^2l\-u2/2du

t-

(108)

with K{ = [m/{eEQR0)]xl\coêÏ9 "Γ, being the

subscript for entry values. After series developments in a and 6 as θ·χ = Θ + a and mvf/2 = t/s(l + 6) we obtain the distance Az between the witness ion entry and exit points

Az = 2R0Jten9 (109)

with

= Ké2i2\K^ Jo

'2 au

The main path deepest point coordinate rM:

ΓΜ = * 0 ^ 2 / 2 (110)

The radial first order matrix elements (giving the extent along z)\

000 = -i (z/a) = 2RQ tan2 Θ[Κ2 + J(K2 - cotan2 Θ)]

(111)

(ζ/δ) = R0 tane[K2 + (1 + K2)J]

(a/a) = - 1 (112)

The TOF matrix elements, including second order effects:

(t/z) = (t/zz) = (t/za) = 0

(if a) = 2a tan 0 ( 1 + / )

(t/6)=a(\+J) (114)

(113)

(t/aa) = ata,n29

x [K2 - cotan2 Θ + J(\ + K2 - cotan2 Θ)}

(115)

(t/αδ) = fltantf[l + K2 + (2 + K2)J] (116)

(t/66)=^[K2-3 + \+K2] (117)

where a = —-—— vs cos Θ.

For ions with velocity components outside the r—z plane there is not yet a reliable analytic descrip-tion available, those developed contain diverging integrals or the final expression does not account for all the terms.

An electrostatic particle guide, included in a drift space of a linear or a mirror equipped TOF instru-ment induces some additional time aberrations which may be severe, as ref. 32 shows. Calcu-lations based on the above time matrix elements seem to confirm that such aberrations arise in cylindrical reflecting systems.

Packet handling and transverse focusing lenses

To perform mass spectrometer tuning and for ease in exploitation, some components, often with only an auxiliary role, are provided. To this cate-gory belong deflecting plate pairs and lenses, mostly "einzel", placed along the ion path. How-ever, lenses may play a major role in concentrating ions on the detector. Thick electrostatic lenses may also induce undesirable time defocusing effects.

Transverse focusing "einzel" lenses are placed in front of the TOF analyzer [33] to efficiently shape ion packets early in their flight. Simple relations can be derived for an almost linear reflectron having a thin lens in front of the mirror. It is also assumed that the ion packet does not pass through the lens on its return, or that when it returns, the lens is deactivated. As the lens is


thin it is instantly traversed. Performing simple matrix multiplications the focal distance / of the lens ensuring transversal focusing at the detector is easily obtained. For single stage mirrors we have

(118)

while for two stage mirrors

f=Lr\ 1 -rfLf

2Ltf-\)

rfLf

(119)

Lf is the source-thin lens distance, Mx the trans-verse magnification in the reflection plane.

To counteract initial velocity spread induced TOF aberration, sample molecules are seeded in supersonic molecular beams directed normally to the extracting field. The ion velocity component normal to the intended flight direction, inherited from the molecular beam, may cause the ion spot to slide partly or entirely from the detector area. An elegant solution to remodel the initial velocity components is the use of a quadrupole doublet [34]. Another solution, technically simpler but without a focusing action on the ions is a plate-pair deflecting device. Such a parallel plate pair of length dp

located at a distance Ld from the detector, deviates the ions by a distance (eEpdp/Us)(Ld + dp/2), Ep

being the field inside the condenser. Such a simple device may be used to suppress background ions originating from molecules with randomly distrib-uted initial velocities [35]. As (vyo/vs)Ld, the distance between the aniline 93 u mass ion and the background spots is 5.6 cm at the detector plane, in a reflectron arrangement of 2 m effective length, Ep can be so selected that the majority of background ions are missing the 4 cm diameter detector.

deflector

Fig. 20. Simplified representation of a three electrostatic deflector TOF system.

Tandem TOF systems

Multiple identical-cell, ion-optical systems may offer outstanding focusing properties, which, how-ever, lead to a concomitant increase in the instru-ment's structural sophistication.

A TOF secondary ion microscope was con-structed from three electrostatic sectors and four electrostatic lenses [36]. The three lenses following along the ion path and the immersion lenses may be activated in pairs. So the magnification can be altered from 60 to 250 the image fields changing accordingly. The electrostatic sectors are spheri-cal, deflecting to 90°. The transverse focusing prop-erties of the lenses and deflectors are identical in the deflection plane as well as normal to it, so stigmatic imaging is ensured. The microscope was designed to accomplish angular and energy focusing in both radial and transverse directions.

We give a first order ion-optical description of a three, equally spaced, 90° spherical deflector sys-tem connected to a lens (Fig. 20). At the detector plane the following instrumental coefficients are of interest: Mx = My radial and axial microscope magnification (objective lens not included): Ca = Cß coefficient of angular aperture (to be can-celled): Q energy aberration coefficient (also to be cancelled) and various TOF coefficients Ty


The coefficients of the system part located between the lens of focal distance / and the detector are denoted by Aj for the transverse and Bj for the longitudinal directions. As a first step we obtain

Mx = My + Ax--2-

Ca — Cß — AXLS + AQI 1 — /

Q = AS

Tx = Bx-B-f

Ta = BxL^Ba[\-J

Ts = B,--^

ΤΊ = ΒΊ + —

To detail the expressions of Aj and B} we use matrix elements from refs. 19 and 31. Therefore

Λν = Lf + d[ 1 -dLj

^bk-JÛ* r2 I \ 'e / \

A6 = rt{\ -Ax)

Äv = 2 h r - 1 (120)

Bn = 2 rP- L,-d[ 1 Y

9ττ

?)] A = re 2 + — - d 1 +

Li+Lf

ΒΊ = ά+ +|7rre

Selecting d = re, Ax = 1, and automatically

As = Bx = Ba = Tx=Ta = Q (121)

Τδ vanishes also if drift space lengths satisfy the relation

9π 2"

L{+Ls + Lf= ( — -2)d

then

Γ7 = 3πα (122)

The lens focal length results from the angular focusing condition

/=A (9π - 6 - 2Ls/d)

3 ( 3 π - 2 )

Thus the total magnification is

3έ/(3π - 2) M, = M, = 1 - 2 L

(123)

(124)

In particle optics a particular attention was paid to such beam transport devices as the four identical cell optical achromat [37]. The time of flight iso-chronous (TOFI) spectrometer for exotic nuclei mass measurements [38] incorporates a four mag-netic sector structure, especially designed to pro-vide good TOF features. The four homogeneous magnetic field sectors (later pole surface coils were also used) each deflect at φνα — 8Γ, the inci-dence and emergence angles being €m = 23.3°, the drift spaces in front of and after each cell magnet being Lm — 0.958rm, rm is the main path radius. Besides first order achromatic stigmatic image for-mation at the detector, TOFI cancels exactly the first order flight time coefficients in x, a and χδ, those of second order in xa, αδ and yß (y the object extent normal to the deflection plane), the others being small.

Based on the conditions to be fulfilled at the cell level in the multiple cell systems, the param-eters of a four homogeneous magnet isochronous achromat were calculated [39]. Its configuration was determined for vanishing fringing field inte-grals (whose role is subject to debate in some cases [40]). Its basic parameters, somewhat differ-ent of those of TOFI, can be summarized as 0m = 82.033°, em = 21.856°, Lm = 0.7444rm, the boundaries curved with a curvature radius ^m = rm x 106/6. When tested with the computer


program GIOS the longitudinal matrix elements were found to be between 2 x 10~6 and 8.6 x i o - 1 3 .

Concluding remarks

At the present time, commercial TOF mass spec-trometers are already able to survey a mass range over 300 000 u. However, the resolution of these instruments is an order of magnitude less. It is up to the users of these instruments, as specialists in biology and biochemistry for instance, to prescribe how detailed a mass spectrum they need in those far-off mass regions. Anyhow, resolution increase still remains a necessity.

To get closer to this permanent objective, not only for TOF mass spectrometry, ion-optical solu-tions will certainly play a distinct role. Much is expected from the leader: the electrostatic mirror TOF mass spectrometer. The contest between homogeneous electrostatic field and gridless mir-ror TOF mass spectrometers has not yet been closed but the latest design seems to have more hidden resources. Parabolic potential as a mirror, despite its wonderful intrinsic focusing properties has some drawbacks when incorporated in a real design (such as the absence of allowed field-free spaces).

Post source focusing, ion compaction and ion bunching procedures may advance also by reunit-ing more focusing features by proper instrumental parameter and impulse shape selection. Some of these solutions could especially benefit instruments working on the usual mass range.

References

1 W.E. Stephens, Phys. Rev., 69 (1946) 691. 2 W.C. Wiley and I.H. McLaren, Rev. Sei. Instrum., 26

(1955) 1150. 3 W.P. Poschenrieder, Int. J. Mass Spectrom. Ion Phys., 9

(1972) 357. 4 K. Walter, U. Boesl and E.W. Schlag, Int. J. Mass Spec-

trom. Ion Processes, 71 (1986) 309. 5 R. Grix, R. Kutscher, G. Li, U. Grüner and H. Wollnik,

Rapid Commun. Mass Spectrom., 2 (1988) 83.


7 U. Boesl, J. Grotemeyer, K. Walter and E.W. Schlag, Anal. Instrum., 16(1987) 151.

8 J. Schwieters, H.-G. Cramer, T. Heller, U. Jürgens, E. Niehuis, J. Zehnpfenning and A. Benninghoven, J. Vac. Sei. Technol. A, 9 (1991) 2864.

9 R. Kutscher, R. Grix, G. Li and H. Wollnik, Int. J. Mass Spectrom. Ion Processes, 103 (1991) 117.

10 R.B. Opsal, K.G. Owens and J.P. Reilly, Anal. Chem., 57 (1985) 1884.

11 D. Ioanoviciu, C. Cuna and S. Cuna, Abstract in Analysis 20 (1992) 95.

12 V. Aviyente, M. Shaked, A. Feinmesser, S. Gefen and C. Lifshitz, Int. J. Mass Spectrom. Ion Processes, 70 (1986) 67.

13 M.L. Muga, Anal. Instrum., 16 (1987) 31. 14 G.R. Kinsel, C D . Mowry, P.J. McKeown and M.V.

Johnson, Int. J. Mass Spectrom. Ion Processes, 104 (1991) 35.

15 B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A. Zagulin, Sov. Phys. JETP, 37 (1973) 45.

16 W. Gohl, R. Kutscher, H.J. Laue and H. Wollnik, Int. J. Mass Spectrom. Ion Phys., 48 (1983) 411.

17 A. Brunelle, S. Della-Negra, J. Depauw, H. Joret and Y. LeBeyec in A. Hedin, B.U.R. Sundqvist and A. Ben-ninghoven (Eds.), Proc. IFOS V, Wiley, Chichester, 1990, p. 39.

18 D. Ioanoviciu, G.E. Yefchak and C G . Enke, Int. J. Mass Spectrom. Ion Processes, 94 (1989) 281.

19 H. Matsuda, T. Matsuo, D. Ioanoviciu, H. Wollnik and V. Rabbel, Int. J. Mass Spectrom. Ion Phys., 42 (1982) 157.

20 X. Tang, R. Beavis, W. Ens, F. Lafortune, B. Schueler and K.G. Standing, Int. J. Mass Spectrom. Ion Processes, 85 (1988) 43.


22 R. Frey, G. Weiss, H. Kaminski and E.W. Schlag, Z. Naturforsch. Teil A, 40 (1985) 1349.

23 H. Wollnik, U. Grüner and G. Li, Annal. Phys., 48 (1991) 215.


25 D. Ioanoviciu, G.E. Yefchak and C G . Enke, Int. J. Mass Spectrom. Ion Processes, 104 (1991) 83.

26 X. Tang, W. Ens, K.G. Standing and J.B. Westmore, Anal. Chem., 60 (1988) 1791.

27 A.L. Rockwood, Proc. 34th ASMS Conference on Mass Spectrometry and Allied Topics, 1986, p. 173.

28 L.D. Hulett, D.L. Donohue and T.A. Lewis, Rev. Sei. Instrum., 62(1991)2131.

29 D. Ioanoviciu, C Cuna and A. Pamula, Abstract in Analysis, 20(1992)95.

30 N.S. Oakey and R.D. Macfarlane, Nucl. Instrum. Meth-ods, 49 (1967) 220.

31 D. Ioanoviciu, in P.W. Hawkes (Ed.), Advances in Electro-nics and Electron Physics, Vol. 73, Academic Press, New York, 1989, p. 1.

D. Ioanoviciujlnî. J. Mass Spectrom. Ion Processes 131 (1994) 43-65 65

32 P.W. Geno and R.D. Macfarlane, Int. J. Mass Spectrom. Ion Processes, 77 (1987) 75.

33 E. Niehuis, P. van Velzen, J. Lub, T. Heller and A. Bennin-ghoven, Surf. Interface Anal.,14 (1989) 135.

34 T. Bergmann, H. Goehlich, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 61 (1990) 2585.

35 A.P.L. Wang and Liang Li, Appl. Spectrosc, 45 (1991) 969. 36 B. Schueler, P. Sander and D.A. Reed, Vacuum, 41 (1990)

1661.

37 K.L. Brown, IEEE Trans. Nucl. Sei., 26 (1979) 3490. 38 J.M. Wouters, D.J. Vieira, H. Wollnik, H.A. Enge, S.

Kowalski and K.L. Brown, Nucl. Instrum. Methods, Phys. Res. A, 240 (1985) 77.

39 D. Ioanoviciu and J.A. Nolen, NSCL Annual Rep. 1989, p. 193.

40 D. Ioanoviciu, in J.F.J. Todd (Ed.), Adv. Mass Spectrom., Vol. 10, Wiley, Chichester, 1986, p. 857.


Design considerations in energy resolved time-of-flight mass spectrometry

A.E. Giannakopulos, David J. Reynolds, T-W. Dominic Chan, A. W. Colburn, Peter J. Derrick*

Institute of Mass Spectrometry and Department of Chemistry, University of Warwick, Coventry CV4 7AL, UK

(Received 28 April 1993; accepted 18 June 1993)

Abstract

We report the design, construction and performance of an energy resolved time-of-flight (ER-TOF) mass spectrometer. The system consists of a linear TOF coupled in parallel to the electric sector of a large double-focussing magnetic-sector mass spectrometer. The use of lensing between the TOF and the electric sector is discussed with respect to the required beam conditions. Ion transmission through the electric sector is treated theoretically, especially with respect to time-of-flight. Preliminary experimental results from the instrument using matrix-assisted laser desorption-ionisation are presented and discussed, with particular regard to the fragmentation of large ions.

Key words: Matrix assisted laser desorption-ionization; Double focussing mass spectrometer; Energy resolved time of flight; Electrostatic lenses; Biomolecules

Introduction

Mass spectrometry is the most powerful tech-nique currently available for the accurate determi-nation of the masses of gaseous ions. In addition, it can, potentially, provide much information as to molecular structure, ion formation and fragmen-tation energetics. Recent years have seen an ever increasing emphasis placed upon the mass range and sensitivity of mass spectrometers. This demand has largely emanated from the areas of biotech-nology and the life sciences, where analytical requirements are characterised by very little sample of a large and complex molecule being available.

The first step in any mass spectrometric experi-ment is the production of ions in the gas phase. For

small molecules this is not a problem and a variety of techniques are available such as electron bom-bardment, chemical ionisation [1,2], field ionisation [3,4], and multiphoton ionisation [5,6]. However, large molecules such as polymers or proteins are, by their very nature, likely to be involatile, having vapour pressures too low to produce detectable ions by the previously mentioned techniques. In addition, thermal sensitivity makes assisted desorp-tion of molecule ions difficult. In recent years, a number of desorption-ionisation techniques have been developed which are capable of acting on large and involatile molecules to produce gaseous ions. Such techniques include field desorption (FD) [4,7,8], thermospray ionisation [9], secondary ion mass spectrometry (SIMS) [10,11], fast atom bom-bardment (FAB) [12-15], plasma desorption (PD) [16,17], electrospray [18,19] and laser desorption (LD) [20,21]. * Corresponding author.

SSDI0168-1176(93)03871-1

68 A.E. Giannakopulos et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 67-86

The applicability of these various approaches depends on the physico-chemical nature of the target material. FAB, FD and liquid-SIMS have been extensively used in the analysis of large bio-logical molecules up to 10 ku, and for larger mol-ecules PD electrospray [18,19] and an extension of LD, matrix-assisted laser desorption-ionisation (MALDI) [22,23], are available.

MALDI

The use of lasers to ablate involatile organic analytes has its roots in the early experiments of Vastola et al. [24,25] in 1968. Subsequent studies in the feasibility and effectiveness of this technique have shown it to be a powerful tool for ion produc-tion for mass spectrometric analysis [26,27]. The laser desorption process imparts energy to the anal-yte either directly from the beam or as heat from thermal excitation of the substrate leading to a phase transition. A plume of material is then ejected into the vacuum, possibly driven by ther-mal vaporisation or sublimation [28], vibrational thermal expansion [29] or Coulombic repulsion [30]. The sample disintegration can lead to the formation of a region of high density and high molecular mobility called the selvedge, in which a number of processes may take place including cationisation, photoionisation, proton transfer, molecular rearrangement and fragmentation.

The use of matrix materials to improve analyte-ion production in laser desorption has been exten-sively studied. Early experiments had shown that the analyte optical absorption coefficient at the wavelength of the laser could have a dramatic effect on the ion yield [31]. Generally speaking, absorbing analytes required lower photon fluxes and gave a stronger yield of molecule ions. It had also been observed that co-desorption of absorbing and non-absorbing amino acids enhanced desorption of the non-absorbing species [32]. By extending this logic further, the use of absorbing materials to improve ion yield has led to the desorption of very large biological molecules. For example, Hillenkamp et al. have

desorbed intact protein molecule ions in excess of 100 ku using a nicotinic acid matrix [33], while our group has observed clusters in excess of 500 ku [34] using this technique.

The energetics of MALDI

In order to desorb an intact involatile and ther-mally labile molecule, it is necessary to introduce energy into the material in such a way as to prevent thermal decomposition. The laser-induced disrup-tion of the molecular lattice is a non-equilibrium process [35] in which molecules are transformed from their solid or liquid phase into a local envelope of ionised gas. In common with other desorption processes this leads to the formation of both ions and neutrals. This scenario should not be compromised by the addition of non-absorbing analyte materials provided that concen-trations are relatively low.

Typically, in a MALDI experiment, excitation will occur over a region of 104μιη2, with energy absorption taking a few nanoseconds. Several theo-ries have been developed to explain the desorption of such large, intact, species from the matrix. The thermal-spike model [35-37] proposes that the matrix molecules sublime from the substrate as a result of local heating. All values of the laser flu-ence would lead to the ejection of material, but a threshold occurs when a certain fluence is reached with a subsequent rapid rise in desorption yield. The ejection of intact molecules is attributed to poor vibrational coupling between the matrix and analyte, leading to a bottleneck in the energy trans-fer from the substrate to the internal modes of the analyte molecule.

The pressure pulse theory [38,39], originally developed to explain sputtering events induced by fast ions such as in PD, has also been applied to MALDI. It is proposed that a pressure gradient is created normal to the surface, and that, on exceeding a certain threshold, molecules are accel-erated into local vacuum. The desorption of large molecules results from a summing of the momen-tum transferred from collisions with fast moving

A.E. Giannakopulos et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 67-86 69

matrix molecules. However, the mechanism by which the pressure gradient is formed is not clearly understood, although coulombic repulsion [40], low-energy secondary electron excitation causing a soft expansion ("pop-corn") effect [41], subli-mation of matrix material [35,36] and repulsive decays [42] may contribute.

Verification of which, if any, of these mechan-isms is acceptable is limited by a lack of precise control over most of the experimental para-meters. For example, the momentum integration proposed by the pressure pulse model predicts different threshold behaviour for molecules of different masses. The number of required colli-sions with small molecules varies as a function of the molecular mass, f(M5^3) assuming a constant binding energy [41]. Experimentally no such depen-dence has been found for ion yield [39]. However, this does not necessarily contradict the model as the predicted yield is not only for ions (detectable by mass spectrometry) but also neutrals.

Likewise, the mechanism of ion production is not well understood. Direct photonisation may be ruled out since MALDI rarely produces many molecular ion radicals [43]; additionally it has been observed that low-energy photons such as 2.94 and 10.6/xm can cause phenomenologically similar desorption and ionisation processes. It is generally thought that ionisation is essentially chemical in nature, possibly a proton transfer or cationisation reaction. It has been demonstrated that the charging pattern depends critically on the matrix-analyte combination but not upon the number of acidic or basic groups a macromolecule may possess [44]. This suggests that a more com-plex interaction of analyte and matrix, rather than a simple acid-base chemistry, is responsible for ionisation. The efficiency of the process is also moot, since whilst, on the one hand, the singly protonated (or deprotonated) nature of the major-ity of ions produced indicates an inefficient charg-ing mechanism, on the other, observations of the virtual removal of matrix-ion signals at specific matrix-analyte ratios in certain matrices suggests a very efficient charging process [45,46].

The ionisation of small organic molecules, including matrix molecules may be explained by the formation of molecular ion radicals. These radicals may act as precursors to a series of chemi-cal reactions culminating in the production of stable molecule or fragment ions [47]. This has been used to explain the predominance of singly charged species in MALDI experiments. The enhanced acidity of matrix molecules upon electro-nic excitation may also play an important role [48]. Whatever the nature of the desorption and ionisa-tion process, the ion packets so generated will pos-ses a distribution of energies in addition to any broadening from subsequent fragmentation pro-cesses in the field-free region of the spectrometer.

Time-of-flight (TOF) mass spectrometry

The pulsed nature of MALDI makes it a natural ion source for TOF mass spectrometry [49,50]. In this type of instrument the mass (strictly speaking mass-to-charge ratio) analysis is carried out on the basis of direct measurement of the ion's time-of-flight through the instrument. The simplest type is the linear spectrometer in which all desorbed ions are accelerated over a short region by an electric field. The terminal velocity so acquired depends on the mass-to-charge ratio of the ion. Subsequent travel through a field-free region provides that the ions are separated in time on the basis of mass-to-charge ratio. For such a device

where / is the time-of-flight, m the mass, q the number of elementary charges, e the charge on an electron and V the accelerating potential. In general, this relationship is not used directly for calibration since V and L are rarely known sufficiently accu-rately. If however, the times of flight of two ions of well-defined mass-to-charge ratios are known then

im \J- = At + B (2) where A and B are calibration constants.


The major disadvantage of TOF spectrometry, however, is the severe degradation of mass reso-lution caused by experimental factors such as delayed ion formation, space charge effects and the energy spread of the generated ion packet.

In order to improve performance a number of approaches have been developed. The ion mirror, or reflectron, time-of-flight instrument [51] makes use of an electrostatic mirror which reflects incident ions through an angle of greater than 90°. Ions with the same mass-to-charge ratio, but possessing higher kinetic energy, penetrate further into the mirror with a subsequent increase in reflection time. The mirror can then be tuned to compensate for the initial energy spread. Instruments of this type have recorded resolutions of up to 10000 [52]. Other approaches, such as impulse field focussing [53,54] and post-source focussing [55] have also been successfully employed.

The energy distribution of the ion packet con-tains important clues to the mechanism of ion production inherent in the energy distribution of the ion packet. It is this potential information which has prompted the development of the energy resolving time-of-flight (ER-TOF) instru-ment. In this paper, we report on the design of an energy-resolved TOF instrument comprising a sim-ple linear TOF device coupled to the electric sector of a high-precision reverse-geometry magnetic-sector mass spectrometer.

Design of the source for ER-TOF experiments

Ion optical considerations

A primary consideration for the combination of a linear TOF and electric sector is the ion source optics. Most types of electric sectors require astig-matic focussing lenses, because a well-collimated beam is necessary in the vertical plane of the instru-ment {x-z direction), to ensure high transmission through the instrument by avoiding collisions with the walls of the spectrometer. In the horizontal plane (y-z direction), focussing is necessary for high transmission through the slits.

Fig. 1. Three dimensional trajectory diagram of the ion source, showing projections of the ion trajectory in the x-z plane (collimated in this plane) and in the y-z plane (focused in this plane).

The M ALDI source is shown in Fig. 1, simulated by a three-dimensional numerical integration pro-gram designed in our laboratory. The effect of astigmatic focussing is also shown. In order to prevent scattering, and have more control over the ion beam, grids were not used in the present design. Individual grid wires have a lensing eifect when a large potential difference is placed across them. The ion trajectories resulting from such len-sing effects are difficult to predict and sensitivity is lost. Figure 2 shows in two dimensions the trajec-tories of ions transmitted through such grids. The SIMION ion optical computer simulation program was used. A collimated ion beam with small kinetic energy starts from a potential of 6 kV and is accelerated towards the grids. Trajectories of ions passing close to the grids experience severe trajectory distortion due to the field.

The effect of a strong electric field above the target

Total ion kinetic energy spectra of samples, con-taining protein analyte and using 3-nitrobenzyl alcohol (NBA) liquid matrix, have shown that a strong electric field at the target results in large

A.E. Giannakopulos et al./Int, J. Mass Spectrom. Ion Processes 131 (1994) 67-86 71

6kV -3kV OV

Fig. 2. Ion optical computer simulation, using the SIMION program, of low-energy ions passing through a - 3 kV grid of 0.8 mm mesh size. A collimated beam of ions starting from a 6 kV potential surface is scattered after passing through the grids. Ions having trajectories

close to the grids are strongly affected by the field.

energy spreads for ions produced from MALDI. By placing a grid 2 mm above the target, with the same potential as the target, the energy spread may be reduced to some electronvolts, which would indicate that the energy spreads are predominantly energy deficits arising from the combination of high electrostatic fields and the spatial distribution of the molecules during ion formation. In Figs. 3 and 4, MALDI kinetic energy spectra, before and after positioning of the grid, are shown. The loss of temporal resolution arising from the length of time the ions spend in the field-free region before accel-eration is a major drawback for grid placement over the target. By reducing the grid-to-sample distance the temporal resolution is expected to improve, however the fringing field from the accel-

erating potential must not extend to the desorp-tion-ionisation region.

The electric sector as a part of an ER TO F mass spectrometer

The most commonly employed electrostatic field for kinetic energy measurements is the radial field between two coaxial cylindrical electrodes. First-order directional focussing for ions emerging from the entrance slit of the electric sector with small angular divergence is offered by such a field in the y - z plane and no focussing at all at the x — z plane. The distance between the two elec-trodes is usually small compared with the mean radius aç of the two electrodes. If an ion enters

5000 6000

KINETIC ENERGY/eV

7000

Fig. 3. MALDI total ion kinetic energy spectrum for NBA + /3-lactoglobulin. The accelerating voltage was 8064 V with the ions being produced within this field.


7800 8000

KINETIC ENERGY / eV

Fig. 4. M ALDI total ion kinetic energy spectrum for NBA + /3-lactoglobulin. The accelerating voltage was 8064 V. In this case a grid at target potential was placed 2 mm above the target. Ions were thus produced in a field free region (compare with Fig. 3).

normal to the electric field, it will describe a circular path through the field only if it has the correct energy to make the centripetal force balance the electrostatic force acting upon it. In order to under-stand and evaluate the parameters affecting the time-of-flight of ions travelling through the elec-tric sector, a numerical integration computer simu-lation program was developed.

The equations of motion of the ions in the cylin-drical polar coordinates (r, φ, z) can be written as

I2> m{r — τφ ) = eE,

mA(r20)=O

d r y

mz = 0

(3)

(4)

(5)

The potential and the field vector was calculated

analytically. The position and the velocity of the ion were calculated numerically, assuming that the forces acting on the ion remained constant for very small time intervals.

Outside the field boundaries the field vector and potential were assumed to be zero. Hence ions entering the field at any radius other than the zero potential surface would have their energy changed by an amount of energy equal to eV, where V is the potential of the field for the specific radius. Gaussian distributions were assumed for initial values for ion energy, position and angle. The study was based on a hypothetical 0.33 m radius electric sector with an electrode separation of 1.1 cm. The object slit and the collector slit on these simulations had a width of 0.165 mm. The object slit was positioned 0.1742 m from the electric


sector. The collector slit was positioned at a dis-tance x" after emerging from the field where the lateral displacement / ' of the beam was mini-mum. The position of the collector slit was calcu-lated analytically from

y = 4 δ + α^-,Ρθ) sinvê

V2l't

+ (px - <5)cos y/ϊφ,

«e(Pl - Ρθ) + x c cos V2<t>e

-λ/2(/9] - δ) sin Λ/20£ (6)

It is assumed that the ions start from a point À with coordinates x = /e' and y = aQp0, and enter the field at x = 0 and y = aQpu where ae is the mean circular radius of the electric sector [56]. The coordinates of a point at the exit of the electric sector are x'\ y" and δ = β + η/2 where β and 7 are the velocity and energy spread respectively, with β = (v - v0)/v0, 7 = (m - m0)/m0 and v0, m0 are

the velocity and the mass of those ions which, when entering the field with this velocity normal to the field boundary at r = ae, describe a circular path within this field.

Positions toward the outer electrode of the elec-tric sector are considered positive, whilst those toward the inner electrode, negative. The same notation is also used for the angle.

In order to investigate the effect of the energy range transmitted by the electric sector upon the TOF of the ions when travelling through the elec-tric sector, a number of trajectories of singly charged ions of different masses between 5000 and 5010 u with different kinetic energies were traced. The ions were started from the object slit of the electrostatic analyser and at three different positions along the >>-axis. One group of ions was started from the centre of the object slit, which coincides with the zero potential surface of the electric sector, and two other groups of ions from either sides of the slit. The ions entered the 81.5° electric sector at an angle normal to the field. The

(0

O)

• o I

Φ

E

43.69

43.68

43.67 H

43.66 H

43.65

43.64 H

43.63

D D D Û D D D O D

°°°ooooo(

++++ ■»■···++<

D D D D ° ° o D D D c

O0°o°ooooooc

h++++ + + +++ + +

x * * *xx Χ Χ * * Χ Χ Χ , < Α * * x * x x

"t"*"****^

o a

+ b

α c

d

• e

« f

* g

Δ h

8053 8058 8063

Energy /eV

Fig. 5. Times-of-flight from object slit to collector slit for ions with different energies and starting positions which enter normal to the field of the electric sector and are transmitted. A number of trajectories have been simulated for ions of masses between 5000 and 5010u. (a), O; (b) +; (c) a: Ions with mass 5010 u starting from the centre, outer and inner side of the slit respectively; (d), ·; (e), ♦ ; (f), x: Ions with mass 5008 u starting from the centre, outer and inner side of the slit respectively, (g), A; (h), Δ; (i), ■: Ions with mass 5000 ü starting

from the centre, outer and inner side of the slit respectively.


Initial angl· /degrt·

Fig. 6. Times-of-flight for ions (m/z 5000) of the same energy but different initial entry angles to the field of the electric sector. All ions start from the centre of the slit which coincides with the zero potential surface of the electric sector: compare with Fig. 5.

results for transmitted ions are presented in Fig. 5 with the times of flight being reported from the object slit to the collector slit which was positioned 0.1232 m from the electric sector. With the trans-

mitted range of energies the resolution, //2di, was found to be better than 2000. In practice it would be difficult to achieve such a collimated beam. The need to increase the signal by increasing the

angl· of dtfltctlon for electric sector /dtg.

Fig. 7. The effect of different deflection angles of electric sectors on the maximum detectable initial angular divergence of monoenergetic ions starting from the middle of the object slit. The total path length is 2.000 m and the collectors slit is adjusted every time to suit the

specific electric sector. The ions are detected by a 1 cm diameter detector.


44 0-

43.B-

43.6-^

43 4-1

K

\ 1 •nglt of arrival at collector silt /dag.

Fig. 8. The relationship between the time-of-flight and the angle of starting with a spread in initial

number of ions transmitted by the electric sector implied that some kind of focussing would have to be used in order to allow a larger number of ions to pass through the object slit. A comparison of the time spread, shown in Figs. 5 and 6, demonstrates how critically the initial angle of the ion upon entering the electric sector affects the total TOF of ions of the same energy and mass starting at the centre of the object slit. Ions following trajectories towards the inner electrode of the electric sector were first accelerated and then decelerated, while the ions of the same energy fol-lowing trajectories towards the outer electrode were first decelerated and then accelerated result-ing in different times of flight. In Figure 7 the maximum angle of entry into the electric sector is plotted against different deflection angles of electric sectors for monoenergetic ions starting from the middle of the object slit. The total flight path is 2 m and the position of the collectors slit is adjusted every time for the specific electric sector. The hypothetical detector has a diameter of 1 cm and the object slit is placed at 0.175 m from the entrance of the electric sector.

To see whether it was possible to improve mass-resolution by some post-sector treatment an ion packet was generated with distributions of initial

arrival at the collector slit for ions of the same mass (m/z 5000) angle, energy and position.

angle, energy and position. The standard deviation in angle was 0.3°, in energy 3eV, and in position, one third of the width of the object slit. This packet was then run through the simulated instrument subject to the constraint that the angular disper-sion dominated the possible temporal resolution. It was found that faster ions arrived at the collec-tor slit with positive angles and the slower with negative angles (Fig. 8). This indicated that the time spread at the collector slit arising from the initial angular distribution could be corrected, if the ions were forced to follow trajectories in a two electric-sector system. The ions which fol-lowed trajectories towards the outer electrode in the first electric sector would follow trajectories towards the inner side of the second electric sector and vice versa, compensating for the different times of flight. The proposed system consists of two iden-tical electric sectors in a C configuration, with a slit equidistant from the end of the first sector and the beginning of the second sector. A similar setup has previously been considered for energy compen-sation and beam steering [57,58].

Ions of different energies were transmitted through both electric sectors of the simulated instrument via the common collector/object slit. The resulting peak widths after the first and sec-


ond electric sectors for singly charged ions of mass 5000 u are shown in Fig. 9. Again a gaussian distri-bution of initial ion conditions was used. The ions were assumed to have an average energy of 8064 eV with a standard deviation of 3 eV, the mean angle being 0° with a standard deviation of 0.3° and the standard deviation in position being one third the width of the object slit.

Experimental

Instrumentation

A schematic diagram of the experimental appa-ratus is shown in Fig. 10. The TOF mass spectro-meter was located in the second field-free region of a large scale, reverse-geometry mass spectrometer and combined with the electric sector of this instru-ment to provide energy-measurement capabilities. The MALDI ion source was located at a distance of 2.4 m from the entrance of the 81.5° angle of deflection, 1 m radius electric sector. Ions formed in the ion source were focussed and steered by an astigmatic focussing double Einzel lens (Fig. 1)

through the object slit prior to an off-axis post-accelerating detector used for tuning purposes. The off-axis post-accelerating detector consisted of a beryllium-copper conversion dynode held at 25 kV and an electron multiplier. After a total flight of 4.6 m the ions were detected by an on-axis post-accelerating detector capable of offering post-accel-eration up to 100 kV. The above detector was coupled via a fibre-optic link to ion-counting electronics. The pressure in the ion source of the instrument was 1 x 10 ~6 mbar and in the rest of the instrument 1 x 10~7mbar. A quadrupled Nd-YAG (Spectron Laser Systems) laser operating at 266 nm, focussed by a 30 cm focal-length spectrasil lens, was used to irradiate the sample at a grazing angle. The electric sector was controlled by the DEC PDP 11 computer of the double focussing instrument [59]. Liquid NBA was used as the matrix in order to provide the necessary shot-to-shot reproducibility, long sample lifetime (some hundreds of shots) and fast and simple sample preparation. ER-TOF spectra for different bio-logical molecules were measured by stepping the electric sector through the required energy range

Initial angl· /dag.

Fig. 9. Correction for the time dispersion caused by the electric sector and arising from the different angles of entry for ions of the same mass, by employing an additional electric sector, subsequent to the first, in a symmetrical C configuration.

A.E. Giannakopulos et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 67-86 11

LENS

Uû. YAQ LASER

OFF-AXIS PAO

A OIIJECT SLIT SOURCE \

\

MIRROR

π •LU'I

7 ; 1111

2.4m

tOOkV rr ' S POST ACCELERATION DETECTOR

o Fig. 10. Schematic diagram of the ER-TOF instrument.

and recording TOF spectra at each point. Ion-counting electronics provided the required sensi-tivity, and a large number of laser shots, in excess of 500, were used in every step of the electric sector in order to provide spectra of the required statisti-cal quality. The TOF spectra were processed using a SUN 3/80 computer after acquisition by a EG&G ORTEC 7100 Multichannel Analyzer.

The source accelerating potential used in all experiments was 8064 V with the on-axis post-accelerating detector operating at — 30 kV. The off-axis post-accelerating detector was used for tuning the ion beam in order to achieve maximum transmission through the object slit and for check-ing the condition of the sample.

Preliminary experiments had shown that the presence of a high electric field above the sample surface caused severe energy deficits (Figs. 3 and 4). This problem was cured by placing a high trans-

mission grid, held at the sample plate potential, over the sample forming a 2 mm field-free region.

Sample preparation

Particular care was taken with the question of the homogeneity of the sample. Initially, 1 μΐ of 1 x 10"5M aqueous analyte solution would be placed on top of 1 μΐ of liquid NBA and dried under a stream of warm air. However, it appeared that there were strong variations in the concen-tration of the analyte-matrix mixture, because analyte peak intensities differed strongly from sam-ple to sample even though shot-to-shot reproduc-ibility was good. ER-TOF spectra, (Fig. 11(a)) obtained at 8064 eV recorded using this method of sample preparation, and a fixed laser pulse energy close to the threshold clearly indicated that there was a variation in the ion peak intensity


time-of-flight /jisec time-of-flight /jisec

500 500

time-of-flight /psec time-of-flight /psec

Fig. 11 (opposite and above), (a) ER-TOF spectra at 8064 eV of bovine insulin showing variation of the analyte peak intensity due to the sample preparation method (see text). Each of the spectra represents a different sample loading and is the average of 180 laser-shots on the particular sample, (b) Improved reproducibility of the ER-TOF spectra at 8064 eV of bovine insulin using superior sample preparation method (see text). Each of the spectra represents a different sample loading and is the average of 180 laser-shots on the

particular sample.

which we interpret as being a variation in the surface concentration of the sample.

For the experiments reported below, samples were prepared by mixing 0.5 ml of 1 x 10 ~5 M aqu-eous analyte solution with 0.5 ml of NBA, and dried using a strong stream of nitrogen gas. For each of the energy steps 1 μ\ of sample was used.

The reproducibility of the ER-TOF spectra was improved with the variation on the analyte peak intensity being much smaller (Fig. 11(b)).

Results and discussion

The ER-TOF experiments were carried out with


time-of-flight /psec time-of-flight /jisec

100-

75 H

so H

25 H

(b)

\*/»W/\i*J

—Γ" 100

WtfM^#jJU

— i — 200

" T — 300 400 500

time-of-flight ^ e c time-of-flight l\xs

Fig. 11 (continued).

three different molecular-mass molecules, bovine insulin (5736 u), /3-lactoglobulin (18 300 u) and bovine albumin (66 500 u). An accelerating potential of 8064 V was used throughout. In Fig. 12, a three-dimensional plot of ER-TOF spectra for NBA + ß-lactoglobulin is shown. On the x-axis the kinetic energy in electronvolts, on the j;-axis the TOF of the ions and on the z-axis the intensity of the ion peaks in terms of ion counts

are plotted. Slicing through various z-y planes, TOF information for ions of specific kinetic energy is provided while z-x planes provide kinetic energy information for specific times of flight. The intensity of the NBA peak is shown partly saturated on the plot to allow the obser-vation of the smaller peaks such as the dimer, tri-mer and the doubly charged /5-lactoglobulin molecule ions.


Figure 13 shows the Time-Selected Energy-Resolved (TSER) spectrum of the molecule ion of bovine insulin (a), together with those of dimer and trimer ions (b). In Fig. 14(a), the TSER spec-trum of the molecule ion region of /3-lactoglobulin

is shown. Doubly and triply charged ions, as well as the dimer of the /3-lactoglobulin molecule, were also observed (Fig. 14(b)). The TSER spectra of bovine albumin are shown in Fig. 15. In addition to the molecule ion, doubly, triply and pentaply

T l M E O F ^ G H T / m S

Fig. 12. Surface plot of ER-TOF spectra for NBA + /?-lactoglobulin. The accelerating potential was 8064 V.


(a)

Energy /*V

-O (410μ3)2Μ+

-· (510μβ)3Μ+

0 · 7950

(b)

Energy /eV

Fig. 13. (a) TSER spectrum of bovine insulin showing the molecule ion region, (b) TSER spectrum of bovine insulin showing dimer (a) and trimer ( ♦ ) peaks.


(a)

Energy /«V

(300μ9) Μ3+

(370μβ) Μ2+

(720μ3) 2M+

(b)

Energy /#V

Fig. 14. (a) TSER spectrum of/3-lactoglobulin showing the singly charged molecule ion. (b) TSER spectrum of /3-lactoglobulin showing the dimer, in addition to the doubly ( ♦ ) and triply (■) charged species.


charged ions (Fig. 15(b)) and the dimer (Fig. 15(c)) were observed.

It is apparent from these spectra that there is some additional broadening of the molecule ion in the TSER spectrum of bovine albumin (Fig. 15(a)) with respect to that of bovine insulin (Fig. 13(a)), which is interpreted in terms of fragmentation of the molecule ions. The alterna-tive explanation, that it is a mass dependent distri-bution in initial ion energies which is being observed, may be discounted, since the instrument is time-selecting the detected ions. In other words, the resultant energy spectrum is for ions of defined veJocity. The immediate (2 mm) region above the sample surface was maintained at zero field, in order to ensure that there was no deficiency in initial ion kinetic energy due to any spatial distri-bution of ion formation. When fragmentation of a molecule ion occurs in the field-free region of a TOF spectrometer, fragment ions with different kinetic energy but almost the same velocity are produced. The velocity of the ions defines the total TOF, and fragment ions with different masses and different energies arrive at the detector of a linear TOF mass spectrometer at almost the same time giving rise to time dispersion and reduc-ing the resolution of the spectrometer. On a TSER spectrum, the energy loss can be interpreted as mass change of the molecule ion.

The TSER results (Figs. 14 and 15), show an apparent broadening of the molecule ion peak. Moreover, the magnitude of this apparent broad-ening increases with the molecular ion mass. This may be understood as a mass dependent instability in the molecule ion that gives rise to increasing levels of fragmentation during its flight through the spectrometer, in other words, a mass-dependent metastability is observed. The molecule ions of ß-lactoglobulin and bovine albumin seem to have undergone extensive fragmentation. This effect is apparent even under threshold conditions in the MALDI process even though fragment ions produced during the desorption-ionisation pro-cess are not themselves observed. This is in agree-ment with Spengler et al. [60] who have reported

similar effects in a range of substances of lower mass (1100-13 000 ku), using a two-stage reflec-tron instrument.

The source of the decay is unclear. It may be a collisionless unimolecular dissociation on a time-scale comparable with the ions' times of flight. Alternatively, it may a collisional activation of the ion by residual gas in the instrument [61]. Theoretical evidence for the unimolecular dis-sociation of such large ions is divided [62] since traditional application of RRKM theories pre-dicts rates far smaller than those observed, supporting the collisional activation hypothesis. Schlag and Levine [63] have suggested, however, that observed rates may be rationalised, if vibra-tional excitation is less than one quantum per mode in which case vibrational relaxation will be much slower than is assumed by RRKM models. They propose a model in which excitation remains localised in large ions, leading to a major reduction in effective degrees of freedom and a proportion of the ions fragmenting more rapidly than pre-dicted by RRKM. It seems likely that at least three factors each contribute to some extent to the observed fragmentation: (a) high internal energy of protein ions formed by MALDI with the more massive ions possessing higher energies, (b) collisional activation of the massive ions by the background gases, (c) slow intramolecular energy redistribution following ion formation and/or collisional activation.

Conclusion

An energy resolved TOF instrument has been constructed using a linear TOF flight device situ-ated in the second field-free region of a reverse-geometry sector mass spectrometer, in tandem with the electric sector of the instrument. It has been shown that particular care must be taken in the lensing system used to combine the two devices in order to preserve time resolution over the total flight path. The protein molecule ions produced by MALDI exhibit a mass-dependent metastability, which may be due to collision effects.


Energy /*V

Energy /§V

Fig. 15 (opposite and above), (a) TSER spectrum of bovine albumin showing the molecular ion. (b) TSER spectrum of bovine albumin showing the doubly, triply ( ♦ ) and pentaply (a) charged molecule ions, (c) TSER spectrum of bovine albumin showing the dimer.

A.E. Giannakopulos et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 67-86 85

(c)

Energy /«V

Fig. 15 (continued).

References

1 M.S.B. Munson and F.H. Field, J. Am. Chem. Soc, 88 (1966)2621.

2 B. Munson, Anal. Chem. A, 49 (1977) 722. 3 M.G. Ingram and R.J. Gomer, J. Chem. Phys., 22 (1954)

1279. 4 P.L. Robert and H.R. Schultern, Anal. Chem., 61 (1989)

61. 5 R.G. Bray, R.M. Hochstrassee and J.E. Wessel, Chem.

Phys. Lett., 27 (1974) 167. 6 E.W. Schlag and H.J. Neusser, J. Am. Chem. Soc, 16

(1983) 16. 7 H.D. Beckey, Int. J. Mass Spectrom. Ion Phys., 2 (1969)

500. 8 L. Prokai (Ed.), Field Desorption Mass Spectrometry,

Marcel Dekker, New York, Vol.9 1990. 9 M.L. Vestal, Int. J. Mass Spectrom. Ion Phys., 46 (1983)

193. 10 K.L. Busch and R.G. Cooks, Science, 218 (1982) 247. 11 F. Lafortune, R. Beavis, X. Tang, G. Stanging and B.T.

Chait, Rapid Commun. Mass Spectrom., 1 (1987) 114. 12 M. Barber, R.S. Bordoli, G.J. Elliot, R.D. Sedgwick and

A.N. Tyler, J. Chem. Soc, Chem. Commun., 7 (1981) 325. 13 C.J. McNeal and R.D. Macfarlane, J. Am. Chem. Soc, 103

(1981) 1609. 14 M. Barber, R.S. Bordoli, G.J. Elliot, R.D. Sedgwick, A.N.

Tyler, B.N. Green, V.C. Parr and J.L. Gower, Biomed. Mass Spectrom., 9 (1982) 11.

15

16

17 18

19

20

21

22 23

24

25

26

27 28

29

30

M. Barber, R.S. Bordoli, G.J. Elliot, R.D. Sedgwick and A.N. Tyler, Anal. Chem. A, 54 (1982) 645. R.D. Macfarlane and D.F. Torgerson, Science, 191 (1976) 191. C.J. McNeal, Anal. Chem. A, 54 (1982) 43. M. Yamaskita and J.B. Fenn, J. Phys. Chem., 88 (1984) 4451. B. Fenn, M. Mann, C.K. Meng and S.F. Wong, Mass Spectrom. Rev., 9 (1990) 37. M. Karas and F. Hillenkamp, A. Benninghoven (Eds.), Ion Formation from Organic Solids, Springer Series in Chemi-cal Physics, Vol.25, Springer-Verlag, New York, 1983. M.A. Posthumus, P.G. Kistemaker, H.L.C. Meuzelaar and M.C. Ten Noever de Brauw, Anal. Chem., 50 (1978) 985. M. Karas and F. Hillenkamp, Anal. Chem., 60 (1988) 2299. K. Tanaka, H. Waki, Y. Ido, S. Akita, Y. Yoshida and T. Yoshida, Rapid Commun. Mass Spectrom., 2 (1988) 151. F.J. Vastola and A.J. Pirone, Adv. Mass Spectrom., 4 (1968) 107. F.J. Vastola, R.O. Mumma and A.J. Pirone, Org. Mass Spectrom., 3 (1970) 101. R.J. Conzzemius and J.M. Capellen, Int. J. Mass Spectrom. Ion Phys., 34 (1980) 197. R.J. Cotter, Anal. Chem. A, 56 (1984) 485. A. Vertes, R. Gijbels and R.D. Levine, Rapid Commun. Mass Spectrom., 4 (1990) 228. R.C. Beavis, J. Lindner, J. Grotemeyer and E.W. Schlag, Chem. Phys. Lett., 146 (1988) 310. P.K. Hohh, Appl. Phys. Lett., 29 (1976) 473.


31 F. Hillenkamp, M. Karas, D. Holtkamp and P. Klusener, Int. J. Mass Spectrom. Ion Phys., 69 (1986) 265.

32 M. Karas, D. Buchmann and F. Hillenkamp, Anal. Chem., 57(1985)2935.

33 M. Karas, D. Buchmann, U. Bahr and F. Hillenkamp, Int. J. Mass Spectrom. Ion Phys., 87 (1987) 53.

34 T.W.D. Chan, A.W. Colburn and P.J. Derrick, Org. Mass Spectrom., 27(1992)53.

35 A. Vertes, R. Gijbels and R.D. Levine, Rapid Commun. Mass Spectrom., 4 (1990) 228.

36 A. Vertes and R. Gijbels, Chem. Phys. Lett., 171 (1990) 284. 37 A. Vertes and R. Gijbels, Scanning Microscopy, 5 (1991)

317. 38 R.E. Johnson and B.U.R. Sundqvist, Rapid Commun.

Mass Spectrom., 5 (1991) 574. 39 B.U.R. Sundqvist, Anal. Chim. Acta, 247 (1991) 265. 40 P.K. Hohh, Appl. Phys. Lett., 29 (1976) 473. 41 P. William and B.U.R. Sundqvist, Phys. Rev. Lett., 58

(1987) 1031. 42 R.E. Johnson and B.U.R. Sundqvist, Int. J. Mass Spec-

trom. Ion Phys., 53 (1983) 337. 43 T.W.D. Chan, Thesis, University of Warwick, (1993). 44 R.C. Beavis and B.T. Chait, K.G. Standing and W. Ens

(Eds.), Methods and Mechanisms for Producing Ions from Large Molecules, NATO ASI, Series B: Physics, Vol.269, 1990.

45 T.W.D. Chan, A.W. Colburn and P.J. Derrick, Org. Mass Spectrom., 26 (1991) 342. T.W.D. Chan, A.W. Colburn and P.J. Derrick, Org. Mass Spectrom., 27 (1992) 188.

46 K. Strupat, M. Karas and F. Hillenkamp, Int. J. Mass Spectrom. Ion Processes, 111 (1991) 89.

47 H. Ehring, M. Karas and F. Hillenkamp, Org. Mass Spec-trom., 27 (1992) 472.

48 M.E. Gimon, L.M. Preston, T. Solouki, M.A. White and D.H. Russell, Org. Mass Spectrom., 27 (1992) 827.

49 W.E. Stephens, Bull. Am. Phys. Soc, 21 (1946) 22. 50 A.E. Cameron and D.F. Eggers, Jr., Rev. Sei. Instrum., 19

(1948) 605. 51 V.l. Karataev, B.A. Mamyrin and D.V. Schmikk, Sov.

Phys.-Tech. Phys., 16 (1972) 1177. 52 R. Grix, R. Kutscher, G. Li, U. Grüner and H. Wollnik,

Rapid Commun. Mass Spectrom., 2 (1988) 83. 53 N.L. Marable and G. Sanzone, Int. J. Mass Spectrom. Ion

Phys., 13 (1974) 185. 54 J.A. Browder, R.L. Miller, W.A. Thomas and G. Sanzone,

Int. J. Mass Spectrom. Ion Phys., 37 (1981) 99. 55 G.R. Kinsel and M.Y. Johnston, Int. J. Mass Spectrom.

Ion Processes, 91 (1989) 157. 56 C.A. McDowell, Mass Spectrometry, McGraw-Hill, New

York, 1963. 57 G.-H Oetjen and W.P. Poschenrieder, Int. J. Mass Spec-

trom. Ion Phys., 16 (1975) 353. 58 W.P. Poshenrieder, Int. J. Mass Spectrom. Ion Phys., 9

(1972) 357. 59 P.G. Cullis, Ph.D. Thesis, University of New South Wales,

Australia, 1987. 60 B. Spengler, D. Kirsch and R. Kaufmann, J. Phys. Chem.,

96 (1992) 9678. 61 B. Spengler, D. Kirsch and R. Kaufmann, Rapid Commun.

Mass Spectrom., 5 (1991) 198. 62 L.L. Griffin and D.J. McAdoo, J. Am. Soc. Mass Spec-

trom., 4 (1993) 11. 63 E.W. Schlag and R.D. Levine, Chem. Phys. Lett., 163

(1989) 523.


Laser ion sources for time-of-flight mass spectrometry

U. Boesl*, R. Weinkauf, C Weickhardt, E.W. Schlag Institut für Physikalische und Theoretische Chemie der Technischen Universität München Lichtenbergstrasse 4, 85747 Garching, Germany

(Received 15 June 1993; accepted 14 September 1993)

Abstract

Different selective and non-selective laser induced ion and anion sources are presented as well as their combination with time-of-flight mass spectrometers either of linear or reflectron type. Resonance and non-resonance enhanced laser ionization methods and their features of photofragmentation are discussed, in their combination with inlet systems for neutral gases. Laser induced VUV- and electron ionization and laser desorption-postionization are illustrated. Tech-niques and applications of multiple laser excitation in the ion source and in the space focus are discussed to achieve synchronicity of two experiments or secondary laser excitation. Examples for laser spectroscopic and mass spectrometric applications are given.

Key words: Linear time of flight; Reflectron; Resonance enhanced laser ionization; Laser desorption-ionization; Photo-fragmentation

1. Introduction

The combination of laser ionization with mass spectrometry results in many sophisticated appli-cations in mass spectrometry and laser spectro-scopy. Therefore, application of lasers in the ion source of a time of flight mass spectrometer introduces features of selectivity into mass spectro-metry, i.e. species-, isomer- or state-selective ion-ization and control of fragment intensities [1-3]. In addition, mass analysis gives a direct correlation between the laser spectrum and the associated mass, thus allowing access to the UV spectra of isotopic species as well as of single components in a mixture, as is usual in cluster spectroscopy.

The first experiments where laser excitation of molecules was performed within mass spectro-meters date back to the 1970s. In 1970, molecular


hydrogen was the first molecule to be ionized by laser and detected in a mass analyzer [4]. One year later, laser ionization and mass selective detection were successfully applied to molecular iodine, heavy water and carbon tetrachloride [5]. All these experiments utilized non-resonant ionization with fixed frequency lasers. Tunable lasers have been used since 1977 to measure UV/VIS spectra of mass selected diatomic molecules such as sodium, potassium [6], iodine [7] and lithium [8]. In 1978, the first laser spectra of mass selected polyatomic molecules by applying multiphoton ionization were obtained [3]. The first mass spectra of polyatomic molecular systems ionized by resonance-enhanced multiphoton ionization were published in 1977 [9] and 1978 [10].

Nowadays, laser mass spectrometers consist mostly of at least one pulsed tunable laser system (typically with a nanosecond pulse width and at UV wavelengths) and a linear or reflectron time-

SSZ)/0168-1176(93)03890-X

88 U. Boesl et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

of-flight mass spectrometer. This setup has the special feature of recording a complete mass spec-trum with each laser shot. In particular, tunable pulsed lasers combine well with reflectron time-of-flight analyzers. Reflectron time-of-flight instruments [11], represent a major breakthrough for time-of-flight mass spectrometry, combining advantages such as high recording speed and high ion transmission with high mass resolution.

In this article we describe a wide range of laser ion sources and their application and coupling to linear and reflectron time-of-flight instruments. Aspects of linear and reflectron time-of-flight instruments are described in section 2. The design of neutral sample sources plays an important role in the resolution and transmission of time-of-flight instruments as well as for resonance enhanced multiphoton ionization or other laser-based ion-ization techniques (section 3). The laser-based ion-ization sources which are compatible with time-of-flight mass spectrometers include non-resonant multiphoton ionization, one-photon VUV ion-ization, laser desorption of ions, laser-induced electron ionization and laser-induced anion formation (section 4).

Multiple laser excitations for each experimental cycle present further fascinating possibilities for which laser time-of-flight mass spectrometry is predestined (section 5). Two laser excitations of one molecular sample allow for excitation of neutrals by primary laser and ionization by a secondary laser, or ionization of neutrals by the primary laser and excitation or fragmentation of the ions by a secondary laser pulse. Multiple excitations are also useful for comparison or cali-bration; either one laser pulse may excite or ionize two different molecular species or two different laser pulses (with different wavelengths or differ-ent pulse widths e.g. nanosecond and femtosecond laser pulse) may ionize the same molecular species, thus excluding errors due to neutral density fluc-tuations (particularly necessary for large laser desorbed molecules). In the latter cases, over-lapping mass spectra are produced allowing reli-able information because they occur under the

same experimental conditions (e.g. the same neutral gas pulse). Ten-kilohertz lasers, recently made available, allow fast sampling of mass spec-tra and thus new access to time-dependent neutral density fluctuations as in, for example, supersonic beam pulses.

Secondary laser excitation may be performed with or without a spatial shift and temporal delay from the primary laser focus and laser pulse. The relative positions of two laser foci may be stable within a tenth of a millimeter even for absolute spatial distances of 100cm and within Ins even for absolute spatial distances of 100 cm and within 1 ns even for absolute temporal delay of some 10-100/is. Thus, even a mass selective secondary exci-tation due to different flight times of ions with different masses is possible. This is the key aspect of one of the most important developments in conventional mass spectrometry: tandem mass spectrometry. Time-of-flight tandem mass spectro-meters (section 6), introducing a new technique into this field, have achieved an important role for laboratory experiments and may achieve such a role as commercial instruments in the near future. This technique includes laser induced metastable ion decay as well as mass selective secondary laser dissociation [12-19].

2. Time-of-flight instruments

The mass separation of time-of-flight analyzers results from the dependence of the flight time on the masses of ions starting at the same time and same point within an electrostatic field. In the most simple time-of-flight spectrometers (two elec-trode ion source, linear drift region) this ion flight time is tA + /D with the time for acceleration iA and the drift time in the field free region iD:

A EA A V U

_ JMmp _ [M ' D - X D V ^ Î / - X D C V T 7

U. Boesl et aL/Int. J. Mass Spectrom, Ion Processes 131 (1994) 87-124 89

where EA is the electric field of the ion source, M is the ion mass in atomic mass units, mp is the proton mass, q is the elementary charge and qU is the final ion kinetic energy. The constant C = (mp/2q)1^2 = 0.71986 /isV1/2 cm- 1) for singly-charged ions. Several factors cause uncertainty in the flight time resulting in a finite mass resolution. Counter measures presented in this section are the use of energy compensation in the space focus of pulsed ion sources or of ion reflecting fields. These methods also allow for tandem mass spectrometric arrangements (see section 6).

2.1. Linear time-of-flight instruments: the space focus

Linear time-of-flight instruments are very simple to operate, can be very small in size and are easy to construct. Usually, the conventional linear time-of-flight instruments suffer from poor resolution, but detailed calculations show that geometrical arrangements and voltages can be chosen to achieve high mass resolution. The major reasons for a broadening of flight time profiles which result in the limited mass resolution of linear time-of-flight mass analyzers are:

(i) the finite time of ion formation or extraction; (ii) the finite volume of ion formation; (iii) the initial velocity of the ions due to the

initial velocity of the neutral molecules or the kinetic energy release of fragment ions formed during ionization.

Limitations due to electronics, data conversion and ion detector response time and space charge effects are neglected for the moment. The first three effects can be reduced to pure initial tem-poral or initial spatial distributions. Reason (i) is due to the finite laser pulse length or the finite rise time of the extraction pulse and therefore a pure temporal effect.

Reason (ii) is due to the finite ionization region, usually the laser focus size, and therefore a pure spatial distribution effect. The initial spatial distri-bution of the ions in the ion source results in dif-

ferent potential energies which (after acceleration) are converted into different kinetic energies causing different flight times. However, the different flight path lengths due to this spatial distribution also cause different flight times. These two flight-time effects are correlated and can compensate for each other in good approximation at a certain point (see "space focus" below) in the field-free drift region. At this point, the remaining effect of the initial spatial distribution is in first order a pure kinetic energy distribution. Ions of the same mass, but formed at different positions, pass this point at the same time but with different kinetic energies.

Reason (iii) ( an initial velocity distribution) for flight time broadening can be separated into a pseudo temporal and a pseudo spatial effect. The former is often called the "turn around time effect". As initial velocities mostly exhibit a distribution, "turn around times" also show a distribution caus-ing a broadening of time-of-flight profiles. While spatial effects can be corrected for by means of energy compensation (space focus, reflectron), tem-poral effects need other means for reduction (shorter ion formation time, high extraction fields against turn-around time effects, HV-pulsed extraction fields, post-ion-source pulsing). Because the spatial effect in linear time-of-flight mass spectrometers with a perpendicular super-sonic beam inlet dominates, it will be discussed more thoroughly below.

The pseudo spatial distribution of ion formation results in a spread of ion kinetic energies in the field-free region. This energy spread is the primary limiting factor of the mass resolution in simple linear time-of-flight instruments, but can be com-pensated for by flight-time correcting means. There is a point in the field-free drift region that intrinsi-cally fulfills this condition of flight-time correction. At this point, the flight-time decrease due to the larger velocities of ions formed at higher potential energies (AU = Ax Eq, where E is the electric field within the ion source) is compensated for by the increase due to the longer acceleration length (Λ: + Ax). ΔΛ: includes the real initial spatial distri-bution due to the spatially-distributed ion


formation as well as the pseudo initial spatial dis-tribution due to initial velocity distribution [20]. Usually, such a flight-time correction is only of first order, as seen from a derivation of the flight time in terms of the energy spread AU: at a certain point in the field-free drift region the linear term in AU will vanish. This point is called the first-order space focus (usually only the space focus).

The correction due to initial space distribution is illustrated in Fig. 1(a). At time t0 - 0 ions (same mass) are formed at different positions and there-fore at different potentials. After a certain time all ions of the same mass pass the space focus simul-taneously and the flight time is corrected (in first order). With an ion detector positioned at the space focus, one can observe a compensated mass spec-trum. However, for a simple two-electrode ion source (as in Fig. 1(a)), the distance A : A + X S F

with xSF = 2xA is much too short for a good

a)

b)

c)

mass selection by flight times. For a two-stage ion source (three electrodes, two electric fields) the position of the space focus can be shifted by vary-ing the electric fields (Fig. 1 (b)), facilitating much longer ion flight times. Besides time lag focusing, this was one of the first methods of enhancing the mass resolution of linear time-of-flight spectro-meters [21].

Recently, we demonstrated that a second-order space focus is also achievable [17,22]. For a well-suited geometry of a two-stage ion source, a set of potentials exists such that not only the linear but also the quadratic term in AU of the flight-time development disappears. The result is a second-order space focus with improved time com-pression of ion clouds as illustrated in Fig. 1 (c). In contrast to the first-order space focus in Fig. 1 (b), the position of the second-order space focus is now fixed; its value is determined by a relation between xAU xA2 and XSF [17,22]:

*A1 — xSF - 2xA2

2(*SF + XM) *SF

\χ$ψ — 2x A2

3x +*A2

SF

The relation between the final ion kinetic energy U and the potential UA2 at the second electrode is

r2(^SF +*A2) UA2 = U- 3x SF

Fig. 1. First-order space focus of a single field (a) and a double field (b) ion source, as well as a second-order space focus (c).

The total ion flight time is tx = tAX + tA2 + tSF

(Withi/A1 = i/-C/A2)

tAX — 2xA\C

VA2

M *SF — *SFCA/ —

By choosing an adequate geometry (xA1, xA2) a

relatively long flight path, XSF t 0 the second-order space focus can be chosen, thus achieving good mass resolution. With a total flight path length *AI + ·*Α2 + *SF of 13 cm, a mass resolution R5()o/a

U. Boesl et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124 91

of 720 (FWHM) for mass 78, and of 840 for mass 106 has been achieved in our group ( 50% = ΜΑίΑΜ=ι/Αί¥ΨΗΜ). Despite this very encourag-ing result, time-of-flight compensation through the space focus of an ion source contains many problems. To obtain a long flight path xS¥ to the space focus, the electric field in the first stage of the ion source has to be very weak, which induces a large spread in flight times due to the "turn around time" effect and space charge effects.

A much better solution is to use the second-order space focus of an ion source as a pseudo ion origin with minimized spatial distribution. In first- (or second-) order, the ions start here with a distri-bution of only kinetic energy. Flight-time broad-ening in the drift region behind the space focus due to the pure kinetic energy effect can now be corrected by second-order flight-time compen-sation means, such as an ion reflector of a refec-tron mass spectrometer. Long flight paths and therefore good mass resolution can now be realized independent of the real ion source. One is now free to optimize the real ion source for mini-mum flight-time broadening (whatever the reasons are) without the need of a long flight path from ion formation to the space focus. In the following section, the energy compensation of reflectron mass analyzers is explained.

2.2. The reflectron time-of-flight mass spectrometer

An important invention in the field of time-of-flight spectrometry is the reflectron time-of-flight mass spectrometer [11] mentioned above. It com-pensates time-of-flight effects due to ion kinetic energy distribution in the space focus. Its charac-teristic feature is an ion mirror consisting of two fields (three electrodes) with a decelerating/acceler-ating and a reflecting field (see Fig. 2). In some cases, also one-field reflectrons are used. For understanding the mode of operation of reflec-trons, let us take the space focus of an ion source as a starting point for ions of one distinct mass, with no spatial or temporal distribution, but with different kinetic energies. After the drift length xD1,

the ions with higher kinetic energy first enter the ion mirror, followed by those with lower kinetic energy. The former penetrate deeper into the reflect-ing field than the latter, resulting in a longer resi-dence time within the reflector for the higher energy than for the lower energy ions. By choosing the appropriate potentials and geometry (xD1,xD2,xs^R5 Us, ^REF)> the shorter flight time of high energy ions in the field-free drift region is compensated by their longer residence time within the reflector according to the following explanation.

Let us assume a symmetric arrangement of the drift regions for the moment (i.e. xD1 = xD2) [23] and take the point in time when ions of one distinct mass are stopped within the ion reflector before being reflected. Starting at this time, the reflector acts as a large ion source. This ions source can be constructed so that (i) it has a second-order space focus and (ii) this space focus lies exactly at the surface of the ion detector. For reasons of sym-metry (xD1 = χΌ2) the pseudo-ion source (space focus of the actual ion source) also has to overlap with the second-order space focus of the reflector, and all ions of the same mass (but different kinetic energy) will be stopped and reflected at the same time within the ion mirror. Due to the slight tilt angle of the reflector axis with respect to the ion trajectories, the flight paths of forward and back-ward ion beam can be separated in space. All ions of the same mass, passing the pseudo ion source at the same time (nevertheless with different kinetic energies) will arrive at the detector simultaneously.

Of course, using a reflectron, only the flight-time spread due to different kinetic energies and not due to pure temporal distributions can be corrected. That means that the time interval during which the ions of one mass pass the pseudo ion source (space focus of the actual ion source) is not affected by reflecting the ions and represents the width of ion peaks in the time-of-flight spectrum (if neglecting contributions of the ion reflector to the peak width). In other words, the ion reflector images the flight-time distribution at the space focus of the actual ion source onto the surface of


+ HV

molecular beam

detector laser focus

/space focus

ion beam

l.slow ions M+ U,on = U- AU 2. fast ions M* U,on = U*AU

Ξ, Uc U Ui

KD2

LU*

-*· — L

Ref

Γ ^ ^ -J )1'">2

X D = XD1 + XD2

Xc h-

rxR -1 ÎR.H

accelerat ion region

f i e l d - f r e e drift region ion mirror

Fig. 2. Scheme of reflectron mass spectrometer. The geometric dimensions and electric potentials used in the formula are indicated.

the ion detector. Now, for long drift lengths χΌ

long flight times without changing flight-time broadening and therefore high mass resolutions can be achieved with the extraction field of the ion source being unchanged. The limiting factors for the quality of this imaging and therefore for the achievable mass resolution will be considered below.

The time interval during which the ions of one mass pass the pseudo ion source is due (i) to not-fully-corrected flight times in the space focus (higher-order terms of the flight time t\\ and (ii) to the ion formation time and (iii) to the time spread within the ion source due to initial velocity distributions called turn around time. These con-tributions have already been discussed above. Usually, turn around times are the largest con-tributions; they can be reduced by high extraction fields EM. A non-optimal energy compensation in the ion source space focus (due to large AU = EAXAx) is negligible because the flight times t\ to this space focus (and thus uncorrected terms At\) is much smaller than the flight times t2

in the rest of the reflectron mass spectrometer (and thus uncorrected terms Δί2).

In conclusion, the salient point is that mini-mization of the turn around times can be achieved by the ion source while kinetic energy compen-sation is now predominantly managed by the ion reflector. In summary, the final ion peak width results from the turn around time, higher-order energy compensation terms and intrinsic time spreads (e.g. ion formation time, detector characteristics. For several other minor effects which limit the mass resolution see Ref. 20).

3. Sources of neutrals for time-of-flight mass spectrometry

The properties of the source supplying the neutrals for ionization considerably affect the transmission and resolution of time-of-flight mass spectrometers. There are simple inlet systems, such as effusive nozzles, which deliver high sample con-centration and are easy to handle, but are not well adapted to the pulsed ionization and detection scheme. Supersonic beams are an interesting alter-native. Here, the sample has to be mixed with a carrier gas (1:100 with usually Ar); their advantage is the possibility of cooling the translational


and internal degrees of freedom of the sample molecules. Thus, special features of narrow band-width laser excitation, e.g. isotope selective or state selective ionization, can be utilized for mass spectrometry.

3.1. Effusive molecular beams

A simple, but effective inlet system of neutrals for time-of-flight mass spectrometers is an effusive molecular beam emitted from a thin canula positioned between the electrodes of the ion source (see Fig. 3(a)). In spite of the continuous inlet, and thus the bad duty cycle (laser repetition rate typically 20 Hz), there is some compensation in sensitivity due to the pure sample inlet and a very small distance of the laser focus to the end of the canula. There is additionally the advantage of easy heating to several hundred kelvin, which may, for example, avoid physisorption effects in the nozzle.

Typical dimensions of the canula are the length of several centimeters, an outer diameter of some 0.3 mm and an inner diameter of 0.1 mm [3,24]. The small width as well as the appropriate voltage applied to this canula guarantee a minimal distor-tion of the electric fields in the ion source. Residual field distortions may be corrected for by an einzel lens refocusing the ions and a reflectron time-of-flight analyzer, which compensates for time-of-flight errors due to different energies. The end of the canula may be positioned as near as 1 mm to the ion optical axis; this results in a high neutral density (up to 1014 cm"3) at the point of ionization and an optimal overlap of molecular beam and laser beam while the total gas flow and thus the strain on the vacuum system is relatively small.

There are molecular flow conditions within the last few millimeters of the canula which cause a collimation of the molecular beam. This reduces the velocity distributions perpendicular to the molecular beam axis and thus collinear with the ion trajectories as well as with the laser beam. This reduces turn around time effects due to initial velocity distribution and, thus causes better mass

a)

Fig. 3. Different kinds of neutral gas inlet systems and ion sources: (a) effusive beam from a thin canula; (b) supersonic molecular beam oriented vertically and (c) in-axis with respect to the ion optical axis; (d) combination of laser desorption and supersonic in-axis molecular beam. Lasers are used for resonance enhanced multiphoton ionization (a)-(d), laser desorption of neutrals (d), electron emission for cationization (e) and electron emission for electron attachment (f ).

resolution; in addition, the doppler width becomes narrower, enhancing laser spectroscopic selectivity. For many analytical purposes, i.e. for gas mixtures, but also experiments on laser photochemistry or spectroscopy, this represents a simple, inexpensive but effective gas inlet system. Of course, the above mentioned features for resolution of mass spectro-metry and laser spectroscopy are significant, but not dramatic. Much better effects, particularly concerning preparation of the molecules for laser

94 U. Boesl et al.I Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

spectroscopy, can be achieved by skimmed super-sonic molecular beams.

3.2. Supersonic molecular beams for time-of-flight mass analyzers

Supersonic molecular beams consist of a mixture of the sample and rare gas (typically 1:100) which is expanded through a small orifice or nozzle into the vacuum chamber. Supersonic molecular beams [25,26] have several features which make them an ideal neutral source for optical spectroscopy and time-of-flight mass spectrometry. They:

(i) can be pulsed; (ii) cause efficient cooling of translational and

internal degrees of freedom; (iii) can very efficiently produce molecular or

atomic clusters; (iv) are well collimated, thus delivering a high

gas density.

Supersonic beams can be run in a pulsed mode by using magnetic or piezo electric valves; thus, they are optimal sources for pulsed ionization and (pulsed) time-of-flight mass analysis. Even at high repetition rates (100 Hz) and, despite the high partial density in each pulse, the pulsed mode reduces the gas load of the vacuum system dramatically by two to three orders of magnitude (valve opening time > 100/xs) in spite of high particle density in each pulse. Nevertheless, due to the high gas backing pressure of up to 10 bar and the dimensions of the nozzle (e.g. length 1 mm, diameter 0.2 mm), a second vacuum system is needed to keep the mean pressure at 10~4mbar, as well as a skimmer to generate a collimated mol-ecular beam and to reduce the pressure in the ion source to 10~6mbar. The overall gain of sample density by the pulsed mode of the nozzle is mainly reduced (a) due to the small ratio of sample to carrier gas and (b) due to large distances between the nozzle and ion source (typically > 5 cm) and the \/r2 dependence of the sample density in front of the nozzle.

The main advantage of supersonic molecular

beams is the extensive cooling of internal and exter-nal degrees of freedom. The cooling of especially rotational (down to 4K) and vibrational motions (down to 100 K) simplifies molecular spectra dramatically by reducing the number of bands as well as the spectral width of their rotational band contours. This enhances or even enables either the assignment of spectra, the selective excitation of single states or the selective ionization of single molecular species. The excitation of single mol-ecular states is important for ion kinetic and spectroscopic experiments as well as for excitation of single species in a mixture for molecular analy-sis. In addition, the ionization yield is enhanced. Due to the concentration of all the molecules to very few molecular ground states, a large number of molecules are able to absorb at one particular wavelength, thus increasing the ion yield.

Due to the high density in the expansion and the efficient cooling, large molecular and atomic clusters can be formed very efficiently. Cluster research working on the missing link between single molecules and bulk material is highly corre-lated with the development of supersonic beam expansions [27].

A further advantage is the good collimation of the supersonic beam. In addition to limiting doppler widths in laser spectra this allows a certain geometrical distance between pulsed nozzle and point of ionization (e.g. the laser focus) leaving space to apply additional experimental techniques to the molecular beam such as combinations with laser desorption, anion production or depletion of a single molecular species by laser excitation.

There are two ways of arranging molecular beams: vertical and collinear to the ion optical axis (see Figs 3(b) and 3(c)). The advantage of a vertical arrangement is a very small velocity com-ponent in the direction of ion trajectories. In an ideal supersonic beam, however, all molecules have the same in line velocity independent of their mass. This results in a mass dependent energy of kinetic motion perpendicular to the ion extraction axis and thus in an increasing deflection of ions out of the usual ion path with increasing

U. Boesl et aljlnt. J. Mass Spectrom. Ion Processes 131 (1994) 87-124 95

mass. This can be partly compensated for by a complex ion optic [28] but causes severe problems for very large masses. There is no such problem for an in line supersonic beam (Fig. 3(c)). But now the initial kinetic energy of the neutrals (in line with ion extraction) varies by the mass (e.g. a neutral mass of 1000 Da seeded in an argon beam may have kinetic energies as large as 6eV). As a con-sequence, large turn around times (actually time shifts due to the unidirectional initial velocity with narrow distribution) effect the mass spectrum and prevent simple time-of-flight mass calibration.

33. Laser desorption of neutrals

Large, involatile and often fragile molecules are very difficult to introduce in a mass spectrometer due to their low vapor pressure. Nevertheless, large molecules of biological importance are of particular interest. Many vaporization techniques for ions have been developed [29], but modern desorption techniques for large neutral molecules and ions (e.g. SIMS [30], FAB [31], 252Cf-plasma [32], matrix-assisted laser desorption-ionization [33] and neutral laser desorption and multiphoton ionization [34]) are of particular interest. All of them are pulsed, compatible with other pulsed techniques and therefore led to a renaissance of time-of-flight mass spectrometry.

While for most of these techniques the ionized molecules are detected, the desorption of neutral molecules is a prerequisite for the combination with resonant laser ionization and time-of-flight analysis. One of the most successful techniques of desorbing large neutral molecules with minimum fragmentation is laser desorption in combination with secondary laser ionization [34]. For second-ary laser ionization of neutral laser-desorbed molecules, kinetic energy distribution due to the spatial spread of ion formation is a major problem for mass resolution. Here, the application of reflectrons allows high resolution time-of-flight mass spectrometry. In particular the combination with a supersonic molecular beam [34-38] is very favorable. The latter serves to cool internal and

translational molecular motions and to transport the molecules into the ion source. Low vibrational excitation in the molecular ground state (and in consequence in the neutral excited and ionic ground state) may even help to reduce the degree of fragmentation after laser desorption in the neutral molecule or after laser ionization in the ion.

An apparatus has been constructed in our group involving laser desorption, a supersonic molecular beam, secondary laser ionization and a reflectron mass analyzer [34-36]. Thus, the advantages of resonant laser ionization (e.g. soft ionization, controlled fragmentation, spectroscopic selection against background) and reflectron mass spectro-metry (high mass resolution, high transmission, experimental flexibility) are available for the mass spectrometry of large, involatile and fragile molecules.

In Fig. 3(d) a scheme of the combination of laser desorption, supersonic beam and ion source is shown. The desorbing laser is focused onto the surface, either of a pure sample or of a mixture of matrix (glucose, polyethylene) and sample. The desorbed molecules enter the supersonic beam of Ar carrier gas downstream from the nozzle. The high energy particles leave the supersonic beam (first energy selection) while desorbed molecules with lower kinetic energy suffer enough collisions for a cooling of translational and internal mole-cular motions. Together with the supersonic beam these large molecules arrive at the ion source where they are ionized by a laser. The so-formed ions are extracted into the field-free drift region and analyzed by the reflectron. Due to the pick up of the large molecules by the rare gas beam about 2 mm downstream of the nozzle, the cooling efficiency is much worse than for volatile mol-ecules. Nothing is known so far about cooling processes in such pick-up sources or the cooling efficiencies in very large molecules. Additionally, the pick-up source can be regarded as a collisional experiment, which may result in mass-dependent deflection into the ion source similar to exper-iments used for neutral cluster separation [39].

In Fig. 4 as an example of laser desorption-laser


post-ionization, mass spectra of a decapeptide with mass 1295 are presented [36]. The upper spectrum has been obtained under soft ionization conditions. It illustrates the clean neutral source of laser desorption combined with the supersonic beam, as well as the high achievable mass resolution. Only the molecular ion appears in the mass spectrum; fragmentation as well as ionization of any kind of chemical background is suppressed. This is a joint effect of our neutral source and the reso-nance enhanced ionization. The inset displays the molecular ion and its natural 13C-isotopomers on

an enlarged mass scale; the mass resolution Ä50% = M AtAM=l/AtFWUM here is nearly 6000 (6 ns laser pulse length). Most of the techniques discussed in the other sections and in Ref. 20 may also be applicable to these large molecules.

4. Ion sources

The application of lasers for ionization in a mass spectrometer allows a lot of different excitation, ionization and dissociation schemes. In addition to being an efficient ion source, the laser ionization

Angiotensin I

Asp - Arg - Val - Tyr - Ile - His - Pro - Phe - His - Leu

SOFT IONIZATION

Ί Γ 1 Γ 1 Γ

1295

y 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 er er Z) o

HARD IONIZATION

I 244

371 272 414

343 i ,

506

534 784

647

619 756

881 927

1000

853

1028

1165

1137

1295

i 1 i i i i i i i i i i i r 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400

MASS —

Fig. 4. Laser desorption-laser postionization mass spectrum of angiotensin. Soft ionization (top) is obtained at low, additional fragmentation (bottom) at higher laser intensities. The inset displays the natural 13C-isotopomers of the molecular ion representing a

mass resolution of Λ50% = 6000.


can be used as preselection technique for mass spectrometry. A variety of filter characteristics in the ionization process can be created by multi-photon ionization (band pass), one photon ioniz-ation (low ionization energy pass) or laser induced photoattachment (only molecules with electron affinity are detected). The photodissociation per-formed with high laser intensities of the ionizing laser or with a second, usually time-delayed, laser delivers characteristic fragments of the molecule which are fingerprints of the molecule and helpful for structural or sequential analysis.

4.1. Resonance enhanced multiphoton ionization (REMPI) and dissociation

With the high intensities available now with pulsed lasers, efficient resonant or non-resonant multiphoton absorption is possible and may be used for the ionization or dissociation of mol-ecules. Two of the most common resonance enhanced multiphoton absorption schemes for molecular ionization are illustrated in Fig. 5(a).

Resonance enhanced multiphoton ionization is intrinsically connected with molecular spectro-scopy [40]. The resonance enhancement in the first excitation step (one photon or multiphoton absorption) can be monitored by the second ion-ization step. The ion signal recorded as a function of the first excitation wavelength delivers the multiphoton ionization spectrum. This multi-photon ionization spectroscopy triggered the intro-duction of resonance enhanced ionization laser mass spectrometry. In fact, the combination of laser excitation and mass analyzers was first realized as a technique of mass selective laser spectroscopy [6-10]. The possibility of involving the spectroscopy of the neutral parent molecule for selective ionization is one of the major advan-tages of multiphoton absorption for mass analysis.

A second important advantage is the possibility of varying the photon intensity of the laser over a wide range. Consequently, during the laser pulse a variety of absorption processes may be induced in the molecular ion or its ion fragments. The special

features of multiphoton ionization will now be discussed in detail.

4.1.1. Species selective ionization. The spectro-scopy of each neutral sample molecule is like a fingerprint with characteristic absorptions. Well-selected wavelengths thus introduce species-selec-tive ionization into mass analysis, thus sometimes substituting for double mass spectrometry exper-iments. Selective ionization of a molecule out of a mixture of species can be performed by tuning the wavelength of the ionizing laser onto a resonance of this molecule. The resonant enhancement then increases the signal of the molecule by several orders of magnitude. This allows, in combination with mass selection, the detection of small mol-ecular traces in a mixture. However, by setting the mass selector to the mass of a specific mol-ecule, mass selective laser spectroscopy from a mixture of molecules with different masses is possible [14-42]. Even isomer selective detection or spectroscopy of a mixture of isomers is possible [43,44], when secondary dissociation is utilized. Selective excitation or spectroscopic character-ization of molecular isotopomers can be achieved for cold, small or medium-sized molecules. How-ever, large and warm molecules exhibit a very con-gested absorption spectrum. In this case, isotope selective excitation is not possible and often also not wanted, because on unchanged natural isotope distribution may support the identification of large molecular species.

4.1.2. Soft ionization. The involvement of a res-onant intermediate state considerably enhances the ionization probability. This means that ionization proceeds efficiently at a low laser intensity; thus, further photon absorption in the molecular ion and subsequent ion fragmentation can be avoided. In summary, an efficient "soft" ion-ization without fragmentation is possible. This is particularly interesting for large fragile molecules or for the analysis of mixtures of species [36,42]. Of course, a necessary condition for "soft" laser ionization is that the excess energy of the ionizing

98

'////*////, M +P- '////&////M*+P-

U. Boesl et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

'////χ///ΛΜ\*- '///////////////y//,

M

'////l////?~^-

fî

'////#//// e

1

'////'

Ulli

///'S

y/y// Mf _

M*

i l lL N _

M

^ Ά ^ .

1

F*

F

b) d)

Fig. 5. Different intramolecular processes during laser excitation, (a) Different schemes of multiphoton ionization such as (1 + 1)-, (2 + l)-(MPI) with one intermediate state, 2-MPI without intermediate state, and prevented (1 + 1)-MPI due to fast intramolecular energy relaxation; (b), (c) ladder switching after ionization of a (b) small and (c) a large parent molecule (the latter with many parallel

decay channels); (d) ladder switching before ionization of the parent ion (only fragment ions appear).

photons above the ionization threshold does not exceed the lowest dissociation threshold of the

ion.

4.13. Variable fragmentation. By increasing the ionization laser intensity, photon absorption within the molecular ions can be induced. If the lowest ion dissociation threshold is exceeded, frag-

ment ions are produced. For small excess energies above these thresholds ( a typical situation in multi-photon ionization due to the commonly-applied laser wavelengths) this fragmentation is meta-stable on a microsecond time scale. Therefore, high yields of metastable ions can be obtained by multiphoton ionization-fragmentation.

A further increase of the laser intensity (laser


intensities smaller than some 10 GW cm- 2 of nano-second laser pulses are only considered here) causes absorption of additional photons within the mol-ecular ion and at a certain excess energy the dis-sociation rate may become considerably faster than the photon absorption rate. Any further absorption can then only take place in the frag-ment ion which again may lead to fragmentation.

At still higher laser intensities, this process will continue to smaller and smaller fragments. Due to the dissociation, the absorption is switched from the molecular ion to fragment ions and again from them to their fragment ions and so forth; accordingly, this process is called a ladder switch-ing". The model of ladder switching has been proposed and proven for aromatic molecular sys-tem [12,17,45,46]. For many molecules, this model explains the variation in the fragmentation peak pattern, which results from multiphoton ion-ization when changing laser intensity. A more general ladder switching model (for a review of multiphoton fragmentation see Ref. 47) results when involving the following considerations.

(i) For larger and larger molecules the possi-bilities of fragmentation and therefore the number of dissociation thresholds within a particular energy interval increase. That means that already for absorption of one or just a few photons within the molecular ions, many more parallel decay channels exist (as illustrated in Fig. 5(c) and dominate the mass spectrum to a greater extent than in the medium-sized molecular ions such as benzene [12,17] (See Fig. 5(b)).

(ii) Fast dissociation of the intermediate neutral state may reduce or even suppress multiphoton ionization. In favourable cases this may induce ladder switching within the neutrals with sub-sequent ionization of the neutral fragments (see Fig. 5(d)). Intermediate dissociation is particularly probable for inorganic and metallorganic mol-ecules [47,48], as well as for hot molecules (e.g. laser desorbed molecules) whose vibrational exci-tation is shifted into an excited electronic state by photon absorption (see Fig. 6) and for many mol-ecules excited to higher electronic states.

(iii) In photodissociation of molecular ions, ionized and neutral fragments appear. In prin-ciple, the latter may also be ionized by multi-photon absorption and contribute to the mass spectrum.

The probability of ionization of the neutral frag-ments resulting from case (ii) or case (iii) depends on their spectroscopic features of absorption. For small aromatic systems and nanosecond laser pulses, it has been shown that in general ionization of neutral fragments do not play any significant role [49]. However, for special wavelengths, carbon atom ionization has been observed at multi-photon ionization of benzene [50] and of other molecules [51]. Secondly, for larger molecular systems, aromatic fragments may be formed which absorb in the near UV (< 260 nm) and are easily ionized. Thirdly, with subpicosecond laser pulses, non-resonant multiphoton ionization is enhanced and even fragments with unfavorable absorption spectra may be ionized, such as hydrogen atoms (see section 5.1).

The possibility of variable fragmentation due to the ladder switching mechanism discussed above is demonstrated in Fig. 4. Such variable fragmen-tation can be used to obtain a set of different fragment spectra of the same molecule or to opti-mize a fragment spectrum for special information (e.g. peptide sequences, discrimination of isomers, identification of isotopomers).

4.1.4. State selective ion preparation. Finally, by selecting special intermediate states, vibrationally-cold cations of many molecules can be formed [52]. This is mainly due to optimized Franck-Condon factors for the transition from the intermediate to the vibrationless ionic ground state and is very useful for ion spectroscopy and ion kinetics exper-iments. The preparation of molecular ions in a few or even in single vibrational levels of their ionic ground state can be tested by measuring the kinetic energy of the emitted photoelectrons and has been shown for molecular ions such as the toluene cation [53], NHf [54], CH3I+ [55], OCS [56], benzene [41,57], monohalogenated benzene


-O processes in cold molecules

■> processes in hot molecules

probable processes

improbable processes

Fig, 6. The different intramolecular processes which may be active for resonance enhanced multiphoton ionization of cold and hot molecules.

cations [41,58] aniline [59] and many others. Alter-natively, rotationally-cold ions can be formed by selective excitation of rovibronic intermediate states or by cooling the neutral parent molecules in a supersonic molecular beam.

4.1.5. Compatibility with time-of-flight mass analysis. Further advantages of using lasers for ionization in time-of-flight mass analyzers are their short pulse length of some nanoseconds and the small achievable foci. As for the ionization efficiency, some percentage of all molecules within the laser focus (at least for (1 + l)-ionization) can be ionized with one laser shot when applying moderate laser intensities (soft ionization) [21,41] and up to nearly 100% when applying broad band lasers with high laser intensities (for which, how-ever, subsequent fragmentation can not be excluded).

4.1.6. Effects reducing multiphoton ionization yields. There are several aspects to be considered

when performing REMPI of molecules. The ionization efficiency may be reduced due to small transition moments to the ionization continuum (especially due to small Franck-Condon factors) or by fast intramolecular processes dissipating the electronic energy more or less into vibrational energy.

As mentioned above, fast relaxation processes within the neutral intermediate state of the mol-ecule may not end up in the ionization continuum but in dissociating states or in electronic states not to be ionized with one-color multiphoton absorp-tion. Particularly, in hot (e.g. due to laser desorption) and large molecules (many degrees of freedom) such fast relaxation processes may dominate due to a high density of accessible vibronic states. Counter measures are higher laser intensities (with the disadvantage of ionic fragmen-tation), short intense laser pulses, cooling in a supersonic beam or two-color excitation. How-ever, for some large molecular systems this high density of states may not be accessible due to a small intervibrational coupling, e.g. some mol-ecules may be dealt with as a chromophore and a residual molecular part with the vibrational coupling between both parts being weak. Known examples are van der Waals clusters, but it may also be applicable to large linear molecules such as linear peptides.

Unfavorable Franck-Condon factors between the neutral ground, neutral intermediate or ionic ground state may be another reason for reduced ionization yields. This may be due to strong changes in molecular structure in the involved neutral and ionic states. Another reason is a too-low-energy intermediate state. At one-color ionization, they may force the use of highly vibrationally excited intermediate states with unfavorable Franck-Condon factors. Two-color excitation may help in these cases.

Finally, ionization yields for very large mol-ecules may be reduced due to the high density of neutral states (vibrationally excited Rydberg states) near the ionization threshold so that these states are excited rather than ionization with low


excess energy photons takes place. Other effects concerning the ionization yield of large molecules are discussed in recent papers of one of the authors [60]. The general question arises as to whether there is a high mass limit of ionizability by multiphoton ionization.

4.2. Non-resonance enhanced multiphoton ionization

In some cases, non-resonant multiphoton ioniz-ation is efficient and more convenient than REMPI. This may be the case for laser evaporated molecules, where nothing is known about the spec-troscopy but it has also been extensively applied to the analysis of inorganic molecules [61].

Even for organic molecules non-resonant ioniz-ation may sometimes be more favorable; e.g. furan [21] is ionized at 370 nm by (2 + 1)-MPI. At a laser intensity of 6 109 W cm- 2, a pulse length of 5 ns and a bandwidth of 1cm"1, an ion yield of 10~8 is achieved, in contrast to the non-resonant 2-MPI at 260 nm and 108Wcm~2 where an ion yield of 10~5 has been observed. Despite the considerably larger laser intensity available at 370 nm, non-reso-nant ionization at 260 nm is obviously much more efficient. The one-photon absorption step in the (2 + 1)-MPI is usually strongly saturated at 109Wcm"2 and 5 ns pulse length, so that 2-MPI and (2 + 1)-MPI are comparable. This is mainly due to a fast relaxation of the (2 + l)-intermediate state as well as a considerably worse two-photon cross section. For subpicosecond high intensity laser pulses especially the non-resonant two-photon absorption cross sections into the ion-ization continuum become comparable to that of resonant enhanced (1 + l)-photon absorption.

4.3. One-photon ionization

Another laser ionization process with large cross sections but which is not enhanced by an inter-mediate state is laser induced VUV photoioniz-ation. By frequency tripling of laser light in noble gases [62] VUV light at high intensities and with short pulse lengths is available (e.g. 118nm by

tripling 355 nm in Xe gas). This light source has recently been used for efficient ionization of mol-ecules, e.g. in time-of-flight mass analyzers [63]. Both methods, ionization by non-resonant multi-photon absorption and by laser induced VUV, do not allow for the high selectivity supplied by involving molecular spectroscopy.

A "species semi-selectivity" is possible for VUV-ionization due to characteristic ionization poten-tials. An example is shown in Fig. 7 [64] which has been measured in a short (13 cm) linear time-of-flight mass analyzer. A complex mixture of gases (the exhaust gases of a combustion engine) is ionized either by electron ionization (see section 4.4) or by laser induced VUV at 118 nm. The latter spectrum shows less selectivity than is possible by REMPI (see Figs. 8 and 9) but still allows discrimination of traces (e.g. methylbenzenes, propylene, NO or N0 2 ) with low ionization thresholds against abundant species (e.g. N2, 0 2 ,

i λ* J

CO,

u El C 7 H 8

CftHfi REMPI

VUV

time-of-flight

Fig. 7. Comparison of electron ionization (El, unselective), resonance enhanced multiphoton ionization (REMPI, highly selective) and laser-induced VUV ionization (semi-selective) of a mixture of gases (emission of an internal combustion engine).

102 U. Boesl et al.I Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

C02) which may prevent mass analysis of traces by saturation of the ion detector, by overlapping to nearby mass channels or by causing space charge effects in the ion source due to very high ion densities. The non-resolved peaks at low masses in Fig. 7 are due to an interference of single masses (e.g. NO) and fragments of methylbenzenes which show flight-time broadening due to metastable decay in the ion source.

4A. Laser induced electron ionization

For some analytical mass spectrometric appli-cations high selectivity of REMPI is not always an advantage. A non-selective ionization is nece-ssary if an overview of components and their relative concentrations within a gas mixture is required or if a neutral ion source for molecules has to be tested for which laser ionization data (molecular spectroscopy, ionization threshold) are not yet available.

An ideal electron ionization source for time-of-flight mass analyzers is electron emission from metal surfaces induced by pulsed lasers [65]. Using pulsed lasers with photon energies larger than the metal work function (e.g. 292 nm for Ta wire), high intensity electron pulses with pulse lengths as short as 10 ns can be generated. Due to the high electron current and the narrow pulse width this electron source is well adapted to pulsed experiments and time-of-flight instruments.

For realization of a laser induced electron ion (LEI) source it is sufficient to hit one of the extract-ing plates by the UV laser pulse. A very efficient way is also to irradiate a wire tip which is positioned in the ion source at the attracting electrode near its axis by the laser (see Fig. 3(e)). In the static field of the ion source laser induced emitted electrons are accelerated opposite to the ion direction and have an energy of some lOOeV when passing and ionizing the neutral molecular beam. This simple arrangement is sufficient for many purposes and has the major advantage to be easily transformed into a REMPI ion source

just by shifting the laser focus by a few millimeters (as in Fig. 3(a) or Fig. 3(b)) and tuning the laser wavelength to a molecular resonance.

In Fig. 8 El (as well as in Fig. 7) a laser induced electron ionization mass spectrum of a mixture of gases (motor vehicle exhaust gas) is shown. In com-parison in Fig. 8 (UV1-UV4), REMPI mass spec-tra for four different wavelengths are shown; here, the ionization of several traces of air pollutents is selectively enhanced, namely NO, acetaldehyde, toluene and three methylated benzenes. For all mass spectra the identical ion source with an effu-sive neutral source (Fig. 3(a)) and with REMPI (Fig. 3(a)) or laser induced electron ionization (Fig. 3(e)) has been used in combination with a very short (13 cm) and compact Wiley-McLaren time-of-flight mass analyzer.

i

<

Ö

C

i

UV1

UV2

^-ni trogen monoxide

-

1

—a

UV3 I

J t 5 n

* UV4 2

El )

h

cetaldehyde

—toluene

U-xylene

trimethyl / benzene

IL 1

, .toluene . J /

1— carbon dioxide 'ôxygen

nitrogen

AASS 1

Fig. 8. The same mixture of gases as in Fig. 6 analyzed for four different wavelengths: UVl(307.6nm), UV2 (301.5nm), UV3 (260.9 nm) and UV4 (264.1 nm); for comparison the El spectrum of Fig. 6 is shown.


Figure 8 nicely illustrates, how both ionization methods complement each other. LEI delivers information about the realistic ratios of concen-trations of the main components, REMPI about details such as traces within the mixture; e.g. NO is detected in Fig. 8 (UV1), while overlapped by the abundant components N2 and 0 2 in Fig. 8 (El), acetaldehyde detected in Fig. 8 (UV2), while over-lapped by the isobaric C 0 2 in Fig. 8 (El). For further details of laser mass spectrometry for exhaust gases with high temporal resolution see Ref. 66. Laser-induced ionization and REMPI even may be combined and performed within one single laser shot if the laser is tuned in resonance (e.g. of toluene) and nevertheless hits the electron emitting wire tip (see Fig. 9, performed in an exhaust gas mixture [64]).

4.5. Anion source using laser induced electron emission from metal surfaces

Anion mass spectrometry, even if not as wide-spread as cation mass spectrometry, has many complementary features to the latter. Many mol-ecules, which cannot be formed as molecular or quasimolecular cations, may be formed as anions. Furthermore, anion fragmentation mass patterns can deliver additional, often complementary structural information. A unique feature of anion mass spectrometry is that neutral molecules can be formed via photodetachment thus supply-ing information about these neutrals. If photo-detachment is performed in a mass selective way (e.g. in the space focus of a pulsed ion source) even mass selected neutrals are available and neutralisation-reionization mass spectro-metry or other secondary experiments on these neutrals may be possible. One such secondary experiment is the analysis of mass-selectively-detached photoelectrons for photoelectron spec-troscopy [67]; thus information is achievable about energetics and kinetics of otherwise not accessible isolated neutrals (e.g. radicals, non-volatile molecules, polymers and clusters of par-ticular size).

_J A J U _ <

CO, C,H 7"8

C7H£

REMPI

1 Λ,

N2

. . I i J

time -of - f l iç

C7H8

EU REMPI

synchronous

ht ►

Fig. 9. The same mixture of gases as in Fig. 6 analyzed with El and resonance enhanced multiphoton ionization as well as a combination of both, synchronously performed at every single laser pulse.

An interesting anion source is the combination of laser induced electron emission (as described above) with a supersonic molecular beam and laser desorption if necessary. For that reason, the electron emitting wire is not mounted in the ion source (as in Fig. 3(e)) but as near as possible to the nozzle of the supersonic molecular beam valve (Fig. 3(f)). Again electrons are emitted by focusing a laser beam onto the wire tip, are decelerated by collisions with neutrals of the supersonic molecular beam and attached to molecules or clusters. Either molecular anions, anionic fragments or products of a reaction or clusters between anions and neutrals are formed. If the anion formation takes place in the high density collision region of the supersonic beam the first vibrationally-hot anions are cooled

104 U. Boesl et al.j Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

and stabilized. The cloud of anions then drifts into the ion source together with the neutral molecular beam and is extracted into the time-of-flight analyzer by applying a HV- pulse.

In Fig. 10 [68] an anion mass spectrum produced by the described anion source is presented; by seed-ing acetylene in the supersonic argon beam a mass pattern of carbon and carbon hydrogen cluster anions are generated. Furthermore, using an iron wire for electron emission allows the formation of FeCxHy anions which are marked in the mass spec-trum of Fig. 10. With different wire material all kinds of metal carbides which are of interest for surface catalytic reactions have been formed [68].

4.6. Quasi continuous ion source and pulsed ion extraction fields

In the previous sections only pulsed ion forma-

tion methods have been described. Nevertheless, in mass spectrometry, many interesting continuous or quasi continuous (with respect to the time scale relevant in time-of-flight mass analysis) ion sources have been developed, such as conventional electron ionization, chemical ionization, electro-and thermospray. These ion sources can also be combined with time-of-flight analyzers when pulsed extraction fields are used. The requirement for the HV pulses is a very high stability but not an extraordinarily fast rise time. For example, with a HV pulse rise time of 50 ns, ion peak widths of 2 ns have been observed [69].

To reduce unfavorable duty cycles, repetition rates of the HV- pulses in the 10 kHz range should be used. Additionally, one should take care that most of the ions formed in the ion source, or drifting into the ion source within the time between two HV- pulses, are extracted

Carbon Clusters

50 55 60 65 70 75 80 85 90

Mass 95 100 105 110 115 120

Fig. 10. Anion carbon clusters and anions of iron carbonyl and iron carbonyl hydride molecules produced in a supersonic beam of Ar seeded with acetylene and laser induced electron emission from an iron wire.


and detected. Simple means of ion storage are sufficient. By conventional electron ionization, mass spectra of large molecules have been measured with a mass resolution of 20000 [70]. For pulsed ion sources the effective resolution may also be significantly increased by pulsed field ion extraction. By such means, with a 6 ns laser pulse, ion pulses with a width of 3.6 ns at the detector of a reflectron and thus a mass resolution of 20000 have been achieved in our apparatus [20]. Others report a record of mass resolution of 35000 achieved with similar means [69].

5. Double excitation schemes within the ion source of time-of-flight mass analyzers

Double excitation schemes may have two goals: (i) synchronous performance of two exper-

iments; (ii) secondary excitation of primarily excited

neutrals or ions. As for goals (i), performing two correspond-

ing experiments and recording their results under identical conditions is absolutely necessary for many applications. Short-term as well as long-term fluctuations of laser pulse characteristics (pulse energy, pulse length, beam diameter), neutral density of the pulsed molecular beam or at laser desorption, space charge effects (due to ion density), mass analyzer transmission and ion detector sensitivity may prevent a successful comparison of the two experiments. This is a major problem with multiphoton ionization at high laser intensities, but also when pulsed neutral sources and pulsed mass analysis are involved. In the optimal case, both experiments are performed at nearly the same time (within a few nanoseconds), in the same region of neutrals and if possible with the same laser pulse or with two tunable lasers but with the same pumping laser.

In the following sections we discuss some examples: comparison of nanosecond and femto-second laser ionization of laser desorbed neutral molecules, the discrimination of isomers, the

calibration of multiphoton ionization by calibra-tion gases, and the characterization of neutral gas pulses by a kilohertz repetition rate laser.

As for goal (ii), many experiments in mass spectrometry involve a secondary excitation of primarily-formed ions. In particular, laser exci-tation is predestined for these kinds of exper-iments. With lasers, primary ions may be formed in small volumes, short times, defined states and often species selectively (equivalent to a primary mass selection). These primary ions have optimal preconditions for a spatial as well as temporal overlap with a second laser focus for further excitation.

Already, multiphoton dissociation after ion formation within a single laser pulse may be considered as such a secondary excitation (see Fig. 5 (b)). This dissociation process can also be induced by a delayed laser pulse supplying the additional features of varying laser inten-sity, wavelength and so forth. Even a rough selection of primary and secondary mass spectra is possible which can be enhanced by means such as HV pulses applied at one of the ion source electrodes.

Furthermore, delayed secondary excitation allows the determination of short ion decay times. Some techniques of mass selective laser spectro-scopy of laser formed molecular ions may also be considered as secondary excitation processes within the ion source. There are many other exper-iments applicable to laser-formed ions, such as photoelectron- [52-59], Threshold photoelectron-and ZEKE spectroscopy [71] or M ATI spectro-scopy [72], PEPICO [73], optical holeburning [74] which belong to secondary excitation experiments within the ions source of time-of-flight analyzers, but are not described in detail in the following sections.

5.7. Multiphoton ionization: comparison of ionization and dissociation using nanosecond and femtosecond laser pulses

As discussed in earlier sections, fast intermediate


relaxation processes may reduce or even prevent REMPI of special small and medium-sized mol-ecules with nanosecond lasers. Due to the com-petition with direct laser desorption-ionization, questions also arose about the efficiency of REMPI of laser desorbed neutral molecules; until now, for very large molecules this so-called post-ionization has not yet been applied successfully above the mass limit of 3000 Da. The question if the main reason may be the decreasing desorption probability of large neutral molecules or decreasing ionization efficiency is not yet answered. While the number of desorbed molecules surely decreases with increasing mass, several other effects are discussed which may influence the ionization yield (see for instance Ref. 60.

One ion-yield-decreasing effect may be due to fast relaxation processes in the resonant inter-mediate state, as it is well known for special classes of smaller molecules. Fast intermediate pro-cesses may be overcome by excitation with very short intense laser pulses thus increasing the opti-cal absorption rate. In fact, for femtoseconds laser pulses (e.g. pulse length of 500 fs, pulse energy 10~6J, focus diameter ΙΟΟμηι) transition prob-abilities of non-resonant two-photon absorption become comparable to resonance enhanced two-photon absorption. Therefore, a comparison of nanosecond and femtosecond laser ionization is of principal interest for multiphoton ionization mass spectrometry.

By focusing both laser beams into the ion source at the same spot within a few tenths of a millimeter, but delaying one of them by several nanoseconds, two mass spectral patterns within one time-of-flight spectrum are recorded simultaneously, separated by the laser delay time only. The advantage of this arrangement is that both experiments are per-formed within one single pulse of neutral mol-ecules. This is of particular importance for laser desorption with its large fluctuations of the surface emitted particle numbers. "Double excitation" within the ion source guarantees a realistic com-parison of both ionization techniques. In Fig. 11 [75] two such double excitation mass spectra

(region of molecular ions, low fragmentation degree) are presented: In Fig. 11(a) tryptophane (mass 204 Da), in Fig. 11(b) /3-Carotene (mass 536 Da) have been ionized. In Fig. 11(a) both laser pulses had the same pulse energy of 10~6 J, corresponding to a difference of 10~4 in intensity. Despite this large intensity difference, both sequences of signals (nano- and femtosecond) have nearly the same ion signal strength. For β-Carotene it was impossible to observe any ion signal for a pulse energy of 10~6 J of the nano-second laser. The amplitude of the molecular ion peaks induced by nanosecond and femtosecond laser pulse ionization become comparable only if the pulse energy of the nanosecond laser is increased up to 4 10"5 J (see Fig. 1 (b)). The same or similar effects have been observed for larger molecules like Gramicidin S (1141 Da), Gramici-din D (1881 Da) and Mesoporphyrin (594 Da). Obviously, femtosecond ionization is much more efficient for these molecules than nanosecond ion-ization. This is probably not a primary effect of larger mass, but of different molecular structure and intermediate state behavior.

In addition, more unfavorable processes at laser desorption (due to different molecular mass, struc-ture, intermolecular forces between adsorbate and substrate and so forth) may cause hotter molecules. This vibrational excitation is preserved in the inter-mediate state by photon absorption inducing there an enhancement of fast relaxation processes.

Ionization and dissociation with femtosecond pulses also shows special features not seen in nano-second ionization; these features are evident in the laser desorption femtosecond postionization mass spectra of/3-carotene in Fig. 12 [75]. At low laser intensities, rather soft ionization with a small degree of fragmentation is achieved; at high laser intensities, strong fragmentation takes place with astonishingly high peaks of Ar and H ions. The ionization of Ar (non-resonant three-photon absorption) demonstrates the high efficiency of non-resonant multiphoton ionization at the extremely high intensities of femtosecond laser pulses (here typically 20 10"5 J, 500 fs, 10"4cm2).

U. Boesl et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

**/ '^photon», ( · — '^photons, ns

ω c D

JÛ L·.

o

ω c

107

94.0 95.0

Ε,, = 1μϋ t-

E ^ l p J ^

fs

•ns 204 205 206

mass /Dal ton

") **photons,fs < < : ^photons, ns

ω C

D

ω c Φ

153.0

Ε„=1.25μϋ

154.0

i - ^ J j - J ^ - ^ 1 1 \ r- Λ Γ* ^ * 4 "" ^ Γ ^ ^ Ι Ε„, = 40μϋ

fs

ns 536 537 538

mass /Dal ton

Fig. 11. Comparison of ionization with nanosecond and femtosecond lasers of (a) tryptophane and (b) /3-carotene. Please note that nanosecond and femtosecond laser pulse energies are equal in (a) and different in (b).


ß-Carotene

ChL ChL CH CH,

7 W a) ns ionization ( I—1 -10 —. cm'

χ10 \ Μ"ιΠπΙί><Η· · ' ■ ' ■ ■ j

/ / / / / / 1

MX • i i I I

100

Uû 9 W

b) fs ionization (out of focus I —10 —-cm'

L //

// //

/ w.

o O -I o

c) fs ionization (in focus. 1—10 —,) car

Ar

H"

C +

H:

c

c x

I I c*x II III il I ' +

536 Dalton

455 Dalton

268 Dalton I»

100 time of flight/με

Fig. 12. Comparison of fragmentation due to (b) low and (c) high intensity femtosecond laser fragmentation for ^-carotene; for comparison, nanosecond laser fragmentation (a) at laser intensities lower than in (b) is shown.


Ar4" appears in the mass spectrum due to its use as the carrier gas of the supersonic beam.

In contrast to Ar, the H atoms must have been formed either as neutral fragments in the neutral intermediate state of the desorbed parent molecules or as neutral fragments of their molecular ions (e.g. /3-carotene in Fig. 12. Furthermore, these frag-ments must have been formed in an extremely short time, namely during the pulse length of the femtosecond laser. This seems to be reasonable because the C-H stretching mode has a time period of about 60 fs. A formation of neutral H in the desorption process is unreasonable because of the lack of a M - H ion signal. Ultrafast fragmen-tation and the absence of fragments with higher masses which would be a hint for near-threshold fragmentation as is typical for nanosecond ladder switching) indicate an excitation up to high energies in the molecular ion or even the neutral above the ionization threshold by femtosecond laser excitation.

5.2. Discrimination of structural isomer s

Several excitation schemes exist to discriminate structural isomers by laser excitation, such as using isomer selective intermediate resonances [42,43] or isomer selective ion dissociation [43,44]. If both methods do not work (e.g. due to hot molecules or fast isomerisation processes) the difference of ionization potentials of both isomers may be used for semi-selective ionization. The problem is then that either both isomers are ionized or selectively the one with the lower ionization threshold, but not selectively the other one. A solution is difference mass spectra due to ionization at two different laser wavelengths. However, such difference mass spectra are very sensitive to fluctuations of any kind and identical experimental conditions are necessary.

This is a typical problem to be solved by time-delayed double excitation in the ion source, thus recording both mass spectra synchronously for every single ionization cycle. In addition, if both laser wavelengths are at least due to two laser

systems with the same pumping laser, effects of laser fluctuations are also minimized as much as possible. An example is shown in Fig. 13 [76]. The isomers xylene and ethylbenzene (both mass 106) had to be separated. With a laser wavelength of 266 nm (YAG-laser) both isomers are ionized; with a second wavelength of 310 nm, xylene mol-ecules excited by the first laser are selectively ionized (see also excitation schemes in Fig. 13). If laser 2 is delayed by 20 ns with respect to laser 1, the so-formed xylene ions also will appear in the time-of-flight spectrum with a delay of 20 ns thus being separated from the common xylene/ ethylbenzene signal. Correspondingly, the 13C-isotopomers of both molecules are separated in time (Fig. 13 on the right). Due to nearly identical

time-of-flight

Fig. 13. Isomer selective ionization (xylene, ethylbenzene) by synchronous one-color/two-color laser excitation involving the difference of ionization potentials.


experimental conditions and recording of both signals in one time-of-flight spectrum the difference signal is easily obtained and fairly stable. Exci-tation schemes as in Fig. 13 are very useful for gas analysis, e.g. fast exhaust gas analysis, where every single laser shot has to supply complete information about particular trace substances [66]. Another application is for substances such as sub-stituted aromatics (as in Fig. 13 or larger ones) which absorb at very similar wavelengths, and for which high resolving excitation (e.g. using cooling in supersonic beams) is not possible. This is the case for molecules with low vapor pressure for which heated inlet nozzles or laser desorption techniques have to be used.

beam axis. At larger distances, the number density is so small that only statistical events are measured causing a low signal-to-noise ratio [77].

A special case for normalization is high-order multiphoton ionization. Saturation of single absorption steps due to high laser intensities or fast intermediate relaxation rates etc. may induce different dependences on laser intensity for sample and calibration gas molecules even if the ionization for both is a multiphoton process of the same order. Therefore, a new kind of calibra-tion has been developed [77]. Two calibration gases, which are ionized by different orders of multiphoton absorption, serve as a kind of sensor

5.3. Synchronous ionization of sample and calibration gases

In order to eliminate short- and long-term fluctuations of laser beam characteristics (pulse energy, pulse length, focus etc.) in experiments where absolute concentrations have to be measured accurately, normalization of multi-photon ionization signals have to be performed. For this purpose, a synchronous ionization of cali-bration and sample gases is very useful. This is achieved by adding one or two suitable calibration gases to the molecular beam and ionizing molecules of interest and calibration gas by the same laser pulse. In this case double excitation concerns the number of molecules rather than the number of laser pulses. Particularly in this case, the exper-imental conditions for both "experiments" (ioniz-ing (i) sample molecules and (ii) calibration gas) are identical and ideal for a confident comparison of both signals, i.e. for normalization.

One possibility is to add calibration gases via a capillary introduced into the canula (e.g. in Fig. 3 (a)) for an effusive molecular beam. The perfect mixing of calibration gases and sample gas is demonstrated in Fig. 14 (a). The ratio of both signals (sample gas:benzene, calibration gas: difluorotoluene) is constant over a distance of 1 mm on each side perpendicular to the molecular

m

-1000

L

1.5

1

0.5

n

(a) gas inlet

cannula axis

-500 0 500 deviation [μππ]

1000

10 20 30 40 concentration (GO / pm

50

Fig. 14. Normalization of resonance enhanced multiphoton ionization signals using calibration gases ionized by the same laser pulse as the sample gas. (a) quality of the mixture of sample (benzene) and calibration (/?-difluorbenzene) gas in radial distance from the beam axis; (b) samples with different amounts of methanol quantitatively analyzed by calibrated laser mass spectrometry and conventional gas chromatography.

U. Boesl et aljlni. J. Mass Spectrom. Ion Processes 131 (1994) 87-124 111

for intensity distribution within the laser focus. For the ideal case of a rectangular or gaussian distri-bution of laser pulse energy in space and time, the ratio of both calibration gas signals is a well-defined function of the intensity conditions within the laser focus. Furthermore, the ionization yield of the sample molecules shows a well-defined depen-dence on the ratio of the calibration gas signals. This dependence is easily determined exper-imentally and may serve as a calibration curve for a particular sample molecule.

This calibration method is easily to realize, very fast, inexpensive and much more reliable than an external measurement of the laser pulse energy. An example for a (3 + l)-multiphoton ionization of methanol is shown in Fig. 14 (b): For compari-son, mixtures of gases, containing different amounts of methanol are analyzed by laser mass spectrometry and by conventional gas chromato-graphy using the same calibration gas. Both methods deliver the same absolute values within a deviation of less than 10% via a large range of concentrations.

Preconditions for ideal calibration gases are easy available, a wide unstructured absorption band and a convenient vapor pressure. Some optimal calibration gases are difluortoluene for (1 + 1)-REMPI in the range of 245-270 nm, perdeuter-ated acetaldehyde and triethylamine for higher-order MPI at 298-315 nm, and at 430-454 nm [77].

5.4. Using kilohertz lasers for sampling neutral gas pulses

For special applications it may be necessary to sample very fast processes. New laser types with kilohertz repetition rates like Nd:YLF lasers (fourth harmonic) or with kilohertz repetition rate amplified mode locked picosecond- or femto-second Titan-Saphire- lasers are ideal sources for double or multiple excitation schemes in the ion source of a time-of-flight mass spectrometer.

One illustrative application is the sampling of neutral gas pulses in 100/is intervals (correspond-ing to a 10 kHz repetition rate). Typical pulse

lengths from pulsed supersonic beam valves (modified Bosch valve) are in the range 100 to 500 μ$. The 10 kHz laser pulses allow the obser-vation of neutral pulse conditions (rise time, pulse length, pulse shape etc.) for every single neutral pulse and therefore delivers very reliable infor-mation. In Fig. 15 [75] a 20 μί 10 kHz laser pulse train of a Nd : YLF laser (TFR, Spectra Physics) has been used to analyze a pulsed valve operating with benzene seeded in Ar. Each 100 /is a new mass spectrum is created. The ion signals in Fig. 15(a) correspond to a neutral pulse smaller than 100 ^s, resulting in one strong time-of-flight spectrum of benzene. In Figs. 15(b) and 15(c) the peaks with lOO^s spacing represent the temporal envelope of neutral gas pulses with considerably larger valve opening times. The time-of-flight spectra, how-ever, contain much more information than only time behavior of a gas pulse; each single peak in Figs. 15(a)-15(c) actually represents a whole mass spectrum, as the inset shows (here soft ionization of benzene).

A further application of high repetition lasers may be a comparison of the neutral gas envelopes belonging to different masses, which may reveal information about the velocity slip of large mol-ecules within the supersonic beam or about frag-mentation within the neutral source (e.g. due to laser desorption). Tuning the laser and measuring spectra of vibrational bands (rotational envelopes) may even allow the determination of the quasi-temperature distribution within the gas pulse and for different masses. Again the advantage here is that all information is sampled for each laser shot, thus averaging out drifts and instabilities in the experimental setup.

5.5. Multiphoton ionization-dissociation

In the sections 5.1 to 5.4, double excitation in the ion source has been used to achieve ion signals which can be compared with high reliability. In the following sections, double excitation means primary and secondary laser excitation of the same sample molecule. A special case of double


i/>

100

+ c ^ t -hb -4-

C S H ;

C 6 H ;

C6»Î

V%H;

%'w

X-Time-of-f l ight [με]

200 Λ00 600 800 1000

100 με Δ 10 KHz Laser exitation

Fig. 15. Gas pulses from a pulsed valve with different pulse lengths analyzed by a 10 kHz laser. The inset shows that every single signal (all of them representing the envelope of the gas pulse) actually consists of a whole mass spectrum.

excitation is multiphoton ionization-dissociation with one single laser pulse: there is a primary process (ion formation) and a secondary process (ion dissociation) which can be described by the different kinds of ladder switching (see section 4.1). As it is possible to perform species selectivity by using REMPI, ionization-dissociation in a time-of-flight analyzer is com-parable to a double mass spectrometry exper-iment similar to collision induced dissociation in conventional tandem mass spectrometry [78]. The conventional primary mass selection used to obtain a fragment-free molecular ion beam now is replaced by the "soft" REMPI without fragmentation. By using an optical delay line and split laser pulses or two laser pulses of different wavelength, ionization and dissociation can be separated in time and space [12,17]. This allows variation of the dissociating laser pulse

intensities independent of the ionization process. Despite the difficulties of a two-laser experiment a very high efficiency of secondary laser excitation can be achieved with the primary ion nearly disappearing totally due to secondary photon dissociation. If the time delay between both lasers is long enough, a limited mass resolution for the secondary excitation is possible (e.g. the groups of fragments with different numbers of carbon atoms due to aromatic parent molecules are separately excitable [12]). In this case a complete MS-MS experiment within the ion source (that means within few millimeters) is performed. In summary, if laser 1 produces a mass pattern of a species-selectively-ionized parent molecule using a second laser delayed in time allows a crude (with respect of mass resolution) but very efficient and extremely simple MS-MS experiment.

U. Boesl et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 87-124 113

5.6. Measurement of short ion decay times in the ion source

The arrangement used in section 5.5, allows not only tandem mass spectrometric experiments but also investigations of short ion decay times. For the determination of ion decay times usually the mass and time dependent kinetic energy dissi-pation to the neutral fragmentation partner is used. The activation of the molecule is usually performed simultaneously with the ionization. Then the time of the decay event is transformed into a position in space by the movement of the precursor ion during its extraction out of the ion source. This position corresponds to a certain ion potential and ion kinetic energy due to the electric fields of the ion source (see Fig. 16(a)). A part of the ion kinetic energy of the intact molecule is carried away by the neutral fragment; this results in a position-dependent and therefore time-depen-dent final kinetic energy, and finally in a flight time of the fragment ion which depends on the decay time. In conclusion, information about the decay rate of a metastable decay in the ion source is deducable from time-of-flight profiles of ion

peaks. This has been used efficiently by several groups for decay time measurements on the micro-second time scale [79].

A technique to measure short decay times (> 15 ns, decay within the ion source) with a reflectron instrument [20] is explained in Fig. 16(b). The decay times of moving molecular ions are transformed into a position x and a special kinetic energy UF of the fragment ion as described above and as illustrated in Fig. 16(a). The relation between this ion kinetic energy UF and the moment of decay /M can be used to determine the decay rate and is displayed in Fig. 16(b). However, for short tM the acceleration of the ion M+ does not last long enough to give the parent ion a reasonable kinetic energy. A short decay time of 20 ns results then in a very small change of UF (as to be seen from Fig. 16(b)) which considerably limits the accuracy of decay time measurements.

Usually, the point of ion formation (x = 0) is also the point of ion excitation. Much higher accuracy, however, can be obtained by delaying the laser excitation of the ions by using a second laser delayed by 200 ns or more (Fig. 16(b)). In this case, a decay time of 20 ns results in a

Ionization

prompt excitation

t M [ , u s e c ]

delayed excitation

Fig. 16. (a) Dependence of fragment ion kinetic energy on the decay time and point of decay respectively within the acceleration region of length xA (iA = time the parent ion M+ needs to pass the length xA). (b) Influence of a decay 20 ns after laser excitation vs. the fragment ion kinetic energy t/F for a prompt excitation (during laser ionization) and for a delayed excitation (second laser delayed by 0.2 s).

114 U. Boesl et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 87-124

considerably larger change OU of the fragment ion energy UF. In Fig. 16(b) OU is about 20eV for the delayed excitation.

Now the question arises of how to measure the distribution of UF exactly and in one laser shot. For these purposes one of the features of reflectron instruments may be exploited. The reflectron voltage C/REF

c a n be detuned in such a way that a nearly linear relation between the ion kinetic energy UF and the time-of-flight tF of the fragment ion is achieved [20]. In our instrument an energy dependence of the time-of-flight is achievable so that the energy change OU = 20eV is transformed into a flight-time difference Δ/Ρ « 100 ns. Figure 17 (1) and (2) displays a measurement of the first excited electronic state (Α(2Π3/2 and A(2U\/2) of OCS+, resulting in decay times of 110 and 80 ns. These values can be compared with fluorescence

decay times: (105 ±3 ) and (77±3)ns [80]; (107 ± 7 ) and (73±7)ns [81] have been deter-mined in very good agreement with our value. The conclusion from this comparison is that the lifetime of the A state of OCS+ (slowest step) determines the overall photodissociation rate and that the intermediate electronic state is a fast decaying non-bonding state. Decay time measurements on the vibrationally-excited ionic A-state revealed still measurable decay times > 15 ns for many vibronic states involving the bending mode (Fig. 17 (3) and (4)), but decay times below 15 ns for vibronic states involving stretching modes (Fig. 17 (5)). This is in contrast to fluorescence decay measurements [80,81] from which decay times far below 15 ns for all vibrational-excited A- state levels have been concluded.

relative t ime-of - f l ight [με]

Fig. 17. Time-of-flight profiles due to different decay times of OCS+; vibrationless (1) and (2), as well as bending (3) and (4) and stretching (5) modes of the first excited ion state have been investigated. On the right-hand side are shown, theoretical time-of-flight

profiles deduced from the parameters of our apparatus.


Figure 17 also shows theoretical peak shapes for different decay times considering the parameters of our apparatus. The accuracy of our method is some 10% with a minimum measurable decay time of 15 ns (with an 8 ns laser pulse). Effects of spatial overlapping of both lasers have been taken into account.

5.7. Separation of secondary and primary fragments in the ion source: separation by HV pulse

Separation of primary and secondary fragments or even of whole mass spectra is necessary for experiments described in sections 5.5 and 5.6, but also for ion spectroscopy as described in section 5.8. Time delay in the order of 100 ns between primary and secondary laser excitation is often the maximum delay possible when using one pump laser only. This short delay is not sufficient to perfectly separate fragments caused by the ionization from those caused by the secondary excitation.

We have developed a technique that allows the separation of the fragment ions due to the ion-ization laser from fragment ions of the same mass due to the secondary excitation process, without increasing the time delay of both lasers beyond the 100 ns limit [82].

In Fig. 18 a time-of-flight mass spectrum of methyl iodide ionized by (2 + 1)-MPI at 369.9 nm is shown. The primary fragmention intensity of the methyl ions would strongly interfere with the methyl fragment ions of a secondary laser exci-tation of the methyl iodide cations. The insets show the methyl ions under different conditions. In Fig. 18(a), the CH3" peak is displayed on an enlarged time scale; in Fig. 18(b), the delayed laser 2 (delayed 60 ns) produces a small CH3" peak that appears as a shoulder of the primary fragment ion peak. In Fig. 18(c), a 3 kV ion extrac-tion pulse is applied onto the repeller electrode of the ion source which has a pulse width of 20 ns and is triggered between the first and second laser pulse. This causes a time-of-flight shift of the secondary fragments selectively due to a change of their initial

momentum. In addition, this momentum transfer induces a large kinetic energy difference of 16eV for mass 15. Considering this energy effect, a special property of reflectron time-of-flight analy-zers is used. The end plate of the ion mirror is set to a potential that is somewhat lower than the energy of primary ions, but higher than that of the second-ary ions. Thus, primary ions strike the end plate and are filtered out of the time-of-flight spectrum and secondary fragments (whole secondary mass spectra) are selectively recorded at the ion detector (Fig. 18(d)). For such a high energy filter action a movable reflector end plate with variable voltage is particularly useful. Thus the electric fields in the ion reflector can be kept constant for optimal energy compensation [83]. One should note that the transfer of momentum for all masses formed before the HV- pulses is the same but results in different energies due to different masses. This restricts the application of this technique to small masses (mass up to 30% of the molecular ion).

5.8. Laser spectroscopy of molecular ions

For the spectroscopy of molecular ions many techniques can be applied. Particularly for the ion ground state, besides conventional photoelectron spectroscopy and fluorescence dispersion spectro-scopy, (ZEKE) [71] with one-photon or multi-photon excitation delivers high resolution spectra which allows even rotational spectroscopy. A similar technique recording ion current com-bines both energy resolution and mass detec-tion, and could be successfully applied for clusters [72].

The first excited states, which are considerably lower in energy for molecular radical cations than for their neutral parent molecules, often exhibit interesting couplings, but thus are subject to very short lifetimes. For this reason, laser induced fluor-escence often fails as a spectroscopic method for these states due to the small radiative yields. How-ever, photoelectron spectroscopy and even ZEKE spectroscopy in this energy region (13-16 eV) suffer, from poor energy resolution (> 30 cm-1). In


0 0 1—

CD a: <

>-

<7> LU

1 CH 3+

.1

Γ

J

a) 1

I I*

Ί J

Laser 1. UR e f = 826V j

h) I u l Laser 1*2, UR e f = 8

I

u 26V

c I *- / Laser 1*2. HV-Pulse,

^ / Laser 1 + 2, HV-

1

| u U R e f = 826V

/ i _ Pulse,URef =810V

■

CH3I+'

L 1 1-

1 1 1 1 1 1 1 1

10 20 30 40 50 60 70 80

TIME OF FLIGHT [ μ ε ]

Fig. 18. Discrimination of primary and secondary fragment ions by high voltage pulse ion extraction. The multiphoton ionization ((3 + 1)-MPI) mass spectrum of methyl iodide is shown exhibiting a strong primary CH^ signal (in (a) on an enlarged time scale), (b) Ch^ fragments induced by a second laser (delayed by 50 ns); (c) application of a high voltage pulse (20 ns) in time between both lasers; in

(d) a reflector end plate is used as a high-energy and therefore primary ion filter.

the latter case, the limitation is mostly due to the low VUV monochromator resolution. Addition-ally, none of these techniques directly allow a mass selective detection (except coincidence measurement), and, often, not mass selective exci-tation either.

There are no such restrictions for photodisso-ciation spectroscopy, where resonances are detected via a subsequent dissociation. It can be applied to radiative as well as non radiative states. Mass selective ion preparation and detec-tion are intrinsic features and there is no principal limitation to spectral resolution other than the bandwidth of the light source, Doppler broaden-ing, or natural linewidths. Photodissociation

spectroscopy of ions with lasers is a typical double excitation experiment in the ion source. A first laser is used for ion formation and a second tunable laser is used for ion spectroscopy detected by photodissociation.

There are two kinds of photodissociation spectroscopy: (i) one-photon dissociation spectro-scopy and (ii) resonance enhanced multiphoton dissociation spectroscopy (REMPDS). In the latter case, one-photon absorption into the on state of interest is followed by absorption of one or more photons to a dissociating state. A sophisticated excitation scheme (for an explanation see section 6) is shown in Fig. 19(a). In both cases of photo-dissociation spectroscopy, the product ion current

U. Boesl et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 87-124 117

(a) y/4W/.·

i.e.

^ 1 MM

Λ\· λ3

——fragment ion F

FRAGMENTATION STEP

SPECTROSCOPIC STEP

molecular ion M+

MULTIPHOTON IONIZATION

molecule M

(b)

B—X transition

M =M0N0FLU0R0-BENZENE CATION

20000 21000 22 000 23000

■LASER 2 FREQUENCY [cm"1]

Fig. 19. (a) Excitation scheme for resonant enhanced multi-photon dissociation spectroscopy; laser 1 is used for a REMPI ion source, laser 2 for wavelength tunable excitation of the mol-ecular ion, laser 3 for selective dissociation of excited molecular ions. The detection step can also be performed by laser 2 with high intensities, but applying laser 3 in the space focus of the ion source (see Fig. 20) allows a separation of the spectroscopic and the detective step. Thus, lower intensities of laser 2 are applicable avoiding broadening effects within the spectra, (b) Resonance enhanced multiphoton spectrum of the first B <— X transition of the monofluorobenzene radical cation.

is measured as a function of the wavelength of laser 2. One-photon dissociation spectroscopy [84,85] is, of course, restricted to predissociating states. This is not the case for REMPDS [82,86,87]. In Fig. 19(b), a REMPD spectrum of the monofluorobenzene radical cation is shown. For this spectrum a special arrangement has been used involving the space focus of the ion source (see section 6).

We shall now consider pulsed ion sources includ-ing their space focus as sophisticated ion source arrangement, and therefore discuss laser exci-tation within such arrangements at the end of this

presentation of laser ion sources for time-of-flight mass spectrometers.

6. Secondary laser excitation in the space focus

For tandem mass spectrometry, decay time measurements, photodissociation spectroscopy of molecular ions and many other "double exci-tation" experiments, a special feature of pulsed ion sources, the so called space focus (see section 2), offers additional new possibilities. In particular, in combination with reflectron time-of-flight analy-zers, a high primary mass selectivity for the second-ary excitation and a high secondary mass selectivity for final registration is available.

6.1. Tandem mass spectrometry

Such a combination is represented in Fig. 20. Laser 1 produces an ensemble of molecular and fragment ions and these are accelerated into the field-free region. When passing the space focus of the ion source they are already separated due to their different mass. The quality of this mass separation depends on the type and quality of the space focus. For a second-order space focus (see section 2) a mass spectrum of xylene and its 1 3 C r

isotopomer is shown in the lower part of Fig. 20. With laser 3, the secondary mass selective exci-tation is achieved: (i) by positioning its focus at the space focus of the ion source and (ii) by firing the laser so that its laser pulse passes the space focus in time with the ion bunch of the desired mass (laser 2 is used for ion spectroscopy; see below and Fig. 19). The mass selectivity of this secondary excitation depends mainly on the size of the laser focus and the quality of the space focus. After secondary excitation the secondary frag-ments due to the selected precursor ion have to be recorded selectively. We developed two tech-niques to obtain such secondary mass spectra [17,19,20,22]: (i) by tuning the electric fields of the ion reflector and (ii) by post-accelerating the ions. The former offers optimum discrimination against decay product ions from other metastable


LASER TANDEM MASS SPECTROMETRY

LASER ION SPECTROSCOPY

MOLECULAR BEAM

LASER 1 ION REFLECTOR

LASER 2

DETECTOR

BUNCHES OF IONS WITH DIFFERENT MASS / J ^

LASER 3

LASER 1

SPACE FOCUS

QUALITY OF SPACE FOCUS

-Ü—L

ION SOURCE SPACE FOCUS

1

j M = 106

U - 6 5 nse

52 nsec I , ^ Λ^Μ=107 |

TIME OF FLIGHT-

Fig. 20. Combination of an ion source with space focus and a reflectron. Molecular or primary fragment ions are formed by laser 1, secondary mass selective excitation and dissociation is achieved by laser 3. With this arrangement tandem mass spectrometry may be performed. The mass selectivity of the excitation by laser 3 is determined by the laser focus size and the quality of the space focus (see bottom of Fig. and section 2.1). For laser spectroscopy of molecular ions a second tunable laser (laser 2) is used (for excitation scheme

see Fig. 19(a)).

precursor ions (a problem, which particularly has to be taken into account for metastable time-of-flight mass spectra). The latter has the advantage of delivering a secondary mass spectrum for every single laser excitation cycle.

The concept of post-acceleration is to raise the kinetic energy of all ions by a fixed amount of energy in order to diminish the relative differences AU/U of all ions. Thus, originally greatly reduced energies (due to a large mass loss during the decay in the field free region) which cannot be com-pensated by the reflectron anymore now fall into the energy region of correction [20]. As an example, toluene has mass 92, its fragment ion C+ mass 12 (Am/m = AU/U = 87%). If the

precursor ions have been formed with an energy of 500 eV, the secondary fragment ions C+ have a residual kinetic energy of 65 eV; this is far beyond the energy compensation capability of reflectrons. After post-acceleration by a potential of 2000 V, £/(C7H^)=2500eV, i /(C+)= 2065 eV; thus AU/U is reduced from 87% to 17% and can be compensated for.

Examples of secondary laser mass spectra for which post-acceleration has been used are shown in Fig. 21. In this figure a secondary mass spectrum of the toluene cation is shown with the wavelength of laser 3 being 266 nm as well as the secondary C6H+-fragments on an enlarged mass scale obtained for 266 and 355 nm of laser 3. The

U, Boesl et al,/Int, J, Mass Spectrom. Ion Processes 131 (1994) 87-124 119

secondary mass spectrum precursor C7HQ

λ3= 266 nm

λ = 266 nm

\~ = 355 nm C 6 H 2

C 6 H 5

Fig. 21. Secondary mass spectrum of primary toluene cations. On the extended mass scale the C6H+ fragment ion group for two wavelengths of laser 3 is shown. These secondary mass spectra have been obtained by the post-acceleration technique (see text).

C7H7" is a very alluding fragment ion in that it may have a benzyl, toluyl or tropylium structure: a question which is still puzzling modern mass spectrometry [88,89]. One fragment ion which has played an important role in the structural discussion is the Q H ^ ion. Laser reflectron mass spectrometry with post-acceleration is sensitive enough to deliver even secondary mass spectra of this weak fragment ion [20]. However, conditions of secondary excitation are easily varied changing wavelengths (Fig. 20), or photon density. The frag-mentation pattern of Fig. 20 (C6H+, 266 nm) is nearly the same as for collision-induced meta-stable decay of toluene at 5 keV collision energy [89]. For multiphoton dissociation of benzene cations, however, very similar patterns of C6H* fragments have been observed in our laboratory for 266 nm as well as for 355 nm. Further work is

necessary to illucidate the role of the C6H+ frag-ment pattern for structural analysis of the C7H7" ion and its precursors.

Secondary mass spectra may not only be due to secondary laser excitation but also to metastable decay induced in the primary laser excitation (laser 1 in Fig. 20). The behavior of metastable decay products within reflectron mass spectro-meters was detected in 1982 [19,20]. To obtain secondary mass spectra of metastable decay products formed in the field-free drift region (typically due to decay rates of l-25^s) the tech-niques mentioned above (reflector field tuning or post-acceleration) can be used as well as for secondary laser excitation. In Fig. 22, a primary multiphoton mass spectrum of xylene (nearly soft ionization) and two secondary mass spectra due to metastable decay of the toluene and the C7H7" cations are shown obtained by the reflector field scanning method. It is interesting to note that the C7H7" cations with low internal energy (metastable decay) show equally intense C5H5" and QH5" frag-ment ion peaks, but a much higher C5H5" ion peak at higher internal energies (see Fig. 18(b) in Ref. 20: secondary excitation with 266 nm or 4.7 eV). In addition, the C6H+ pattern in Fig. 20 resembles the pattern in Fig. 21 (355 nm). These observations give some hints as to which decay channels of C7H^ open up at increasing internal energies.

For some experiments with laser excitation in time-of-flight mass spectrometers, some means of mass filtering may be necessary in addition to mass discrimination techniques presented in sections 2, 4.1, 5.5 and 5.7 and this one. The reflector end plate (in particular the moveable one [83]) has already been mentioned as a high kinetic energy filter in section 5.7. If high abundant masses show cross talking on neighboring mass channels a filter-ing before secondary excitation with good mass selection is presented [22]. It consists of a mesh of linear thin wires with small distances (e.g. less than 1 mm) and alternating voltages (see Fig. 23) applied to them. The whole arrangement is positioned at or near the space focus of the ion source, the voltages are switched on during a HV pulse at the moment


precursor: C7H7

secondary

mass spectra

C5H5 C6H5

JL JL

C 0 H;

CH, C6H UL

CH

primary mass spectrum C 7 H ;

mass

Fig. 22. Primary (nearly soft) resonance enhanced multiphoton ionization mass spectrum of /?-xylene and secondary mass spectra due to metastable decay of the primary molecular ion and of its primary fragment C7H7". For these secondary mass spectra no laser 3 is used. The metastable decay in the field-free drift region is due to laser 1 excitation. Here, the reflector field tuning technique has been applied as an alternative to the post-acceleration technique.

when the mass to be filtered out passes. The small distance of the wires gives rise to a strong electric field between them even for small amplitudes of the HV pulse. However, in combination with the alter-nating voltages it results in a near-zero field shortly before and after the wire arrangement and thus a minimum distortion of other masses. The filtering of the abundent 3 2S+ ion for the sake of the rare 33 S+ ion by this arrangement is shown in Fig. 23(b). Secondary excitation by a laser now can take place near this mesh.

These experiments (e.g. Figs. 20-23) illustrate the wide range of new possibilities laser ion sources in time-of-flight mass analyzers allow for tandem mass spectrometry.

6.2. Laser spectroscopy

A further development of the two-laser exci-tation scheme (lasers 1 and 3 in Fig. 20) even allows a sophisticated way of performing laser spectroscopy on molecular ions (e.g. monofluoro-benzene (Fig. 19). As an on source, multiphoton ionization via the vibrationless neutral Sj inter-mediate state in a supersonic beam was used. Thus, the preparation of rotationally and vibrationally cold monofluorobenzene ions was possible [87]. As for the spectroscopy of the ion, in this experiment the spectroscopic excitation step and the photodissociation step have been separated by using two different lasers (lasers 2 and 3 in Fig. 20) delayed in time and shifted in space. The excitation scheme is shown in Fig. 19(a). This enables one to use a low-intensity laser pulse (laser 2) for the spectroscopy, avoiding spectral broaden-ing. However, a very intense laser pulse (laser 3) can be applied for multiphoton dissociation to effectively fragment and therefore detect the excited ions. To avoid spectroscopic structure due to laser 3, its wavelength has to be chosen in such a way that no absorption of photons of laser 3 from the ionic ground state is possible. For monofluoro-benzene cations this is easily achieved by the second harmonic of a YAG laser. Due to the spatial distance between the foci of lasers 2 and laser 3 it takes 4μ$ for the ions to arrive at the focus of laser 3 and the space focus of the ion source. This arrangement allows mass selective multiphoton dissociation and thus an elimination of background ions produced by lasers 1 or 2. The observed spectrum of the B <— X transition of the monofluorobenzene cation is shown in Fig. 19(b).

7. Conclusion

A variety of arrangements and their features of laser excitation for ion sources of time-of-flight mass analyzers is discussed in the previous sections. Possible methods for supplying neutrals are included as well as the many characteristics of


HV pulse time-of-flight [με] Fig. 23. (a) Trajectories of ions passing a mesh of linear wires with alternating polarity, (b) Mass spectra obtained with this mesh positioned at the space focus (wire spacing 0.5 mm; applied HV pulse 100 V, 5 ns); spectrum (HV off): sulfur fragment ions of OCS; spectrum (HV on): HV pulse switched on at the moment the ambient 3 2S+ isotope is passing the mesh (the ion current axis is enlarged by

a factor of 4).

resonant multiphoton ionization (e.g. state and species selectivity, features of fragmentation, effects reducing multiphoton ionization yields) and of non-resonant laser ionization (e.g. by laser induced VUV or electron emission). Much emphasis has been put on double excitation schemes either for confident comparison of two experiments (by performing them under nearly identical conditions) or for secondary excitation of primarily-formed neutrals or ions. Finally, the combination of ion source optics, space focus, laser ionization and secondary laser excitation was discussed as a sophisticated ion source with special features for tandem mass spectrometry, laser spectroscopy of molecular ions and other experiments. All arrangements discussed in the previous sections are explained with exper-iments performed in our laboratory. This presen-tation illustrates the great variety of new exper-imental possibilities laser excitation combined

with time-of-flight analysis supplies for mass spectrometry.

Acknowledgements

The authors gratefully acknowledge the many contributions of Mrs Carsten Baßmann and Ger-hard Drechster.

References

1 For a review see: V.S. Letokhov, Laser Photoionisation Spectroscopy, Academic Press, Orlando, FL, 1987.

2 (a) D.M. Lubman (Ed.), Lasers and Mass Spectrometry, Oxford University Press, New York, 1990. (b) D.M. Lubman, Mass Spectrom. Rev. 7 (1988) 535; 555.

3 U. Boesl, HJ. Neusser, E.W. Schlag and Z. Naturforsch. Teil A, 33 (1978) 1546; in H. Walther and K.W. Rothe (Eds.), Laser Spectroscopy, Springer Series in Optical Science, Vol. 21, Springer Verlag, Berlin, 1979, p. 164.

4 N.K. Berezhetskaya, G.V. Varonov, G.A. Delone, N.B. Delone and G.K. Pisskova, JETP, 31 (1970) 403.


5 S.L. Chin, Phys. Rev. A,4 (1971) 992. 6 A. Herrmann, S. Leutwyler, E. Schuhmacher and L. Wöste,

Chem. Phys. Lett., 52 (1977) 418. 7 L. Zandee, R. Bernstein and D.A. Lichtin, J. Chem. Phys.

69 (1978) 3427. 8 E.W. Rothe, B.P. Mathur and G.P. Reck, Chem. Phys.

Lett., 53 (1978) 74. 9 V.S. Antonov, I.N. Knyazev, V.S. Letokhov, V.M. Matiuk,

V.G. Movshev and V.K. Potapov, Opt. Lett., 3 (1978) 37. 10 U. Boesl, H.J Neusser and E.W. Schlag, Z. Naturforsch.

Teil A, 33 (1978) 1546. U. Boesl, H.J. Neusser and E.W. Schlag, in H. Walther, K.W. Rothe (Eds.), Laser Spectroscopy, Springer Series in Optical Science, Vol. 21, Springer Verlag, Berlin, 1979, p. 164.

11 First order compensation: S.G. Alikanov, Sov. Phys., JETP, 4 (1957) 452. Second-order compensation (reflectron): B.A. Mamyrin, Avt. svid. (USSR Patent), No. 198034, (1966). B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A. Zagulin, Sov. Phys. JETP, 37 (1973) 45.

12 U. Boesl, H.J. Neusser and E.W. Schlag, Chem. Phys. Lett., 87(1982) 1.

13 M.A. Johnson, M.L. Alexander and W.C. Lineberger, Chem. Phys. Lett., 112 (1984) 285.

14 K. LaiHing, P.Y. Cheng, T.G. Taylor, K.F. Willey, M. Peschke and M.A. Duncan, Anal. Chem., 61 (1989) 1460.

15 S. Della-Negra and Y. Le Beyec, Anal. Chem., 57 (1985) 2035.

16 T.J. Cornish and R.J. Cotter, Anal. Chem., 65 (1993) 1043. E.R. Williams, L. Fang and R.N. Zare, Int. J. Mass Spectrom. Ion Processes, 123 (1993) 233.

17 R. Weinkauf, K. Walter, C. Weickhardt, U. Boesl and E.W. Schlag, Z. Naturforsch. Teil A, 44 (1989) 1219. U. Boesl, R. Weinkauf, K. Walter, C. Weickhardt and E.W. Schlag, J. Phys. Chem., 94 (1990) 8567. U. Boesl, R. Weinkauf, K. Walter, C. Weickhardt and E.W. Schlag, Ber. Bunsenges. Phys. Chem., 94 (1990) 1357.

18 B. Spengler, D. Kirsch and R. Kaufmann, Rapid Commun. Mass Spectrom, 5 (1991) 198.

19 U. Boesl, R. Weinkauf, H.J. Neusser and E.W. Schlag, J. Phys. Chem., 86 (1982) 4857.

20 U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spectrom. Ion Processes, 112 (1992) 121.


22 U. Boesl, K. Walter, R. Weinkauf and E.W. Schlag, MS-MS-Time-of-Flight Mass Spectrometer, US-Patent Nr. 5,032,722, (1991) German Patent Nr. DE 3920566 C2, (1993).

23 It can be shown that for xD) unequal xD2, but *D = ^DI + *D2 = constant reflector potentials and flight times for optimal energy compensation stay constant: The important geometric factor is xD.

24 U. Boesl, H.J. Neusser and E.W. Schlag, Ion-source, US-Patent Nr. 4,433,241, (1984). German Patent, Nr. 2,942,386, (1983).

25 R.E. Smalley, D.H. Levy and L. Wharton, J. Chem. Phys., 64 (1976) 3266. D.H. Levy, Science, 214 (1982) 263.

26 H.L. Selzle and E.W. Schlag, Chem. Phys., 43 (1979) 111. 27 R. Pool, Science, 248 (1990) 1186.

O.F. Hagena, Rev. Sei. Instrum., 63 (1992) 2374. 28 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei.

Instrum., 60 (1989) 347; 792; 61 (1990) 2585; 2592. 29 For reviews: see H.R. Morris, Soft Ionization Biological

Mass Spectrometry, Heyden, London, 1980. A. Benninghoven, Ion Formation From Organic Solids, Springer Series in Chemical Physics, Vol. 25, Springer, Berlin, 1983. Ph. A. Lyon, Desorption Mass Spectro-metry, American Chemical Society, Washington, D.C., 1985.

30 A. Benninghoven and W. Sichtermann, Anal. Chem., 50 (1978) 1180.

31 M. Barber, R.S. Bordoli, G.J. Elliot, R.D. Sedgwick and A.N. Tyler, Anal. Chem., 54 (1982) 645A.

32 R.D. McFarlane and D.F. Torgerson, Int. J. Mass Spectrom. Ion Phys., 12 (1977) 461.

33 M.A. Posthumus, P.G. Kistemaker and H.L. Meuzelaar, Anal. Chem., 50 (1978) 985. K. Tanaka, H. Waki, Y. Ido, S. Anita, Y. Yoshida and T. Yoshida, Rapid Commun. Mass Spectrom., 2 (1988) 151. M. Karas and F. Hillenkamp, Anal. Chem., 60 (1989) 2301. M. Karas, U. Bahr, A. Ingendoh and F. Hillenkamp, Angew. Chem., Int. Ed. Engl., 28 (1989) 760. R.C. Beavis and B. Chait, Rapid Commun. Mass Spectrom, 3 (1989) 233; 432; 436. B. Spengler and R.J. Cotter, Anal. Chem., 62 (1991) 793. I.K. Perera, E. Uzcategui, P. Hâkansson, G. Brinkmalm, G. Pettersson, G. Johansson and B.U.R. Sundqvist, Rapid Commun. Mass Spectrom., 4 (1990) 285.

34 U. Boesl, J. Grotemeyer, K. Walter and E.W. Schlag, in J.F.J. Todd (Ed.), Advances in Mass Spectrometry, Wiley, New York, 1986, p. 985. J. Grotemeyer, U. Boesl, K. Walter and E.W. Schlag, Org. Mass Spectrom, 21 (1986) 645. J. Grotemeyer and E.W. Schlag, Acc. Chem. Res., 22 (1988) 399.

35 H. von Weyssenhoff, H.L. Selzle and E.W. Schlag, Z. Naturforsch. Teil A, 40 (1985) 674.

36 U. Boesl, J. Grotemeyer, K. Walter and E.W. Schlag, Anal. Instrum., 16(1987) 151.

37 D.M. Lubman and L. Li, Ref. 2(a), p. 353. M.S. de Vries, H.E. Hunziker and H.R. Wendt, Ref. 2(a), p. 383. T.R. Rizzo and D.H. Levy, Ref. 2(a), p. 402.

38 R.J.J.M. Steenvoorden, T.L. Weeding, P.G. Kistemaker and J.J. Boon, in K.G. Standing and W. Ens (Eds.), Methods and Mechanisms for Producing Ions from Large Molecules, NATO ASI Series, Plenum, New York, 1991, p. 315. D.M. Lubman, in K.G. Standing and D.M. Lubman (Eds.), Methods and Mechanisms for Producing Ions from Large Molecules, Plenum, New York, 1991, p. 297.


39 U. Buck and C. Lauenstein, J. Chem. Phys., 92 (1990) 4250.

40 For reviews see: P.M. Johnson, Appl. Optics, 19 (1980) 3920. S.H. Lin, Y. Fujimura, H.J. Neusser and E.W. Schlag, Multiphoton Spectroscopy of Molecules, Academic Press, Orlando, FL, 1984. V.S. Letokhov, Laser Photoionization Spectroscopy, Academic Press, Orlando, 1987.

41 U. Boesl, J. Phys. Chem., 95 (1991) 2949. 42 C. Weickhardt, R. Zimmermann, U. Boesl and E.W.

Schlag, Rapid Commun. Mass Spectrom., 7 (1993) 183. 43 D.M. Lubman, R. Tembreull and C.H. Sin, Anal. Chem.,

75(1985) 1084; 1186. T.Ch. Chang and M.V. Johnston, J. Phys. Chem., 91 (1987) 884. H. Kühlewind, H.J. Neusser and E.W. Schlag, J. Phys. Chem., 89 (1985) 5593 and 5600.

44 S.T. Fountain and D.M. Lubman, Anal. Chem., 65 (1993) 1257.

45 U. Boesl, H.J. Neusser and E.W. Schlag, J. Chem. Phys., 72 (1980) 4327.

46 W. Dietz, H.J. Neusser, U. Boesl and E.W. Schlag, Chem. Phys., 66 (1982) 105.

47 A. Gedanken, M.B. Robin and N.A. Kuebler, J. Phys. Chem., 86(1982)4096. T. Baer, Com. At. Mol. Phys., 13 (1983) 141.

48 M.A. Duncan, T.G. Dietz and R.E. Smalley, Chem. Phys., 44(1979)415. S. Leutwyler, U. Even and J. Jortner, Chem. Phys., 58 (1981)409. D.P. Gerity, L.J. Rothberg and V. Vaida, Chem. Phys. Lett., 74 (1980) 1. G.J. Fisanick, A. Gedanken, T. Eichelberger, N.A. Kuebler and M.B. Robin, J. Chem. Phys., 75 (1981) 5215. J.B. Morris and M.V. Johnston, J. Phys. Chem., 89 (1985) 5399. Y. Nayano, Y. Achiba and K. Kimura, J. Phys. Chem., 90 (1986) 1288. M. Stuke, Appl. Phys. B,38 (1985) 209. J. Dannon, H. Zacharias, H. Rottke and K.H. Welge, J. Chem Phys., 76 (1981) 2399.

49 R.S. Pandolfi, D.A. Gobeli and M.A. El-Sayed, J. Phys. Chem., 85(1982) 1779.

50 E. Sekreta, K.G. Owens and J.P. Reilly, Chem. Phys. Lett., 132(1986)450.

51 R.L. Whetten, K.J. Fu. R.S. Tapper and E.R. Grant, J. Phys. Chem., 87 (1983) 1484. P. Chen, J.B. Pallix, W.A. Chupka and S.D. Colson, J. Chem. Phys., 84 (1986) 527. A. Bolovinos, S. Spyrou and A.C. Cefalas, J. Chem. Phys., 85 (1986) 2335.

52 For a review see: K. Kimura, Int. Rev. Phys. Chem., 6 (1987) 195.

53 J.T. Meek, S.R. Long and J.P. Reilly, J. Phys. Chem., 86 (1982) 2809.

54 J.H. Glownia, S.J. Riley, S.D. Colson, J.C. Miller and R.N. Compton, J. Chem. Phys., 77 (1982) 68.

W.A. Conaway, R.J. Morrison and R.N. Zare, Chem. Phys. Lett., 113(1985)429.

55 W.A. Chupka, S.D. Colson, M.S. Seaver and A.M. Wood-ward, Chem. Phys. Lett., 95 (1983) 1237. A.M. Woodward, S.D. Colson, W.A. Chupka and M.G. White, J. Phys. Chem., 90 (1986) 274.

56 B. Yang, M.H. Eslami and S.L. Anderson, J. Phys. Chem., 89 (1988) 5527. R. Weinkauf and U. Boesl, J. Chem. Phys., 98 (1993) 4459.

57 S.R. Long, J.T. Meek and J.P. Reilly, J. Chem. Phys., 79 (1983) 3206.

58 X. Ripoche, I. Dimicoli, J. Lecalvé, F. Biuzzi and R. Botter, Chem. Phys., 88 (1988) 305. K. Walter, K. Scherm and U. Boesl, J. Phys. Chem., 95 (1991) 1188.

59 J.T. Meek, E. Sekreta, W. Wilson, K.S. Wiswanathan and J.P. Reilly, J. Chem. Phys. 82 (1985) 1741.

60 E.W. Schlag and R.D. Levine, J. Phys. Chem., 96 (1992) 10608.

61 C.H. Becker, Ref. 2(a), p. 84. C.H. Becker, Ref. 38, p. 293.

62 C.R. Vidal in L.F. Mollenauer and J.C. White, (Eds), Topics in Applied Physics, Vol. 59, Springer, Berlin 1987, p. 57. R. Hilbig, G. Hilber, A. Lago, B. Wolff and R. Wallenstein, Com. At. Mol. Phys., 18 (1986) 157. G. Hilber, A. Lago and R. Wallenstein, J. Opt. Soc. Am. B,4 (1987) 1753.

63 K. Rademann, B. Kaiser, U. Even and F. Hensel, Phys. Rev. Lett., 59(1987)2319. U. Schuhle, J.B. Pallix and C.H. Becker, J. Am. Chem. Soc, 110(1988)2323. J. Kanofsky and D. Gutman, Chem. Phys. Lett., 15 (1972) 236. J. Boyle, L. Pfefferle, J. LoBue and S. Colson, Combust. Sei. Technol., 70 (1990) 187.

64 C. Bäßmann, C. Weickhardt and U. Boesl, unpublished results.

65 O. Botstanjoglo and F. Heinricht, J. Phys. E, 20 (1987) 1491. E.R. Rohwer, R.C. Beavis, C. Köster, J. Lindner, J. Grotemeyer and E.W. Schlag, Z. Naturforsch. Teil A, 43 (1988)1151. J.G. Boyle, L.D. Pfefferle, E.E. Gulcicek and S.D, Colson, Rev. Sei. Instrum., 62 (1991) 323.

66 C. Weickhardt, U. Boesl and E.W. Schlag, Anal. Chem., submitted. U. Boesl, C. Weickhardt, R. Zimmermann, S. Schmidt and H. Nagel, SAE Technical Paper Series 930083, (1993) 61.

67 M.K. Gilles, W.C. Lineberger and K.M. Erwin, J. Am Chem. Soc., 115(1993) 1031. D.G. Leopold, K.K. Murray, A.E. Stevens Miller and W.C. Lineberger, J. Chem. Phys., 83 (1985) 4849. D.W. Arnold, S.E. Bradforth, Eun Ha Kim and D.M. Neumark, J. Chem. Phys., 97 (1992) 9468. A. Weaver, R.B. Metz, S.E. Bradforth and D.M. Neumark, J. Phys. Chem., 92 (1988) 1031.


G.F. Gantefor, D.M. Cox and A. Kaldor, J. Chem. Phys., 97 (1992) 8990. G.H. Lee, S.T. Arnold, J.G. Eaton, H.W. Sarkas, K.H. Bowen, C. Ludewigt and H. Haberland, Z. Phys. D, 20 (1991) 9. For reviews see: J.P. Maier (Ed.), Ion and Cluster Ion Spectroscopy and Structure, Elsevier, New York, 1989. R.D. Mead, A.E. Stevens and W.C. Lineberger, in M.T. Bowers (Ed.), Gas Phase Ion Chemistry, Academic Press, New York, 1984 p. 214.

68 G. Dreschsler, C. Bäßmann and U. Boesl, unpublished results.


70 R. Grix, R. Kutscher, G. Li, U. Grüner and H. Wollnik, Rapid Commun. Mass Spectrom., 2 (1988) 83.

71 K. Müller-Dethlefs and E.W. Schlag, Ann. Rev. Phys. Chem., 42 (1991) 109. T. Baer, W.B. Peatman and E.W. Schlag, Chem. Phys. Lett., 4 (1969) 243.

72 L. Zhu and P. Johnson, J. Chem. Phys., 94 (1991) 5769. H. Krause and HJ. Neusser, J. Chem. Phys., 97 (1992) 5923.

73 T. Baer, in M.T. Bowers (Ed.), Gas Phase Ion Chemistry Vol. 1, Academic Press, New York, 1979, p. 153.

74 M.R. Topp, Int. Rev. Phys. Chem., 12 (1993) 149. W. Scherzer, O. Krätschmer, H.L. Selzle and E.W. Schlag, Z. Naturforsch. Teil A, 47 (1992) 1248.

75 R. Weinkauf, P. Aicher, G. Weseley, J. Grotemeyer and E.W. Schlag, J. Phys. Chem., submitted.

76 C. Weickhardt, H. Nagel and U. Boesl, unpublished results. 77 U. Boesl, C. Weickhardt, S. Schmidt, H. Nagel and E.W.

Schlag, Rev. Sei. Instrum., submitted. 78 For a review see: K.L. Bush, G.L. Glish and S.A. Luckey,

Mass spectrometry/Mass Spectrometry, VCH, Weinheim, 1988.

79 A.S. Werner and T. Baer, J. Chem. Phys. 62 (1975) 2900. T. Baer, Adv. Chem. Phys., 64 (1986) 111. J.A. Booze and T. Baer, J. Chem. Phys., 98 (1993) 186. D. Proch, D.M. Rider and R.N. Zare, Chem. Phys. Lett., 81 (1981)430. J.L. Durant, D.M. Rider, S.L. Anderson, D. Proch and R.N. Zare, J. Chem. Phys., 80 (1984) 1817. H.J. Neusser, H. Kühlewind, U. Boesl and E.W. Schlag, Ber. Bunsen-Ges. Phys. Chem., 89 (1985) 276.

H. Kühlewind, H.J. Neusser and E.W. Schlag, J. Phys. Chem., 88 (1984) 6104. A. Kiermeier, H. Kühlewind, H.J. Neusser, E.W. Schlag and S.H. Lin, J. Chem. Phys., 88 (1988) 6182. H.J. Neusser, J. Phys. Chem., 93 (1989) 3897.

80 P. Erman, M. Larsson, Phys. Scr., 23 (1981) 1052. 81 D. Klappstein and J.P. Maier, J. Chem. Phys., 83 (1981)

590. 82 R. Weinkauf, K. Walter, U. Boesl and E.W. Schlag, Chem.

Phys. Lett., 141 (1987) 267. K. Walter, R. Weinkauf, U. Boesl and E.W. Schlag, J. Chem. Phys., 89 (1988) 1914.

83 U. Boesl, K. Walter, R. Weinkauf and E.W. Schlag, Reflectron time-of-flight mass spectrometer with movable reflector end plate, application for German Patent, P 3943688.8-33, (1993).

84 R. Kakoschke, U. Boesl, J. Herrmann and E.W. Schlag, Chem. Phys. Lett., 119 (1985) 6. A.M. Woodward, S.D. Colson and W.A. Chupka, J. Phys. Chem., 90 (1986) 274. E.J. Bieske, M.W. Rainbird, I.M. Atkinson and A.E.W. Knight, J. Chem. Phys., 90 (1989) 752. R. Weinkauf and U. Boesl in G.S. Hurst and N. Omenetto (Eds.), Resonance Ionization Spectroscopy 1990, Institute of Physics, Bristol, 1990.

85 For reviews see: J.T. Moseley, Adv. Chem. Phys. 40 (1985) 245. J.A. Syage and J.E. Wessel, J. Appl. Spectrosc. Rev., 24 (1988) 1.

86 K. Walter, R. Weinkauf, U. Boesl and E.W. Schlag, Chem. Phys. Lett., 155 (1989) 8. K. Walter, U. Boesl and E.W. Schlag, Chem. Phys. Lett., 162(1989)261.

87 P.O. Danis, T. Wyttenbach and J.P. Maier, J. Chem. Phys., 88(1988)3451. X. Ripoche, I. Dimicoli, J. Lecalvé, F. Piuzzi and R. Botter, Chem. Phys., 124 (1988) 305. Y. Tsuchiya, M. Fujii and M. Ito, J. Chem. Phys., 89 (1989) 6965.

88 L.M. Bass and M.T. Bowers, Org. Mass Spectrom., 17 (1982) 229.

89 C.I. Proctor and F.W. McLafferty, Org. Mass Spectrom., 18 (1983) 193.


Photoemission electron impact ionization in time-of-flight mass spectrometry: an examination of experimental consequences

Steven M. Colby, James P. Reilly* Department of Chemistry, Indiana University, Bloomington, IN 47405, USA

(Received 24 February 1993; accepted 24 June 1993)

Abstract

The use of photoemission electron impact ionization in time-of-flight mass spectrometry is examined. It is found to complement laser ionization by providing many of the advantages of electron impact ionization. Some ions are more easily produced by photoemission electron impact ionization. Under certain conditions, these ions can act as an impedi-ment to laser ionization experiments. Methods for distinguishing between ions produced by laser ionization and photo-emission electron impact ionization are presented. A new source, designed for simple conversion between these two ionization methods, is presented. This source is easily adapted to the many laser ionization time-of-flight instruments already in use.

Key words: Electron impact ionization; Photoemission electron impact ionization; Laser ionization

1. Introduction

Time-of-flight mass spectrometry (TOF-MS) is finding an increasingly wide variety of appli-cations. Its sensitivity, simplicity, and unlimited mass range make it one of the most versatile mass spectrometric techniques. The popularity of the method has accelerated in the past decade with the development of high power, short pulse lasers. These provide TOF-MS with a sensitive and selective ionization source. Electron impact ionization is now rarely used in TOF-MS. It is, however, the most common ionization method in other types of mass spectrometers [1,2]. Electron impact ionization sources are usually continuous. For use with TOF-MS an electric field must be pulsed to establish the start of the time-of-flight.


Recently, however, Rohwer et al. introduced a new pulsed electron impact ionization method. Their approach, which we call photoemission electron impact ionization, uses a pulse of photo-generated electrons for inducing ionization [3]. The electrons are produced by directing a laser pulse at a wire within the source of a time-of-flight instru-ment. The resulting short pulse of photoelectrons is accelerated across the source to an energy great enough to cause ionization. This method provides the advantages of electron impact ionization to the large number of time-of-flight instruments that currently use laser ionization. These advantages include the abilities to ionize all types of com-pounds, to produce mass spectra whose character-istics do not strongly depend on instrumental parameters, and to use the large database of elec-tron impact reference spectra currently available.

Since its introduction, the photoemission

SSDI0168-1176(93)03 878-P

126 S.M. Colby and J.P. Reillyjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 125-138

electron impact ionization method has been applied only sparingly [4,5]. This paper expands on the initial work by exploring advantages of the technique and pointing out how the phenomenon can affect conventional laser ionization experi-ments. We start by showing that photoemission electron impact ionization can sometimes be much more efficient than laser ionization. We then demonstrate that, because of this, it also can cause severe interference in laser ionization time-of-flight experiments. Methods are suggested for detecting this interference. Finally, we describe a new ion source designed for both photoemission electron impact ionization and laser ionization. This source is easily adaptable to the many laser ionization time-of-flight instruments already in use and has some distinct advantages over the sources proposed to date. The availability of a dual, easily interconvertible source will further expand the range of applications to which TOF-MS is suited.

Laser ionization and electron impact ionization are complementary techniques. The advantages of laser ionization are that it provides good temporal and spatial resolution, controlled fragmentation, and high spectrometric selectivity. There are also several drawbacks to laser ionization, each of which can be overcome by electron impact ion-ization. The first is that the high selectivity of laser ionization makes it inappropriate for the analysis of unknowns. Some analytes may be diffi-cult to laser ionize because they require the energy of many photons, do not absorb at convenient wavelengths, or fragment before ionizing. Ion yields and fragmentation patterns of different neutrals may vary widely under the same set of experimental conditions. The second drawback to laser ionization is that the exact conditions under which experiments are performed strongly affect the observed mass spectra. This makes it virtually impossible to form a large database of reference spectra.

Laser induced photoemission has been used as a source of electrons in several applications other than TOF-MS [6-23]. These include microwave generators [10-12], particle accelerators [13,14],

and free electron lasers [14-18]. All require very short electron pulses with narrow kinetic energy and angular distributions. Research in these areas involves examining the effects of various wave-lengths, laser pulse widths, and types of irradiated targets. The latter include conductors, semiconduc-tors, and organic materials. One particularly imaginative use of laser induced photoemission is the measurement of potentials in ultra-high-speed electronic circuits. This is difficult because any probe making contact with a wire would introduce a perturbation to the circuit. However, since the number of electrons produced in photoelectric emission depends on the field above a surface, a laser beam directed on a wire can be used as a non-intrusive probe of its potential. This has been demonstrated experimentally on a picosecond time scale [19,20].

The mechanism of the laser induced photo-emission process is primarily based on the photo-electric effect. However, thermal emission also may occur if the electron source is significantly heated by either absorption of energy from the laser light or through other means. The density of electrons emitted thermally depends on the temperature through the Richardson-Dushman equation: J = AT2 exp(-<E>/A;r) where / is the current den-sity in A cm- 2, A is a constant, and Φ is the work function of the emitter [24]. Thermal emission is not desirable in applications requiring short pulses of electrons, because it may take place over a time longer than the temporal duration of the laser pulse [25]. Another physical phenomenon that can influ-ence the photoemission process is the Schottky effect, the reduction of a work function due to the presence of an external field [24,26,27]. It lowers both the tunneling barrier controlling the rate of thermal emission and the energy needed for electron emission. The Schottky effect becomes important as the local field approaches tenths of a millivolt per Angström [10].

For some ionization processes photoemission electron impact ionization can be much more effective than direct laser ionization. For example, through this process it is possible for a single

S.M. Colby andJ.P. Reilly/Int. J. Mass Spectrom. Ion Processes 131 (1994) 125-138 127

near-UV photon to produce ions whose appear-ance potentials are tens of electron volts. This can occur because photoemitted electrons, which require only a few electron volts to produce, can easily be accelerated to hundreds of electron volts. Under some conditions, the conversion of photons into electrons may even have an efficiency greater than unity [10, 28]. The relative efficiency of the two ionization processes is a function of photon flux, electron production rate, ionization cross-sections, and energetics. Photoionization and elec-tron impact cross-sections tend to be similar (i.e. in the realm of 10~16 to 10~18 cm2) but laser fluence is generally much higher than electron fluence. Con-sequently, resonantly enhanced laser ionization can be significantly more sensitive. However, helium, argon, and other substances whose high lying excited states are beyond the range of powerful tunable lasers can be more effec-tively ionized by electron impact. This is particu-larly true for the production of multiply charged ions.

An important difference between photoemission electron and conventional electron impact ion-ization is the magnitude of the electron currents involved. Conventional filament sources typically work in the 100/xA to 1mA range [1]. However, if a 100/xJ laser pulse were converted into elec-trons during 10 ns, with a conversion efficiency of 1 %, then a 16 A electron pulse would be produced. Such high currents would exist only during the short period of the laser pulse. Although the time integrated number of electrons may be similar to that of a filament source, short bursts of high elec-tron current may yield more extensive fragmenta-tion or even multiple ionization because a molecule or fragment may be hit by more than one electron. In our experiments most of the laser light passes through the instrument and we produce only a fraction of the maximum possible electron current.

We obtained the data presented in this paper with a combined gas chromatograph-time-of-flight mass spectrometer (GC-TOF-MS). Our experiments were originally directed toward developing new techniques for the ultrasensitive

detection of organometallics [29,30]. However, we found that photoemission electron impact ion-ization was the dominant ionization process under the conditions of some of our experiments. We have therefore, to an extent, focused on how photoemission electron impact ionization affects GC-TOF-MS. Our choice of reagents was also determined by our original goals. They served well, however, for illustrating the photoemission electron impact ionization process and its effects.

2. Experimental

2.1. Reagents

Tetraethyltin was purchased from Aldrich Chemical Company. Samples were prepared by dilution in "spectranalyzed" méthylène chloride from Fisher Scientific. Final concentration was lOOngmL-1 of tetraethyltin. Zero Grade helium and argon were purchased from Air Products. They were passed through a Varian filter before entering the gas Chromatograph at a flow rate of « 0 . 6 m L m i n - 1 Λ-butylbenzene, used to demon-strate the dual source, was purchased from Aldrich Chemical Company and introduced with-out dilution.

2.2. Instrumentation

Figure 1 shows the three TOF-MS sources used in this experiment. Source A had two acceleration regions adjusted to Wiley-McLaren space focusing conditions [31]. Source B was shaped like a cup in an attempt to partially confine the Chroma-tographie effluent and thereby increase instrument sensitivity. The cup had holes to allow the laser beam to pass through and the chromatography column to enter. It will be described further else-where [29]. Both of these sources were designed for laser ionization but photoemission electron impact ionization can occur in them when electrons are produced at the first (nickel) grid and then acceler-ated across the source to the bottom plate or cup.

128 S.M. Colby andJ.P. Reilly/Int. J. Mass Spectrom. Ion Processes 131 (1994) 125-138

B C

G2 — Gl - . G0. ■ hv

G 2 ~ G l ~ G 0 - H

o /o-

s CI

o HB

G 3 ^ -G2—-Gl —

G 0 - · hv

4 Fig. 1. The three instrument sources: (A) standard; (B) cup; (C) dual ionization. CI, column inlet; HB, heating block; Gj_3, grid holders;

G0, solid plate or cup.

Source C was specifically designed so that it could be used for either laser or photoemission electron impact ionization. This allows the two complemen-tary techniques to be used interchangeably without modifying the instrument. Source C was similar to A except that an additional grid, G3, was added to enable the use of both modes. For gas phase laser ionization, G2 is grounded and the source func-tions just as it does in configuration A with ion-ization taking place between G0 and Gj . For photoemission electron impact ionization the laser is directed at the plate G0 of source C. The field between G0 and G! accelerates the resulting photo-electrons toward and through G,. Electron impact ionization then takes place just above Gj. The dis-tance the electron can travel beyond Gj depends on the electric fields in the acceleration regions. The fields between Gi and G2, and G2 and G3 are adjusted to space focusing conditions. Figure 2 shows the trajectory of the photoemitted electron.

The cup and the plate G0 were manufactured from aluminum while the grid holders Gj through G3 were stainless steel. The grids were 90% trans-mitting nickel mesh (Buckbee Meers). The work functions of aluminum and nickel are 4.28 and 5.15eV respectively [32]. For some experiments, 8.4 cm long deflection plates were mounted parallel to the flight tube axis just above the source (not shown in Fig. 1.) Table 1 lists the voltages and

dimensions used for different instrument con-figurations. The remainder of this paper will refer to these as A, B, C, and C' as defined here and in Fig. 1.

For tetraethyltin analysis chromatography was performed using a Varian 3700 Gas Chromato-graph with on-column injection and a 30 m long, 250/zm i.d. fused silica capillary column (Supelco SPB-1). The Chromatographie temperature pro-gram ranged from 40 to 200°C. The interface between the gas Chromatograph and the TOF-MS, was held at 250°C and has been previously described [33]. When in use, the cup was also heated to approximately 270°C. The end of the capillary column was positioned within the first acceleration region. The mass spectrometer was evacuated through a liquid nitrogen trap by a Varian VHS-6 diffusion pump. The base pressures were approximately 1.0 x 10~4 and 4.6 x 10~5

Torr with the helium or argon carrier gas flowing and approximately 2.0 x 10~6Torr without carrier

Fig. 2. Schematic of photoemission electron impact ionization process.

S.M. Colby and J.P. Reilly/Int. J. Mass Spectrom. Ion Processes 131 (1994) 125-138 129

Table 1 Instrument configurations

Distances (mm) G0-Gt

G,-G2

G2-G3

G2-MCP G3-MCP

Voltages (V) Go G, G2

G3

Configuration

Aa

12.52 4.27 -

626.6 -

2096.82 1828.82 Ground

-

Bb

12.60 8.56 -

632.5 -

2092.60 1834.47 Ground

-

Cc

10.13 9.39 9.29 -

623.6

2119.38 1880.62 Ground Ground

c'd

10.13 9.39 9.29 -

623.6

1907.76 2007.76 1820.84 Ground

a Standard source. b Cup shaped source. c Dual source with photoionization. d Compare with Fig. 1.

gas flow. Ions were detected by a pair of tan-dem microchannel plates (Galileo). Their out-put was amplified (LeCroy VV101ATB) and then digitized by a high-speed waveform recorder (Biomation 6500). The data were transferred to a microcomputer where software was used to record mass spectra and integrate the ion signal over the isotopic mass envelope of tin.

The light source used for these experiments was a Lumonics TE-861 ArF excimer laser with a pulse duration of about 8 ns and a repetition rate of 20 Hz. It has previously been shown that this 193 nm wavelength can produce tin ions [34] and induce the photoemission of electrons [28]. The laser beam was apertured and then focused into the source of the mass spectrometer using a 1000 mm focal length lens. The total distance from the laser to the instrument was approxi-mately 2 m. The rectangular focal spot produced by this lens was 4.5 mm wide and 0.6 mm high. This was measured by passing a sharp edged blade through the focal plane and monitoring the transmitted light intensity. The energy of the laser beam was often adjusted by either changing the laser discharge voltage or the size of the aperture. The laser energy also declined as the laser gas mixture matured. The lens was positioned so that

the narrow edge of the focal spot intercepted the Chromatographie effluent approximately 1.0 mm from the tip of the Chromatographie column. The laser energy was measured before the instrument with a Gentec ED-500 joulemeter or a Laser Precision Rm-660 Universal Radiometer.

In this experiment the energy of the ionizing electron depended on the point of ionization. In sources A and B this point is not well defined because the electrons are produced at the first grid and then continuously accelerated across the source to the bottom plate or cup. Once the elec-tron has sufficient energy it can induce ionization at any position. However, since the average energy of the electrons during their acceleration is equal to half of their final energy, we can assume that the typical electron/molecule collision occurs when the electron has half of its total possible energy. Under conditions A and B this corresponds to an energy of approximately 130eV. Since this is above the energies used in conventional electron impact ionization, our instrument may produce more fragmentation.

When using source C' the electron energy is still dependent on the point of ionization. How-ever, the range of possible ionization positions can now be controlled by varying the potential on G0. As shown in Fig. 2, electrons decelerate as they enter the region between Gj and G2. For all of the G0 potentials listed in Table 1 their energy is low enough for them to be turned back toward G0. Ions formed between G\ and G2 are acceler-ated toward the detector while those that are produced between G0 and G\ are accelerated toward G0 and cannot be detected. It is therefore possible to limit the production of detectable ions to the lower part of the region between G\ and G2

by limiting the potential drop between G0 and G^

3. Results and discussion

5.7. Efficiency of photoemission electron impact ionization

Photoemission electron impact ionization is able

130 S.M. Colby andJ.P. Reillyjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 125-138

A r 4 + A r 3 + Ar2+ Ar+

x10

10 20 30

Mass/Charge

40 50

Fig. 3. Mass spectrum showing multiply charged argon ions. (Other noteworthy peaks are m/z 14, 16, 28, and 32, corresponding to N4

0 + , N j , and O j respectively.)

to produce some ions more efficiently than laser ionization. We have done two experiments to illustrate this. They demonstrate the formation of multiply charged ions and the uncharacteristic production of ions at extremely low light levels.

3.7.7. Multicharged argon Photoemission electron impact ionization's

ability to produce ions with high appearance potentials is illustrated by the degree to which it can generate multiply charged argon. Figure 3 shows a mass spectrum recorded when argon was used as the GC carrier gas. The data represent the summation of 200 laser shots, with a typical intensity of 32 MW cm"2, taken with mass spectro-meter configuration A. Argon ions with mass-to-charge ratios of 40 (Ar+), 20 (Ar2+), 13.33 (Ar3+), and 10 (Ar4+), along with some ions attri-butable to air, are seen. (The ion with a mass-to-charge ratio of eight could be 0 2 + and therefore cannot positively be identified as Ar5+.) The data were obtained using the most sensitive range of the waveform recorder. The largest peaks are therefore off scale. Spectra taken on less sensitive ranges displayed these peaks in their entirety. Table 2 contains data relevant to the ionization of argon. This includes the energy required to produce multi-

ply charged argon ions [35], the number of 193 nanometer photons that would be required if the ions were produced by photoionization, and the electron impact cross-sections for the production of the ions [2,36-39]. Clearly, photoemission elec-tron impact ionization is able to produce ions with very high appearance potentials. To produce these by laser multiphoton ionization would require an extremely intense light source. It has previously been shown that, even with « 1014W cm of 193nm light, argon is not ionized beyond the 2 + state [40]. Table 3 lists the relative peak areas in our mass spectra. These are compared with elec-tron impact cross-sections from Table 2, normal-ized to the cross-section for producing Ar3+. Their remarkable similarity is further evidence that most of these ions are produced via photo-emission electron impact ionization. Conventional 70 eV electron impact ionization produces only Ar+ and Ar2+ in a ratio of approximately 1 to 0.13 [41].

3.1.2. Signal at ultra low light levels A second illustration of the efficiency of photo-

emission electron impact ionization is the very low amount of light needed to produce ions. We explored this by using very small amounts of

S.M. Colby andJ.P. Reillyjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 125-138 131

Table 2

Ionization, energetic and cross-section data

Ion Appearance potential (eV)a Photons Mechanism Cross-section (m )

He+

He 2 +

Ar+

Ar2 +

Ar3 +

Ar4 +

24.587

54.416

15.759

27.629

40.74

59.81

4

9

3

5

7

10 Ar^

Ar3

Ar4

He + e -

He + e -

Ar + e -

Ar + e -

Ar + e -

Ar + e -

Ar + + e ■

Ar 2 + + e

Ar3 +

H e + + 2e

He 2 + + 3e

Ar + + 2e

Ar2 + + 3e

Ar 3 + + 4e

Ar4 + + 5e

-+ Ar 2 + + 2e

-» Ar3 + + 2e

+ e -► Ar 4 + + 2e

3.7 x 10"21

1.4 x 10~24

2.5 x 10- 2 0

1.6 x 10-21

8 x 10"24<

9 x 10"21

3 x 10"21

1.6 x 10-21

a From ref. 35. b From ref. 2, for lOOeV electrons. c 4 x 10~2 3at 150eV.

incident light. Instead of sending the laser beam into the instrument, we directed it to a point on the exterior of the vacuum chamber approxi-mately 8 cm from the 19 mm diameter input window. The only light that entered our mass spectrometer was a small amount scattered from the primary beam as it passed through the one meter lens 90 cm away from the instrument. Figure 4 shows a mass spectrum of helium carrier gas taken under these conditions. It was recorded with 4000 laser shots and mass spectrometer con-figuration B. The spectrum shows a strong He+ ion signal. To produce this by multiphoton ionization would require the absorption of four ArF laser photons; however, the incident light was so weak that it was not possible to measure it without a calibrated photomultiplier. We conclude that the few 193 nm photons that do enter the instrument generate electrons at an interior surface and the helium ions are thereby produced by photo-Table 3

Ionization data

Ion

Ar+

Ar 2 +

Ar3 +

Ar4 +

Measured

peak area3

676.6 64.4

1 0.27

Relative cross-sectionb

625 40

1 -

a From Fig. 3. b From Table 2.

emission electron impact ionization. Since this depends linearly on light intensity, it is possible to observe ions at even very low light levels.

3.2. Photoemission electron impact ionization as a cause of interference in laser ionization TOF-MS

The additional ions produced by photoemission electron impact ionization can act as an impedi-ment to the detection of ions produced by photo-ionization. This occurs through two processes: space charge effects and detector saturation. We performed experiments to observe both of these.

3.2.1. Detector saturation Following the arrival of a large current of ions,

microchannel plate response becomes nonlinear. It is possible for ions produced by photoemission electron impact ionization to cause this effect. For example, Fig. 5 shows the relative He* signal as a function of laser energy for instrument con-figuration A. Different sets of points represent data taken at different times or on different days. The He+ ion yield is not exactly reproducible because any variation in the laser beam profile changes the amount of light scattered off the tip of the chroma-tography column. Figure 5 shows that the He+

signal increases linearly with laser energy and then levels off as saturation begins.

The very large He+ signal reduces the sensitivity and linearity of the microchannel plates for many

132 S.M. Colby andJ.P. Reillyjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 125-138

10 20 30

Mass/Charge

40 50

Fig. 4. Mass spectrum obtained at very low light levels. The principal peak is He4

microseconds. Figure 6 shows three mass spectra of Sn+ ions from tetraethyltin taken with instrument configuration A. These ions arrive approximately 10//s after the He+ signal. The He+ photoemission electron impact ionization signal (not shown) was much stronger in the second spectrum than in the first. The isotope ratio in Fig. 6(A) displays the expected distribution. This includes the stable iso-topes at 112u (1.01%), 116u (14.8%), 117u (7.75%), 118u (24.3%), 120 u (32.4%), 122 u (4.56%), and 124 u (5.64%) [42]. In contrast, Fig. 6(B), recorded with 25 mJ of light, shows a dis-torted isotope distribution. The microchannel plate saturation effect was so strong that under

CO

O

o > σ CD

et:

2500 r

2250 V

2000 V

1750 h L

1500 V

1250 V

1000 f-

750 V

500 f-

250 f-

o L

Ü D

D° D Ü

some conditions the Sn+ signal actually increased as the laser light intensity and He+ peak decreased. An example of this is shown in Fig. 6(C) which was recorded with 9.1 mJ of light. Although this is about one-third of the fluence used to obtain the spectrum in Fig. 6(B), the signal intensity is clearly stronger.

It is possible to reduce the detector saturation by deflecting the carrier gas ions out from their normal flight path. We did this by placing a + 1000 V potential on one of the deflection plates and then dropping the potential to 0 V after the carrier gas ions had passed and before the ions of interest entered the deflection region. Deflection He+ increases the Chromatographie signal from a

o> o %

o x Δ

0 O OX Δ

χ θ

10 15 20

Laser Energy (mJ)

25 30

Fig. 5. Relative He+ signal as a function of laser energy for instrumental configuration A. Different symbols correspond to data taken at different times.


B

c À I . I . I i 1 1 1

90 100 110 120 130 Mass/Charge

Fig. 6. Tetraethyltin mass spectra showing tin isotope distribution as indicator of nonlinearity: (A), the normal isotope distribution; (B), the isotope distribution when He+ ions are saturating the detector; (C), when the laser power is reduced by two-thirds. Only (B) and (C)

are plotted on the same vertical scale.

100 pg sample of tetraethyltin by a factor of 20. The effect is even greater for smaller samples whose signal is unobservable without deflection.

3.2.2. Space Charge Effects Space charge effects are also exacerbated by

photoemission electron impact ionization. Excess space charge reduces both the sensitivity and the resolution of the instrument. Sensitivity is reduced when analyte ions are pushed radially away from the path that would take them to the detector. This problem can be reduced by confining the ions to a narrow tube [29] or placing a wire with a negative potential down the length of the flight tube [43,44]. It is not possible to eliminate space charge effects with deflectors because they are strongest before the ions are separated. Resolution is impaired because space charge increases the width of the ion's kinetic energy distribution; conditions that maximize carrier gas signal eliminate the structure in later parts of the mass spectrum. For example, there are usually several background peaks arising from pump oil. However, when the helium or argon signal is extremely high the pump oil signal remains strong but appears extremely broadened and unrecognizably distributed.

4. Methods for distinguishing photoemission electron impact ionization from photoionization

Photoemission electron impact ionization can clearly have serious consequences for experiments that depend on the wavelength selectivity of laser ionization. It is, therefore, important to be able to detect its presence. We have developed two methods for identifying the ions produced via this process.

The first method relies on varying the electric field in the ionization region. As the potential of G0 approaches that of G] the maximum kinetic energy of the photoemitted electrons is lowered. When this energy drops below the appearance potential of an ion, the ion's signal disappears. Figure 7 shows this for He2+ and He+. The He2+

ion signal is the first to disappear. As expected, it does so around the point where the maximum electron energy is near the ion's 54.4 eV appearance potential. These data were taken with instrumental conditions B and 0.44 mJ of light.

The second method for identifying photo-emission electron impact ionization is to exchange the potentials on G0 and G\. Under normal con-ditions, positive ions are formed in the first acceleration region and accelerated toward the detector. Electrons are ejected from the grid G{


c

■e < "o c σ> CO

c

+ Q>

X

80000

70000

60000

50000

40000

30000

20000

10000

-i 1400

1200

1000 3

+ o

-f

-H-? Φ H».

o o +

0

800 <

σ

600 | c o

400 Γ

200 q>

x

-400 - 3 0 0 - 2 0 0 - 1 0 0 0 100

G0-G1 Voltage Difference

200 300

Fig. 7. The effect of GQ-GJ acceleration field on ion signals: ( + ), He2+; (O), He4

and accelerated toward the plate G0. When the grid potentials are interchanged, positive ions formed in the first acceleration region are directed away from the detector. One would therefore not expect to see any ions produced by photoionization under these conditions. However, ions are still detected. These must be produced by photoemission electron impact ionization. This occurs because electrons formed at G0 can now be accelerated through G^ After passing through the grid, into the second acceleration region, these electrons can collide with neutrals to form positive ions that are able to reach the detector. As referred to above, this

process is shown in Fig. 2. Figure 8 presents spec-tra taken with instrumental conditions B. The sig-nal arises from helium introduced as the carrier gas. The spectrum in Fig. 8(A) was recorded using normal grid potentials, while that in Fig. 8(B) was obtained with those potentials switched. The shift in ion flight times is approximately that expected for ions produced just above G\. The reduction in signal is due to the fact that the elec-trons travel a much smaller distance in the second acceleration region than they do in the first and the amount of light striking each grid may be sub-stantially different. Both spectra are plotted on

A

B

10 20 30

Mass/Charge

40 50

Fig. 8. Effect of ion acceleration field direction on observed mass spectrum of helium: (A), normal field direction; (B), grid potentials switched. The peak at m/zO is due to the detection of scattered photons.


the same vertical scale. Switching the grid poten-tials is the easiest method for detecting photoemis-sion electron impact ionization because the changes and measurement are simple.

5. Advanced source design

The instrumental conditions C and C' represent a new ionization source designed for simple con-version between photoionization and photo-emission electron impact ionization. It has some distinct advantages over previous designs [3-5]. These include better field uniformity, reduced ther-mal emission, better space focusing, and the ability to change modes without disturbing the instru-mental setup. Previous designs involved the place-ment of a thin wire within the source region. This wire was used as the electron source and perturbed the uniformity of the electric field. Reducing the spatial distribution of the ionization by using the wire to form a potential well contributes even more to this problem [4]. The lack of field uniformity reduces the effectiveness of space focusing con-ditions since the field potential is no longer simply a function of distance between the grids.

Thermal emission is undesirable because it may continue long after the end of the laser pulse. With the sources previously proposed it may occur when too much laser light is directed on the thin wire. In the new source thermal emission is less likely because the light is distributed over the larger sur-face area and volume of the plate G0. This should allow the use of greater photon fluxes and the gen-eration of higher currents.

Space focusing in this new source is as effective as that in traditional laser ionization sources. Under the conditions used with the new source, ionization by electrons, whose energy is at least 50 eV, is only found within a region 2.5 mm above G^ Calculated times-of-flight for He+ ions produced within this region vary by only 14 ns. This is substantially below the time variation resulting from initial ion velocity distributions [45].

The greatest advantage of the new source is the ease of switching between ionization methods. No

physical changes need to be made within the instru-ment. Only the three grid voltages and, if desired, the path of the laser beam need to be adjusted. We have found that for photoemission electron impact ionization, redirecting the laser beam, so that it hits the plate G0 instead of passing directly between G0

and G b is not absolutely necessary because there is sufficient scattered light. However, doing so sub-stantially increases the signal intensity. To move the beam we either adjust the 1 m lens or insert a second 400 mm focal length lens in front of the instrument. The second lens makes it possible to quickly return to the same laser ionization conditions.

5.7. Results with advanced source design

To demonstrate the new dual source we analyzed a sample of w-butylbenzene using both ionization modes. This molecule was chosen because there are clear differences between its electron impact and laser ionization mass spectra. The sample was allowed to diffuse into the instrument through a needle valve and the gas Chromatograph was dis-connected. Figure 9 shows the resulting mass spectra. Figure 9(A) was obtained with conven-tional laser ionization. It shows the relatively soft ionization possible with even this short wavelength of light. The principal peaks are at 134, 119, 105, 91, and 92 u. These masses correspond to the parent ion and fragments resulting from the loss of parts of the alkyl chain. The data were obtained using configuration C with 1000 laser shots, 45/xJ per pulse, and a i m focal length lens. The reso-lution is limited by the rather large laser focal spot.

The second mass spectrum, (Fig. 9(B)), was obtained with photoemission electron impact ion-ization under instrumental conditions C'. It shows the greater fragmentation expected in this case. Significant signals are seen for all the lower mass fragments. The resolution is slightly worse than that of the laser ionization mass spectrum but much better than that of the spectrum shown in Fig. 8(B), which was obtained without space focusing conditions. Figure 9(C) displays a 70 eV


A

B

ALJVJL_A_AJ

c

ft » i i . i-i*- jJLun. .fcilL I

0 20 40 60 80 100 120 140 Mass/Charge

Fig. 9. Three mass spectra of «-butylbenzene with: (A), laser ionization; (B), photoemission electron impact ionization; (C), 70 eV electron impact ionization.

electron impact ionization spectrum taken from the literature [46]. As expected, there is greater frag-mentation in Fig. 9(B) than in Fig. 9(C) because of the higher range of electron energies and currents. The only changes made between record-ing the spectra in Figs. 9(A) and 9(B) involved adjusting the lens and setting power supply voltages. The same light fluence was used.

6. Summary and conclusions

A primary goal of our research has been to maximize the analytical sensitivity of laser ion-ization GC-TOF-MS [30]. We found, however, that photoemission electron impact ionization produced the major limitation to developing a more sensitive technique. For example, our attempts to increase tetraethyltin signal included increasing the laser power and confining the GC

effluent with the cup shaped source. However, both of these steps led to more photoemission electron impact ionization, which actually reduced our sensitivity to tetraethyltin. As shown, this reduction resulted from photoemission electron impact ionization of the GC carrier gas that is present in much greater abundance than the analytes. Carrier gas ionization is not ordinarily a problem in laser ionization GC-TOF-MS because the carrier gas or laser wavelength can be selected to prevent it.

We expect that problems may be found when-ever an experiment depends on specific proper-ties of laser ionization such as selectivity or con-trolled fragmentation. Photoemission electron impact ionization can occur whenever there is the possibility of light striking the source and high intensities, high fields, or short wavelengths are used. Short wavelengths are not necessarily a requirement because multiphoton absorption can also induce photoemission [47,48]. The degree to which photoemission electron impact ionization occurs is a function of instrument geometry. The more confined the ionization region, the more likely it is that scattered light will strike a surface. In our experiments this was exemplified by the cup source, which had only small holes (2.5 mm x 13.5 mm and 3 mm diameter) for the laser beam and capillary column to pass through.

Photoemission electron impact ionization greatly expands the versatility of TOF-MS. It is a universal ionization method that combines electron impact ionization's advantages of reproducibility and generality, with a short temporal duration and reasonable spatial definition. One of its most attractive characteristics is that it can be adapted to the large number of time-of-flight instruments already in use. Our new dual source should be particularly convenient. It requires only slight modification of standard laser ionization time-of-flight instruments.

Photoemission electron impact ionization is an excellent complement to laser ionization. It should be useful for the analysis of unknowns because the mass spectra obtained can be compared with those

S.M. Colby andJ.P. Reilly/Int. J. Mass Spectrom. Ion Processes 131 (1994) 125-138 137

from conventional electron impact ionization. It also makes it possible to detect molecules that, because of low cross-sections, very short excited state lifetimes [49], or high ionization thresholds, are difficult to laser ionize.

With this method of photoemission electron impact ionization, the advantages of time-of-flight are retained. These include low cost, high through-put, low weight, simplicity, and the ability to obtain a complete mass spectrum with each laser shot. In the past, for many applications, these advantages have been outweighed by the lack of a convenient electron impact ionization source. With photoemission electron impact ionization this is no longer the case.

It is important to be aware that photoemission electron impact ionization can also be a major source of background signal in laser ionization experiments. We have pointed out several ways of identifying signal from photoemission electron impact ionization. Once identified, steps can be taken to eliminate it by reducing scattered light or adjusting laser parameters.

Acknowledgment

This work has been supported by the U.S. Environmental Protection Agency.

References

1 F.A. White and G.M. Wood, Mass Spectrometry, Appli-cations in Science and Engineering, Wiley, New York, 1986, p. 18.

2 T.D. Mark and G.H. Dunn (Eds.), Electron Impact Ion-ization, Springer-Verlag, New York, 1985, pp. 176, 306.

3 E.R. Rohwer, R.C. Beavis, C. Köster, J. Lindner, J. Grotemeyer and E.W. Schlag, Z. Naturforsch. Teil A, 43 (1988) 1151.

4 J.G. Boyle, L.D. Pfefferte, E.E. Gulcicek and S.D. Colson, Rev. Sei. Instrum., 62 (1991) 323.

5 K. Zipperer, R.C. Beavis, E.R. Rohwer, J. Lindner, C. Kostner, J. Grotemeyer and E.W. Schlag, Proc. 37th ASMS Conference on Mass Spectrometry and Allied Topics, Miami Beach, FL, 21-26 May 1989, pp. 568-569.

6 R.E. Honig and J.R. Woolston, Appl. Phys. Lett., 2 (1963) 138.

7 A.Sh. Airapetov, G.A. Gevorgyan, I.V. Levshin and B.N. Yablokov, Sov. Techn. Phys. Lett., 15 (1989) 428 (and references therein).

8 O. Prokoviev and P.G. Seiler, Nucl. Instrum. Methods Phys. Res. A, 278 (1989) 608.

9 C.H. Lee, Appl. Phys. Lett., 44 (1984) 565. 10 M. Boussoukaya, H. Bergeret, R. CheHav, B. LeBlond and

J. Le Duff, Nucl. Instrum. Methods Phys. Res. A, 279 (1989) 405.

11 M. Yoshioka, M. Mutou, Y. Fukushima, T. Kamei, H. Matsumoto, H. Mizuno, S. Noguchi, I. Sato, T. Shidara, T. Shintake, K. Takata, H. Kuroda, N. Nakano, H. Nishimura, K. Soda, M. Miyao, Y. Kato, T. Kanabe and S. Takeda, Proc. 1984 Linac Conference, Gesellschaft fur Schwerionenforschung, Darmstadt, Germany, Report GSI-84-11, 1984, p. 469.

12 E.L. Garwin, W.B. Hermannsfeldt, C. Sinclair, J.N. Weaver, JJ. Welch and P.B. Wilson, IEEE Trans. Nucl. Sei., 32 (1985) 2906.

13 W. Peter and M.E. Jones, Part. Accel., 24 (1989) 231. 14 Y. Kawamura, K. Toyoda and M. Kawai, Appl. Phys.

Lett., 45 (1984) 307. 15 Y. Kawamura, K. Toyoda and M. Kawai, Appl. Phys.

Lett., 45 (1984) 1287. 16 C.H. Lee, P.E. Oettinger, E.R. Pugh, R. Klinskowstein,

J.H. Jacob, J.S. Fraser and R.L. Sheffield, IEEE Trans. Nucl. Sei., 32 (1985) 3045.

17 J.S. Fraser, R.L. Sheffield and F.R. Gray, Nucl. Instrum. Methods Phys. Res. A, 250 (1986) 71.

18 J.S. Fraser and R.L. Sheffield, IEEE J. Quantum Electron., 23 (1987) 1489.

19 J. Bokor, A.M. Johnson, R.H. Storz and W.M. Simpson, Springer Ser. Chem. Phys. (Ultrafast Phenom. S), 46 (1986) 123.

20 R.B. Marcus, A.M. Weiner, J.H. Abeles and P.S.D. Lin, Appl. Phys. Lett., 49 (1986) 357.

21 T. Anderson, I.V. Tomov and P.M. Rentzepis, J. Appl. Phys., 71 (1992)5161.

22 N.K. Sherman, P.B. Corkum and T. Srinivasan-Rao, Proc. SPIE-Int. Soc. Opt. Eng., 664 (1986) 74.

23 L.H. Luthjens, M.L. Horn and M.J.W. Vermeulen, Rev. Sei. Instrum., 57 (1986) 2230.

24 A. Modinos, Field, Thermionic, and Secondary Electron Emission Spectroscopy, Plenum, New York, 1984, pp. 20, 22.

25 S.D. Moustaizis, M. Tatarakis, C. Kalpouzos and C. Fotakis, Appl. Phys. Lett., 60 (1992) 1939.

26 W. Schottky, Z. Phys., 14 (1923) 63. 27 J.G. Potter, Phys. Rev., 58 (1940) 623. 28 P.E. Oettinger, Proc. SPIE-Int. Soc. Opt. Eng., 998 (1988)

24. 29 S.M. Colby and J.P. Reilly, Anal. Chem., submitted. 30 S.M. Colby, M. Stewart and J.P. Reilly, Anal. Chem., 62

(1990) 2400. 31 W.C. Wiley and I.H. McLaren, Rev. Sei. Instrum., 26

(1955) 1150. 32 R.C. Weast (Ed.), Handbook of Chemistry and Physics,

64th edn., CRC Press, Boca Raton, FL, 1983, p. E-76.


33 R.B. Opsal, Ph.D. Dissertation, Indiana University, (1985). 34 R. Larciprete, S. Fontana and E. Borsella, Chemtronics, 4

(1989) 184. 35 R.C. Weast (Ed.), Handbook of Chemistry and Physics,

64th edn., CRC Press, Boca Raton, FL, 1983, p. E-63. 36 K. Stephen, H. Helm and T.D. Mark, Measurement of

Partial Electron Impact Ionization Cross-Sections of Rare Gas Atoms and Dimers, in A. Quayle (Ed.), Advances in Mass Spectrometry, Vol. 8A, Heyden, London, 1979, p. 122.

37 D.C. Gregory, P.F. Dittner and D.H. Crandall, Phys. Rev. A, 27 (1983)724.

38 A. Müller, E. Salzborn, R. Frodl, R. Becker, H. Klein and H. Winter, J. Phys. B, 13 (1980) 1877.

39 K. Stephen, H. Helm and T.D. Mark, J. Chem. Phys., 73 (1980) 3763.

40 T.S. Luk, H. Pummer, K. Boyer, M. Shahidi, H. Egger and C.K. Rhodes, Phys. Rev. Lett., 51 (1983) 110.

41 F.W. McLafferty and D.B. Stauffer (Eds.), The Important

Peak Index of the Registry of Mass Spectral Data, Vol. 1, Wiley, New York, 1991, p. 100.

42 G. Friedlander, J.W. Kennedy, E.S. Macias and J.M. Miller, Nuclear and Radiochemistry, 3rd edn., Wiley, New York, 1981, p. 624.

43 N.S. Oakey and R.D. Macfarlane, Nucl. Instrum. Methods, 49 (1967) 220.

44 R.D. MacFarlane, D.F. Torgerson, Y. Fares and C.A. Hassell, Nucl. Instrum. Methods, 116 (1974) 381.

45 R.B. Opsal, K.G. Owens and J.P. Reilly, Anal. Chem., 57 (1985) 1884.

46 E. Stenhagen, S. Abrahamsson and F.W. McLafferty (Eds.), Registry of Mass Spectral Data, Wiley, New York, 1974, pp. 179, 263.

47 P. Martin, R. Trainham, P. Agostini and G. Petite, Phys. Rev. B, 45 (1992) 69.

48 E.M. Logothetis and P.L. Hartman, Phys. Rev., 187 (1969) 460.

49 C.W. Wilkerson Jr., S.M. Colby and J.P. Reilly, Anal. Chem., 61 (1989) 2669.

International journal of Mass Spectrometry and Ion Processes 131 (1994) 139-148 139 0168-1176/94/507.00 © 1994 - Elsevier Science B.V. All rights reserved

High-resolution mass spectrometry in a linear time-of-flight mass spectrometer

Johann M. Grundwürmer^, Michael Bönisch, Gary R. Kinsel*, Jürgen Grotemeyer^*, E.W. Schlag

Institut für Physikalische und Theoretische Chemie der Technischen Universität München, Garching, Germany

(Received 17 May 1993; accepted 20 July 1993)

Abstract

This paper describes a new method to achieve high-resolution mass spectra in a linear time-of-flight mass spectrometer. Both direct and matrix assisted laser desorption/ionization of large molecules are performed on this instrument. Using post-source pulse-focusing a mass resolution of 4600 (FWHM) has been achieved. This technique can correct high initial ion translational energies as well as long periods of ion formation. Therefore it is predestinated to laser desorption/ ionization applications. Experimental results will be shown demonstrating the capability of this new instrument to produce high-resolution ion signals. In addition, a simple way to calibrate masses is discussed.

Key words: Resolution enhanced time of flight; Matrix assisted laser desorption-ionization; Direct laser desorption-ionization; Resolution; Mass calibration

Introduction

Since the early 1960s direct laser desorption/ ionization (DLD/I) is used to transfer molecular ions of pure samples into the gas phase for mass spectral analysis [1]. But the upper limit of this method is a molecular weight of normally less than a few thousand daltons. Using matrix-assisted laser desorption/ionization (MALD/I) Karas et al. [2], as well as Tanaka et al. [3], suc-ceeded in introducing ions with a molecular weight of more than 100 000 Da. Both DLD/I and MALD/I methods are commonly coupled with a time-of-flight mass spectrometer (TOF-MS) for

* Corresponding author. ' New address: Institut für Physikalische Chemie der Julius Maximilians Universität Würzburg, Germany. Î Present address: Texas A & M University, Department of Chemistry, College Station, TX 77843, USA.

mass analysis of the resulting ions. Research and application have continued to extend the obtain-able molecular weight, the accuracy of the mol-ecular weight detection and the mass resolution. The obtainable molecular weight now exceeds 100 000 Da [4]. The mass accuracy for ions with a molecular weight of less than 30 000 is about 0.01 % [5]. In coupling MALD/I to a FT-ICR mass spec-trometer the best mass resolution which can be achieved is approximately 12000 (FWHM) for cytochrome c (m/z 12 300) [6]. Using a linear TOF-MS Beavis and Chait [7] demonstrated that the mass resolving power can be increased to about 700 (FWHM) through a careful control of the sample preparation and the desorbing laser power. Nevertheless, none of these methods accomplishes all three aims. FT-ICR mass spectro-meters still have a limited mass range: up to about m/z 40000. High mass molecules usually show

SSDI0168-1176(93)03882-M

140 J.M. Grundwürmer et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 139-148

wide, unresolved isotopic peak patterns. For that reason there is still a need for higher mass resolu-tions to be obtained.

A large number of reports and reviews are devoted to TOF mass spectrometry and the mass resolving power of this instrument type. Three components in the ion formation conditions deter-mine the resolution. These are the spatial width over which the ions are produced, the distribution of the initial ion velocities, and the time span in which the ions are formed. For laser desorption/ ionization (LD/I) experiments the spatial width is generally assumed to be very small. Various results have been published on the efforts to measure the initial ion kinetic energy in MALD/I [8] and DLD/I [9]. Although the experimental details vary, energies of the desorbing ions ranging from several to a few tens of electron volts have been found. In general it appears that the matrix assist-ance limits the kinetic energies to the lower values while direct laser desorption produces ions with larger kinetic energies up to a few electron volts. Clearly, the kinetic energy of the desorbed sample ions increases with the size of the sample molecules [10]. It is often assumed that ions are only formed within the duration of the desorbing laser pulse, but the time component of ion formation in LD/I experiments is not really defined. This assumption has not yet been experimentally verified. It has been suggested that one reason for the low resolution of ion signals observed in LD/I experiments using a resonance enhanced time-of-flight mass spectro-meter (RETOF-MS) is due to the ion formation time. This time can be longer than that for the desorbing laser pulse [11].

Since the initial demonstration of TOF-MS, some theoretical studies of the factors determining mass resolution in this instrument have been per-formed that have led to a variety of approaches for enhancing the mass resolution [12]. The RETOF-MS has been perhaps the most successful of these approaches. Either large initial ion velocities or spatial distributions can be corrected with this instrument [13]. Resolutions of more than 30000 (FWHM) [14] have been achieved using this

instrument design in combination with supersonic jet sample introduction or short ionizing laser pulses or both. However, it should be recognized that no static field TOF-MS, including resonance enhanced time-of-flight mass spectrometry, is capable of correcting long ion formation times. One way to solve this problem is the use of a time-dependent acceleration of the ions. Several attempts implementing time dependent ion accel-eration have been described in the literature. These are time-lag focusing [15], velocity compaction [16], impulse field focusing [17], dynamic field focusing [18], and post-source pulse focusing (PSPF) [19]. Each of these methods has the potential capacity to correct for ion formation times.

PSPF in particular presents several attractive features [20] that may be considered. Figure 1(a) shows examples of different initial ion velocities and different ion formation times for molecules with the same mass in a static field TOF mass spectrometer. Ions with a smaller initial kinetic energy, corresponding to their velocity, arrive later at the detector. Ions with a different ion for-mation time keep this difference until they reach the detector. The first implementation of PSPF is relatively simple. The ions produced in a conven-tional two-stage static electric field enter a short, initially field-free pulse-focusing region. After the ions have entered this region a voltage is applied to the region that compresses the individual mass ion packets, yielding higher resolution ion signals at the detector (Fig. 1(b)).

In an earlier paper [21], results of calculations showed an approximately ten-times-better mass resolution using PSPF in a linear TOF compared with a static field linear TOF. The instrument in this paper is very similar. For that reason, a simi-lar improvement has been calculated for this instru-ment. The capabilities of the presented instrument have been predicted with calculations on the spread of flight times of cytochrome c. These calculations show that even with a ten-times-longer ion forma-tion time the full width at half of maximum of the ion signal is only twice that with the short ioniz-ation time.

J.M. Grundwürmer et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 139-148 141

Static Field

a) Initial ion velocity v

V

v-dv *>

E1 ; E2 ;

b) Ion formation time t

t

t+dt o

• >

• >

·- -—>

o >

-->

E1 E2

Detector

+ >

Detector

Static field and Post Source Pulse Focusing

a) Initial ion velocity v

· ■ >

v-dv ·>

E1 · E2 ·

b) Ion formation time t

t · >

t+dt o

E1 · E2

• > r - E3

Detector

> E3 Detector

Fig. 1. Principle of PSPF. (a) Different initial kinetic energies and ion formation times of molecules with the same mass in a static field TOF-MS. (b) Static field TOF-MS with an additional PSPF region; the PSPF voltage is applied after the ions have entered this region.

Implementation of PSPF requires nothing else than the addition of a short region next to the ion source region, a pulse voltage supply, and a time-delay generator to offset the voltage pulse from the ionizing laser pulse. Such focusing is not able to cover the entire mass spectrum. A relatively large fraction of the mass spectrum may be focused without serious sacrifices in ion collection or resolution for the remainder of the mass spec-trum. The part of the mass spectrum focused depends on the length of the pulse-focusing region and the ion formation parameters. This part may be 70% or more, relative to the highest mass ion focused using conventional static instrumental

conditions. A major drawback of this technique is, at first sight, that accurate mass calibration appears to be impossible. This paper demonstrates that, by keeping the voltages of ion source and PSPF-area constant and fixing the delay time of the PSPF-voltage pulse, it is possible to obtain an analytical function for mass calibration.

The results presented in this paper are collected with a home-built linear TOF system with a 2 m flight path and an 11 cm post-source pulse focusing area located after a conventional LD/I ion source. This instrument will be used for investigations of both DLD/I and MALD/I. This paper demonstrates significant enhancements of the mass resolving power of such an instrumental design.

Experimental setup

The schematic of the instrumental arrangement is presented in Fig. 2(a). The two basic components of the TOF-MS structure are a cylindrically bored, cubic aluminum source region chamber and a 2 m stainless steel flight tube. These are isolated by using a 6 in gate valve installed between the two parts. Both the source chamber and the flight tube are pumped by a 330 Ls"1 turbo molecular pump. Two ionization gauges mounted separately in the source and flight tube show typically an operating vacuum of about 10~6Torr in both chambers.

Figure 2(b) shows in detail the ion source and acceleration region assembly. The acceleration region consists of a series of 15 round, stainless steel acceleration plates, assembled parallel by three ceramic rods. These ceramic rods, furnished with grooves to fix the plates, are mounted on a horizontally and vertically micrometer-adjustable Teflon flat. Each of the acceleration plates has a diameter of 65 mm, is 1 mm thick, and has a 10 mm hole in the center. Plates 1 and 2 are separated by 9 mm and define the ion source region. Plates 2 and 3 are separated by 15 mm and define the secondary acceleration region. Plates 4-15 are each separated by 9 mm and define


Detector

Plate 1 2 3

Teflon flat

+/-H\*-+/-H\fc-

GNDo-

M

5 6 7 8 9 10 11 13

12

+/- Pulsed Voltage

14 15 (b)

r Ceramic Rods

GND

Ion Source

Fig. 2. (a) Schematic of the instrument, (b) Ion source and acceleration region assembly in expanded detail.

the pulse-focusing region. The drift region from plate 15, which defines the end of the pulse-focusing region, to the grounded grid in front of the ion detector measures 1900.0 mm. The plates in the pulse-focusing region are connected via a series of \AkQ resistors. Acceleration plates 2-15 have a 10 mm hole bored through their centers for trans-mission of the ions. Copper, 90% transmission grids are fixed with conducting glue over the

holes in acceleration plates 2, 3 and 4 to define the initial and post acceleration electric fields.

Two methods of introducing samples into the source chamber are possible. Volatile sample species may be introduced through a molecular leak inlet connected to a needle tube in the vacuum chamber. The needle serves to direct the sample vapor to the ion source region. A direct insertion probe is available that enters the source chamber

J.M. Grundwürmer et al.j Int. J. Mass Spectrom. Ion Processes 131 (1994) 139-148 143

from behind the acceleration region assembly. The spring mounted, polished stainless steel probe tip enters the ion source region through an 8.0 mm hole bored in acceleration plate 1. When ionizing gas phase samples, the uncoated probe is placed in position to fill the hole in acceleration plate 1 and better define the source region electric field.

Five power supplies, a digital delay generator, a fast high voltage power switch and a pulsed voltage power supply are the electronic components required. Two power supplies with maximum out-puts of 35.0 kV are used for static voltage biasing of the first and second acceleration plates. The third power supply has a maximum voltage of 6500 V and is used for the pulsed-focusing region. An assembly of two 10 cm long deflection plates about 1 m before the detector in the mass spectro-meter flight tube is supplied with the fourth power supply, with a maximum voltage of 1250 V. The fifth power supply is used for biasing of a dual microchannel plate detector with a total post-acceleration of -2.0 kV. The ion deflection plates may be used to eject low-mass matrix ions pro-duced during MALD/I experiments. Pulse focus-ing is achieved by switching a voltage of typical 2500 V with the high voltage power switch in less than 25 ns and for a duration of 30 /is. TTL pulses produced by the digital delay generator define the switch-on and the switch-off time by their rising and falling edges independently for both pulsed high-voltage fields. In practice the voltage pulse is delivered to acceleration plate 4, whereas acceler-ation plate 15 is maintained to ground, generating a constant accelerating electric field across the pulse-focusing region.

At the end of the 2 m flight tube a standard dual microchannel plate detector is positioned, which detects the ion. The distance from the grounded grid defining the end of the field-free flight tube to the detector surface is 11.0 mm. The transient current from the ion signals is terminated through a 50 Ω impedance collector and fed to a 500 MHz digital oscilloscope for continuous monitoring of the ion signals. This signal may be further trans-mitted to a 200 MHz transient digitizer for spectra

collection and storage. The output of a fast photo-diode, positioned to detect a reflection of the desorbing laser pulse, triggers the digital delay generator, the oscilloscope, and the 200 MHz tran-sient digitizer.

The Nd:YAG laser used has an output pulse length of 5-6 ns after passing through appropriate doubling crystals producing the third harmonic wavelength of 355 nm. The output laser radiation is turned 90° through a Pellin-Broca prism that acts to separate the various fundamental and har-monic wavelengths. The Nd:YAG laser may be operated either in the single shot mode or at a repetition rate of 10 Hz. Each laser shot produces a plume of desorbed ionic and neutral material that may contain both intact and fragmented molecules of the sample material. The ions produced this way are accelerated and detected in the linear TOF-MS.

The zinc mesoporphyrin sample was prepared by dissolving méthylène chloride solution of the sample, aspirating the solution through a syringe needle, and air-spraying the solution in the form of small droplets onto the stainless steel probe tip. Cytochrome c was codissolved with 2,5-dihydroxybenzoicacid (DHB) in a solution of ethanol/water (v: v 1:1). The molar ratio of cyto-chrome c to DHB in the solution was approxi-mately 1 to 2000. Five to ten microliters of this solution of matrix and probe were applied to the front side of the probe tip and dried in a stream of air.

Experimental results

The usual experimental procedure is first to establish the desired electric acceleration voltages. This will lead to a LD/I ion signal obtained without the focusing voltage pulse. Next, the focusing voltage pulse is applied to the pulse-focusing region, but the pulse is timed to occur after the ions to be focused have entered the focusing region. An equivalent mass spectrum to the spec-trum produced without the focusing voltage pulse is observed. The delay time for the focusing voltage pulse is then gradually reduced to shorter times


8

CO

a

657 657

46.2 Flighttime [ \is

M l » » ' P » " f l . ""»■■ ",' ■'lllwl*«*>PWllwJMLijMM 0.0 20.0

Flighttime [ μ8 40.0

Fig. 3. Zinc mesoporphyrin (MW 656), DLD/I focused with PSPF; the inset shows the parent ion region.

until an optimum mass resolution is observed. This optimum is theoretically performed if the complete ion bulk of the highest mass of interest is located in the pulsed-focusing region, just at the time when the focusing voltage pulse is applied. Finally, the mass resolution is further improved through fine adjustment of the voltage pulse.

This relatively simple procedure was applied to a number of medium and high molecular weight ion signals produced using both DLD/I and MALD/I to test the newly constructed linear TOF-MS and to demonstrate the effectiveness of the PSPF method. With an electrostatic field a resolution of only about 600 for zinc mesoporphyrin and less than 400 for high mass molecules was achievable in the described instrument. Figure 3 shows a single shot DLD/I spectrum of zinc mesoporphyrin (MW 656 Da) with PSPF. The wavelength used for this spectrum was 355 nm. The static source region accelerating field at plate 1 was biased to 6770 V, acceleration plate 2 at 6500 V and plate 3 at ground. Initially plate 4 was at 0 V. A focusing

voltage pulse of 714 V was applied 2.9 μϊ after firing the desorbing laser at plate 4 for pulse focus-ing of the zinc mesoporphyrin parent ion. If PSPF was used, the resolution of the molecular signal was increased to 4600 (FWHM), allowing clear distinc-tion of the individual isotopic peaks in the mol-ecular ion signal.

Figure 4 shows a MALD/I spectrum of cyto-chrome c (MW 12 360 Da) in DHB matrix with PSPF. The spectrum is a sum of 50 lasershots at a desorbing wavelength of 355 nm. The static elec-tric field for the cytochrome c spectrum was 20000 V at acceleration plate 1, 19 500 V at accel-eration plate 2, and ground at plate 3. A focusing voltage pulse of 2570 V was applied to the pulse-focusing region 8.3/xs after the irradiation with laser light. The PSPF technique allows the resol-ution of the molecular ion to be improved to 900 (FWHM). An [M - H + 23] tentatively identified as the sodiated parent ion can be separated from the [M + H]+ ion clearly. The molecular peak pattern of cytochrome c has a natural width at

J.M. Grundwürmer et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 139-148 145

Righttime [ μβ

Fig. 4. Cytochrome c (MW 12 360 Da), MALD/I focused with PSPF; the inset shows the parent ion region.

half height of 10 Da. With a resolution lower than 10000 it is not possible to resolve the isotopic peak pattern of a molecule with a masses high as cyto-chrome c. The ion transmission of the described instrument is reduced by approximately 36% by the four grids installed. Eliminating two of these should improve the ion intensity.

Mass calibration

The aim of mass spectrometry is to identify unknown substances based on their masses and/ or the mass of their fragments. This requires a well-defined mass calibration. The relationship between time-of-flight (/) and the mass-to-charge ratio (m/z) of an ion in a conventional TOF-MS with only static electric fields is given by

yj{m/z) = at + b

Using post-source pulse focusing changes this simple relation to a complex one, because different masses are accelerated by different post-source fields. Ions of different masses pass a different dis-tance of the pulsed region before the electrical field

is turned on. These ions ultimately achieve different kinetic energies in the drift region. Three cases can be distinguished

(i) Before the PSPF voltage is turned on, the low mass ions pass the post-source pulse focusing region. These are accelerated as in a static electric field.

(ii) After the ions of interest entered the pulsed field region, the voltage is applied to the pulsed area.

(iii) Finally, masses higher than the highest mass of interest will be decelerated first, because they have to run from a ground field to the total pulsed voltage before they are accelerated again to their initial kinetic energy.

Only the masses of the ions, which are in the pulsed region, are usually of interest for mass spectrometry.

These different behaviors make it difficult to estimate the precise mass of the ions focused by the PSPF through the relation given above. There-fore, the following equation has been extrapolated from the theoretical term for the flight time. This equation allows the calibration of the post-


Table l Comparison between flight times, theoretic masses and measured masses using different calibration points

(CsI)„Cs+

(«)

13 14 15 16 17 18 19 20 21 22 23 24

m/z (cluster)

3510.4 3770.2 4030.0 4289.9 4549.7 4809.5 5069.3 5329.1 5588.9 5848.7 6108.5 6368.3

/a

101.36 104.43 107.41 110.30 113.11 115.86 118.52 121.11 123.65 126.14 128.56 130.94

Calculated curveb (m/z)

(1)

3510.1 ±0.8 * 4030.3 ± 0.9 * 4549.4 ± 0.9 4810.3 ±1.0 * 5327.9 ±1.0 5587.8 ±1.0 * 6107.25 ±1.1 6369.30 ±1.1

(2)

3516.0 ±0.8 3772.1 ±0.9 * 4288.7 ± 0.9 4548.1 ±0.9 * 5069.2 ±1.0 5328.6 ±1.0 * 5849.8 ±1.0 6108.7 ±1.0 *

(3)

3533.1 ±0.8 3783.1 ±0.8 4036.5 ± 0.9 4292.2 ± 0.9 * 4810.0 ±1.0 * 5328.5 ±1.0 * 5849.8 ±1.1 * 6367.6 ±1.1

a Flight time, taken from the mass signals in Fig. 5 without further evaluation of their centroids. b See text: the four peaks marked with an asterisk are taken from the spectrum in Fig. 5 to define the calibration curve.

source-pulsed TOF-MS with three or more cali-brating masses:

/ = c0 + cxml/4 + c2m

x/1 + c3m3/4

The constants q can be found by gaussian mini-mization of the square of the errors. Essential for an improved mass calibration is the use of the same

conditions as for the probe. The best results should give an internal standard or a probe tip with both the calibrating substances and the probe. An opti-mum calibration can be achieved if the calibrating substances have similar masses to the probe. The highest mass of the calibration substances and the probe determines the delay time for the focusing

n-12

n-13

n-16

|llwHPMBflflBH^^BBBWn|fflMl BM

n = 22

91.7 111.7 Flighttime [ μβ ]

Fig. 5. (CsI)„Cs+ from a Csl: sucrose mixture (1 mol : 1 mol) focused with PSPF

131.7

J.M. Grundwürmer et al./Int. J. Mass Spectrom. Ion Processes 131

voltage switch-on time. The calibration curve is only exact for ions which are present in the post-source pulse focusing region during the switch-on of the voltage. Changing the delay time of post-source pulse focusing demands a new mass cali-bration, because the ions are then subject to differ-ent post-accelerating potentials and get more or less kinetic energy.

A mass spectrum of Cs(CsI)* cluster has been used to test the calibration function. The spectrum shows ions produced by irradiation of a 1:1 mix-ture of Csl and sucrose with a laser wavelength of 355 nm. A voltage of 9.0 kV at plate 1, 8.8 kV at plate 2 and a voltage of 1.175 kV for post-source pulse focusing, applied 15.1 s after the laser pulse, accelerates the ions. The flight times in Table 1 are taken from the mass signals in Fig. 5 without further evaluation of their centroids. The four peaks in Table 1 marked with an asterisk are taken from the spectrum to define the calibration curve. The masses of the other peaks have been recalculated from the calibration curves:

t = 128.31 - 39 .6W / 4 + 6.10m1/2 - 0.18m3/4 (1)

t = -68.96 + 30.77m1/4 - 2.27m1/2 + 0.15m3/4

(2)

/ = -249.2 + 93.49m1/4 - 9.55m1/2 + 0.43m3/4 (3)

The data in Table 1 show that an accuracy of mass calibration better than 0.02% can be reached if the calibrating substances have a mass differing 4% up to 20% from the unknown mass. Smaller differ-ences between unknown and calibrating sub-stances should lead to a dramatic improvement in mass accuracy.

Conclusions

The incorporation of PSPF in a linear TOF-MS offers a simple method for significantly improving the resolution of a large fraction of a recorded mass spectrum even without careful sample preparation and accurate laser power control. These fea-tures can be achieved with good ion collection

131 (1994) 139-148 147

efficiencies and without degrading the resolution of the remaining mass spectrum. The PSPF TOF-mass spectrometry technique has the capabil-ity to correct for both large distributions in initial ion translational energies and long durations of ion formation. These features may be particularly valu-able for both DLD/I and MALD/I applications. The calibration function shows that high accuracy in mass determination can be achieved. The experi-ments show that the isotopic peak pattern of mol-ecules with a medium molecular weight can be resolved as well as sodiated molecular ion peaks for higher mass molecules. The possibility of using PSPF with high static extraction voltages is clearly demonstrated. Furthermore, it is antici-pated that improvements beyond the mass resolution values achieved in these experiments will be realized through a number of relatively minor experimental modifications.

Acknowledgments

This work was supported by grants from Deutsche Forschungsgemeinschaft (GR917/6-2) and the Bundesministerium für Forschung und Technologie (13N5307-2). G.R.K. gratefully acknowledges the fellowship support of the Alex-ander von Humboldt Stiftung, Bonn, Germany.

References

1 (a) R.E. Honig and J.R. Woolston, Appl. Phys. Lett., 2 (1963) 138. (b) For an early review see: R.J. Conzemius and J.M. Capellen, Int. J. Mass Spectrom. Ion Phys., 34 (1980) 197. (c) A. Benninghoven (Ed.), Springer Series in Chemical Physics. Vol. 25: Ion Formation from Organic Solids, Springer Verlag, Berlin, 1983, pp. 190-244. (d) A. Benninghoven (Ed.), Springer Proceedings in Physics. Vol. 9: Ion Formation from Organic Solids (Proc. IFOS III), Springer Verlag, Berlin, 1986, pp. 142-162.

2 M. Karas, D. Bachmann, U. Bahr and F. Hillenkamp, Int. J. Mass Spectrom. Ion Processes, 78 (1987) 53.

3 K. Tanaka, H. Waki, Y. Ido, S. Anita, Y. Yoshida and T. Yoshida, Rapid Commun. Mass Spectrom., 2 (1988) 151.

4 M. Karas, U. Bahr, A. Ingendoh and F. Hillenkamp, Angew. Chem., Int. Ed. Engl., 28 (1989) 760.

5 R.C. Beavis and B.T. Chait, Anal. Chem., 62 (1990) 1836.


6 J.A. Castoro, C. Köster and C.H. Wilkins, Anal. Chem., 65 (1993) 784.

7 R.C. Beavis and B.T. Chait, Rapid Commun. Mass Spec-trom., 3(1989) 233.

8 R.C. Beavis and B.T. Chait, Chem. Phys. Lett., 181 (1991) 479.

9 Y. Pan and RJ. Cotter, Org. Mass Spectrom., 27 (1992) 3.

10 (a) W. Ens, Y. Mao, F. Mayer and K.G. Standing, Rapid Commun. Mass Spectrom., 5 (1991) 117. (b) R.C. Beavis and B.T. Chait, Chem. Phys. Lett., 181 (1991)479.

11 R.C. Beavis and B.T. Chait, in A. Hedin, B.U.R. Sundqvist and A. Benninghoven (Eds.), Ion Formation from Organic Solids (Proc. IFOS V), Wiley, Chichester, 1990, pp. 125-130.

12 (a) D. Price and G.J. Milnes, Int. J. Mass Spectrom. Ion Processes, 99 (1990) 1. (b) J.E. Campana, Anal. Instrum., 16 (1987) 1. (c) Y. Le Beyec, Adv. Mass Spectrom., 11 (1989) 126. (d) U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spectrom. Ion Processes, 112 (1991) 121.

13 (a) B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A. Zagulin, Sov. Phys. JETP, 37 (1973) 45. (b) V.l. Karataev, B.A. Mamyrin and D.V. Shmikk, Sov. Phys. Tech. Phys., 16 (1972) 1177.

(c) B.A. Mamyrin and D.V. Shmikk, Sov. Phys. JETP, 49 (1979) 762.

14 (a) H.J. Neusser, U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spectrom. Ion Phys., 60 (1984) 147. (b) T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 60 (1989) 792.

15 (a) W.C. Wiley and I.H. McLaren, Rev. Sei. Instrum., 26 (1955) 1150. (b) E.D. Erickson, G.E. Yefchak, C.G. Enke and J.F. Holland, Int. J. Mass Spectrom. Ion Processes, 97 (1990) 87.

16 (a) M.L. Muga, Anal. Instrum., 16 (1987) 31. (b) M.L. Muga, U.S. Patent 4,458,149, July 3, 1984.

17 (a) N.L. Marable and G. Sanzone, Int. J. Mass Spectrom. Ion Phys., 13 (1974) 185. (b) J.A. Browder, R.L. Miller, W.A. Thomas and G. Sanzone, Int. J. Mass Spectrom. Ion Phys., 37 (1981) 99.

18 G.E. Yefchak, C.G. Enke and J.F. Holland, Int. J. Mass Spectrom. Ion Processes, 87 (1989) 313.

19 G.R. Kinsel and M.V. Johnston, Int. J. Mass Spectrom. Ion Processes, 91 (1989) 157.

20 G.R. Kinsel, C D . Mowry, P.J. McKeown and M.V. Johnston, Int. J. Mass Spectrom. Ion Processes, 104 (1991) 35.

21 G.R. Kinsel, J.M. Grundwürmer and J. Grotemeyer, J. Am. Soc. Mass Spectrom., 4 (1993) 2.


149

The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer

Benjamin M. Chien, Steven M. Michael, David M. Lubman* Department of Chemistry, The University of Michigan, Ann Arbor, MI 48109, USA

(Received 1 March 1993; accepted 8 June 1993)

Abstract

The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer (IT-RETOF MS) combination have been explored for detection of ions generated from both internal and external ionization sources. Ions were generated inside the trap using laser-induced resonance enhanced multiphoton ionization at 266 nm, while ions were generated external to the trap using either an atmospheric pressure or low pressure d.c. plasma source or an electrospray ionization source. It is demonstrated herein that the ion trap provides an effective means of storing ions from 10 μ8 up to 10 s prior to mass analysis via pulsed d.c. extraction into the RETOF MS device. In addition, it is shown that the storage capabilities of the device provide enhanced resolution and sensitivity as the storage time is increased. A resolution of nearly 2100 at m/z 93 is demonstrated using external injection from the atmospheric pressure d.c. plasma source when a storage time of 9 s is used before ejection. Further, a resolution of «3300 at m/z 1000 is demonstrated using external injection from an electrospray ionization source into the trap with a storage time of 931 ms. The IT-RETOF storage capabilities are shown to provide the potential for nearly 100% duty cycle in converting a continuous ion beam into a pulsed source for TOF. The detection limit of the device is demonstrated with liquid injection techniques for a typical sample and found to be in the low femtomole range. In addition, the r.f. voltage was shown to be an effective means of eliminating low mass background ions from the trap and, thus, from the TOF mass spectra obtained.

Key words: Ion trap storage; Reflectron; Resonance enhanced multiphoton ionization; Electrospray ionization; Atmos-pheric pressure ionization

Introduction

In this article we present the design and perfor-mance of a novel ion trap storage-reflectron time-of-flight (IT-RETOF MS) mass spectrometer combination. This hybrid instrument provides cap-abilities which combine some of the best features of the ion trap and time-of-flight devices. The time-of-flight mass spectrometer has recently become widely used because of several distinct advantages [1-17]. These advantages include the fact that TOF devices can rapidly measure an entire mass spec-

corresponding author.

trum on every ionization pulse. This property of time-of-flight is particularly important in exper-iments where the ionization source has a low duty cycle as in laser ionization [3,5,6,8] or in experiments where rapid monitoring of transient species may be required. Other advantages include the potentially high resolution that can be achieved with TOF devices. A number of methods including super-sonic jet cooling [5,6,8] and an ion reflecting mirror, i.e. a reflectron [7-11], have been used to achieve resolution well in excess of several thousands using a variety of ionization sources, including laser ionization and electron impact. An increasingly important property of TOF devices

SSDI0168-1176(93)03877-0

150 B.M. Chien et ai/Int. J. Mass Spectrom. Ion Processes 131 (1994) 149-179

has become the ability to measure high mass with high transmission, especially in matrix-assisted laser desorption ionization (MALDI) experiments where masses in excess of 200 000 u have been detected using TOF mass spectrometers with high voltage acceleration [14]. In addition, TOF devices are relatively simple, rugged and inexpensive and require only d.c. power supplies with essentially no current load to bias the acceleration region. There are no moving parts, scanning electric fields or slits to limit the throughput.

A major limitation of TOF devices is that there is no storage of ions in the acceleration region. Ion storage is particularly important in studies invol-ving ion/molecule chemistry, for studies of very slow formation of metastables, for enhancing the detection limits by integration of the signal, and for interfacing a continuous ion beam to a TOF device. Although there have been several attempts to store ions using d.c. fields, the ion storage time via this method is limited [15]. However, a more versatile method for achieving ion storage and tandem MS is the ion trap mass spectrometer.

The quadrupole ion trap in itself is a powerful tool for mass analysis and storage of ions over a wide mass range with excellent sample detection limits [18-20]. The ion trap has been used with numerous ionization sources, including electron impact [19], chemical ionization [21] and photo-ionization [22], which can create ions directly inside the trap. Alternatively, techniques which require ion introduction from an external source into the trap such as atmospheric pressure sam-pling glow discharges and electrospray [23] have been interfaced to ion traps. The trap has been used to analyze ions in excess of 70 000 Da and recently has been able to achieve extraordinarily high resolution [24]. In addition, an important fea-ture of the trap is its ability to perform multiple stages of tandem mass spectrometry in combina-tion with collisional or photodissociative fragmen-tation techniques [24]. A key feature of the ion trap is its ability to obtain high sensitivity through ion storage and integration of the signal over an extended period of time [25],

The ion trap also has several inherent disadvan-tages. Although the trap can store high mass ions, it is often difficult to scan the radiofrequency to a sufficiently large value of the voltage in order to scan out high mass ions. A technique has been developed known as axial modulation to scan out high mass; however, this usually occurs at the expense of the accuracy of the mass calibration [24,25]. In addition, very high mass resolution can be achieved in the trap by scanning the r.f. voltage or the r.f. frequency very slowly [26,27]. However, the rate at which the mass is scanned, i.e. 4 m/z per second in the latter work, is impractical for many applications. The commercial Finnigan ion trap at this time has a resolution of only 185 at m/z 69 and 1700 at m/z 502 and can only scan to m/z 650 at a scan rate of 5000us"1 [25].

The goal of our work has therefore been to inter-face an ion trap storage device as a front end source for a TOF MS [28]. This device combines the stor-age capabilities of the ion trap with the relatively high resolution, speed and high mass capabilities of the TOF to potentially produce a hybrid instru-ment with unique capabilities. This article describes the design of an ion trap-time-of-flight device in which ions are initially produced either inside the trap via resonance enhanced multi-photon ionization (REMPI) or external to the trap using an atmospheric pressure d.c. plasma source, low pressure glow-discharge source and electrospray ionization (ESI). Once ions are pro-duced in the trap, they are stored for a time inter-val ranging from 10 μ$ to 10 s and subsequently ejected into a RETOF for mass analysis. The ejec-tion method utilizes a d.c. pulse which is applied to the exit endcap of the ion trap. This d.c. pulse destabilizes the trap and the entire contents of the trap are simultaneously ejected for analysis into the RETOF.

In the present article we demonstrate the design and operation of an IT-RETOF MS. The IT is shown to be an effective means of storing ions prior to mass analysis by the RETOF and conse-quently the storage properties of the IT provide enhanced resolution and sensitivity for analysis

B.M. Chien et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 149-179 151

by TOF. It is also shown that as the storage time is increased from 10^s to 10 s the resolution is enhanced significantly with the addition of a buf-fer gas. A resolution of nearly 2100 at m/z 93 can be obtained using the extended storage time pro-vided by the trap, while at m/z 1000 a resolution of nearly 3300 can be obtained using an electrospray source. In addition, the IT-RETOF is shown to be an effective means of interfacing a continuous ion beam source to a TOF, achieving a high duty cycle using a low pulse-out extraction rate. A detection limit in the low femtomole regime can be typically obtained for injection from external ionization sources. Further, it is demonstrated that the trap is an effective means of ejecting low mass back-ground prior to analysis by TOF.

Experimental

The experiment consisted of two modes of operation, trapping internally produced ions and trapping externally produced ions. The experimen-tal setup of the internal ion production experiment is shown in Fig. 1, while that of the external ioniz-ation experiments is depicted in Figs. 2-5. Both configurations consisted of a differentially pumped angular RETOF MS (Model D850) inter-faced to a quadrupole ion trap storage device (Model C-1251 manufactured by R.M. Jordan Co., Grass Valley, CA) [28] and a liquid injection sample ionization source.

For the internal ionization experiment, a liquid injection sample source was used to deliver the sample, dissolved in a solvent, through a heated stainless steel tube to the first vacuum chamber. In this chamber, pumped to approximately 150 mTorr, the sample and solvent are vaporized, most of the solvent is removed and the sample passed through a skimmer cone to the ion trap region. In the ion trap, the sample is ionized via 266 nm radiation generated by a Quanta Ray DCR-3 Nd:YAG laser which passes directly through the trap. The resulting ions are then stored in the trap until the trapping potential

is shut off and subsequently an extraction pulse is applied to the exit endcap of the ion trap. The timing and characteristics of the ionization-trapping-extraction processes are described in detail below. This extraction pulse triggers the start for the TOF mass analysis. Upon leaving the trap, the ion packet enters the field-free drift region « 1 m long, at the end of which its velocity is slowed and reversed by the ion reflector. The newly focused ion packet then retraverses the drift region and is detected by a dual 40 mm micro-channelplate detector with a gain of about 106

to 107.

Internal ionization sample injection system

An ISCO Model μΙΧ-500 micropump LC syr-inge pump was used to deliver the sample through a 50 μηι fused silica capillary directly to the vaporiz-ation chamber. The flow rates were typically 50 \*L min- 1 and the solvent used was methanol degassed in an ultrasonic bath. The 50 μηι fused silica capillary was inserted through a zero-dead-volume-tee into a 1 in long, -^ in diameter stainless steel (SS) tube with 0.02 in i.d. A tightly fitting \ in copper tube was placed over this SS tube and was tightly wound with -^ in thermocoax cable heater. This copper tube was added to the sample intro-duction assembly to reduce the stress on the wound thermocoax cable heater. The working temperature of the nebulizer was controlled by a thermocouple and a temperature controller. Working tempera-tures were generally 90-110°C; however, the actual sample temperature is expected to be much lower due to the expansion and heat carried away by the solvent vapor. This assembly is directly inserted and o-ring sealed to the back of the vapor-ization chamber (Fig. 1). The vaporization cham-ber is pumped to a pressure of approximately 150mtorr by a 650Lmin_1 mechanical pump. The sample vapor is injected from the vaporiz-ation region through a skimmer fitted with a 3 in long stainless steel tube (0.04 in i.d.) into the ion trap region via the pressure difference.

ton

ref

lect

or

40 m

m M

CP

de

tect

or

Cu

tub

e

ion

gau

ge

| tu

be

fligh

t tu

be

liner

. V

L

VR

1 V

R2

arid

, VA

■Ein

zel

len

slÏ

VX

Ï <1' —

— _

j

ss tu

be

mec

hani

cal

6" d

iffus

ion

V

L

pum

p

pum

p

VX

2

bea

m

stee

ring

pl

ates

tin

VF

OC

4

-dW

i usi

on

pum

p

1.0

m

Fig

. 1.

The

IT

-RE

TO

F M

S w

ith

inte

rnal

ion

izat

ion

inte

rfac

e an

d sa

mpl

e in

trod

ucti

on r

egio

n sh

own.

Whe

re K

r f

= ra

diof

requ

ency

vol

tage

, F

extr

= d.

c. e

xtra

ctio

n vo

ltag

e,

VA

= ac

cele

rati

on v

olta

ge,

VL

= f

light

tub

e li

ner

volt

age, V

FO

C =

foc

usin

g vo

ltag

e, V

x2

= be

am s

teer

ing

volt

age,

VR

X =

ion

rep

elle

r vo

ltag

e, V

K2

= io

n re

flec

tor

volt

age.

(R

epri

nted

wit

h pe

rmis

sion

fro

m r

ef.

28.)


insulated tungsten needle

ss tube

fused silica capillary

heated pneumatic nebulizer assembly

ion beam condenser, V,

lens

skimmers

IT/ReTOF MS

Vaporization chamber

Glow Discharge API chamber

Differentially pumped dual orifice interface

Fig. 2. The sample introduction source including nebulizer, vaporization chamber, API glow discharge chamber, ion beam condenser lens, and differentially pumped dual orifice interface. (Reprinted with permission from ref. 34.)

Internal ionization and trapping

The sample vapor present in the ion trap is ionized by UV laser radiation (λ = 266 nm) which is directed through 3.1 mm holes in the ring elec-trode of the ion trap. Ions which are selectively produced by laser-induced REMPI are trapped by the applied r.f. potential, while other ions and unionized sample and solvent vapor are pumped away by the 6 in diffusion pump located directly below the ion trap.

The ion trap consists of two endcap electrodes with a ring electrode between them. These elec-trodes have hyperbolic surfaces and are config-ured as shown in Fig. 6. The ion trap was completely enclosed with ceramic spacers placed between the ring and endcaps except for an inlet and exit aperture 3.1 mm in diameter on the end-caps and two (3.1 mm) holes for the laser beam to pass through. A ~ in stainless-steel tubing with 0.02 in i.d. was tightly fitted into a hole on the

ring electrode in order to introduce helium or other gases into the trap to increase the local pres-sure when needed. A Vernier needle valve was used to finely control the amount of gas admitted into the trap. Typical pressure in the ion trap ranged from 5 x 10"4 to 10~3Torr. During operation, both endcaps are held at 0 V while an r.f. signal of constant frequency (1.0 MHz) and variable amplitude (0-460 Vpp) is applied to the ring elec-trode. This applied r.f. field serves to trap ions present within the volume of the trap. Varying the r.f. amplitude varies the m/z range of ions that are stable within the trap. Ions with appropri-ate m/z for a particular r.f. amplitude have a stable trajectory within the trap and, therefore, are trapped [25]. The mass range of the ions that will be trapped was approximated by computer simula-tion [29]. After a selected m/z range has been trapped, a d.c. pulse was applied to the exit endcap to simultaneously extract all ions from the trap for TOF analysis as detailed below.

ion

ref

lect

or

40

mm

MC

P

dete

ctor

io

n g

auge

| tu

be

fligh

t tu

be

liner

. V

L

VR

1 V

R2

Vex

tr

acce

lera

tion

inze

l len

s <S

——

—i

f VX

2

bea

m

stee

ring

Vap

oriz

atio

n

AP

I Glo

w D

isch

arge

C

ham

ber

Cha

mbe

r 6"

diff

usio

n

\/i

v/F

OC

pu

mp

■

¥

rv

A#

4-d

Ii Fusio

n

pum

p

1.0

m

f ft

? Y

Fig

. 3.

Ext

erna

l io

n in

ject

ion-

ion

trap

ref

lect

ron

TO

F M

S w

ith

exte

rnal

d.c

. pl

asm

a io

niza

tion

sou

rce.

Fp

ush

= A

PI

cham

ber'

s d.

c. p

ush

volt

age,

FA

, K

B =

pre

trap

Ein

zel

lens

vol

tage

s, V

rSm

= r

adio

freq

uenc

y po

tent

ial,

Kex

tr =

d.c

. ext

ract

ion

volt

age,

VL

= V

xx

= f

light

tub

e li

ner

volt

age, V

FO

C =

foc

ussi

ng v

olta

ge,

VX

2 =

bea

m s

teer

ing

volt

age,

F

R1

= io

n re

pell

er v

olta

ge,

VR

2 =

ion

refl

ecto

r vo

ltag

e. (

Rep

rint

ed w

ith

perm

issi

on f

rom

ref

. 34

.)


o-ring

z.d.v. tee Cutube thermocoax

f.s. capillary

\ ^ l ^ i ^ i ^ i ^ i & ^ S ^ ^ s ^ ^

ί^Τ^Τ^Τ^Τ^Γ^Γ^^Τ^^Τ^Τ^

27 gauge ss needle

skimmer

nebulizing gas inlet

mechanical pump

Fig. 4. Detail of the ESI source.

The ionization, trapping, extraction, and detec-tion processes were timed as follows: the Q-switch output pulse of the laser was used to trigger a Systron-Donner Corp. Datapulse 100A Pulse Gen-erator (PG) and a California Avionics Labora-tories, Inc. Model 112AR Digital Delay Generator (DDG). The PG output pulse was used to trigger the r.f. power supply (which pro-vided the trapping potential to the ion trap). The DDG output pulse triggered the extraction pulse supply and the LeCroy Model 9400A digital oscilloscope (DOSC) data collection system. In Fig. 7, a typical experimental cycle is depicted and is described in detail with the timing of the internal ionization experiments.

The external ionization and trapping exper-iments consisted of both a plasma ionization source and an ESI source. For plasma ionization, the liquid injection source was used to deliver the sample, dissolved in a solvent, through a heated pneumatic nebulizer assembly to the vaporization chamber where the sample is vaporized and the solvent removed. The sample then passes through a channel to a separate second chamber and is ionized via a d.c. plasma source in 1 atm He. The resulting ions are injected through a pair of differ-entially pumped skimmers (« 1.5 Torr), which

sample the on-axis component of the ion beam. Alternatively, the vaporized sample can be sampled through the first skimmer and ionized in the differentially pumped region via a glow dis-charge mechanism under reduced pressure. A third method of external ionization used was elec-trospray which has its own detailed description below. The ions produced externally, regardless of method, were transported into the mass spectro-meter region and collimated by a set of Einzel lens into the ion trap device. The ions were stored or accumulated until the trapping potential is shut off and subsequently an extraction pulse is applied to the exit endcap of the ion trap, starting the TOF mass analysis as described above. The timing and characteristics of the trapping-extraction pro-cesses are described in detail below.

External ionization sample injection

In order to inject a variety of samples, including samples of relatively low volatility, into the TOF system, a liquid chromatography-atmospheric pressure ionization interface (LC-API) using a relatively high current plasma source in 1 atm He was employed [30]. The system consists of a heated pneumatic nebulizer, a vaporization chamber, a

ion

refle

ctor

40m

mM

CP

dete

ctor

C

ion

gaug

e I

tube

fli

ght t

ube

liner

. VL

VR1

VR2

Vext

r ac

cele

ratio

n

ther

moc

oax

moc

oax

r—p—

88

tube

I

|J

f W

Einz

el le

ns

w

<s—

... i

▼ T

fTf

Elec

trosp

ray

Asse

mbl

y m

echa

nica

l pu

mp

+++ f VX

2

beam

st

eerin

g pl

ates

6-di

ffusi

on

VL

.

VFO

C

pum

p 4-

dtff

i usio

n

P"m

P 1.

0 m

Fig

. 5.

Ext

erna

l io

n in

ject

ion-

ion

trap

ref

lect

ron

TO

F i

nstr

umen

t w

ith e

lect

rosp

ray

sour

ce.

Det

ails

are

the

sam

e as

tho

se f

or F

ig. 3

.


quadrupole ion trap cross section top view

ceramic spacers

3.1 mm to detector

r*7.1 mm z » 10.0 mm

x exit endcap

ring electrode

Fig. 6. Detailed construction of the quadrupole ion trap.

plasma source ionization chamber, and a differen-tially pumped dual orifice interface. The details of the system are as follows (Fig. 3).

An LC syringe pump (Varian Model 8500) was used to deliver the sample through a 180/im fused silica capillary to the vaporization chamber. The flow rates were typically varied from 30 /iL min"1

up to 0.5 mL min"1 and the solvent used was methanol degassed in an ultrasonic bath. The 180 /im fused silica capillary was inserted through a zero-dead-volume tee into a coaxial ^ in stain-less-steel tube with a 0.01 in i.d. The liquid was pneumatically nebulized into fine droplets using a high velocity jet of helium (1.2 L min"1) at the end of the fused silica capillary [30,31]. The generated mist was then swept through a | i n heated brass tube with another helium gas stream (« 200 mL min"1) into the vaporization chamber. The helium flow prevents the sample from deposit-ing on the surface of the heated tube and conducts heat to vaporize the solvent. Two fine control flow meters were used to control the gas flow of the nebulizer in order to achieve the best signal. A ^ i n thermocoax heater was used heat the | in brass tubing. The heated section is a \ in long, în

in diameter cylindrical brass block tightly wound with ^ i n thermocoax cable heater through which the γζ in tube is inserted. This assembly is directly o-ring sealed to the back end of the vaporization chamber. The vaporization chamber consists of a 1.0 in x 1.0 in. diameter cylindrical brass chamber and a heated brass block with a 0.5 in x 0.03 in diameter channel connected to the ionization chamber. The vaporized samples were transferred through the heated narrow channel into the ioniza-tion chamber. The temperatures of the pneumatic nebulizer and vaporization chamber can be con-trolled independently using a thermocouple and temperature controller. The typical working tem-perature was 150-250°C. However, the actual working temperature of the sample is expected to be much lower than the body temperature of the vaporization chamber or pneumatic nebulizer since heat is carried away by the vaporized solvent and carrier gas. Because of the solvent vapor and high He flow used in the nebulizer, the actual pressure in this chamber is slightly greater than 1 atm.

External ionization

Ionization was produced at atmospheric pres-sure or at reduced pressure in the differentially pumped interface with the plasma ionization source or at atmospheric pressure with the ESI source. The atmospheric pressure ionization pro-cess provides soft ionization, where the proto-nated molecular ion generally is the predominant peak observed. In the reduced pressure ionization process relatively soft ionization or fragmentation can be controlled by altering the experimental conditions.

API. The ionization chamber (Figs. 2 and 3) is directly attached to the skimmer and consists of 1.0 in x 0.8 in (diameter) brass cell with two în Cajon ports with quartz windows for observation and one | in Cajon port for the discharge electrode. The glow discharge is formed with the use of a 0.04 in diameter tungsten rod which is ground to a sharp tip. The rod is covered with glass or Teflon

Pul

se T

imin

g of E

xper

imen

tal C

ycle

Q s

witc

h ou

tput

pul

se

LeC

roy

9400

A

DO

SC

(Trig

ger p

ulse

sta

rts T

OP

mas

s

1 j

TOFM

S

ι 1

1 e

xtra

ctio

n 1 1

1

1 D

ufts

esuo

olv

1 1

r1 * 1

IDDG

" f"

L—

digg

er

|

1 '

*!

^-

"-

—"

■■

—^

'^

r.f.

pwr.

L

i A

POG

A

4

Ml

1 sa

mpl

e j «

in

let

hv

π

|Nd:

YA

O

1 be

er

loe

wte

h

LA

JU

LA

J M

OO

■ v

/

ooec

[^

^■

"•

^■

"•

^

Y

1 Pl

otte

r

^ 1

PC

/AT

/ ^

V

) i—

i

ΤΓΠ

l· m

mm

m^m

X

—

j

Fig

. 7.

Tim

ing

dia

gra

m a

nd

sch

emat

ic o

f th

e ex

per

imen

tal

set

up f

or t

he

inte

rnal

io

niz

atio

n e

xp

erim

ent.

(R

epri

nte

d w

ith

per

mis

sio

n f

rom

ref

. 34

.)

B.M. Chien et all Int. J. Mass Spectrom. Ion Processes 131 (1994) 149-179 159

insulation and only the tip is exposed to prevent an unstable discharge from occurring on the side of the rod. In order to initiate the plasma source - 1 to -2 kV was applied to the electrode. When a glow was obtained, the actual operating voltage was dropped as much as 100-500 V from the initial breakdown voltage and a current of 0.1-0.6 mA was maintained. The resulting plasma appeared as a white and blue glow extending from the tip of the electrode. The most stable glow was main-tained in 1 atm of helium. The vaporized sample is ionized via ion/molecule reactions induced by the He d.c. plasma source as described in previous work [32]. The ionized sample is transported into a dual-orifice differentially pumped interface. The first sampling orifice of this atmospheric pressure interface is 275 μτη and the second orifice inlet to the mass spectrometer is 325 m. The region between the two orifices is pumped by a 65Lmin~l mechanical pump to a pressure of « 1.5 Torr. The pressure in the mass spectrometer acceleration region under these conditions is « 1 x 10"6 Torr. An ion beam condenser with vol-tage Flens is placed between the*" two orifices to enhance the transmission of ions into the mass spectrometer. Kpush is a small positive voltage placed on the entire atmospheric presssure inter-face so that there is a voltage drop between the two orifices.

Reduced pressure ionization. The vaporized sample traversed through the ionization chamber and was sampled into the differentially pumped region. Instead of initiating a glow discharge at the tip of the electrode in the ionization chamber, -200 to -500 V was applied to the entire API chamber to initiate the glow discharge between the first skimmer and the ion beam condenser (maintained at ground potential). Under these con-ditions a bright glow was formed with a current of 0.1-1.0 mA and an operating voltage about 100 to 300 V lower than the initial breakdown voltage. When operated under ambient air conditions, some fragmentation was normally observed in the mass spectrum since higher voltage was required to

induce breakdown and maintain the glow. How-ever, if helium was introduced into the system, voltage as low as —60 V is sufficient to maintain a glow and thus soft ionization can be achieved. By increasing the discharge voltage, fragmentation can be induced for structural information.

ESL Another API external ion trapping mode of operation was ESI. In this experiment the ISCO LC syringe pump was used to deliver the sample through a ΙΟΟμηι fused silica capillary directly inserted into the 27 gauge SS needle. This needle was supported in a SS zero-dead-volume tee through which a nebulizing gas flow could be applied. The tee-needle sprayer assembly was maintained at 3.5-4.5 kV relative to the inlet capil-lary tube. The flow rates were typically 0.5-ΙΟμί ιηΐη - 1 and the solvent used was a mixture of methanol and water of various proportions (50/50-100/0 MeOH/H20), determined by solubi-lity requirements of any particular sample. The solvent was degassed in an ultrasonic bath prior to use. All samples were made acidic by addition of acetic acid, as is standard to enhance ionization in positive ion mode electrospray (Figs. 4 and 5).

The sprayer assembly was directed onto a ^ i n 200 mm long 0.5 mm i.d. heated capillary inlet tube. A tightly fitting \ in copper tube was placed over this SS tube and was tightly wound with -^ in ther-mocoax cable heater. The working temperature of the inlet was controlled by a thermocouple and a temperature controller. Working temperatures were generally 70-180°C to assist in desolvating the sampled electrosprayed droplets. This assem-bly is directly inserted into the first vacuum cham-ber which was pumped to a pressure of approximately 1.5 Torr by a 650 L min"1 mech-anical pump. The exit of this heated tube is direc-ted onto the previously mentioned 325/im skimmer, i.e. the second skimmer of the set-up in Fig. 3. The inlet tube and skimmer typically had potentials of 50-350 V and 0-120 V applied to them, respectively. The distance between the tube and skimmer is adjustable and is usually main-tained at 2-7 mm. There is a cylindrical lens

160 B.M. Chien et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 149-179

around the space between the tube exit and skim-mer which assists in focussing the ions into the skimmer. This lens has an applied potential of 70-200 V.

The analyte solution is electrosprayed producing highly charged droplets which are pushed toward the inlet tube by electrostatic forces. Some of these droplets enter the tube and pass through it with the help of the tube's applied potential and the ambient pressure gradient. The ions exiting the tube are then drawn into the skimmer via the potential dif-ference between the tube and the skimmer. In this space the ions undergo many collisions and by altering the potential difference, the pressure and the gap length, fragmentation can be enhanced or reduced [33].

Trapping

The ions produced by the external ionization sources were sampled through the second skimmer (325 μιη) and entered the mass spectrometer region. The ions underwent a supersonic jet expan-sion and cooling, then were collimated by a set of Einzel lens into the ion trap device.

The ion trap is the same as in the internal ioniz-ation experiment except the two holes that allowed the laser beam to pass through the trap are closed with Teflon plugs. In some cases, if the energy of the ions from the ionization source is high, the ring electrode and both endcaps can be biased with positive voltage to slow down the ions and improve the trapping efficiency. Using this method, ions with energy as high as 300 eV can be injected into and stored in the trap as will be discussed in more detail later.

The trapping, extraction, and detection pro-cesses were timed as follows. A Global Specialties Co., 4001 pulse generator (PG) was used to trigger two California Avionics Laboratories, Inc. model 112 AR digital delay generators (DDG1 and DDG2). The DDG1 was used to trigger the r.f. power supply. An R.M. Jordan Co. r.f. power supply operated at 1.0 MHz, 0-460 Vpp was used to trap ions up to m/z 185, while a modified EAI r.f.

power supply with variable amplitude 0-2200 Vpp, 1.1 MHz output was used to trap ions with m/z greater than 200. The DDG2 output pulse trig-gered the extraction puiser and this puiser passed its pulse to trigger the Lecroy model 9400A DOSC or Precision Instrument Inc. model 9825 Signal Averager (SA) (see Fig. 8). The signal from the detector was amplified by a Stanford Research Systems Model SR445 DC-300MHz amplifier when needed.

In Fig. 8, a typical experiment cycle for external ionization is depicted: the PG (+10V) triggered both DDGs. The r.f. trapping potential remains on while this sequence occurs, thus trapping the ions continuously produced from the external ion-ization source using the buffer gas to collisionally relax the energetic ions produced by this source. The r.f. potential remains on until the DDG1 pulse triggers the r.f. power supply. Then, 2//s (internal delay of r.f. power supply) after the rising edge of the DDG1 pulse arrives, the r.f. potential is shut off and remains off for the entire pulse width of the DDG1 pulse. DDG1 output pulses (20 ns rise time, + 10V amplitude, 10^s FWHM) can be delayed from a few microseconds to as long as 10 s after triggering. Therefore, the delay of the DDG1 determined the duration of ion trapping, i.e., how long the ions could be stored or accumulated in the trap. After its set delay, the DDG2 outputs a pulse (20 ns rise time, +10 V amplitude, 10 /is FWHM) which triggers the extraction puiser. The extraction puiser serves the dual purpose of both providing a start time reference for the TOF mass analysis and providing an extraction pulse to the exit endcap of the ion trap, thereby ejecting the ions from the trap. Upon the arrival of the rising edge of the DDG2 pulse, the extraction puiser passes this trigger pulse, the DDG2 output pulse, to the DOSC or SAs external trigger. This provides the start time refer-ence for the TOF mass analysis. Following an interval of 1.5/xs after the arrival of the rising edge of the DDG2 output pulse, the extraction puiser sends the extraction pulse to the exit endcap of the ion trap. This extraction pulse was a d.c. square wave -150 V in amplitude and 2 /is in width

Pul

se T

imin

g o

f Exp

erim

enta

l Cyc

le

Pul

se g

ener

ator

tri

gger

pul

se

Var

iabl

e re

p ra

le o

f pul

se g

ener

alor

DD

G1

Del

ay

DD

G2

Del

ay

2|ftS

8C

(Trig

ger p

ulse

sta

rts T

OF

mas

s

1 R

eTO

FMS

r* ^}

_1!Γ

i i

1 1

pids

or

1 1

r1 * 1

IDD

G2

1 i i

l_

r.f.p

wr.

1

* m

DO

G1

A

Mgg

er

1 P

Use

1

Gen

erat

or

Ci

\ l*

Cro

y IA

JO

-J M

OO

DO

SC

or

RS

232

L^

»^ 1

PN

825S

A

m

m

i

386

PC

1 '

' T

IHM

j "

^1

Fig.

8. T

imin

g di

agra

m a

nd s

chem

atic

of

the

expe

rimen

tal

set u

p fo

r th

e ex

tern

al i

oniz

atio

n ex

perim

ent.

(Rep

rint

ed w

ith p

erm

issi

on f

rom

ref

. 34

.)

162 B.M. Chien et ail Int. J. Mass Spectrom. Ion Processes 131 (1994) 149-179

with 10 ns rise and fall times. The delay of DDG2 was set to coincide with the delay of DDG1, there-fore, extraction of the ions occurred simulta-neously with the r.f. trapping potential being triggered off. The repetition rate of the pulse gen-erator can be set at 0.1-10 KHz, limited by the ions with the longest flight time, therefore the repetition rate of the entire experimental cycle varies from 0.1 to 10 KHz.

In the internal ionization experiment the timing details were the same with the following exceptions (see Fig. 1). The Q-switch output pulse from the Nd : YAG laser triggered the PG and the DDG. After a 50 ns internal delay the laser pulse fires (7 ns FWHM at 266 nm) and ionizes the sample vapor in the ion trap. The r.f. trapping potential remains on while this sequence occurs, thus trap-ping the ions produced by the laser beam. The PG pulse triggers the r.f. power supply so that the delay of the PG determined the duration of ion trapping. The DDG triggered the extraction pulse supply which serves the dual purpose of both providing a start time reference for the TOF mass analysis and providing an extraction pulse to the exit end-cap of the ion trap, thereby ejecting the ions from the trap.

TOF Operation

The ions upon exiting the ion trap, pass through a set of accelerating plates and an Einzel lens which serve to focus the ion packet and accelerate it into the field-free flight tube region through a potential difference of about -1400 V. A pair of beam deflecting plates are then used to steer the ions towards the ion repeller-reflector assembly, where the ion packet is more tightly focused, reversed in direction, and reaccelerated through the flight tube (with angular displacement from its initial axis of trajectory) onto the 40 mm dual microchannel plate detector (Model C726, R.M. Jordan Co.). In more recent work, the detector has been updated to a 40 mm triple microchannelplate detector (Model C726-T, R.M. Jordan Co.) for enhanced sen-sitivity. The reflectron flight tube is pumped by a

Varian VHS 4 diffusion pump while the main chamber is pumped by a Varian VHS 6 diffusion pump. A restriction of 1 in tubing is placed between the flight tube and the main chamber, which pro-duced typical operating pressures of 8 x 10"6 and 1 x 10~6Torr, respectively. The actual pressure in the ion trap during sample introduction though was between 5 x 10~4 and 10~3Torr.

The TOF of the extracted ion packets was measured on the DOSC. Signal averaging was used to enhance the signal-to-noise ratio and reported spectra are averages of 100 single waveforms unless noted otherwise. In the experiment, DDG2 simultaneously triggered the DOSC and the extrac-tion puiser. The ion signals from the detector were sent to the input of the DOSC, and the time differ-ence between various ion peaks and the trigger (/ — 0) reference provides the time of flight of each ion.

The TOF spectra in the DOSC are then trans-ferred to a 386 IBM compatible PC using an RS 232 interface bus established between the DOSC and the computer. A user written QUICKBASIC pro-gram was used to control the transfer processes. The size for each data point was 16 bit for our experiments. The raw data from DOSC were in ASCII form. It was converted into signed decimal form by a user-written program.

A study of system resolution performance found that the spectral resolution is limited by the time resolution of the Lecroy 9400A DOSC (10 ns). Later experiments used a Precision Instrument Inc. PI9825 SA in order to take advantage of its 5 ns time resolution. The PI9825 model is a com-plete SA on two PC-AT cards inserted onto a 386/ 486 PC motherboard with a maximum digitization rate of 200 M sample per second. Software accom-panying this product provides a graphic control panel used to acquire and store the data on a real-time basis. The stored data in binary form was converted into ASCII form by a user-written program and loaded into commercial spreadsheet software such as EXCEL or LOTUS for further data analysis.

Mass calibration was performed by measuring


Aniline No trapping

Aniline Trapping time: 1 msec

10Ö

c Φ

1 °"

OC

Ü -1

(A)

)

(B)

Α ^ j , , , !

J 1 1 1 1—

<λ J 1 1

110

32 33 34 35 36 37 38 39 40 41 42 Time of Flight(usec)

Fig. 9. IT-RETOF mass spectra of aniline with no r.f. trapping potential applied. (A) 20.10^s delay before extraction; (B) no delay before extraction. Both spectra were obtained with the following operating conditions. VL = Vxx = -1100 V, VX2 = -980 V, Kpoc = -400 V, VA = -300 V, Kextr. = -150 V, VKl = +1200 V, VK2 = +1950 V. (Reprinted with per-mission from ref. 28.]

the time of flight (T) of a few known masses (such as background water or methanol cluster ion signal) to find out the constant x and y in the empirical equation by linear regression analysis, m/z = xT2 + y.


Internal ionization

Mass spectra. The capabilities of the ion t rap-time-of-flight device are demonstrated in the mass specra of Figs. 9-12. These spectra were taken with 7 x 10~4M aniline dissolved in methanol. Sample and solvent were pumped into the interface cham-ber via the syringe pump set at a flow rate of 50 L min"1. Vaporization, delivery to the ion trap region, and ionization occurred as described above. Figures 9(A) and 9(B) are mass spectra of aniline without ion trapping, i.e. no r.f. applied to the ring electrode of the ion trap. Figure 9(B) shows the extraction and TOF mass analysis of aniline ions by applying a -150V d.c. pulse to the exit endcap of the ion trap 0.10 /is after laser ionization

32 33 34 35 36 37 38 39 40 41 42 Time of Flight (usec)

Fig. 10. IT-RETOF mass spectrum of aniline obtained by laser induced REMPI at 266 nm with 1.0 ms trapping time. R.f. frequency = 1.19 MHz, r.f. voltage = 180 Kpp, and all other operating conditions the same as those of Fig. 9. (Reprinted with permission from ref. 28.)

has occurred. The m/z 93 peak is assigned to the aniline molecular ion. The 13C isotope peak is pre-sent at m/z 94 at approximately 6% of the mole-cular ion peak's intensity. The m/z 77 peak is the C6Hj fragment. Increasing the laser power increases the relative intensity of the QH5" peak. In Fig. 9(A) the spectrum is of the same sample and conditions as Fig. 9(B) except that the extraction pulse is delayed 20.10μ8. This spectrum shows that with no r.f. applied to the ring electrode of the ion trap, all ions produced are pumped away by the 6 in diffusion pump, thus, only a flat line background was detected. Figure 10 shows a spectrum of the same sample with the r.f. potential applied to the ring electrode of the ion trap and an extraction delay of 1 ms (1 ms trapping time). This spectrum clearly illustrates that the ions are successfully trapped and the trapping results in an increase in resolution.

The ability of the IT-RETOF mass spectro-meter to trap ions for an extended period of time is shown in Fig. 11. In Fig. 11 an aniline spectrum with 2 s trapping time using a laser repetition rate of 0.5 Hz results in a similar mass spectrum with similar resolution as that with a 1 ms trapping time.


100

ANILINE Trapping time: 2 sec

34 36 38 40 Time of Flight (usée)

44

Fig. 11. IT-RETOF mass spectrum of aniline obtained by laser induced REMPI at 266 nm with 2 s trapping time. All operating conditions were the same as those in Fig. 7 except the laser repetition rate was reduced to 0.5 Hz. (Reprinted with permis-sion from ref. 28.)

100

ANILINE Trapping time: 5 msec

25 30 35 Time of Flight (usec)

Fig. 12. IT-RETOF mass spectrum of aniline obtained by laser induced REMPI at 266 nm. The laser power has been increased to induce fragmentation which is stored in the trap for 5 ms and detected by the RETOF MS. All operating conditions were the same as those in Fig. 7. (Reprinted with permission from ref. 28.)

Although the mass spectra in Figs. 10 and 11 are similar, the relative peak height after 2 s of storage typically decreases by a factor of 5 to 10 relative to storage after 1 ms as ions are lost from the trap. In Fig. 12 is shown a laser REMPI mass spectrum of aniline obtained when the laser power is increased to produce extensive fragmentation. This mass spectrum demonstrates that the molecular ion and its accompanying fragments can all be stored for an extended time, i.e. 5 ms, by the trap and detected by the RETOF. The laser REMPI mass spectrum is essentially the same as that obtained without trapping. The resolution in this figure is slightly degraded compared to the other figures due to the extended mass range processed by the digitizer. An expanded view of any of these peaks reveals a resolution very similar to that obtained in Figs. 9-11.

Experimental parameters

R.f. frequency and voltage. The r.f. potential applied to the ring electrode determined the m/z range that resulted in stable trajectories in the trap, that is, the applied r.f. frequency determined

the m/z range trapped. Varying the r.f. frequency was experimentally determined to move the observed m/z range that was trapped. As the fre-quency was lowered, a region of lower m/z ions were trapped and as the frequency was increased this "window" shifted toward higher m/z. Other than a shift in m/z values over which the "win-dow" was located, frequency variations showed no changes in instrumental performance in the experiment. The experimental results are reported with 1.19 MHz as the selected r.f. frequency, as this setting was compatible with our chosen m/z range (i.e. the "trapping window" roughly centered on m/z 93). Frequencies in the range 800 kHz-1.19 MHz were studied. Varying the amplitude of the r.f. voltage applied at a fixed frequency also served to move the "trapping window." A limita-tion observed here was that r.f. voltage amplitudes less than 160Vpp were too low to trap the ions studied in this work.

D.c. extraction pulse. The extraction voltage was varied from - 5 0 to -350 V. It was found that — 150V provided both the best signal intensity


and resolution. If the extraction voltage was increased significantly, both the resolution and sig-nal were degraded. This is due to the fact that the reflectron performance is very sensitive to the exact voltages used on the grids. Various extraction pulse widths were also attempted. A pulse width of approximately 0.5/xs or less was too short to allow significant extraction of ions from the trap, resulting in decreased signal intensity or no observed signal. A pulse width from > 0.5 /is to «5.0/xs resulted in the best signal intensities, with « 2.0/xs yielding the best resolution.

Resolution and sensitivity. The peak width (FWHM) of the molecular ion peak in Fig 9(B), where there was no trapping, was 30 ns. This cor-responds to a resolution of nearly 700. In Fig. 10, the molecular ion peak width was 16 ns, which is a resolution of 1300 at m/z 93, when the trapping capability is utilized. This enhanced resolution pro-vided by the ion trapping is probably due to improved spatial resolution of the ions. As shown in recent computer simulations, the extended trap-ping of ions results in their accumulation in the center of the trap [29]. Upon d.c. ejection, this accumulation results in a marked improvement in the spatial resolution factor in the TOF device, as opposed to the untrapped situation where the ions are produced over the diameter (« 2 mm) of the laser ionization beam. A well-defined ion packet in a narrow region of space might yet prove to be particularly suited for use in obtaining high resol-ution in a gridless reflectron device.

A measurement of the sensitivity was also per-formed in the IT-RETOF using laser induced REMPI at 266 nm for benzene. The benzene was dissolved in methanol in a concentration of 10~5 M and a series of successive dilutions down to 10~7 M were made. The sample was injected as before using the microsyringe pump at a rate of 30/iLmin"1

into the trap. The 266 nm radiation was used at a power density of 5 x 105Wcm~2 which was the highest power at which ionization could be pro-duced without significant fragmentation. In the case where no storage was used, a lower limit of

detection o f « 600 fmol (S/N = 3) was obtained. If a storage time of 90 ms was used following a single laser pulse then the detection limit was « 200 fmol, i.e. at least a factor of 3 improvement over no trapping. This is probably partially due to the smal-ler ion cloud following trapping which is more efficiently ejected along the z-axis of the trap. How-ever, if the ions are trapped for 1 s and the laser pulsed ten times during this period, then a detec-tion limit of « 60 fmol is achieved. Thus, the pro-duction and storage of ions over several laser pulses enhances the sensitivity by a factor of nearly 10 compared to the situation where no storage is used. Of course, during this storage time some fraction of the ions will be lost from the trap and at some point saturation of the trap with resulting space charge losses will result. Indeed it is found that a saturation point is reached after storage of about 4 pulses. Additional pulses do not appear to increase the detection limit while inducing space charge broadening at the given laser power. How-ever, at lower laser power where less ions are pro-duced per laser pulse or in a sample molecule where ionization is much less efficient than benzene, the use of additional laser pulses within a storage time interval will be increasingly important for enhan-cing the sensitivity for detection. Nevertheless, in the case illustrated herein, the use of ion trap stor-age with the RETOF can provide excellent sensi-tivity and may enhance the sensitivity even further where low signal levels are present.

External injection

API. The capabilities of the IT-RETOF for detection of ions generated externally by the API d.c. plasma source are demonstrated in Figs 13-15 [34]. Figure 13 shows the API mass spectrum of the molecular ion region of pyridine injected into the IT-RETOF using a 10~6M pyridine solution in methanol. The vaporization and ionization of the liquid sample and the transmission of the resulting ions to the ion trap were performed as described previously. The spectrum obtained reveals three fully resolved peaks at m/z 78, 79 and 80 which


100 η

90 A

80 A

70 A

$ 60 -J * I

> I

1

ί 40-1

30 A

20 A

10 J

Atmospheric Pressure lonization of Pyridine, MW 79, 950 ms trapping time. Resolution > 1500

M'

J

MH

11 <0 en

Tim· of Right ( utec )

Fig. 13. API mass spectrum of pyridine with Kpush = 0V, F,ens = +200 V, VA = +90V, FB = +190V, Frf =+400 FpP, VcxtT. = -300 V, VL= VX] = -1500 V, K F O C =0V, VX2 = -1385 V, VKl =+110V, FR2 = +1000V, Fbias = + 2 8 0 V. (Reprinted with permission from

ref. 34.)

are assigned to the molecular ion, the protonated molecular ion and its 13C isotopic peak respec-tively. This spectrum was obtained for a storage time of 950 ms and the resulting resolution was > 1500. In this case the R.M. Jordan r.f. power supply was used to supply r.f. voltage to the ring electrode. A voltage of « 400 Vpp was used at a frequency of 1.0 MHz in this spectrum. The r.f. voltage is pulsed off before the ions are ejected into the RETOF which results in a peak width FWHM of Uns and thus relatively high resolu-tion is observed. The sensitivity here is also excel-lent where only 160fmol of analyte was used to obtain this signal. The ability to pulse off the r.f. power supply prior to d.c. ejection provides better resolution in this spectrum than if the r.f. is not

pulsed off. However, the maximum voltage of 460 Vpp available for this supply limits the ability to trap externally injected ions [34] from the d.c. plasma source to m/z 185.

In Fig. 14 is shown API d.c. plasma source IT-RETOF mass spectra of amitriptyline. The spectrum in Fig. 14 was obtained using 10"6M amitriptyline dissolved in methanol which was vaporized and ionized as described previously. The mass spectrum obtained consists of the proto-nated molecular ion peak at m/z 279 and its 13C isotopic peak at m/z 280. This spectrum was obtained using a modified EAI r.f. quadrupole power supply operating at 1.1 MHz and 530 Vpp. This unit could supply the additional r.f. voltage, up to 2200 Vpp, required to trap m/z > 185. This


Atmospheric Pressure lonization of Amitriptyline, MW 278, 90 ms trapping time

100 j

90 4-

80 4-

70 4-

ξ βο 4-

H î 40 j

30 4-

20 4-

10 4-

10

MH

H h H H H 1 1 h

δ δ δ δ δ δ δ δ δ δ δ δ δ δ δ δ CO

IO 8 S s s Time of Right ( usée >

Fig. 14. API mass spectrum of amitriptyline with Kpush = +75V, Klens = +25V, KA = -140 V, KB = -70V, VIÎ = +S30V?p9

Kextr = -170V, VL = KX1 = -1500V, VFOC = -1250V, VX2 = -1385V, VKl = -100V, VK2 = +1000V, Fbias = OV. (Reprinted with permission from ref. 34.)

unit could not be pulsed off before ejection of ions into the RETOF because of the nature of the RF circuit involved. The inability to trigger the r.f. off was shown to result in some peak broadening due to ejection of ions over different phases of the r.f. pulse [28]. Nevertheless, the FWHM width ofm/z 279 is « 20 ns with a resulting resolution of > 1500 using an ion storage time of 90 ms.

In further work, we investigated the effect of the ion storage time on the sensitivity and resolution in the RETOF. This is shown in Figs. 15(a)-(e) for a continuous ion beam where the mass spectrum of pyridine in the IT-RETOF is monitored as a func-tion of trapping time. The pyridine ions were pro-duced using the API d.c. plasma source as in Fig. 13. Figures 15(a)-(e) show that with the use of a

continuous ion beam both the resolution and inten-sity improve with increased trapping time. For example, in trace (e), the ions are trapped for only a very short period and the result is a spec-trum with relatively poor resolution and weak sig-nal intensity similar to that obtained with no trapping under the same conditions. The next spec-trum, (d), shows improvement in resolution and signal intensity with only 1 ms of ion trapping while trace (c), with a moderate trapping time of 90 ms, shows a marked improvement in both of these signal characteristics where a peak of 16 ns FWHM is observed. Trace (b) shows an extension to long trapping time (900 ms) and the resultant improvement in resolution and signal intensity where the observed peak is only 10 ns FWHM.


Effect of trapping time on resolution and signal intensity

i

I <

FWHM - 8 ne

( · )

-JL·

trapping time - 9 sec

( b ) 10 ns

900 ms

( c )

90 ms

( d )

1 ms 21 ns

( e )

28 ns 10 us

J |\ «.*< " - - ~ —■ " - -

IO

s CO CO

1 0

s en 3

m m ιο S oo 8 * * ιο CO CO CO

Time of Right (used

IO o c* IO CO

IO

IO CO

IO

s IÔ CO

Fig. 15. API mass spectra of pyridine as a function of trapping time: (a) 9 s; (b) 900 ms; (c) 90 ms; (d) 1 ms; (e) 10/is. Kpush = +220 V, Vieris = +170 V, VA = -205 V, KB = -150 V, Krf = +460 Fpp, Kextr = -225 V, Kbias = +345 V, and all other conditions the same as

Fig. 13.

Finally, an extremely long trapping time of 9 s in (a) shows some increase in resolution, « 2100 (8 ns FWHM), over the previous case with similar signal intensity. The 8 ns FWHM observed begins to reach the time resolution limit of the PI 9825 sig-nal averager. Thus, an increase in ion trapping time significantly enhances the resolution and signal intensity even up to seconds of storage.

The increase in resolution with extended storage

time is presumed to be due to the improved spatial and energy resolution of the ions in the trap using a collisional buffer gas. The extended trapping of ions results in their accumulation in the center of the trap which, upon extraction, yields a more spatially resolved ion packet and thus enhances the spatial resolution of the RETOF. In addition, the large number of collisions with the He buffer gas as a function of storage time collisionally relax


the translationally hot ions, thus improving the energy component of the resolution. In previous work described earlier [28] using laser-induced REMPI as a means of producing ions inside the trap, the resolution was found to improve rapidly as a function of storage time. In this previous study the resolution reached its optimal value in less than 100 ßs and little improvement was observed for longer storage times. This indicates that the ions were indeed being rapidly relaxed into the center of the trap. However, these experiments involved ions produced by laser-induced REMPI which resulted in an ion energy distribution of < 0.1 eV. In the present study the energy of the ions entering the trap has been measured to be generally around 20 eV although some ions of higher energy may also enter the trap and be trapped. Thus, one might expect a significant enhancement in resol-ution for these translationally hot externally gener-ated ions over an extended trapping time as indeed is observed. Most of the collisional cooling and thus improvement in resolution, though, does appear to occur within the first several hundred milliseconds.

In order to further study the enhancement of resolution as a function of storage time, the ion beam was pulsed so that an ion beam of a defined duration could be injected into the trap and stored for a variable length of time. In this case ion beams of 100 ^s, 500/is, 1 ms, and 10 ms were pulsed into the trap and stored from 100^s up to several sec-onds. At storage times above 2 ms there was essen-tially no difference between the peak widths obtained using either a continuous or pulsed ion beam. At storage times less than 2 ms a small dif-ference in peak width could be observed. This dif-ference amounted to « 10% at « 1.5 ms, where the pulsed ion beam produced a narrower peak width. The most likely explanation for this observation is that, in the case of the continuous ion source, the ions that enter the trap early in the storage cycle will undergo more cooling than the ions that enter towards the end of the cycle. At longer storage times a greater fraction of the ions in the trap will have undergone extensive cooling as compared to

shorter storage times. Thus, at longer storage times the ion peak widths of continuous ion sources will be very similar to that of pulsed ion sources. How-ever, at shorter storage times with continuous ion beams, where the number of ions entering the trap at the end of the cycle may be a significant fraction of the ions stored in the trap, one might expect to observe some broadening of the ion peak in the RETOF as compared to a pulsed ion beam. In these experiments such an effect is indeed observed, although the difference between the ion peak width is small even at short storage times and at long storage times there is no measurable differ-ence. In addition, an attempt was made to study the enhancement of resolution as a function of storage time at different He buffer pressures. However, the work reported herein was performed at a He buffer pressure that optimized the signal. A change in the He buffer pressure resulted in a rapid decrease of the signal so that a study of the storage properties as a function of time at various pressures was not possible.

In addition, the signal intensity was also found to increase as a function of storage time. This is due to the integration of the ion signal which occurs over the entire trapping time; hence, for longer trapping time, enhanced ion accumulation occurs and the signal intensity increases. The ion accumu-lation benefit of the ion trap has a limit in its ability to enhance signal intensity, for at some point saturation of the trap with resulting space charge losses will be observed.

Reduced pressure ionization. External ion injec-tion into the IT-RETOF was also investigated for other ion sources. In particular, a low pressure glow discharge was produced between the two skimmers in the differentially pumped orifice region which was used to ionize organic molecules sampled from the API [35]. The results obtained using this low pressure glow with detection by the IT-RETOF are shown in Fig. 16 for detection of dibenzothiophene sample from liquid injection into an atmospheric pressure source. This figure illus-trates an important advantage of the IT-RETOF


Reduced Pressure Glow Discharge Ionization of Dibenzothiophene« MW 184, 5 ms trapping time.

i I 1

( a )

( b )

fc A à ■

V

I

1 II

(M#OH)H *

R.F. Vpp = 400 V

(H3O) N 2

/

1

R.F. Vpp * 800 V

Expansion View 185

184 " * \ ^

v» i 2 2 m^t m2S *>22 *o 1 Φ Φ * ^ Φ '^^T #^T^T

■ » I » « ■ « M ■ k ~—l>. 1 ^ - » fc — 1

1 1 1 — 1 1 1 1 r -

+ +

1 1

Time of Right ( usec )

Fig. 16. Reduced pressure glow discharge ionization mass spectra of dibenzothiophene. (a) Krf = 400 Vpp; (b) Frf< = 800 Vpp, and VA = +100V, VB = +200V, Fextr. - -160V, VL = VX{ = -1500V, V¥0C = 0V, VX2 = -1385V, VK] = +150V, VR2 - +1000 V.

(Reprinted with permission from ref. 34.)

in its ability to eliminate background ions based upon the selected trapping conditions. This is a potential problem in the presence of large back-ground ion peaks which may saturate the trap or the detector, especially when attempting to amplify the smaller signal peaks. Figures 16(a) and (b) demonstrate the effect of changing the r.f. poten-tial applied to the ring electrode of the ion trap on the observed spectrum and its ability to eliminate background ions. These spectra are of dibenzothio-phene in 10~4 M methanol solution, ionized via the low pressure glow discharge in He. In Fig. 16(a), the r.f. applied was 400 Vpp and the observed spec-trum is dominated by the background ion peaks with a trace analyte peak observed. However, by increasing the applied r.f. to 800 Vpp the trapped

m/z range is shifted to higher m/z, eliminating the background ion peaks from the observed spec-trum as shown in Fig. 16(b). Because the trap under this condition was not saturated by the background ion peaks, more of the analyte was stored and accumulated, resulting in the enhancement of ana-lyte signal intensity. Thus, the r.f. potential applied to the ring electrode determines the m/z range that can be trapped. By varying the r.f. amplitude, the m/z range trapped can be selected to eliminate low mass background.

An important problem encountered in inter-facing an atmospheric pressure generated ion beam to the ion trap is selecting the correct para-meters to optimize the kinetic energy of the beam. The ion trap device is most efficient in trapping ions


of relatively low energy, i.e. 0-20 eV [36-38]. It was found that the pressure in the differentially pumped region is particularly important in the ion trans-mission. In the API experiments performed herein the first skimmer orifice is 275 μιη and the second skimmer orifice is 325 /im. This results in a pressure of « 1.5 Torr in the differentially pumped region. Under these conditions the ion beam energy is very dependent on Klens, i.e. the focussing ring electrode voltage. The signal was found to be most intense with a Klens « 3 0 V.

The key parameters deciding the ion energy and transmission efficiency from the ionization cham-ber through the dual orifice interface were Kpush

and Flens. Fpush aided in pushing the ions from the API chamber into the dual orifice interface. Although the pressure differential was the main factor influencing the drawing of ions into and through the interface region, a Kpush of +70 to + 150V was experimentally determined to signifi-cantly enhance the ion current traversing the inter-face region. This parameter was optimized with each individual experimental run. Limitations were that Fpush < +70 V provided no significant increase in ion throughput and Fpush > +150V caused a reduced pressure glow discharge to form in the differentially pumped interface region. Klens

served to condense the ion beam emanating from the first skimmer and focus it upon the second skimmer's orifice. This parameter was also found to be a critical factor influencing the energy of the ions entering the pre-trap Einzel lens. The Kiens

typically was run at +10 to +35 V. A setting within this range was found to provide effective ion beam focussing without imparting too much energy to the ions. Greater Klens settings create an ion beam with too much energy to effectively trap ions with-out a d.c. bias being applied to the trap.

In further work, it was found that high energy ions could also be stored within the trap. If the first skimmer was changed to 350 μιη, the pressure in the differentially pumped region was typically 5-10 Torr, under our operating conditions. Under these conditions the ion transmission efficiency is very low and only ions with higher energy were

sampled into the mass spectrometer region. Indeed, a high voltage on Flens was needed to efficiently focus ions into the second skimmer. In addition, the He flow rate in the API chamber was found to be critical, where a high He flow rate enhanced the resulting transmission of ions but also appeared to increase the energy of the ions. An ion energy measurement in the mass spectro-meter was performed using a blocking grid and the ion energy was found to be as high as several hun-dred electronvolts. However, using the technique described in the experimental section whereby high energy ions can be slowed down and trapped by using a positive voltage bias on the ring elec-trode and both end caps, ions of up to 300 eV could be injected from an external source and trapped.

ESI. The results of interfacing this IT-RETOF mass spectrometer with an ESI source are demon-strated in Figs. 17-20. Figure 17 shows the ESI mass spectrum of singly and doubly charged Argi-nine molecular ion injected into the IT-RETOF using 1.2 x 10"6M Arginine dissolved in MeOH/ H20/HOAc (80/20/5) mixed solvent. The pro-duction and transmission of ions to the trap were performed as described before in the experimental section. The spectrum obtained was a single wave-form spectrum with a storage time of 931 ms. Thus, only 1 s is needed to acquire this spectrum with S/N of the molecular ion peak greater than 20. The actual Arginine sample consumed is 200fmol. This clearly demonstrates the ability of the IT-RETOF to detect trace transient species or rapidly eluting species. The FWHM of the MH +

peak is 10 ns which corresponds to a mass resol-ution of « 2700. Because the production of ions of different charges is a statistically distributed process, the relative intensity of singly charged vs. doubly charged molecular ion varied from one single waveform to another. However, as shown in Fig. 18, the spectrum obtained after being averaged 50 times showed a reproducible signal intensity pattern of both charged ions under the same exper-imental conditions. The FWHM of the MH +

ion of Arginine in Fig. 18 is 16 ns which corre-


Electrospray lonization of Arginin·, MW 174, 931 ms trapping time, single waveform spectrum

12 - r

10 4-

8 +

I , •fi « -L

4 +

2 +

( M ♦ 2H ) + 2

[ J j L i l l ^ ^ Jj,.L lly.i, JJilJpi J. J.L·

10 ns

Timt of Right ( usée )

Fig. 17. ESI single waveform spectrum of Arginine with VrS. = 500Vpp, Kcap = +50 V, Kskimmer = +32 V, Klens = +90 V, VA = -17 V, j / B = -65V, Fextr. = -400V, VL = Vxl = -1500V, VFOC - -1250V, VX2 = -1385V, VKÏ = -197V, VK2 = +410V.

sponds to a resolution of > 1700. The degradation of resolution in the averaged spectrum was the result of using the modified EAI quadrupole r.f. power supply which could not be pulsed off. The actual sample consumed in obtaining Fig. 18 was 1 pmol. A voltage of 530 Vpp was used at a fre-quency of 1.1 MHz in obtaining the above two spectra.

The ability to obtain enhanced resolution at higher molecular weight through the use of ion storage is demonstrated in Fig. 19 for the molecu-lar ion region of bradykinin. Figure 19 shows the expansion region of the MH + peak of bradykinin obtained using ESI produced ions from

2 .3xlO~ 6 M bradykinin dissolved in MeOH/ H20/HOAc (80/20/5 (v/v)) solvent injected into the IT-RETOF. The spectrum obtained was a 400 waveform average using a storage time of 931ms per cycle. The r.f. voltage used was 1975 Vpp and only 12 pmol of sample was used in obtain-ing this spectrum. The peak of highest intensity observed in Fig. 5 is the bradykinin MH + peak at mjz 1061. The FWHM of this peak is only 20 ns which corresponds to a mass resolution of « 3300. This spectrum clearly shows the completely resolved 13C isotope peaks with mjz of 1062 and 1063. The ions with mjz of 1060 and 1059 were also observed. These two ions are the result of loss of


Electrospray ionization of Arginine, MW 174, 931 ma trapping time, 50 waveforms average

90 -r

80 +

70 +

60 +

è S 50 H-

1 1 40

30 +

20 +

10 +

( M ♦ 2H ) 4-2

■- l | ·-■■· >\ ' - ■ | - i i - > | i Î . i - - · ij-JÎ-Î.- . p - . . . . . | . j l . i . . J.

16 ns

o m

o o CO CO

o CO

o 5

o 10

o 0>

o CO II)

o m

o S

Tim· of Flight ( ustc )

Fig. 18. Fifty waveform average of ESI Arginine spectrum. All conditions same as those of Fig. 17.

one and two protons respectively. The appearance of these ions is more prominent when higher r.f. voltage was employed suggesting that the loss of H atoms most probably occurred from collisions inside the trap.

In Figs. 20 and 21 are shown ESI-IT-RETOF mass spectra of Gramicidin S illustrating the use of different storage time on the S/N quality of the IT-RETOF spectra observed. The spectra were obtained using 10"6M Gramicidin S dissolved in MeOH/H20/HOAc (80/20/5) mixed solvent. In Fig. 20, the spectrum was obtained using a 3.1 ms storage time and was averaged 50 times. This cor-responds to a total analysis time of only 1.5 s. The

Gramicidin S sample actually consumed in obtain-ing this spectrum is 250fmol. The major peak observed in this spectrum is the singly charged Gramicidin S molecular ion with a m/z of 1142. The doubly charged molecular ion with a m/z of 571.5 was also observed as the second largest peak in the spectrum. A r.f. voltage of 2200 Vpp was used to obtain this spectrum. Under the same condi-tions, another spectrum was obtained using a 931 ms storage time and was averaged 50 times as shown in Fig. 21. The S/N ratio is greatly improved in this spectrum because of the ability of the ion trap to accumulate the ions through longer trapping time. This capability is especially


Expansion Spectra of Bradykinin. Resolution " 3300

70 -r

60 4-

50 +

£ 40 m

c

1 >

1 30

20 +

10 +

MH

[ ^ À fy A Λ jv ■*■■ «■ A

O CN

en

* CM CO

00 Ol

co

CN

CO

CD CO CO

o CO

* CO

œ CO

c* Iß CO

CO Iß CO

Time of Flight ( usec )

Fig. 19. ESI spectrum of Bradykinin showing expanded view of molecular ion peak with VTf = 1975 Vpp, Vcap = +50 V, Skimmer = +10 V, Flens = +195 V, VA = - 92 V, VB = - 70 V, KextFi = -330 V. All other conditions same as those of Fig. 17.

prominent when interfaced to an ESI source as opposed to a plasma source where the ion trap is easily saturated because of the high ion current (1-10 m A) produced in the ionization source. Gen-erally, the ion current measured in the ion trap region is usually less than a few picoamperes in our current experimental setup. This results in the reduction of space charge effects inside the ion trap. Thus, the resolution of all the spectra obtained using the ESI source is generally improved over

that obtained from the plasma source. The FWHM of Gramicidin S doubly charged molecu-lar ion peak in Figs. 20 and 21 is « 35 ns which corresponds to a mass resolution of « 1900. The amounts of Gramicidin S actually consumed in these two spectra were 240fmol and 7.2pmol respectively.

The present limit for trapping externally gener-ated ions via electrospray is « 3000. Neverthe-less this has allowed us to trap and detect large


Electrospray ionization of Gramicidin S, MW 1141, 3.1 ms trapping time, 50 waveforms average

«I

c

s .fi

12 -r

10 +

8 +

β +

4 +

+ 2

(M + 2Η)

Time of flight ( usec )

Fig. 20. ESI spectrum of Gramicidin S showing the doubly charged molecular ion peak with Krf =2200Vpp, Fcap = +50V, Skimmer = + 10V, Klens = +195 V, VA = -92V, VB = -70 V, Kextr. = -330 V. All other conditions same as those of Fig. 17.

multiply-charged peptides including Melittin (m/z 2845), bovine cytochrome C (m/z 12 327), horse heart myoglobin (m/z 16 950) and bovine albumin (m/z 66 266). This is shown for the case of cyto-chrome C in Fig. 22 where multiply charged ions from (M + 11H)11+ to (M + 20H)20+ are observed. This spectrum was obtained using 4.8 x 10_7M cytochrome C dissolved in MeOH/H20/HOAc (50/50/2.5) mixed solvent. The storage time used here was 931ms and the spectrum was averaged 200 times. The amount of cytochrome C sample actually consumed in obtaining this spectrum is 400fmol. The maximum r.f. voltage of 2200 V was used in trapping these externally generated

ions. In order to extend the mass range a higher r.f. voltage will yet be required.

An important potential advantage of the IT-RETOF is the duty cycle of the device. The actual pulse out time of the device is « 2 μβ. Thus, if one uses a storage time of > 10 ms per cycle for sampling a continuous ion beam, the duty cycle approaches nearly 100%. Other methods have been used to sample continuous ion beams into TOF such as beam modulation and pulsed extrac-tion. In the beam modulation technique an ion beam is swept past a slit to obtain time resolution in the TOF device. In previous work with an API beam, this method was shown to provide excellent


Electrospray ionîzation of Gramicidin S, MW 1141, 931 ms trapping time, 50 waveforms average

loo -I 90

* 80 * 70 c 2 60 « 50 V £ 40 ·§ 30 Œ 20 -

10 \ o -V ^

c c <C

100 -90 -80 -

£ 70 M

§ 60 -♦ *

"S 50 · V

1 40 ' | 30 -

20 -10 -

0 ■ c

I

I i i ■ I qi i ti ■ i

> o > ^ > CD

, > CD

o i ^ > o

■ L.,1L.I

δ CO CO

. . L . , l l . . J

o CO

o

ii il yufciidli

o CM f**

L I . . . . . . .-, kl ' ·|''·' '■'

o CM * ■ "

J i ^

^ o CD h*

·"'·■ Ί ■

O CD

·"

ii-Ul, ,

o 6 00

MA_ ,

o ό CM

... ^ o * 00

T

o * CM

. .

o CO 00

1 1 ,

i l k rf|ln lik I

o CO CM

(M + 2H)+2

,., oo en σ> σ> ψ^ LO

σ> o

MH +

.LiiijÎLt|fc.il.h,liu. t i _ .

^ » ^ O O CM CD CO en

Time of Right ( usée )

Fig. 21. ESI spectrum of Gramicidin S showing both the singly and doubly charged molecular ion peaks. All conditions are same as those of Fig. 19.

resolution in a linear TOF-MS [39]; however, the sensitivity was rather limited due to the poor duty cycle. It is estimated that even with a beam mod-ulation rate of 10 kHz, a maximum duty cycle of « 0.02% can be achieved. An alternative method for interfacing a continuous ion beam to the TOF is the pulsed extraction technique. In this method an ion beam is transmitted between the acceleration plates of the TOF and then rapidly pulsed out with an extraction pulse into the TOF drift tube which is transverse to the ion beam. The duty cycle is limited by the length of the extraction plates and velocity of the ion beam, but has been estimated to be as high as « 2.5% based upon a repetition rate of 10 kHz [40]. The duty cycle of these two tech-niques will decrease when detecting ions with larger m/z where a lower repetition rate must be used.

An advantage of the ion trap is that it can achieve an excellent duty cycle for continuous ion beams independent of repetition rate. As shown in this work, a trapping time of 100 ms can be used to enhance the signal and resolution of ions generated by an API source. Under these conditions, a pulse-out rate of only 10 Hz is required, which can easily be processed even with relative!} modest elec-tronics and software. When fast spectrum acquisi-tion must be performed, such as applications in GC or LC when detecting rapidly eluting species or monitoring transient species, a lOO^s trapping time and a pulse-out rate of 10 KHz can be used to obtain the same 100% duty cycle. In contrast, the beam modulation and pulsed extraction meth-ods require a high extraction rate (10 kHz) to achieve a reasonable duty cycle. Although this is


90 -r

80

70

60 +

2 50

I 40 ce

30 +

20 +

10 +

Electrospray ionization of horse heart Cytochrome C, MW 12,327, 931 ms trapping time, 200 waveforms average

+ 16

+ 17

+ 18

+ 19

+ 20

L i lli nily!

+ 15

ψι

+ 14

+ 13

+ 12

+.11

iifiihiiiilÊlliie

* «~ d o *"

* ■

^ ^t o *~

«* *— 00

o *"

* *-CM r—

*~

* ^ CO

^ r -

* *— d CM f—

Time of Flight (

*t T—

<* CM

*— usec )

* *— 00 CM

*-

*· ^ CM CO

«~

«· r -

co CO

«-

Fig. 22. ESI of cytochrome C showing multiply charged peaks with VrS. = 2200 Vpp, V^p = +100 V. Fskimmer = +30 V, Klens = +160 V, and VtxXx ■= -330 V. All other conditions same as those of Fig. 17.

possible to process with modern digitizers and soft-ware, the use of the low repetition rate provided by the IT-RETOF greatly simplifies data collection and processing.

There are other potential advantages of the IT-RETOF over pulsed extraction methods with-out storage. Simulations performed on the SIMION

program show that orthogonal extraction of ions becomes difficult above 50 eV of energy where the ions can not be easily turned around and trans-

mitted down the flight tube. In comparison, the IT-RETOF is capable of slowing down and trap-ping ions of high energy with resulting high resolu-tion due to the storage properties of the trap. Also, the orthogonal pulsed extraction method will be limited in the mass range that will be observed due to the transverse velocity component of the ion beam. However, the mass storage range of the IT-RETOF is determined by the voltage and frequency applied to the trap and can be made

178 B.M. Chien et al.j Int. J. Mass Spectrom. Ion Processes 131 (1994) 149-179

extremely large in conjunction with the use of a buffer gas in the trap. In recent simulations [28] it has been shown that a storage range of several hundred thousand atomic mass units should be possible under appropriate conditions. In the pre-sent work the ability to trap externally generated ions is limited to m/z < 3000 by the voltage range of the r.f. power supplies available. However, sto-rage of ions over a relatively large mass range has been demonstrated in the trap [41,42]. The subse-quent pulse-out of the ions from the trap into the RETOF occurs on-axis so that any further energy difference in the kinetic energy of ions of different masses in the extraction process is no longer impor-tant. In addition, a further possible advantage of the IT-RETOF in these experiments that has not been demonstrated herein is the possibility for MS-MS studies in the trap and for the studies of long-lived metastable decay in the IT-RETOF combination.

A measurement of the limit of detection attain-able for liquid injection-nebulization into the plasma source was examined using pyridine dis-solved in methanol solution. An initial solution of 6.55 x 10"6M was prepared and successive dilu-tions were made down to 6.55 x 10~10M. These samples were run as detailed in this work, but with the storage time optimized for each concen-tration. A lower limit of detection of 2-3 fmol was determined using a S/N = 3 as our limit of detec-tion criterion. Similar limits of detection were observed for liquid injection into the ESI source using Arginine, Leucine-Alanine and Gramicidin S dissolved separately in MeOH/H20/HOAc (80/20/5 (v/v)) solvent. Initial solutions of / x 10_5M were prepared and successive dilutions were made down to 6.55 x 10~8 M. These samples were run as detailed previously but with the con-ditions optimized for each analyte and concen-tration. A lower limit of detection of 35, 20 and 80 fmol respectively were determined using S/N = 3 as our limit of detection criterion. Later work on ESI incorporated a 40 mm triple micro-channel plate detector in our IT-RETOF mass spectrometer where limits of detection of 4.5, 0.7,

and 0.5 fmol were achieved for Gramicidin S, Argi-nine, and Leucine-Alanine respectively. Future work will involve improving the ion transmission effeciency in the interface region, which should allow even lower detection limits.

In conclusion, the capabilities of an IT-RETOF MS combination have been demonstrated for detection of ions generated by liquid injection into plasma source atmospheric pressure ioniz-ation mass spectrometry and detection of ions generated from an ESI ionization source. The ion trap has been shown to be an effective means of storing externally generated ions for up to 10 s, prior to mass analysis in the RETOF MS. The IT-RETOF can provide nearly 100% duty cycle in converting a continuous ion beam into a pulsed source for TOF, which will be important for inter-facing to Chromatographie applications. It is also shown that the storage capabilities of the device provide enhanced resolution and sensitivity as the storage time is increased. The detection limit of the device was also demonstrated with liquid injection techniques in the plasma source and found to be in the low femtomole range, while similar low femto-mole detection limits were also observed for ESI of peptides. In addition, the r.f. voltage was shown to be an effective means of eliminating low mass back-ground peaks from the trap and, thus, from the TOF mass spectrum obtained.

Acknowledgements

We thank Bruce Thomson of Sciex for the loan of an electrospray source. This work received par-tial support from the National Science Foundation under grant BIR-9223677.

References


2 D. Price and J.E. Williams (Eds.), Time-of-Flight Mass Spectrometry, Pergamon, Oxford, 1969.

3 D.M. Lubman and M.N. Kronick, Anal. Chem., 54 (1982) 660.


4 J.M.B. Bakker, J. Phys. E., 6 (1973) 785. 5 D.M. Lubman and R.M. Jordan, Rev. Sei. Instrum., 56

(1985) 373. 6 R.B. Opsal, K.G. Owens and J.P. Reilly, Anal. Chem., 57

(1985) 1884. 7 B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A.

Zagulin, Sov. Phys.-JETP, 37 (1973) 45. 8 U. Boesl, H.J. Neusser, R. Weinkauf and E.W. Schlag, J.

Phys. Chem., 86(1982)4851. 9 R. Frey, G. Weiss, H. Kaminsky and E.W. Schlag, Z.

Naturforsch. Teil A, 40 (1985) 1349. 10 D.M. Lubman, W.E. Bell and M.N. Kronick, Anal. Chem.,

55 (1983) 1437. 11 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei.

Instrum., 60 (1989) 347; 60 (1982) 792. 12 R. Grix, R. Kutscher, G.Li, U. Grüner and H. Wollnik,

Rapid Commun. Mass Spectrom., 2 (1988) 83. 13 U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spec-

trom. Ion Processes, 112 (1992) 121. 14 R.C. Beavis and B.T. Chait, Rapid Commun. Mass Spec-

trom., 3 (1989) 233. 15 J.G. Boyle, CM. Whitehouse and J.B. Fenn, Rapid Com-

mun. Mass Spectrom, 5 (1991) 400. 16 R.J. Cotter, Anal. Chem. A, 64 (1992) 1027. 17 X. Tang, R. Beavis, W. Ens, F. Lafortune, B. Schueler and

K.G. Standing, Int. J. Mass Spectrom. Ion Processes, 85 (1988) 43.

18 W. Paul, Angew. Chem., Int. Ed.Engl., 29 (1990) 739. 19 B.D. Nourse and R.G. Cooks, Anal. Chim. Acta, 228

(1990) 1. 20 R.G. Cooks and R.E. Kaiser, Jr. Ace. Chem. Res., 23

(1990) 213. 21 J.S. Brodbelt-Lustig and R.G. Cooks, Spectra, 11 (1988)

30. 22 J.N. Louris, J.S. Brodbelt and R.G. Cooks, Int. J. Mass

Spectrom. Ion Processes, 75 (1987) 345. 23 G. Van Berkel, G. Glish and S. McLuckey, Anal. Chem., 62

(1988) 1284. 24 R. Cooks, G. Glish, S. McLuckey and R. Kaiser, C & E

News, March 25 (1991) 26.

25 R. March and R. Hughes, Quadrupole Storage Mass Spec-trometry, Wiley, New York, 1989.

26 R. Cooks, J. Williams, K. Cox, K. Kaiser and J. Schwartz, Rapid Commun. Mass Spectrom., 5 (1991) 327.

27 D.E. Goeringer, S.A. McLuckey and G.L. Glish, Proc. 39th ASMS Conference on Mass Spectrometry and Allied Topics 1991, Nashville, TN, p. 532.

28 S.M. Michael, M.Chien and D.M. Lubman, Rev. Sei. Instrum., 63 (1992) 4277.

29 C.Ma, H. Lee and D.M. Lubman, Appl. Spectrosc, 46 (1992) 1769.

30 J. Zhao, J. Zhu and D.M. Lubman, Anal. Chem., 64 (1992) 1426.

31 R.C. Willoughby and R.F. Browner, Anal. Chem., 56 (1984) 2626.

32 I. Sofer, J. Zhu, H.S. Lee, W. Antos and D.M. Lubman, Appl. Spectrosc, 44 (1990) 1391.

33 S.M. Michael, B.M. Chien and D.M. Lubman, 65 (1993) 2614.

34 B.M. Chien, S.M. Michael and D.M. Lubman, Anal. Chem., 65 (1993) 1916.

35 S.A. McLuckey, G.A. Glish, K.G. Asano and B.C. Grant, Anal. Chem., 60 (1988) 2220.

36 J.F.J. Todd and R.M. Waldren, Int. J. Mass Spectrom. Ion Phys., 29 (1979) 301.

37 C.S. O, H.A. Schuessler, Int. J. Mass Spectrom. Ion Phys., 40(1981)67.

38 C.S. O, H.A. Schuessler, Int. J. Mass Spectrom. Ion Phys., 40(1981)77.

39 C. Ma, S.M. Michael, M. Chien, J. Zhu and D.M. Lubman, Rev. Sei. Instrum., 63 (1992) 139.

40 J. G. Boyle and CM. Whitehouse, Anal. Chem., 64 (1992) 2084.

41 R.G. Cooks and R.E. Kaiser Jr., Ace. Chem. Res., 23 (1990) 213.

42 R.G. Cooks, J. Williams, K. Cox, K. Kaiser and J. Schwartz, Rapid Commun. Mass Spectrom., 5 (1990) 327.

International Journal of Mass Spectrometry and Ion Processes 131(1994) 181 -192 181 0168-1176/94/S07.00 © 1994 - Elsevier Science B.V. All rights reserved

Pulse amplitude analysis: a new dimension in single ion time-of-flight mass spectrometry

P.V. Bondarenko, P.G. Grant, R.D. Macfarlane* Department of Chemistry, Texas A&M University, College Station, TX 77843-3255, USA

(Received 21 June 1993; accepted 27 July 1993)

Abstract

Using single secondary ion detection, it has been shown that the amplitude of pulses formed by a microchannel plate electron multiplier array is dependent on the number of electrons and other ionizing species that are ejected into the microchannel when the ion impacts at the entrance of the microchannel. We have measured the pulse amplitude distribution for impacts where a single electron is ejected and have observed the onset of multiple electron emission when the incident ion velocity is increased. We have shown that pulse amplitude analysis is sensitive to the velocity and number of atoms in a molecular ion. By using pulse amplitude windows, we have measured 252Cf-PD mass spectra where the 1+ or 2 + ion of insulin can be selectively chosen or rejected on the basis of pulse amplitude. A new component in the 252Cf-PD mass spectrum in the high mass region has been identified as being due to low velocity neutral species formed by metastable decay in the acceleration region. Elimination of this component should enhance the sensitivity for low intensity high molecular mass species.

Key words: MicroChannel plate; Pulse amplitude analysis; 252Cf-plasma desorption mass spectrometry; Background reduction

Introduction

When an energetic ion strikes the dynode of an electron multiplier ion detector, an electron avalanche is triggered which can be converted into a transient electronic pulse. In a time-of-flight (TOF) mass spectrometer, this pulse signals the arrival of an ion at the end of the flight tube and, when coupled to a fast clock, is used to electroni-cally measure the TOF of the ion.

The 252Cf-plasma desorption mass spectrometry (252Cf-PDMS) method uses the single ion TOF methodology. It was the first of the particle-induced desorption methods to demonstrate feasi-bility for the study of proteins [1] and the mass


range has been extended up to 45kDa. With increasing molecular weight, molecular ion yields decrease and at the 45 kDa level the molecular ion intensity is barely discernible above background. The background in the high mass region for 252Cf-PDMS is a consequence of complications that are associated with single ion TOF measure-ments. This background has always been a problem and several studies have been devoted to understanding its origin. Some of the background in this region is due to uncorrelated events where a single fission fragment ejects a burst of ions but no time zero signal is generated for that ion packet [2]. There is also a significant energetic neutral species component arising from metastable ion decay in the acceleration region [3]. Considerable effort has been directed toward reducing this back-

SSDI0168-1176(93)03881-L

182 P.V. Bondarenko et aljlnt. J. Mass Spectrom. Ion Processes 131 (1994) 181-192

ground level for Cf-PDMS studies in order to increase the signal-to-noise ratio for high mass species. The most recent study used a pulsed electrostatic particle guide to simultaneously reduce the uncorrelated ion and neutral species level [4]. But despite these efforts, a significant residual background exists. The study described in this paper is part of this continuing effort.

When an ion impacts a surface at a velocity below 105ms- 1 , electrons are ejected by potential emission [5]. Recently, it has been shown that pro-tein molecular ions can also eject light negative ions (e.g. H~) or positive ions as a result of surface-induced dissociation [6], Above 105ms_ 1, kinetic ejection of electrons becomes the dominant process and more than one electron can be ejected. Many studies have been carried out which document this feature and which show that the electron multiplicity is dependent on ion velocity and the number of atoms in the ion [7].

With the recent successes of matrix-assisted laser desorption-ionization (MALDI) for producing protein molecular ions in a higher mass regime [8], a considerable amount of activity has been directed toward the problems of ion detection [9]. What we can conclude from these studies in terms of what constitutes the ideal detector is that the MCP is a remarkably good detector. It has a moderately low work function, the small angle of incidence presented to the incident ion by the microchannel structure enhances kinetic electron ejection [10], and if photons or light ions (e.g. H~) are emitted, they may still be able to initiate an electron cascade in a microchannel.

The focus of this study is on details of the electron cascade that a single microchannel in the microchannel array generates and whether the number of electrons exiting a microchannel is dependent on the number of electrons ejected when the incident ion collides with the wall at the entrance of the microchannel. The multiplication process is a stochastic event producing a broad distribution of electronic pulse amplitudes [11]. If the electron current emanating from a single micro-channel is saturated, then the pulse amplitude

distribution is independent of the electron multi-plicity associated with the impact of the incident ion. If, however, there is no saturation, then the pulse amplitude distribution might reflect the number of electrons ejected in the ion/dynode interaction. This effect offers the possibility of selecting secondary ions for TOF analysis accord-ing to the number of electrons they eject when they strike the MCP.

Experimental

Instrumentation

A 252Cf-plasma desorption mass spectrometer was used in this study. Details of the system are described elsewhere [4]. The 252Cf source mounted on a 1 μηι thick Ni backing was positioned 2 mm behind the sample which consisted of an aluminized Mylar film (1.5 μιη thick) mounted over a 1 cm diameter aperture on a 1 mm thick stainless steel disc. The acceleration voltage was applied to the sample disc and 252Cf-source. The pulse for the start signal for the TOF measurement was generated by detecting the electrons ejected from a 2 cm diameter, 1.5//m thick aluminized Mylar conversion foil located at ground potential 1 cm from the 252Cf source. A 90% transmission Ni electromesh acceleration grid (Buckbee-Meers, Minneapolis, MN) was located 8 mm in front of the sample foil. Two sets of electrostatic deflection plates were mounted 5 cm in front of the accelera-tion grid. In one part of this study, these plates were used to investigate the metastable neutral component of the mass spectrum. The length of the field-free region was 80 cm.

The detector located at the end of the flight tube consisted of two 40 mm diameter microchannel plates (Galileo Electro-optics, Sturbridge, MA) mounted in a chevron configuration. A grounded 90% transmission grid was mounted 5 mm in front of the detector. A voltage of - 3 kV was applied to the front face of the first MCP. This voltage was distributed to the other components of the detector through an internal resistor divider network

P.V. Bondarenko et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 181-192 183

designed to give 1000 V potential difference across each plate. The two MCPs were separated by a distance of 2 mm with a potential difference of 200 V between the MCPs. A 50 Ω impedance-matched anode at ground potential was used as the electron collector.

The pulses from the start detector were fed into a fast pre-amplifier (Ortec Model 9301, Oak Ridge, TN) and then to a constant fraction discriminator (CFD) (Canberra Model 2128, Meriden, CT). The discriminator level was set to reject smaller ampli-tude pulses that are due to alpha particles from 252Cf decay. The fast logic pulses generated by the CFD were transferred to the start input of a custom-designed time-to-digital converter (TDC) [12].

A schematic of the configuration of the stop detector electronics is shown in Fig. 1. For the purpose of illustrating the details of the function of this configuration, we present a scenario where two ions strike the surface of the detector at differ-ent times. Ion #1 produces a low amplitude pulse (e.g. 0.01 V) while ion #2 produces a higher ampli-tude pulse (e.g. 0.05 V). The pulses are amplified in two stages (to 0.1 and 0.5 V) using a fixed gain fast pre-amplifier (Ortec Model 9301, Oak Ridge, TN) and to 1 and 5 V using a variable gain timing-filter amplifier (TFA), (Ortec Model 454, Oak Ridge, TN). The larger pulse is followed by a set of dis-cernible small reflection pulses with diminishing amplitude extending to 500 ns. (The smaller pulse #1 also has this feature but the amplitudes of the reflection pulses are below threshold levels for detection.) The output of the TFA is then divided into two equal branches which are then fed into two constant fraction discriminators, (CFD-A and CFD-B), (Ortec Model 473A, Oak Ridge, TN). For CFD-A, the discriminator is set low (e.g. 0.1 V) resulting in the generation of fast logic pulses for pulse #1 and pulse #2 plus an extra pulse from the first reflection of pulse #2 (which has an amplitude of 0.5 V). The discriminator level of the CFD-B is set higher (e.g. 1.5 V). Only the primary component of pulse #2 can trigger CFD-B at this discriminator level. The three fast logic pulses

STOP DETECTOR

-v-

#1 T #2

I PREAMPLIFIER

ΛΓ

#1 T #2

f AMPLIFIER

1

Y

»1 ♦ #2

τηρτ- Y

CFD-A

Ï #1

w v -

#1 * #2

CFD-B

w «2 Y w

#1 «2

Γ

Ï #2

GDG

START

CFD

Y TDC

"1 Γ

(TOF)1

Fig. 1. Schematic diagram of the configuration of electronic modules used in the pulse-amplitude analysis of signals coming from a MCP electron multiplier array that is detecting single secondary ions. CFD-A and -B are two constant fraction discriminators operating in parallel but with different discrimi-nator level settings. CFD-B is coupled to a gate-and-delay generator (GDG) that generates a blocking pulse which gates the stop input of a time-to-digital converter.

generated by CFD-A for these two ions are sent through a 700 ns delay line into the stop input of the TDC. The single fast logic pulse from CFD-B is sent to a gate-and-delay generator, GDG, (Ortec Model 416A, Oak Ridge, TN) which generates a blocking pulse with a variable width and delay. This blocking pulse is sent directly to a gate input associated with the stop pulse circuitry of the TDC. The blocking pulse is delayed using the GDG so that the fast timing pulse for ion #2 from CFD-A and the gate pulse for ion #2 overlap in time. The time width of the blocking pulse is lengthened to cover the reflected pulses associated with pulse #2 (e.g. 900 ns). With the gate circuit activated, only stop pulses with amplitudes within a pre-selected window (e.g. 0.1-1.5 V) contribute to the TOF

184 P.V. Bondarenko et al.j Int. J. Mass Spectrom. Ion Processes 131 (1994) 181-192

spectrum. For our scenario, even though three pulses are presented to the input of the stop pulse circuitry, only one is processed, the signal from ion #1. The signal from ion #2 triggers both CFD circuits and generates a block pulse that disables the stop pulse circuitry during the time that pulse #2 arrives at the TDC and for an extended period of 900 ns to block the reflection pulses. If the width of the gate pulse is too narrow (e.g. 100 ns), the first reflected pulse associated with pulse #2 will also appear in the TOF spectrum because its amplitude is within the pre-selected window. When the gate circuit is turned off, a normal TOF spectrum is obtained that includes contributions from the entire spectrum of pulse amplitudes. Pulse-ampli-tude dependent spectra were obtained by varying the pre-selected window from 0.1 to 5 V using a constant width of 0.5 V.

Sample preparation

Bovine insulin and gramicidin S were obtained from Sigma (St. Louis, MO) and used without further purification. Nitrocellulose (Schleicher & Schuell, Inc., Dassel, Germany) was dissolved in acetone (7uguL_1) and electrosprayed onto aluminized Mylar using a microprocessor-controlled syringe pump electrospray system (Selmi Model UNP, Sumy, Ukraine) [13]. Peptide samples were prepared by adsorption from 10~4 M aqueous solutions (0.1% TFA) onto nitrocellulose.


Pulse amplitude distribution for H~

The first objective of this study was to determine the pulse amplitude distribution for an incident ion which ejects a single electron on impact with a MCP. The prominent H~ ion in the negative ion 252Cf-PD spectrum of the Mylar sample backing was selected. With a voltage of -3.00 kV on the front face of the first MCP on the stop detector, the acceleration voltage was increased from zero to

a threshold value of -3.10kV for H" detection. Under these conditions, the H" ion has an incident energy of lOOeV when it impacts the face of the MCP. The amplitude of the signal from H~ was adjusted to an average of 1 V using the timing-filter amplifier. A pulse amplitude analysis was carried out following the procedure described above covering a range of amplitudes from 0 .1-5 V using a window of 0.5 V. The measurement was repeated for a range of incident kinetic energies from 100 to 12000eV by varying the acceleration voltage. The results are shown in Fig. 2.

At lOOeV, the FT ion has a velocity of 1.4 x 105ms~1 which is above the 1.0 x 105ms~l

threshold for kinetic ejection. The pulses formed from these electrons have an amplitude distri-bution starting at threshold and sharply decreas-ing in intensity to zero at 2 V pulse amplitude. At 200 eV, the overall FT intensity increases by a factor of two but the pulse amplitude distribution is essentially the same as at lOOeV. But when the kinetic energy is increased to lOOOeV, the pulse amplitude dramatically shifts to an average value of 2.2 V and when the kinetic energy of the H~ ion is increased further, the pulse amplitude distri-bution continues to shift to higher values. At

o iH X

3 Ç 3 o 3 >> w a a>

HN

6.5 H

6 -

5.5 -

bj 4.5 \

4 -

3.5-31

2.5 \

9.\

l . ö -

l i

0.5 j o-i

CFD threshold (volts)

Fig. 2. Pulse amplitude distributions for FT ions impacting at different incident energies on an MCP electron multiplier detector: + , lOOeV; ·, 200eV; *, lOOOeV; O, 2000eV; D, 7000 eV;A, 12000eV.


12000eV, most of the distribution lies above 3 V approaching the 5.0 V cut-off limit of the CFD.

These observations essentially give validity to the concept that was being tested, that the amplitude of pulses generated by a MCP as a result of electron ejection from single ion impacts is sensitive to the number of electrons ejected. The observation that the pulse amplitude distribution is invariant for lOOeV and 200 eV FT ion impacts is taken as proof that this is the distribution for ion impacts where a single electron is ejected. This conclusion is consistent with what is known about the kinetic ejection process [10]. In this velocity regime, most ion impacts do not eject any electrons and for those impacts where electron ejection occurs, single electron ejection is the dominant process. Previous experimental studies and theoretical calculations showed that with increasing incident velocity, the probability for kinetic ejection of electrons increases sharply and multiple electron ejection becomes more probable. We observe a dramatic increase in the pulse amplitude distribution for H~ ions at lOOOeV and this increase can be attri-buted to multiple electron ejection in the initial ion impact. This observation also means that electron multiplication is not saturated in the MCP for single electron ejection by 100-200 eV H" ions. We cannot draw any conclusions about what the electron multiplicity is from pulse amplitude analysis because of the possibility of non-linearity in the electron multiplication process within a microchannel when more than one electron initiates an electron cascade.

Influence of reflection pulses

One of the difficulties encountered in designing electronic circuitry for MCP electron multipliers is that the fast pulse that is generated at the anode is difficult to transmit with no distortion because of the high frequency components of the signal. Small impedance mismatches in signal transmission result in the generation of a series of reflection pulses following the primary pulse. The use of a conical anode with an impedance that matches

the transmission line and input electronics greatly reduces the amplitude of the reflection pulses but it is difficult to totally eliminate that component [11].

We incorporated a 50 Ω impedance conical anode into the stop detector to minimize the con-tribution from reflection pulses because these pulses interfere with the resolution of primary small and large amplitude pulses when performing a pulse amplitude analysis. But even with the use of the impedance matched anode, some pulse reflec-tion occurred resulting in a pattern of satellite pulses following the primary pulse at 100 ns inter-vals. The amplitude of the first reflection pulse was 10% of the amplitude of the primary pulse while the amplitudes of pulses from higher order reflec-tions were decreasingly smaller.

In our pulse amplitude analysis studies, this problem only appears when the primary pulse has an amplitude greater than 1 V and the lower level discriminator on CFD-A is set below 0.2 V. The consequence of the presence of this reflection pulse is that if we wish to reject pulses with ampli-tudes greater than IV, the primary pulse is rejected, but its first reflection pulse is accepted and appears in the TOF spectrum delayed by 100 ns relative to the TOF associated with the primary pulse. This effect is depicted in Fig. 3. Figure 3(a) shows the normal negative ion spec-trum from the polymer side of the aluminized Mylar backing foil in the low mass region recorded at — 5kV acceleration voltage. The average pulse amplitude for the H" ion (2keV) is 2.2 V (Fig. 2). Intense peaks for H", C^, C2H", and C2H2 are observed with no evidence of any satellite peaks even though the pulses for these ions show a significant reflection component. The reason why these reflection pulses do not appear in the normal TOF is due to a fortuitous feature of our custom-designed TDC. After each stop pulse is processed, there is a 780 ns dead time interval during which no additional stop pulses are accepted. This interval covers the time span where the reflection pulses appear. Figure 3(b) shows the spectrum obtained using only stop

186 P.V. Bondarenko et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 181-192

Mass (m/z)

9.00-

1 7 . 0 0 -"03 c c J ce

x: u o.oo--c

a : 3.00 -

1.00 J

o 100 -

"03 c

I0.75:

J 0 · 5 0 : 0.25-

7.001

1 C

hann

el x

lO L

S

llllllllllllll

Cou

nts

8 1

1 1

1 1

1 .

! 1

1.00-

--

10 20 1 . 1 L J 1 ■ ■ 1 . ,

30 , , 1 ,

fa | H'

b

l·

c

Y

CJi

. , ll

Γ

. - - -* „

C2H-

1 1

. JLJJJ C2H-

f

11 Mi., .1

1000 1500 2000 2500 3000 Time of Flieht (ns)

3500 4000

Fig 3 252Cf-PD negative ion mass spectra in the m/z 0-34 mass region for a Mylar target aluminized on the back side using different secondary ion pulse amplitude distributions and which demonstrate the function of the gate pulse: (a), > 0.01 V, no gate pulse; (b), 0.10-

0.50 V, 300 ns wide gate pulse; (c), 0.10-0.50 V, 900ns wide gate pulse.


pulses that fall within a window of 0.1-0.5 V. The intensity of the H~ ion has been reduced by a factor of 60 because the majority of these ions have ampli-tudes greater than 0.5 V. However, a new peak appears at a time shifted by 300 ns with an inten-sity that is 14% of the primary H~ peak. This peak is due to the third reflection pulse whose amplitude falls within this pre-selected window. Since the primary pulse has been eliminated, these pulses now are recorded because they are no longer in the time shadow of the primary pulse.

Since one of the objectives of this study is to develop pulse amplitude analysis as a means of reducing the contribution of fast ions to the back-ground in the high mass region, a scheme for eliminating the contribution from these reflection peaks had to be found.

When we first began these studies, we used what should have been the ideal circuit for fast pulse amplitude analysis, a CFD with a built-in single channel analyzer (Ortec Model 583). However, when we discovered the problem with the reflec-tion pulses, we had to devise the strategy discussed in the experimental section to circumvent this problem. CFD-A serves as the lower level discrimi-nator and generates the train of fast logic pulses for the TOF measurement. CFD-B serves as the upper level discriminator and provides a fast logic pulse that is used to generate a blocking pulse at the input of the TDC stop pulse circuitry. Using the GDG, the time duration of the blocking pulse can be expanded to encompass the reflection pulses. Thus, when CFD-B is triggered, that pulse (which is also present in the pulse train coming from CFD-A) and its reflection pulses are blocked at the TDC and do not contribute to the TOF spectrum.

Figure 3(c) shows the TOF spectrum for the same region as Fig. 3(b) which demonstrates the effectiveness of the blocking pulse. The blocking pulse essentially eliminates the H" peak because, as in the case shown in Fig. 3(b), the pulses from this ion have amplitudes greater than 0.5 V. But now the blocking pulse has been lengthened to 900 ns which means that the reflection pulses from the high amplitude pulses are eliminated as

well. The cluster of three ions centered around C2H~ is also attenuated in intensity by the selection of the lower amplitude pulses but not as much as for H~. This difference means that the pulse amplitude distribution for these ions is shifted to lower voltages relative to the H~ spectrum. The velocities of these ions is a factor of five smaller than that for H~ which means that their pulse height distributions have a large single electron ejection component.

Pulse amplitude analysis for insulin molecular ions

In a 252Cf-PD positive ion mass spectrum of insulin, the dominant molecular ion is (M + H)+

and a detectible (M + 2H)2+ ion is also present. We measured the pulse amplitude distributions for these ions using acceleration voltages of 11 and 25 kV and - 3 kV post acceleration. Figure 4 shows these results. The distribution for the 14keV (M + H)+ ions is identical to what we obtained for the 100 eV H" ions (Fig. 2) which means that the pulses formed were initiated by single electron ejection. At 28keV, this molecular ion produces considerably larger pulses with an average ampli-tude of 2.7 V. At this velocity, the ion impact is

0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75 4.25 4.75


Fig. 4. Pulse amplitude distributions for the 1 + and 2+ mole-cular ions of insulin recorded using 14kV and 28 kV total accel-eration voltage: + , (M + H)+, 14keV; · , (M + 2 H ) 2 \ 28 keV; *, (M + H)+, 28keV; O, (M + 2H)2+, 56keV.

188 P.V. Bondarenko et aljlnt. J. Mass Spectrom. Ion Processes 131 (1994) 181-192

considerably more complex, ejecting a multiplicity of electrons, photons, and probably light ions (e.g. H~) which contribute to generating a much larger electron avalanche. The 2+ ion, accelerated through a potential of 14kV has an energy of 28keV but its velocity is the same as the 14keV singly-charged ion and it has the same number of atoms. Thus, we would anticipate that the two ions would have the same pulse amplitude distribution. As shown in Fig. 4, this was the result that was obtained. The 2+ ion at 28 kV acceleration has an energy of 56keV and also has the highest pulse amplitude distribution (average value, 4 V). As the velocity of the ion increases, the multiplicity of particles ejected into the microchannel on ion impact also increases. The enhancement of the size of the pulses generated by these events is a consequence.

Influence of number of atoms in a molecule on pulse amplitude

Electron multiplicity in kinetic electron ejection is dependent on two properties of the impacting ion, velocity and the number of atoms in the mole-cule [7]. When an energetic molecular ion impacts a surface, each atom within the molecule can kineti-cally eject an electron. Thus, at a given velocity, the greater the number of atoms in a molecule, the higher the ejected electron multiplicity. We wished to determine whether pulse amplitude analysis would be sensitive to this effect. For this study, we selected gramicidin S (MW 1141) and bovine insulin (MW 5733). For the gramicidin S measurement, the kinetic energy of the (M + H)+

ion was 5.36 keV and for the singly-charged insulin ion, 28keV. Both ions have a velocity of 3 x 104ms- 1 . Figure 5 shows a comparison of the two pulse amplitude distributions. The grami-cidin S spectrum has a large one electron compo-nent and a high amplitude tail indicating that some of the collisions are ejecting more than one elec-tron. But the insulin molecular ion at the same incident velocity produces a distribution of pulse amplitudes that are much larger than what is

0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75 4.25 4.75


Fig. 5. Pulse amplitude distributions for the molecular ions of gramicidin S and insulin that have the same impact velocity of 3 x 104ms_ I . O, Gramacidin S (M + H)+; □ , Bovine insulin (M + H)+. The higher amplitude distribution for insulin is due to the influence of the number of atoms in the molecular ion on kinetic electron ejection.

observed for gramicidin S. This enhancement we attribute to the contribution of more atoms to kinetic electron ejection for incident insulin mole-cular ions.

Application of pulse amplitude analysis to 252Cf-PDMS

The motivation for carrying out the pulse ampli-tude analysis study was to use this feature to reduce the background level in the high mass region of a 252Cf-PD mass spectrum. Figure 6(a) shows the mass spectrum of insulin in the molecular ion region (total acceleration voltage, 14 kV) utilizing the full distribution of stop pulse amplitudes (>0.1V). The spectrum shows three significant peaks corresponding to the singly- and doubly-protonated molecular ion and the B chain fragment ion. Note that the background rises sharply with decreasing mass and is relatively flat beyond the TOF of the insulin molecular ion. Figure 6(b) shows the spectrum obtained selecting stop pulses having amplitudes of 0.1-1 V. Most of the intensi-ties of the (M + H)+ ion and B-chain fragment ion are retained but the intensity of the 2 + molecular

P.V. Bondarenko et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 181-192 189

9.00.

_7.00.

5.00 A

3.<XH

1.00-3

7.00 J

5.00 A

a 3 · 0 0 '

1.00.

Mass (m/z) 2000 3000 4000 5000 6000 7000 8000 9000

_ _ J i i ι I , , , , I , , , , I , , , , I , , , , I , ,

(M + H) +

(Μ + 2ΗΓ

B-chain

w^H«^K*^Mw

(M + H) +

B-chain

(Μ+2ΗΓ

.r*^^ W M A ^ H N W ^ ^

T-.0J5À

i2 0.50H

0.25 H

(M + 2H)

B-chain

(M + H) +

20000 25000 30000 35000

Time of Flight (ns)

40000

Fig. 6.252Cf-PD mass spectra of insulin in the molecular ion region showing the influence of pulse amplitude selection on the recorded mass spectra: (a), > 0.1 V; (b), 0.1-1 V; (c), 2-4 V. Data acquisition time for (a) and (c) was 5 min and for (b), 15 min.


ion has been greatly diminished because the ampli-tudes of the pulses for these ions are above the 0.1-1 V window at 14kV acceleration voltage (Fig. 4). The background level has been reduced a factor of two. This reduction is primarily due to electronic deletion of pulses from uncorrelated ions. Uncorrelated ions come from fission-frag-ment and alpha particle induced excitations of the sample that are not correlated in time with any of the start pulses that are accepted by the TDC. The mass spectrum of these uncorrelated ions is identi-cal to the correlated spectrum which means that most of these ions have m/z values between 1 and 100. Because they are fast ions, they primarily form high amplitude pulses which are filtered out by the 0.1-1 V window.

When the acceptance window was shifted to 2-4 V, the spectrum dramatically changed as shown in Fig. 6(c). The 2+ ion is now the domi-nant peak and the 1+ ion of insulin is greatly diminished because ions ejecting more than a single electron are being selected. Several small peaks in the region of m/z 1000-2500 are also more prominent in this spectrum compared to the full spectrum (Fig. 6(a)). These ions are fragments of the B-chain that have high amplitude and are superimposed on a background containing a significant low amplitude pulse component that is filtered out when the window is set at a high level. The origin of these low amplitude background pulses is the subject of the following section.

We can see from this study the usefulness of pulse amplitude analysis in learning more about the ionic composition of the TOF spectrum. How-ever, the prime reason for introducing this concept, to dramatically reduce the background in the high mass region was only partially realized. Let us focus on the region m/z 7000-10000 in Fig. 6. The background level in this region is 200 ionss"1 (Fig. 6(a)). By selecting only pulses with amplitudes 0.1-1 V, the background is reduced to 82ionss"1. Approximately 40% of the pulses appear to be coming from impacts on the MCP which produce an electron cascade resembling that produced by the impact of a low velocity

species. What is the nature of these species? We can rule out the uncorrelated ions because the majority of these species generate high amplitude pulses.

Identification of a new component to the background in the high mass region

A characteristic feature of the background of a 252Cf-PD mass spectrum which is well known but rarely referred to is the spectrum of ions between m/z 40 and 100 in the positive and negative ion spectrum with the general composition C^H ,. They come from a hydrocarbon film which forms on the surface of the sample in vacuum. These ions which often dominate the background in this region are relatively unstable and a significant fraction undergo metastable decay in the field-free region forming fast ionic and neutral species with the velocity and TOF of the parent molecular ion. These events are filtered out by the 0.1 -1 V window so they cannot contribute to the background in the m/z 7000-10000 region. A signature of these events is the residual spectrum that is observed when the ions are electrostatically deflected. Figure 7 shows some of these spectra obtained using the insulin target and 14kV total acceler-ation voltage. The ions were deflected past the stop detector by applying 2000 V to two parallel plates located 5 cm behind the acceleration grid. Figure 7(a) shows two neutral particle spectra obtained using only the lower level CFD-A, an integral pulse amplitude analysis. One of these spectra was recorded using all stop pulses greater than 0.1 V in amplitude and the other with the level set for pulses greater than 2 v amplitude. The sharp peaks are due to energetic neutral species from metastable decay in the field-free region of ions with m/z 50-100 but there is a tail in the distri-bution that extends into the high mass region of the spectrum. The fact that the sharp metastable peaks are not noticeably broadened suggests that the decay mode involves loss of a light fragment, probably a hydrogen radical. The tail of the distri-bution is attributed to those ions which undergo metastable decay in the acceleration region. The


2.50 J

I.5CH

0.50 H

1.25 J

U0.75-

0.251

Mass (m/z) 1000 2000 3000 4000 5000 6000 7000 8000

1 ' I i i I i ■ ■ ■ I ■ ■ ■ ■ I ■ ■ ■ ■ I . ■ ■ ■ I . ■ . , I , , , , I , , , , | ,

Window from 0.1V to 2V

10000 15000 20000 25000

Time of Flight (ns)

30000 35000 40000

Fig. 7. Neutral particle 252Cf-PD mass spectra recorded using pulse amplitude analysis and with a total acceleration potential of 14 kV showing the contribution of low velocity neutral particles to the background in the high mass region: (a), spectra recorded in the integral

mode with the pulse amplitude discriminator set at 0.1 and 2 V; (b), differential mode with a 0.1-2 V window.

final velocity attained by these species is depend-ent on where in the acceleration field the disso-ciation occurred. This mode of decay is a source of time-correlated low velocity neutral species. Verification that these species have a low velo-city came when it was found that 50% of the pulses from these species had amplitudes less than 2 V. The measurement was repeated using a 0.1-2V differential window in the pulse ampli-tude spectrum which is equivalent to subtracting the integral >0 .1V spectrum from the > 2 V spectrum. The differential spectrum is shown in Fig. 7(b). The spectrum clearly shows a contri-

bution extending beyond the m/z 7000-10000 background from this source. By comparing the background level for the insulin spectrum shown in Fig. 6(b) with this spectrum we find that 90% of the residual background after high pulse amplitude discrimination is due to these low velocity neutral species. We have identi-fied the source of the residual background in the 252Cf-PD mass spectrum: low velocity neutral species that are formed by metastable decay of ions in the acceleration field that originate from the hydrocarbon film that is present on the surface of the targets.


Conclusion

These studies have shown that under conditions of single ion analysis, the amplitude of electronic pulses formed by a microchannel plate electron multiplier array is dependent on the number of electrons and other ionizing species that are ejected in the initial ion impact at the entrance of a microchannel. This observation implies that more than one electron cascade can be triggered within a microchannel at one time. Pulse ampli-tude analysis of the pulses from secondary ions gives the possibility of identifying that an ion is multiply charged, gives a relative measure of the number of atoms in the molecular ion and offers a new approach to lower the background in the high mass region. When coupled with other back-ground reducing methods, the high mass back-ground in a 252Cf-PD mass spectrum can be reduced by at least a factor of 10.

Acknowledgements

We thank Dennis Shelton for his contributions to the design, interfacing and implementation of the gate circuit that was used in this study and Mark Miller for computer software assistance.

We also gratefully acknowledge the financial support of NIH (GM 26096) and the Welch Foundation (Grant No. A-258).

References

1 B. Sundqvist and R.D. Macfarlane, Mass Spectrom. Rev., 4 (1985)421.

2 A. Hedin, P. Hâkansson and B. Sundqvist, Int. J. Mass Spectrom. Ion Processes, 70 (1986) 203.

3 B. Sundqvist, I. Kaminsky, P. Hâkansson, J. Kjellberg, M. Salehpour, S. Widdiyasekera, J. Fohlman, P. Peterson and P. Roepstorff, Biomed. Mass Spectrom., 11 (1984) 242.

4 B. Wolf and R.D. Macfarlane, J. Am. Soc. Mass Spectrom., 3 (1992) 706.

5 H. Hagstrum, Phys. Rev., 96 (1954) 325. 6 W. Aberth, Anal. Chem., 62 (1990) 609.

J. Martens, W. Ens, K.G. Standing and A. Verentchikov, Rapid Commun. Mass Spectrom., 6 (1992) 147.

7 R.J. Beuhler and L. Friedman, Int. J. Mass Spectrom. Ion Phys., 23(1977)81.

8 M. Karas, A. Ingendoh, U. Bahr and F. Hillenkamp, Biomed. Environ. Mass Spectrom., 18 (1989) 841.

9 R. Kaufmann, D. Kirsch, H. Rood and B. Spengler, Rapid Commun. Mass Spectrom., 6 (1992) 98.

10 E.S. Parilis and L.M. Kishinevskii, Sov. Phys. Solid State, 3 (1960)885.

11 J.L. Wisa, Nucl. Instrum. Methods B, 73 (1979) 587. 12 B. Turko, R.D. Macfarlane and C.J. McNeal, Int. J. Mass

Spectrom. Ion Phys., 53 (1983) 353. 13 P.V. Bondarenko, R.A. Zubarev, A.N. Knysh and B.V.

Rozynov, Biol. Mass Spectrom., 21 (1992) 323.

International journal of Mass Spectrometry and Ion Processes 131 (1994) 193-209 0168-1176/94/S07.00 © 1994 - Elsevier Science B.V. All rights reserved

193

Mass analyzed threshold ionization: structural information for a mass spectrum and mass information for ionic spectroscopy

Philip M. Johnson*, Langchi Zhu Department of Chemistry, State University of New York, Stony Brook, NY 11794, USA

Mass analyzed threshold ionization spectroscopy is a technique for simultaneously recording a mass spectrum and an optical spectrum for each parent ion present. In this method a laser is tuned through the ionization continuum of the sample, while the instrument is set up to record a given mass only when the light is tuned to an ionization threshold. The presence of an ionization threshold is signaled by the presence of long-lived high Rydberg states which are field ionized after they are separated from directly produced ions. A reflectron mass spectrometer then provides mass information to go with the optical spectrum of the ions provided by scanning the laser. Vibrational structure of the ionic ground state is easiest to obtain, but future development may enable rotational information to be gained as well. The technique should find use in resolving isomeric ambiguity in mass spectra, in determining the structure of species present in mixtures, and for examining the photophysics of both the neutral and ionic states of the molecule of interest.

Key words: Mass analyzed threshold ionization spectroscopy; Rydberg states; Reflectron; Mass spectra; Optical spectra

(Received 4 February 1993; accepted 8 April 1993)

Abstract

1. Introduction

Imagine, if you will, that a mass spectrometer would be able to present the infrared spectrum of each ion present in a mass spectrum. How much easier the analysis of the spectrum would be, and how much more information would be present about the sample and about the structure of the ions. The vibrational frequencies of the ions would identify the ionic isomers at a given mass and tell much about their structures, while inten-sities and hot bands could provide such infor-mation as the temperature and vibrational frequencies of the neutral precursor. We would

like to describe a technique which has the promise of providing the equivalent of just such infor-mation. At the present time, the instrumentation required for this technique has only just approached a generally usable form, and only a couple of systems have been studied. Also much of the photophysics involved in the method is still being unravelled. Therefore this must be regarded as a preliminary report on a new spectroscopic tool which is still in its infancy, but shows great promise.

One of the most perplexing things about a mass spectrum is that there is often no clue as to the isomeric identity of a given mass. For electron beam ionization there is a large literature in which such information has been gained by crack-ing patterns, etc., but in general there is no defini-* Corresponding author.

SSDI0168-1176(93)03875-M

194 P.M. Johnson and L. Zhujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 193-209

tive way of knowing the bonding in a species appearing at a given mass. One would like to com-bine mass spectrometry with some other kind of spectroscopy which gives structural information. Nuclear magnetic resonance would be nice, and one could envision that some clever radio fre-quency induced ion beam interaction could be devised, but at present this is not an option. That leaves optical spectroscopy, particularly the kind which reports on the vibrational structure of the species. And as long as photons are being used anyway, why not combine the spectral method with the ionization process itself. In other words, while using photons to ionize the sample of inter-est, use the spectral selectivity of the photons to produce the structural information at the same time.

Normally, ionizing a molecule with a photon is not very spectrally selective. Since the electron can leave with any kinetic energy consistent with energy conservation, absorption in the ionization region of a molecule is continuous, even though the ion is left in discrete quantum states. If one measures the kinetic energy of the electron and subtracts that from the photon energy, the energy of the quantum state is then known for the ion left behind whose mass can be determined by any mass spectrometric method. This is a technique called photoelectron, photoion coin-cidence (PEPICO) spectroscopy and produces information similar to the method presented here. Its problem is that as a coincidence technique where each electron has to be identified with an individual ion, the signal acquisition rate is very low and inconsistent with low repetition-rate lasers.

Another strategy for getting spectral informa-tion in an ionization process is to collect electrons of only a single energy and scan the photon energy over the ionization continuum. Usually the chosen electron energy is as low as possible so that signal will only appear when the photon energy exactly matches the energy difference from some neutral state to a quantum state of the ion. This is called threshold ionization spectroscopy because the

ionization potentials, or thresholds which are the energy levels of the ions, show up as lines in the resulting spectrum. Alternately, because there are Rydberg series converging on each threshold, the presence of high Rydberg states is a marker for an ionic state and one can scan the photon energy looking for these high Rydberg states. It is the combination of threshold ionization spectroscopy with mass spectrometry, called mass analyzed threshold ionization (MATI) spectroscopy, which is the subject to be treated here. To understand MATI, however, we must first discuss the ideas behind threshold ionization spectroscopy.

2. Threshold ionization spectroscopy

2.7. Development

The idea of scanning a light source across the ionization continuum looking for ionization thresholds dates back to Villarejo et al. in 1967 [1]. They used a sector analyzer tuned to pass only very low energy electrons while scanning a monochromatized VUV source, and managed to measure the spectrum of Ar, Kr, and Xe ions. Shortly thereafter, Baer et al. [2] found a new way of discriminating against energetic electrons which involved the fact that electrons with no appreciable velocity can be more easily directed through a small hole. This was called steradiancy discrimination, and has been quite successfully used with continuous or quasicontinuous light sources such as synchrotron radiation.

With the development of pulsed laser ionization sources, a new method of discrimination against energetic electrons was possible. One could simply wait for them to leave the ionization vicinity on their own accord and pulse the remaining near-zero energy electrons into the detector with a small electric field. This was called zero kinetic energy electron (ZEKE) spectroscopy. In addition to the near-zero energy electrons, the extraction field inevitably field ionizes any high Rydberg states which are formed and this adds to the signal. The detailed mechanism of ZEKE was

P.M. Johnson and L. Zhujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 193-209 195

outlined in 1988 [3], although the basic technique was developed in 1984 by Müller-Dethlefs et al. [4].

To take a threshold ionization spectrum, one scans a tunable light source (usually a laser) across the ionization continuum of the molecule of interest. New ionization thresholds, and therefore the rotational, vibrational and electronic states of the ion, occur regularly and are accompanied by Rydberg series which converge on each threshold. Exactly and only at each threshold, all of the photon energy can be contained in the ion core and electrons are produced with near-zero energy. By detecting only these very low energy electrons a threshold ionization spectrum is obtained. Faster electrons can be discriminated against by sector analyzers [1] or by limiting apertures [2]. In the ZEKE method discrimination is made simply by waiting (typically 1-2/is) for any energetic electrons to escape the ionization region before extracting the desired electrons with a small electric field. From the very beginning the ZEKE technique provided a dramatic improvement in the resolution obtainable from photoelectron spectro-scopy. The energy resolution did not depend upon the dispersion of an electron energy analyzer, but was more comparable to the resolution of the light source, which for a laser can be very good.

With the remarkably high resolution being obtained by the ZEKE technique, it became apparent that the ionization potentials obtained by measuring the location of the origin peak did not agree with those gotten by extrapolating Rydberg series. Consideration of this problem led to a much greater understanding of the physics involved in the ionization [3]. It was soon realized that in most instruments there are small stray fields which sweep out all free electrons within less than a microsecond, and the ZEKE signal usually arises from the field ionization of high Rydberg states. The name of the technique is therefore not the most descriptive, but will probably continue for historical reasons. For certain systems such as anions which do not have Rydberg states, the ZEKE method can be used (with care) to collect truly free electrons, but this is an exceptional case.

With this understanding of the ZEKE mechan-ism it became easy to understand the factors affect-ing the signal strength and resolution of the method. The physics of the field ionization are at least semiquantitatively known, so it is easy to estimate how small the pulsed extraction field must be in order to have the desired resolution. Also it is now obvious that one of the major things affecting the signal strength is the lifetimes of the high Rydberg states in comparison to the delay between the excitation pulse and the extraction pulse.

2.2 Rydberg physics

The rule of thumb to estimate how far down into a Rydberg manifold a given voltage will cause ionization is

A = 6VËcm-[ (1)

where E is the electric field in volts per centimeter [5]. This has been shown to approximately hold for atoms, but no careful studies have ever been done for molecules. In order to obtain the highest resolution, one would only like to measure the very high Rydberg states quite close to the limit. From the above formula, it is apparent that if the whole manifold of Rydberg states lives until the field ionizing pulse, one can use a maximum extrac-tion field of about 30mVcm_1 if subwavenumber resolution is desired. This presents some challenges to experimental design. A Rydberg orbital C wave-numbers below an ionization limit will have a principle quantum number n of about 331/V/C and a radius of Ra^/C or 5.8/C/xm.

A most important consideration in thinking about pulsed field ionization of Rydberg states is their lifetimes. There will be a loss of those states which do not live long enough to be around when the extraction pulse is applied. Decay mechanisms can include autoionization, radiationless transi-tions, radiation, and collisions, and the timescale to be considered is the 1-5/is delay typically used in a ZEKE experiment. Vulnerability to each of these loss pathways varies with the principal


quantum number of the Rydberg state and the location of the level with respect to the first ioniza-tion limit and other Rydberg manifolds.

We can deal with radiative decay first and easily because there is very little. Because of the minimal overlap of a Rydberg orbital wavefunction with any core function and the small energy gap to neighboring levels, high Rydberg states have very long radiative lifetimes which increase as n3 [6].

When a molecule is excited to a state which lies above the first ionization potential, it is always possible for that state to decay by transferring some energy from the core to a Rydberg electron, thereby causing it to leave. The core energy which is transferred to the departing electron can be rotational, vibrational or electronic. The auto-ionization lifetime varies greatly depending upon the type of energy transferred and character of the orbital of the excited electron. Of particular interest here is that vibrational autoionization can be relatively slow, particularly if the excited electron is in a high Rydberg orbital. The time that a large-H electron spends in the vicinity of an ion core drops off roughly as 1/tf3, and when the electron is at large distances the vibrational motion of the core does not affect its motion. In a MATI experiment if a molecule autoionizes before the directly produced ions are separated from the Rydberg states, it is lost from the experiment.

Radiationless transitions can occur either above or below the first ionization limit, so they will affect the 0-0 signal as well as all the other lines. It is perhaps not well known that a significant amount of the photon absorption for a large molecule just above the first ionization potential does not produce ionization. These highly excited mole-cules are apparently spreading the excitation among the many degrees of freedom of the mole-cule in an irreversible way. Either some other dissipative path such as dissociation can compete with ionization or the lifetime is long enough that the molecule is lost from the experiment before ionization can occur. Although the mechanism of radiationless transitions in highly excited molecules is not known in detail, presumably high Rydberg

electrons can only transfer their energy to the core when in the core vicinity, so radiationless transition lifetimes should also scale as n3.

Collisional losses could almost be considered another form of autoionization or radiationless transition, depending upon whether the electron gains or loses energy. Since the size of a high Rydberg orbital is very large, collisions are much more probable than in a sample of ground state molecules. For example the pressure (room temperature density) at which the mean distance between molecules is equal to the radius of an n = 200 orbital is 3.3 x 10"6 Torr. Quite frequently therefore, there are other molecules inside the orbit of a high-« electron at normal sample pressures. However, there is a lot of space there, and the electron may never encounter an interloping molecule. Even if it does, the scattering may be elastic and no loss of the Rydberg state may occur. There is evidence that the dissociation of a van der Waals molecule and the subsequent transit of an argon atom through a Rydberg orbit does not disturb it much [7]. Little is known in detail about this subject, however, and we are not able to predict the effects of density on Rydberg state life-times.

Except possibly for collisional effects most decay mechanisms favor a longer lifetime for higher values of n. While the use of large voltages to field ionize the Rydberg states will cause broad lines in a ZEKE spectrum because a broad range of Rydberg states with different energies can be ionized, there is a limit to the breadth. This limit will be dependent on the pulse delay and arises from the lower Rydberg states decaying before they can be field ionized. However, forced autoio-nization due to the Stark mixing of Rydberg mani-folds can cause additional structure and should be watched for if high fields are used. To date almost no systematic work has been done to investigate the fate of Rydberg states and the only information we have is that for larger molecules like substituted naphthalenes, the largest the linewidth seems to get is about 6 cm"1 [8]. That implies that Rydberg states with n < 100 have lifetimes of less than 1 /xs.

P.M. Johnson and L. Zhu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 193-209 197

Due to the fact that the molecules can undergo radiationless transitions at all thresholds but auto-ionization occurs only above the lowest, the origin of a spectrum may have a different linewidth variation with respect to field ionizing voltage than the rest of the spectrum.

3. MATI

While ZEKE is an extremely valuable technique, the fact that electrons are measured means there can be an ambiguity about the nature of the species giving the signal. Since the ZEKE tech-nique does not involve actually measuring the kinetic energy of an electron, why not simply measure the ion instead of the electron? Then one could get the desired mass information. This is not as simple as in ZEKE, where the discrimination against the directly produced electrons relies upon their attaining appreciable velocities and leaving the ionization vicinity. Ions are not at all mobile and cannot be relied upon to go anywhere of their own volition.

3.1. Rydbergjion separation

The basic idea behind MATI is that directly produced ions can be separated from Rydberg molecules by putting them in a small electric or magnetic field. This field should be kept as small as possible in order that field ionization does not deplete the entire upper Rydberg manifold. In most of what follows we will refer to an electric field as the separating element, but magnetic separation schemes can be envisioned and will be subject to the same considerations.

In order to perform a MATI experiment, ions resulting from the field ionization of Rydbergs must be separated in time from the directly produced ions at the detector. This means that these groups must either be separated in space in the source region before the field ionization or a differential velocity component be given to one group so the separation can occur during the transit through the apparatus. The size of the

originally produced cluster will expand due to the thermal components of the molecular velocities, and it is instructive to examine the sorts of fields and times necessary to overcome the thermal diffusion.

For convenience, let us assume the original distribution of ions and Rydbergs is a gaussian of the form

£(*o)=- i72 exp(-b2x20] (2)

This can be convoluted with a Maxwell-Boltzmann velocity distribution to yield the spatial distribution at any time /:

1 G{x, t) =

xexp

2kTV

m

1/2

J_ 2kTV (3)

This is seen to be a gaussian with full width

„ / 1 2kTt2\ 1/2

m (4)

so the distribution spread is linear in time. If one is trying to separate two distributions of the above type, it must be done rapidly enough that the induced separation beats the thermal spread. The distance that an ion will move in an electric field E is given by d—qEt2/2m, so separation (defined such that the distributions overlap at their \\e points) can be achieved in time / if

„ A m I 1 2kTV 1/2

qt2 \bA m

(5)

It is instructive here to assume a vanishingly small initial distribution and to put in some numbers for the constants. With E in volts per centimeter and m in atomic mass units, separation can be attained if

Et > 5.2 x 10~8(7>ηβ)1/2

(6)

198 P.M. Johnson and L. Zhu/Int. J. Mass Spectrom. Ion Processes 131 (1994) 193-209

Since both E and / should be kept as small as practical, it is seen that for a given mass the temperature should be kept as low as possible. However, even at room temperature a IV cm- 1

field will separate mass 80 ions from a group of excited neutrals in about 8//s. This corresponds to a distance of about 3.8 mm.

Although one can envision MATI schemes which take advantage of an initial physical separation of the ions from the Rydbergs, it is not this initial spatial division which is most important for resolution at the detector. Rather it is the relative velocity of the packets induced by the applied voltage which is most effective since this results in continued separation after the original field is removed. Still, however, a relative velocity should be attained which is larger than the thermal spread of the packets. The thermal velocity at the \/e point of the Maxwell-Boltzmann distribution is v = (AkT/m)xi2, and this velocity is attained at / = m/qE in an applied field. This means that a relative velocity of the packets exceeding the thermal spread will be obtained when

^ ^ ( ^ Y ^ U x l O - 8 ^ ) ' / 2 (7)

It is seen that this critical velocity occurs before the packets have separated physically. The time t is that at which the field ionization occurs, since after that any applied field will affect the packets equally.

In order to achieve the best separation low temperatures are desired, and these are most easily obtained in supersonic beams. While the mean energy per particle of a gas expanding in a supersonic beam does not change, the velocity distribution narrows considerably as the random motion is converted into directional flow. In the end the mass flow velocity is (5/3)1//2 = 1.29 times the most probable speed of the gas molecules before expansion, and the temperature which is a measure of the width of the velocity distribution is [9]

T Γ = 1 + Ι ( 7 - 1 ) Μ 2 ( 8 )

where M is the Mach number of the gas the 7 is the heat-capacity ratio Cp/Cv. The terminal Mach number is

M r = \33{P0D)0A (9)

for argon when the nozzle diameter D is in centimetres and the reservoir pressure P0 is in atmospheres. The exponent may be somewhat higher for helium [9].

In order to get the lowest temperatures in a supersonic expansion, it is necessary to expand the molecule of interest in some light monatomic carrier gas, usually helium or argon. Since the mean velocity of the carrier is different from that of the sample gas due to the mass difference, depending upon the beam conditions, the sample may be travelling with the carrier at its own mean velocity, or something in between. When the sample moves slower than the carrier, it is called "slippage", and is more likely to occur for large mass differences [10]. A helium beam at room temperature has a mean speed of 1.26 x 105 cm s_1

or 1.26 mm ßs~l. A molecule of mass 80 moving at this speed has a kinetic energy of 0.66 eV, up from the value for that mass at room temperature of kT, or 0.025 eV. The faster the sample speed for a given separating field, the more voltage drop the molecule will see during the several microseconds allotted for the separation. Therefore it is advanta-geous to have the sample seeded in a helium expansion without any slippage. Fortunately, this is also the condition which gives the narrowest velocity distribution.

Any applied retarding voltage could eventually stop an ion in a supersonic beam dead in its tracks. However, if one wishes to keep the field less than 1 Vcm"1, that would take at least 6 mm of travel, which takes about 1 ms because the ions are slow-ing down. For many systems this may be longer than the Rydbergs want to wait, so one often has to be content with merely slowing the ion speed with respect to that of the Rydbergs.

When the Rydberg states are physically or velocity separated from the directly produced ions, a small voltage pulse is applied to ionize the

P.M. Johnson and L. Zhujlnt. J. Mass Spectrom. Ion Processes 131 (1994) 193-209 199

highest surviving Rydbergs. In order to have decent spectral resolution, it is best to keep this pulse small, as is done in ZEKE. Unlike ZEKE, where the excitation is done in an almost fieldless region, in MATI the highest Rydbergs have not formed due to the applied separation field. One might think that this loss of some of the sample would result in a lower signal level in a MATI experiment. When the Rydberg manifold is sufficiently long-lived this is not the case, however.

It is well known that the absorption cross-section per unit energy is constant for the higher Rydberg manifold. This is because the increase in the density of states (as n3) as one goes to higher energy is exactly compensated by a decrease in the oscillator strength (as n~3 [11]) of each individual state. Therefore near an ionization threshold where the absorption is quasicontinuous, a laser of a given bandwidth will excite the same number of molecules regardless of the distance from the threshold. A photon 6 cm"1 below the threshold will produce just as many Rydbergs as a photon at the threshold, and even if a 1 Vcm - 1 field has destroyed all of the Rydberg levels closer to the threshold, an ionizing electric field pulse of 1.4 Vcm"1 will still provide 1cm"1 resolution at the same signal level as in a ZEKE experiment. If Δ is the resolution desired and E0 is the separation field, the ionizing field Ex is given by

£i~(f+\/^) (10)

In practice, the MATI signal level may be decreased because of the shorter lifetimes of the lower Rydberg states being used, and of course the absolute position of the whole spectrum will be shifted to lower photon energy by the initial field ionization.

After the field ionization pulse, one still has lower Rydberg molecules mixed in with the newly created ions and these must be separated by another small field before a high voltage can be applied to drive the ions into the mass spectro-meter. There are no real time constraints on this separation if the loss mechanism for them involves

E: 0,0 \l GO (V/cm)

Fig. 1. Schematic of the simple MATI apparatus used to obtain first spectra. Here there are no switched voltages and field ionization occurs when the Rydbergs pass through a grid by virtue of their velocity in a supersonic beam traveling down the axis of the flight tube. The electric fields in the various regions are given, with the flight tube being field free and the detector being on the right.

a neutral product since they are not desired any-way. However, if these states tend to autoionize it might be better to affect a separation before too many unwanted lower Rydbergs join the field-ionized higher Rydbergs. After this secondary separation and the application of the high voltage which will ionize all remaining Rydbergs, there should in theory be three packets of ions. In prac-tice we have never seen more than two, however, and it is assumed that the directly produced ions are mixing with the field-ionized lower Rydbergs in the geometries we have tried. When the three pack-ets of ions have been produced and separated, it is then necessary to propagate them through the mass spectrometer without having them spread.

Our original MATI apparatus established the various voltage regions with grids and apertures and relied upon the motion of the ions through these regions to separate and field ionize the excited molecules at the appropriate times. A diagram of this simple scheme is shown in Fig. 1. The molecular beam is directed through an aperture and then three grids which define regions of 0.9, -2 , and -60 Vcm- 1 . The laser is focused into the first region at a variable distance from the first grid, depending upon the amount of delay that is desired between creation and field ionization of the Rydbergs. Ions created directly by the laser are retarded in this initial field and arrive at the first grid well after the Rydbergs. The jump from 0.8 to 2 Vcm- 1 in the second region field ionizes some of the Rydbergs and the accelerating field separates them from the excited molecules with lower


U)

100 -A

80

■e 60 <

2 40

G 20 O

CRHC

C6H6»Ar2

+ (C6H6)2

C6H6.Ar

— i 1 1 1 1 1 1 1 1 —

500 1000 1500 Channel Number (50 ns)

Fig. 2. A mass spectrum of a benzene-Ar sample taken on the apparatus of Fig. 1. The light is tuned to the lowest ionization potential of the benzene-Ar van der Waals complex. The MATI signal is the small darkened peak which appears before the larger direct ionization

peak.

100 H

3

2 ΊΗ £

80 -

60 -

40 H

20 -

J — i — i — i — i I i i i i I i i ι i I ι ι ι ι I ι i i i _ l ι ' ■ '

- i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — r

227 226 225 224 223 222 221

w 100

G 3 80

CÖ 60

40 2

£ 20

o 0

I 1 I I I I I I I l _

i — i — r — T

227 1 — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — i — r

226 225 224 223 Probe Laser Wavelength (nm)

222 221

Fig. 3. The vibrational structure of the ground state of the pyrazine ion. The upper trace is a ZEKE spectrum, while the lower is a MATI spectrum taken on the apparatus of Fig. 1.


a 4 a s

Γ " M " I f L^

cl, d , ( I , I-,

Fig. 4. Schematic of a tandem focusing MATI apparatus. In this device the voltage is switched in the source region d\ to cause field ionization and provide an acceleration αλ. Ions are brought to a space focus after the flight region c/3 and accelerated/deflected into a reflection stage which does velocity focusing. The distance L2

referred to in the text is the total field-free flight length before and after the reflection grids and is not to scale in this diagram.

quantum numbers. Upon passing through the second grid, the remaining Rydbergs are ionized and all ions are accelerated into the flight tube which is maintained at a negative high voltage. The sequence of retarding and accelerating fields were designed to ensure that the near-threshold ions of a given mass get to the detector before the unwanted ions, which may be in much greater abundance. In that way detector saturation or ringing, and tailing of the mass peak will not obscure the smaller MATI signal.

This scheme has the great advantage of simplicity. There are no switched voltages and the apparatus is easy to construct. However, there is no space or velocity focusing of the ions on the detector, which severely limits the mass resolution obtainable this way (Fig. 2). Figure 3 shows, however, that the threshold ionization spectrum obtained with such a simple method agrees favorably with a spectrum obtained with ZEKE. A lower signal-to-noise in the MATI spectrum has been found to be attributable to a lack of transverse focusing in this original apparatus. In order to improve the collection efficiency and mass resolution it is necessary to pay greater attention to space and velocity focusing.

3.2. Mass focusing

There are two primary causes of the time spread

of ions reaching the detector in a time-of-flight mass spectrometer: the spatial spread of the ions as they are created, and the spread of velocities in the sample as the ions are formed. Each of these must be compensated in order to achieve the high-est mass resolution in an instrument. Fortunately techniques for focusing both space and velocity distributions are well known [12,13]. The only difficulty with MATI is that the voltages in the source region should be kept much lower than is ordinarily used in a TOF mass spectrometer. Our present spectrometer is one in which space focusing [12] is followed by a reflectron region [13] to give velocity focusing. This is somewhat in the same spirit as an instrument reported by Weinkauf et al. [14] which was used to investigate photo-dissociation of ions.

3.2.1. Space focus The concept of compensating for the initial

spatial spread of ions in a TOF mass spectrometer was introduced by Wiley and McLaren in 1955 [12]. The idea is to impart different velocities to the ions originating from different source locations in such a way that they all arrive at the detector at the same time (if they have the same mass). In this way the size of the original sample of ions does not matter to the mass resolution and this is crucial if high resolution is desired. Space focusing is imple-mented by using two acceleration regions and a drift region as shown in Fig. 4. In MATI, however, one should be concerned that having separated two groups of ions, a space focusing scheme might put them back together again and void the experiment. What saves the day is that not only have the ion groups been separated spatially, but they have been given different velocities.

Although Wiley and McLaren [12] provided an equation that the dimension and voltage parameters must satisfy in order to get space focus-ing, it was derived assuming that the initial velocity of the ions is zero along the axis of the flight path. When high voltages are used to accelerate the ions,


this is a good assumption. However, in MATI low voltages are used and the initial velocity can be a substantial perturbation on the space focusing. We will therefore provide an equation for the focusing which includes the initial velocity and is somewhat more convenient for the determination of the necessary voltages than that given pre-viously.

We will use the notation that the distances and times spent in the three regions of the ion flight are d\,d2,d3 and tx,t2,t3. The initial velocity is v0 and the accelerations in regions 1 and 2 are ax = qEx/m and a2 = qE2/m, where Ex and E2 are the fields in the regions. Thç times in the three regions can then be given:

U = -v0 + (v2

0 + 2axdxY/2

ax (H)

-(VQ + 2axdx)x'2 + (vl + 2axdx + 2a2d2)

x/2

a2

t* = (vl + 2axdx + 2a2d2)

{/2

(12)

(13)

Space focusing is achieved when the derivative of the total time with respect to dx is equal to zero. After some algebra and making the substitutions, q = a2/ax, r = d3/d2, and

2d2ax d2

one gets

q4 + {3x - 2)q3 + (3x2 - 6x + \)q2 +

(14)

x3 -6x2

+ (2 + r - — )x q-2x3 + {r+\)x2 = 0 (15)

Equation (14) simply states that the initial energy can be ignored if it is substantially less than the energy gained in the first region. Otherwise Eq. (15) can still be used to calculate the required voltages. There is usually one real root to the quartic equation which is greater than unity.

When the space focus is operated in tandem to the velocity focus, the total time difference at the detector between the ions produced from the Rydberg states and those produced directly by the light is entirely due to the Wiley-McLaren stage. The reflector is set up to compensate for velocity differences in that section so substantial further separation of the ion groups does not occur there. This is mitigated by the fact that a compromise must be made in setting the voltages to account for the substantially different velocities of the two ion groups. In the source region, the transit time difference arises from the differences in position and velocity produced by the initial field, which gives an acceleration a0 to the directly produced ions for a time t0.

The time difference of arrival of the two ion groups at the space focus can be determined by realizing that the source distance for the directly produced ions is

dt = d,-^ (16)

and the velocity is VQ = v0 + a0t0. Then the velocities of the two groups at the two grids are

vx={vl + 2axdx)x>2 (17)

υΐ = [vl-\-2aidi +2a0v0t0 + (al - α ,α 0 ) /ο] Ι / 2

(18)

v2 = (v2l+2a2d2)

l/2 (19)

vî=[(vî)2 + 2a2d2]l/2 (20)

From these one can get the time difference from

AT=T+_T=z?n!± + {v+-Vl)(l-l) ax \ax a2J

+dn.i)+(AÂ (2I) \vj v2J a2

For a negative a0 and under space focusing condi-tions, this separation is almost proportional to 2a0t0/ax and is only weakly dependent upon the other parameters. Therefore, the time separation at the detector will be a few microseconds for a


5/xs separating field with ax/a0 — 2. Interestingly, as tff3 increases, the separation gets less since V2 > v2- In general, since the mass resolution is adversely affected by the thermal spreading in this stage, it is advantageous to keep it relatively short.

3.2.2. Velocity focus The general principles of a reflectron mass

spectrometer have been extensively discussed [13]. The basic idea is that ions which start out at the same time with different energies travel different distances which are carefully planned to make them end up at the detector at the same time. This is an excellent means of compensating for the thermal velocity spread and/or the velocity spread at the first grid caused by the initial spatial spread of the ions and has allowed TOF spectrometers to attain extremely high mass resolution.

The geometry of a reflectron stage is shown in Fig. 4. The voltages are set in the turn-around region so that ions with higher velocity travel a longer path. In order to establish the proper voltages, one can get both first and second order focusing by setting to zero the first and second derivatives of the flight time with respect to the initial velocity s0. With the distance Lx being that for an original acceleration a3, L2 being the flight length in the field free regions before and after the reflector, and L3 being the distance between the reflector grids with acceleration <z4, the velocities at grids 5 and 7 can be written

sx={sl+2a,Lx)xl2 (22)

s2 = ( 4 + 2fl3L, + 2a4L3)l/2 (23)

The total TOF in the reflection stage is then

T = — + — + - ^ (24) a3 si a4 a5

By setting the first derivative of the time with respect to s0 to zero one can determine the field

in the turn-around region of the reflector from

1 1 s2

«5 #4 2

x[- — + MSl-2A+2L,-L2)] (25) L a3ô s\ \a3 a4 ) \

where a4 is determined from likewise setting the second derivative equal to zero and eliminating a5:

_ a3s2l[sl(2L3 + L2)+s\Ll]

0 4 - L3[sl(s>-3a3L2)-sl] ( 2 6 j

This is a remarkably simple relationship consider-ing the complexity of the intermediate equations. The function T(s0) is very flat in the region of s0

used in the above equations, with the first deriva-tive άΤ typically being \0~9Tds0. It should be noted that these equations can be applied to any reflectron mass spectrometer and are easier to use than any previous treatments of which we are aware.

A TOF scan from an apparatus such as just described is shown in Fig. 5 for the molecule pyrazine (mass 80) expanded in a mixture of argon and helium. The dimer and the argon van der Waals complex are also seen. For this experi-ment the separation voltage was only -0.9 V cm- 1, the field ionization pulse was 3.15 VcnT1, and the extraction field was less than 0.5 Vcm- 1 . It is seen that the separation between the ion groups is a substantial fraction of the total flight time and the mass resolution is of order 50. This is con-sistent with the modeling we have done using the above equations, which also shows that better conditions could be found.

The primary cause of loss of mass resolution in this system is due to the thermal velocity spreading during the long time spent in the low voltage space focusing stage. Wiley and McLaren [12] commen-ted that it is impossible to get both space and energy focusing in this type of arrangement and things are made worse by the differing characteris-tics of the two ion groups. The most important parameter affecting the mass resolution in the spectrometer described here is the extraction field E\. The thermal time spread at the detector falls off


loo A

m 80

Ö P

>-

60 H

40

20 H

0 H

- | 1 1 1 ( 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r

0 50 100 150 200

Time of Flight (με)

Fig. 5. A mass spectrum of pyrazine taken on the apparatus of Fig. 2. The light was tuned to the lowest ionization potential. Pyrazine-Ar and dimer masses are also seen in the mass spectrum and the field ionized ions are designated F.I. This spectrum was taken with a very

low extraction voltage (< 0.5 Vcm- 1) indicating that moderate mass resolution can be obtained even under those conditions.

nearly exponentially with an increase in this field. This effect is clearly shown in Fig. 6. Since E\ also affects the optical resolution, a trade-off between optical and mass resolution is necessary. We are currently exploring different source configurations and there is reason to believe that even room temperature MATI can be done with moderate mass resolution.

Of course, if a high extraction voltage can be used, the time spent in the source region becomes negligible and MATI can be done in a normal reflectron mass spectrometer with very high mass resolution. This has been shown to be the case by Jouvet et al. [15], and by Krause and Neusser [16]. The high extraction field should affect the spectral resolution, and the spectrum of Jouvet et al. shows the « 10 cm- 1 resolution typically obtained with high field pulsed field ionization of aromatics. However, this broadening was not found to be the case in the spectra of Krause and Neusser. The reason for the narrowness of their spectral lines may point to some very interesting Rydberg

physics. From what we now understand, good optical resolution with high voltage extraction should not be the general case, however.

3.3. Other geometries

Electric field separation of the Rydbergs from the ions collinear to the molecular beam is not the only way that can be envisioned. In many instances, transverse extraction of the field ionized ions would be much more convenient, and magnetic fields could be used for the separation.

Geometries can be considered with the separation field either still collinear to the mole-cular beam, or transverse to it. If the separating field is transverse to the molecular velocity, the directly produced ions can be given a sideways velocity component which, after a given flight time, would prevent them from going through an aperture which defines the entrance to the mass spectrometer. A significant flight time after the


1500 -,

3 looo A \

1 \

U - | , . 1 1 < ■ 1 , , , , 1 , 1 , 1 1 1 , 1 1 ,

0 2 4 6 8 10 Et (V/cm)

Fig. 6. Calculated width of a mass peak in a tandem apparatus as a function of the extraction voltage. The total flight time is about 70 /is, the temperature of mass 80 is 10 K, and the other voltages are optimized for a space focus flight length of 6.5 cm.

field ionization would allow the different ion pack-ets with different velocities to separate even if the groups had not completely divided at the time of the field ionization. However, this type of scheme requires complete spatial separation in the source region, which is not required in the above method with velocity selection in the reflectron.

When the ion packets have become transversely disconnected, it is possible to accelerate them longitudinally as above, or transversely. By first moving the directly produced ions transversely in one direction, then field ionizing and applying the same voltage in the opposite direction for the same amount of time, the desired ions can even be pushed above a repeller plate while the others continue on in the beam direction (Fig. 7(A)). We tried this method, without too much success, allowing about a 10 cm flight for the separation. From our experience, we feel that much attention will have to be paid to keeping the ions together in the low voltage environment for this method to work.

Longitudinal acceleration after a transverse separation would have the advantage that the ion packets would not have to be completely separated in the source region since a defining aperture could be placed well inside the flight tube at an appropriate ion focus (Fig. 7(B)). With appro-

A H

-/ov .n.ov — " o v

C I) Reflectron

Fig. 7. A few of the many possible ways of separating directly-produced from field-produced ions: (A), give the direct ions a transverse velocity component and then deflect the field ions into a TOF mass spectrometer; (B), give the direct ions a small side-ways velocity component so that they encounter a beam stop far down the flight path or appear at a different spot on the detector; (C), deflect the directly-produced ions out of the beam with a magnetic field; (D), have a tandem focusing apparatus oriented perpendicular to both the laser and supersonic beam.

priate ion optics, it may even be possible to have the different ion packets arrive at different places on a multichannel plate detector and the various signals simultaneously recorded using a segmented collector plate.

The transverse velocity component need not be applied to the ions by an electric field, but a magnetic field may do as well (Fig. 7(C)). The problem with magnetic fields is that they are very difficult to confine to a small region. Therefore, it may be necessary to rapidly turn off the magnetic field in order to make it affect the directly produced ions without drastically affecting the ions produced later by the field ionization. However, the cyclotron orbit time is independent of the velocity of an ion, so all ions will turn the same amount in a given magnetic field applied for a given time. To turn an ion of mass 80 through a 45° arc in 2 /xs requires a magnetic field of 0.33 T, and smaller angles require linearly smaller fields. Due to the inductance of electromagnets, it is difficult to turn them off rapidly. However, if one were able to produce about a 0.1 T field which could be turned off in less than 1 ^s, magnetic separation is an attractive option.

Not enough is known about the effect of


magnetic fields on molecular Rydberg states to safely predict what a varying magnetic field would do to autoionization and radiationless tran-sition lifetimes. This is something that would have to be considered (and would be an interesting object of study) in such a M ATI scheme.

Two transverse flight tube schemes are also worthy of mention. In the first, one applies a voltage longitudinal to the molecular beam in order to separate the ions as in the standard scheme, but then one applies a nonuniform trans-verse voltage. By making a repeller plate out of some resistive material and applying a voltage across it longitudinal to the molecular beam, there would be a transverse voltage on the ions which varied linearly with distance down the beam. If the packets were completely separated, they would then be accelerated by different amounts and arrive at the detector at different times.

A final, fully transverse geometry (Fig. 7(D)) essentially consists of the standard reflectron apparatus described above but rotated by 90°. Again there would be a low voltage Wiley-McLaren source region and the same sort of voltage variations applied. The voltage after the ions have gone through the first grid could be a little higher in this case because there would be no lower Rydbergs to worry about, and that grid would be placed closer to the laser focus because the beam will be going parallel to it. The different velocities and positions of the directly produced ions and the field produced ions in the source region would provide those packets with different space and velocity focusing in the subsequent reflector and should therefore provide similar, or even superior behavior to the collinear geometry.

This sort of arrangement, without the low voltage source, has been demonstrated by Jouvet et al. [15]. They used about a 0.6 Vcm- 1

separation field and a 65 Vcm- 1 extraction pulse to record the MATI spectrum of paradifluoro-benzene in a commercial reflectron TOF mass spectrometer.

3.4. Light sources

3.4.1. Pump-probe A pulsed laser is therefore the most practical

source for a ZEKE or a MATI experiment. Since the ionization potentials of most molecules are in the 7-14eV range, single photon excita-tion in the near UV region is not possible and therefore one must resort to either some multiple photon excitation scheme or produce light in the VUV region by some nonlinear mixing method.

To date, all MATI experiments have used two photon pump-probe excitation to get the total energy of the molecule to the ionization limits. This works quite well with molecules which have stable low lying states. In this method one laser is tuned to a resonance in some bound molecular electronic state. With high enough resolution, the excited molecule can therefore be prepared with specific rotational and vibrational quantum numbers. A second laser is then wavelength scanned over the photon energy range which promotes the excited molecule into the electronic continuum region. This has the obvious advantage that VUV light is not needed, but there are other benefits as well. One is that it is possible to get a great deal of information about the bound intermediate state from the ZEKE or MATI spectrum.

Since two pulse lasers are used in the excitation, they can be either fired simultaneously in time, or the ionization laser can be delayed. If the delay is systematically varied, the lifetime of the inter-mediate state can be obtained from the decay of the signal with respect to the time between the laser pulses. This technique has long been used while collecting the total ion signal, and with laser pulse widths ranging from nanoseconds to femtoseconds the dynamics of a large number of molecular systems have been explored. Using the ZEKE signal, new selectivity provides additional information about the dynamic processes [17]. The loss of ionization signal in a pump-probe experiment may be either because of a radiative


or a radiationless process, or commonly a combi-nation of both. In a radiationless "decay" in the collisionless gas phase, the molecule does not lose energy, but simply redistributes it among its many degrees of freedom. This can be envisioned as sort of an entropy increase, but is not because in theory the energy will have a Poincaré recurrence to the original state if given enough time. For any mole-cule of more than a few atoms, however, the recurrence times can be so long that the energy degradation can be considered permanent. Since angular momentum of the molecule must be conserved, most energy degradation involves con-verting electronic energy to vibrational energy or vibrational energy involving high frequency modes to vibrational energy involving a larger number of low frequency modes.

The point of this degradation concerning a pump-probe MATI scheme is that when a radiationless transition occurs, particularly one which involves changing electronic states, the Franck-Condon factors for the ionization process are drastically altered. If a molecule has many low frequency vibrational modes populated after a radiationless transition in the intermediate, a probe photon which only reaches the lower part of the ionization continuum is unlikely to cause ionization because the Franck-Condon factors for going from the hot intermediate to the cold ion will approach zero. With ZEKE or MATI, one can not only observe the decay of the signal with respect to the delay between the lasers, but one can also examine the evolution of the Franck-Con-don factors, giving an indication of how energy is evolving in the intermediate state.

Another advantage of a pump-probe scheme is that it allows one to circumvent unfavorable Franck-Condon factors occurring between the neutral ground state and the ion. If there is a substantial geometry change in a molecule upon removing an electron, ionization in the lower part of the continuum is going to be an improbable event. By using a vibronic state as an intermediate which has a geometry more similar to the ion, it is often possible to observe the lower

ionization thresholds which would otherwise be inaccessible.

3.4.2. VUV A single photon scheme using VUV radiation

has the advantage that it is completely general for any molecule irrespective of its bound state structure. One needs to know nothing about the near-UV absorption structure or the photo-dissociation properties of the electronically bound excited states. It would therefore be very attractive for routine use were it not for the difficulty of generating the tunable VUV laser light. Fortunately, with modern high powered pulsed lasers that has recently become possible with useful amounts of power [18].

Photons in the VUV are easily generated with current laboratory lasers by frequency tripling or four wave sum or difference mixing in various rare gases and metal vapors [19], primarily mercury [20] and magnesium [21]. All of the gases and vapors provide more efficient conversion using resonant wave mixing requiring two different colors, and the most attractive systems appear to be difference frequency mixing in krypton and sum mixing in mag-nesium. Wavelengths up to the VUV can be gene-rated using sum mixing in barium borate crystals. Conversion efficiency in the gases is quite low (10~6

to 10~3), but with tens of millijoules from the dye lasers, the microjoules of energy in the VUV beam are quite sufficient to produce a spectrum. Resonant frequency subtraction in krypton can have efficien-cies exceeding 10~4 of the input power and high resolution ZEKE by even the less efficient frequency tripling in rare gases has been shown to produce excellent results [22]. It should be emphasized that because the molecular transitions are one photon processes and the ion collection efficiency is very high, large light fluxes (millijoules) are not needed.

Because of the nonlinearity of the frequency mixing process, it is desirable to have as much power in the input laser beams as possible. The most efficient processes are the ones where one fixed frequency input beam is resonant with a two photon transition in an atom while a second


scanned frequency input is added to or subtracted from that two photon energy. The desirable components of a VUV laser system are therefore a very large pump laser which drives two dye lasers, plus various frequency shifting crystals and cells. For example one of the most attractive schemes is the difference frequency mixing in krypton where one dye laser must produce 216.6 nm light (e.g. by tripling 649.8 nm or doubling 543.9 nm and mixing with the YAG fundamental), while the other is tuned over the entire visible and near-UV range to produce light from 120 to 200 nm. Since difference mixing does not require phase match-ing, scanning ranges can be as long as the dye tuning range allows.

If light above the lithium fluoride window cutoff is needed, it is possible to triple light in a supersonic jet of rare gas expanding in a vacuum. Such a scheme has been used to advantage in using ZEKE to examine the rotational structure of ions [22] and could equally be used for M ATI.

3.5. Samples

MATI will be particularly useful for studying mixtures of any sort. These arise in many, if not the vast majority, of the situations which are encountered in chemical research. The products of most chemical reactions are not pure, and it is often inconvenient or impossible to separate the components before analysis. This is particularly true if the chemistry takes place in the supersonic beam itself, as can happen in the photolysis of molecules to produce radicals.

The presence of the supersonic beam in a MATI experiment naturally suggests experiments related to clusters and van der Waals complexes since they are commonly formed in the expansion. Although it is possible to find beam conditions which favor some distribution of cluster sizes, it is impossible to get a pure beam. Even to characterize the cluster distribution with mass spectrometry is difficult because most ionization methods deposit enough energy into the clusters to fragment them, chang-ing the initial distribution. The fact that MATI is a

threshold technique means that once the ionization has occurred there is no excess energy in the ion unless one wishes there to be by exciting a higher vibra tional level. At the origin, fragmentation of the ion cannot occur and therefore the ion mass measured in the instrument is a true measure of the molecule that absorbed the light.

In a pump-probe experiment, an intermediate resonance can sometimes be used to choose a particular species out of a mixture. Then it is possible to use MATI to investigate the dissocia-tion of the ion with respect to its internal vibrational energy. Such a study has recently been published by Krause and Neusser [16]. They used an intermediate resonance to selectively ionize the benzene-argon van der Waals complex as a function of resulting vibrational energy in the ionic complex. They found that the mass of the complex disappeared as the ion energy is increased, with the resultant appearance of the mass of the benzene monomer ion. From this a limit on the dissociation threshold of the ion can be determined, and working back through the ionization potentials of the two species the binding energy of the neutral ground state complex can also be calculated. This is an excellent example of the sort of structural problem which can be approached by this technique.

4. The future

The MATI technique is still in its infancy and it is not clear what various directions will be taken as it matures. A few can be envisioned, however.

By gating out specific groups of ions, MATI can be used as a source of state selected ion beams. One will be able to put a specific amount of a given type of vibrational energy into the ion by tuning the light to the appropriate threshold. This sample of ions in a pure excited vibrational state can be then used to investigate the effects of vibrational excita-tion on reactivity. This is very difficult to do with neutral molecules because pure excited state samples are not easily obtained. Other methods


of state selecting ions are not as versatile as MATI, which will be used for this purpose in the very near future.

Could one also study the excited electronic states of an ion with MATI? To date we are not aware of any ZEKE studies of such a type. The more rapid autoionization of electronically excited states may preclude such a feat, but it may be possible to find some systems in which the conditions are right to see at least the first excited electronic state in ZEKE or MATI.

How about MATI as a general purpose tool in mass spectrometry? For now, this method will probably remain in the domain of the optical spectroscopist because of the complicated laser systems which must be used. However, the rapid development of laser techniques would indicate that easily used VUV light sources are not too far in the future. By the time these appear, there should be a reasonable database of knowledge which will enable the mass spectrometrist to acquire the additional information that MATI can provide. Other than the laser and the supersonic beam there is nothing involved in the MATI scheme which is more complicated than what is currently commercially available.

Can the supersonic beam be done away with? There is nothing theoretically impossible about building a MATI source which will focus the thermal velocities. Then room temperature samples would become viable. Already, if one is willing to use high extraction voltages the designs in existence would suffice for this. If a lower data acquisition rate is allowable or only one mass is to be monitored, the MATI source could be directed into a magnetic sector mass analyzer to take advantage of the vast background of development on those systems. Mass resolution would then not be a factor. The ability to do room temperature samples would open up the whole range of mass spectrometric applications to the advantages of finding out more about the structure of the ions being observed.

Whatever the future, this new technique should prove to be useful.

Acknowledgment

We would like to acknowledge the support of the US Department of Energy under grant ED-FG02-86ER13590.

References

1 D. Villarejo, R. Herrn and M. Inghram, J. Chem. Phys., 46 (1967) 4995.

2 T. Baer, W. Peatman and E. Schlag, Chem. Phys. Lett., 4 (1969) 243.

3 G. Reiser, W. Habenicht, K. Müller-Dethlefs and E. Schlag, Chem. Phys. Lett., 152 (1988) 119.

4 K. Müller-Dethlefs, M. Sander and E. Schlag, Z. Naturforsch. Teil A, 39 (1984) 1089; Chem. Phys. Lett., 112 (1984)291.

5 W. Cooke and T. Gallagher, Phys. Rev. A, 17 (1978) 1226. 6 H. Bethe and E. Salpeter, Quantum Mechanics of One- and

Two-Electron Atoms, Springer Verlag, Göttingen, 1957, p. 269.

7 X. Zhang, J. Smith and J. Knee, J. Chem. Phys., 97 (1992) 2843.

8 J. Smith, C. Lakshminarayan and J. Knee, J. Chem. Phys., 93 (1990) 4475.

9 R. Smalley, L. Wharton and D. Levy, Ace. Chem. Res., 10 (1977) 139.

10 J. Anderson, R. Andres and J. Fenn, Adv. Chem. Phys., 10 (1966) 275.

11 H. Bethe and E. Saltpeter, Quantum Mechanics of One-and Two-Electron Atoms, Springer Verlag, Göttingen, 1957, p. 263.

12 W. Wiley and I. McLaren, Rev. Sei. Instrum., 26 (1955) 1150. 13 V. Karataev, B. Mamyrin and D. Shmikk, Sov. Phys.-Tech.

Phys., 16(1972) 1177. 14 R. Weinkauf, K. Walter, C. Weickhardt, U. Boesl, and E.

Schlag, Z. Naturforsch. Teil A, 44 (1989) 1219. 15 C. Jouvet, C. Dedonder-Lardeux, S. Martrechard-Barra

and D. Solgadi, Chem. Phys. Lett., 198 (1992) 419. 16 H. Krause and H. Neusser, J. Chem. Phys., 97 (1992) 5923. 17 S. Hillenbrand, L. Zhu and P. Johnson, J. Chem. Phys., 95

(1991) 2237. 18 J.W. Hepburn, in C.Y. Ng (Ed.), Applications of coherent

vacuum ultraviolet to photofragment and photoionization spectroscopy, Recent Progress in Vacuum Ultraviolet Ionization and Dissociation of Molecules and Clusters, World Scientific, Singapore, p. 435.

19 R. Hilbig, G. Hilber, A. Lago, B. Wolffand R. Wallenstein, Comments At. Mol. Phys., 18 (1986) 157.

20 R. Hilbig and R. Wallenstein, IEEE J. Quantum Electron, 19(1983) 1759.

21 H. Junginger, H. Puell, H. Scheingraber and J. Vidal, IEEE J. Quantum Electron, 16 (1980) 1132.

22 R. Tonkyn, J. Winniczek and M. White, Chem. Phys. Lett., 164(1989) 137.


Decay energetics of molecular clusters studied by multiphoton mass spectrometry and pulsed field threshold ionization

HJ. Neusser*, H. Krause Institut für Physikalische und Theoretische Chemie, Technische Universität München, Lichtenbergstrasse 4, 85748 Garching, Germany


Abstract

New developments in time-of-flight mass spectrometry combined with multi-photon ionization provide hitherto unachievable information on the decay of weakly bound cluster ions. Dissociation thresholds of cluster ions are import-ant for the understanding of their binding and structure. We present two experimental methods that allow the deter-mination of dissociation energies of weakly bound ionic and neutral clusters. Besides the breakdown technique observing the metastable ion efficiency curves, a novel method is presented and described which is based on delayed pulsed field ionization of long-lived Rydberg states close to the ionization energy. In this way state-selected benzene-noble gas ions are produced and their decay is observed in a reflectron mass spectrometer. Results for dissociation thresholds of various molecular dimers and larger complexes of aromatic molecules are presented. The pronounced differences in dissociation energy between the neutrals and the ions are interpreted in terms of charge-transfer resonance interaction in the ionic complex leading to structural changes after the ionization process.

Key words: Clusters; Multiphoton ionization; Pulsed field ionization; Reflectron; Charge transfer resonance

1. Introduction

Most of the information on the dissociation kinetics and energetics of molecular ions has emerged from mass spectrometric investigations. The basic principle is the observation of the frag-mentation pattern of molecular ions as a function of the energy of the ionizing electrons or photons. In this way, appearance energies for different decay channels of a variety of ions have been determined from breakdown graphs of the fragment intensity [1]. With appropriate kinetic models, dissociation energies can be deduced from the appearance energy. Here statistical models like quasiequili-


SSDI0168-1176(93)03884-0

brium and RRKM have been successfully applied [2-4]. The metastable decay of ions, i.e. their dis-sociation during their flight through the mass spec-trometer, plays an important role as it leads to a clear identification of the parents and the daughters of a particular decay process. Furthermore, the decay time of the ions can be determined from the shape of the metastable mass peaks when the typical flight times in the mass spectrometer are known [5].

For the investigation of decay dynamics it is necessary to observe the decay of ions whose inter-nal energy is well defined. Clear tests of theoretical models are more feasible with energy-selected ions, since, in this way, experiment and theory can be compared with a minimum of averaging. Produc-

212 HJ. Neusser and H. Krause/Int. J. Mass Spectrom. Ion Processes 131 (1994) 211-232

tion of energy-selected ions turns out to be a non-trivial problem: even if photons of well-defined energy are used for ionization the ions are gener-ally produced with a broad distribution of internal energies spanning the range from zero to the maximum energy above the adiabatic ionization energy. The emitted electrons take away a variable amount of energy and leave behind ions with a broad energy distribution. The shape of the internal energy distribution of the ion is not known apriori since it is a complicated function of Franck-Condon factors and autoionization resonances.

A clear situation exists exclusively at the first ionization threshold. Here, no excess energy is left for the emitted electron and the threshold ions are inherently energy-selected. At higher vibrational energy thresholds, in addition to the energy-selected threshold ions, a great number of non-energy-selected ions due to the lower ioni-zation potential are produced. This gives rise to a step-like behavior of the total ion current, each step indicating a new ionization threshold. For experi-ments with energy-selected ions, the separation of threshold ions from the strong background of non-energy-selected ions, is inevitable. One of the techniques that has been successfully applied is the photoion photoelectron coincidence technique [6,7]. Here, ions that are in coincidence with elec-trons of low or zero kinetic energy are monitored exclusively. Naturally, the energy resolution of this technique is limited by the monochromator resolu-tion and the electron energy resolution. The hitherto sharpest energy resolution obtained with the technique was some lOOmeV («800cm"1). A better resolution is expected if high repetition rate lasers in the UV range become available for ionization.

In this report we describe a new technique to produce state-selected ions that is based on the pulsed field ionization of long-lived Rydberg states. Pulsed field ionization allows the observa-tion of highly resolved ion spectra by detecting photoelectrons, as has been demonstrated by Reiser et al. [8], or by monitoring threshold ions

[9]. In this work we discuss its virtue for the production of state-selected ions, particularly for cluster ions [10]. It will be demonstrated that the production of ions in various vibrational states is possible and that the decay of cluster ions with low dissociation energy can be observed.

The selectivity of the ion production is supported by resonance enhanced two-photon ionization. Resonance-enhanced two- and multiphoton ioni-zation was shown to have particular advantages for mass spectrometry [11-14]. Efficient ionization is achieved by stepwise two- or three-photon absorption via a resonant real intermediate state. This resonance enhancement of the multiphoton ionization process leads not only to a drastic increase of the ionization yield but also results in a high selectivity of the ion production. In contrast to a conventional one-photon ionization into the ionization continuum, in resonance-enhanced multi-photon ionization the ionization is controlled by the resonant intermediate state spectrum which often displays sharp structure, even in large poly-atomic molecules and small clusters of these molecules. In addition, it is species selective, this being of particular importance for the study of clusters.

The goal of this work is to demonstrate that mass spectrometric techniques provide new information on the energetics and kinetics of weakly bound clusters with dissociation energies of less than 1 eV. In section 2.2 we give a description of our experimental setup. Then, results on the dissocia-tion energies of dimers of aromatic molecules and of larger homogeneous clusters are presented. These have been deduced from the measured breakdown of the metastable decay process as a function of the two-photon energy and the ionization energies. Several conclusions on the type of interaction and the resulting structures of the complexes are discussed. In section 3 the pulsed field ionization technique used in this work is described and its application for production of state-selected cluster ions presented. The results yield upper limits for the binding energy of dimers of benzene with noble gases.

HJ. Neusser and H. Krause/Int. J. Mass Spectrom. Ion Processes 131 (1994) 211-232 213

2. Breakdown measurements

2.1. Introductory remarks

The measurement of breakdown graphs is a well-established technique in the mass spectrometry of molecular compounds. This technique has produced a variety of data on the energetics of molecular systems (see e.g. ref. 15). In this work the selectivity of the resonance-enhanced two-photon ionization is combined with the observa-tion of the metastable decay of small homo and hetero clusters of polyatomic molecules. It will be shown that a wealth of information on the disso-ciation energy of neutral and ionic dimers and even larger complexes is obtained with this technique.

In a dissociation reaction without reversal acti-vation energy, simple relations between the dissociation energy D0 of the neutral dimer X2, the dissociation energy E0 of the ionic dimer X j , and the ionization energies of the monomer X and the dimer X2 exist:

Z>o(X2)=AE-IE(X) (1)

£ 0 ( X + ) = A E - I E ( X 2 ) (2)

Here IE(X) and IE(X2) are the ionization energies of the monomer and the dimer, respectively: AE is the appearance energy for the dissociation of the dimer ion X j into X+ and X. Using these relations the dissociation energies of the neutral and the ionic dimer can be determined from the measured ionization energies of the monomer and dimer and the appearance energy. Eqs. (1) and (2) can be easily extended to hetero dimers and larger complexes.

The main problem with the determination of the dissociation energy is a kinetic shift of the appear-ance energy from this threshold. Generally, this kinetic shift can be expected to be smaller when a slow decay of the ion is monitored. Particularly, for benzene dimers with a relatively small dissociation energy below 1 eV, we have shown that the restricted phase space of the van der Waals modes [16] leads to negligible kinetic shifts [17].

A general problem in a cluster experiment with

supersonic cold molecular beams is the variety of complexes of different sizes and the monomers that are produced simultaneously. It is difficult to dis-tinguish between monomer ions that have been produced by ionization of a neutral monomer and monomer ions originating from the rapid dis-sociation of the dimer ion. The situation is clear when a metastable decay channel is observed, since monomer ions produced in this way can be well separated from the monomer ions produced by direct ionization of the neutral monomers, e.g. in a reflectron mass spectrometer operated in the par-tial correction mode [18,19] (see below).

Moreover, the selectivity of resonance enhanced two-photon ionization provides a means of selec-tively ionizing, e.g. the dimer, and of suppressing the signal from a direct ionization of the monomer. A necessary precondition for this is that the inter-mediate states of both species are sharp and well separated from each other, this being the case for a variety of molecules.

Another crucial point for the application of the relation in Eqs. (1) and (2) is the ionization energy. Strictly speaking, the adiabatic ionization energy has to be known to find exact dissociation energies rather than lower limits. This is no problem for the monomers since here the ionization thresholds are sharp and can be found with an accuracy of a few reciprocal centimeters. However, it turns out to be a problem for dimers, particularly if geometrical changes from the neutral to the ionic dimer occur (e.g. from a T-shaped to a sandwich structure [20]). Here vertical transitions to higher vibrational states in the intermolecular potential are the strongest ones and the adiabatic transition is weak and its position cannot be identified. As a general rule we may assume that a very sensitive measurement of the ion current leads to ionization energies close to the adiabatic ionization energy.

In order to confirm this we recorded photo-ionization efficiency curves after two-color two-photon ionization using different intermediate states. Principally, one expects that the Franck-Condon factors for transitions to the electronic ground state of the cluster ion should change for


different intermolecular modes of the cluster in the intermediate state. Additionally, it should be men-tioned that autoionization resonances, as observed for several van der Waals dimers, [21-23] could strongly enhance the transition probability to the ionic ground state. However, in some clusters with small geometrical changes during the ionization process, sharp onsets of the ionization are observed [24,25]. Here, the technique of pulsed field ionization which will be described below can be applied to find very accurate values of the adia-batic ionization energy.

2.2. Experimental

The details of the experimental setup were described in our previous work [17]. A scheme of the setup is shown in Fig. 1. Benzene homo clusters or heterogeneous van der Waals clusters of benzene (B), /?ara-difluorobenzene (F), toluene (T) and cyclohexane (C) are produced in a supersonic jet expansion of a gas mixture, consisting of 20 mbar of each monomer component seeded in noble gas

SIGNAL i

(He) at a pressure of 1-5 bar. The gas mixture is expanded through a pulsed 200 /xm diameter noz-zle orifice and the central part of the beam is selected by a skimmer before it enters the mass spectrometer in a collinear configuration.

The clusters are ionized by resonance enhanced two-color two-photon ionization in the acceleration region of a linear reflectron time-of-flight (RETOF) mass spectrometer [17]. In the linear RETOF the metastable ion intensity for the different decay channels of a hetero cluster ion can be measured with high precision for different decay channels and clusters, since the ion trajectories are independent of the kinetic energy of the observed ions.

In order to determine the ionization energy of the clusters, photoionization efficiency curves were recorded by observation of the ion signal of the respective cluster for a fixed photon energy of the first laser, (tuned to the resonant intermediate state) but a varying photon energy of the second laser. To find the appearance energy of a dissocia-tion channel we recorded the intensity of daughter ions that are produced by slow, metastable disso-

I0N DETECTOR

SKIMMER

GAS INLET DRIFT REGION ι ,

! (REFLECTOR • — — — I i_

— — zi zz zz ~ ~ ~ I L LZ

ION BEAM i |

U~^ 'ref

PUMP 3

Fig. 1. Schematic drawing of the linear reflectron mass spectrometer that has been used for the experiments described in this work. The molecular beam produced by a pulsed nozzle is skimmed and collinear to the ion flight paths. The ions are produced by multiphoton ionization by either one laser beam or two laser beams of different color. After acceleration they pass a small hole in the channel plates

and are reflected by 180° in the reflecting field towards the channel plates.


dation of the parent cluster ion in a field-free drift region of the RETOF instrument. The onset of the metastable signal yields the appearance energy of the observed dissociation channel.

For the investigation of the appearance energy of a selected decay channel of a cluster ion the linear RETOF is operated in the partial correcting mode. In this mode, daughter ions produced by a cluster decay in the field-free drift region of the RETOF can be observed as separated peaks (drift peaks). This is shown in Fig. 2 for the dissociation of the B2" dimer ion into a B+ ion and a neutral B mol-ecule. In this experiment a delay of 20 ns between the first and the second laser pulse was chosen. Therefore, the ion signal originating from two-color two-photon ionization is well separated

from the signal produced by one-color two-photon ionization through the first laser alone. Both drift peaks due to one- and two-color ioniza-tion are marked by arrows in the upper mass spec-trum showing the drift peaks on an enlarged scale. Similar mass spectra were recorded for all investi-gated homo and hetero dimers. The expansion con-ditions were chosen such that the intensity of larger clusters, e.g. the trimer ion, is 1000 times smaller compared to the dimer intensity. Thus, dissociation from larger clusters into the dimer ion can be neglected. The laser power is chosen so that photo-ionization of the dimer is maximized and at the same time the ion yield for smaller masses result-ing from fragmentation after three-photon absorp-tion is minimized.

B+

B= : C6H6

B — Bj (2ν,)

B - B* (v1+i/2

J Vwvw__

I ' ' > ' I 58 59

JL

B;

I ~i—i—1—1—1—i—1—i—1—j—1—1—1—1—1—1—1—1—1—1—1—1—1—1~

50 55 60 65 70 75 TIME-OF-FLIGHT [μϊ] -

Fig. 2. Part of the time-of-flight mass spectrum of the benzene dimer (B2) when operating the reflectron in the partial correction mode. The small peaks at 58.5//s (see inset) are due to a metastable decay of the benzene dimer cation (Bj) in the drift region of the reflectron mass spectrometer depicted in Fig. 1. Ionization was achieved in a two-laser, two-color experiment with the second laser pulse delayed by 100 ns. The first mass peak originates from an excitation of two photons with hi/j, while the second stronger peak is a two color signal

due to the absorption of one photon from each laser beam {hvx + hv2).


| l l l l | l l l l | fl I I I I | I I 1 1 | 1 I I 1 | I I I I | I ff 1 I I | I I I I | I I H | I I I I | 1 I

8.9 9.0 9.0 9.1 9.3 9.4

TWO-PHOTON ENERGY [eV]

Fig. 3. Breakdown of the intensity of the metastable ion peaks for three dimers, benzene-toluene (BT), toluene-toluene (T2) and benzene-benzene (B2). The relative intensities are plotted as a function of the two-photon excitation energy. The appear-ance energy of the metastable decay is taken as the intercept of the breakdown graph with the constant noise level baseline produced by three-photon absorption.

2.3. Dimers of aromatic molecules

The technique described above has been applied to a variety of homo and hetero dimers. A typical result for the breakdown of the metastable inten-sity is shown in Fig. 3 for the benzene (B2) and toluene (T2) homo dimer and the benzene-toluene hetero dimer (BT) [17]. The appearance energy is determined with an accuracy of about lOmeV. Figure 4 displays the onset of the ion current of B2 near the threshold. The arrows indicate the smallest photon energy leading to an ion current above the noise level. The lower trace shows the ion current after excitation of the 6 l, S ] vibronic state with the first photon. In order to vary the Franck-Condon factors for the transition from the inter-mediate to the ionic ground state in the upper spectrum, an intermediate state including the exci-tation of an intermolecular vibration [26] was used. The identity of this state is not clear. For this

c

b o

o

l

hv2

1

hvj

61 w1

■YAA/WW*V

I

hv2

i

hvj

^ψφ** '«**»¥» ιΛ&γ+γΚνΛ/Ιβ,ιΛΕ*

8.45 8.5 8.55 8.6 8.65 8.7 8.75 8.8 8.85

Two-Photon Energy [eV]

Fig. 4. Ionization efficiency curves as a function of the two-photon energy hv\ + hv2 for two different intermediate states of the resonance enhanced two-photon ionization process: bottom, the u6 intramolecular mode is excited in the S\ inter-mediate state {hv\ = 38564 cm- 1); top, in addition to the v6

mode, an unassigned intermolecular (van der Waals mode) is excited {hv\ = 38586cm"1). In both cases the same ionization energy is obtained. For interpretation see text.

reason we utilized other van der Waals intermedi-ate states as well, but obtained basically the same result as shown in Fig. 4. Here, both intermediate states yield the same ionization threshold within the error limits of the experiment. Because of this result and the high sensitivity of our experiment we believe that this value is close to the adiabatic IE.

All results for the ionization energies and appearance energies are listed in Table 1 together with the deduced dissociation energies D0 and E0 of the neutral and ionic dimers, respectively. We would like to mention that, for the determination of the neutral dimer dissociation energy D0, the very accurate ionization energies of the monomer are used (see Eq. (1)). Thus, an error due to an uncertainty of the ionization energy can be excluded in this case.

It is interesting to discuss the results in a sys-tematic way in terms of the different properties, e.g. polarizability, dipole moment etc., of the

Tab

le 1

D

isso

ciat

ion

ener

gies

of

dim

ers.

Mea

sure

d io

niza

tion

ene

rgie

s (I

Es)

, app

eara

nce

ener

gies

(A

Es)

of

the

met

asta

ble

diss

ocia

tion

cha

nnel

s di

scus

sed

in th

e te

xt, a

nd d

isso

ciat

ion

ener

gies

D0

and

E0

of t

he r

espe

ctiv

e ne

utra

l an

d ch

arge

d cl

uste

rs (

for

defi

niti

on o

f D

0 an

d E

0 se

e te

xt);

the

err

ors

repr

esen

t th

e re

prod

ucib

ility

of

the

mea

sure

d va

lues

Neu

tral

D

imer

a

Ben

zene

-be

nzen

e B

enze

ne-

cycl

ohex

ane

Tol

uene

-to

luen

e B

enze

ne-

tolu

ene

Par

a-O

FB

-pa

ra-O

FB

B

enze

ne-

para

-OF

B

IE(e

V)

8.65

±

0.01

9.12

±

0.02

8.34

±

0.01

8.42

±

0.01

8.87

± 2

0

8.75

±

0.02

Do(

meV

)

70 ±

10

80 ±

20

150

±1

0

130

±1

0

90 ±

20

80 ±

20

Mai

n in

term

olec

ular

in

tera

ctio

ns

Dis

pers

ion

Dis

pers

ion

Dis

pers

ion

dipo

le-d

ipol

e D

ispe

rsio

n di

pole

-ind

. di

pole

D

ispe

rsio

n

Dis

pers

ion

Dim

er

cati

on

(Ben

zene

-be

nzen

e)+

(Ben

zene

-cy

cloh

exan

e)+

(Tol

uene

-to

luen

e)+

(Ben

zene

-to

luen

e)+

(par

a-O

FB

-pa

ra-O

FB

Y

(Ben

zene

-pa

ra-O

FB

Y

AE

(eV

)

9.31

±0

.01

9.32

± 0

.02

8.97

±0

.01

8.95

±0

.01

9.25

± 0

.02

9.24

± 0

.02

£ 0(m

eV)

660

± 20

200

± 40

630

± 20

530

± 20

380

± 40

490

± 40

Mai

n in

term

olec

ular

in

tera

ctio

ns0

CT

R

Ele

ctro

stat

ic

CT

R

CT

R

CT

R/e

lect

rost

atic

CT

R

a Par

a-O

FB

: /7

ara-

Dif

luor

oben

zene

. b C

TR

: ch

arge

tra

nsfe

r re

sona

nce

inte

ract

ion.


molecules. Since benzene (a =10.6 À [27] and /?0ra-difluorobenzene (a =10.3 À [27]) have no dipole moment and nearly the same polarizabil-ity, the dissociation energies D0 of B2, F2 and BF are expected to be nearly the same. This is in full agreement with our experimental data. Toluene has a small dipole moment (μ = 0.38ϋ [28]) and a slightly larger polarizability (a =12.6 À [27]) compared with those of benzene and /7-difluoro-benzene. This agrees with the experimental finding that D0 is larger for the BT and T2 dimers.

For ionized van der Waals clusters, electrostatic and charge transfer resonance interactions are strongly enhanced, leading to an increased disso-ciation energy compared to that of the neutral dimers. For instance, for the benzene dimer, the binding energy increases by nearly one order of magnitude from 70meV to 60meV when the dimer is ionized. Thus, ionization goes along with a strong structural rearrangement of the complex. While the neutral benzene dimer is supposed to have a T-shaped structure, the dimer cation, stabi-lized by charge transfer resonance interaction, should have a sandwich structure with a parallel arrangement of the two benzene planes. Charge transfer resonance interaction is known to stabi-lize charged dimers of aromatic molecules. For heteroclusters the strength of this interaction decreases as the difference in the ionization energy of the monomers increases. A large difference in the ionization energy leads to a localization of the charge at the component with the lower ionization energy, and thus to a small charge transfer resonance interaction. Therefore, bonding of the homodimer ions is expected to be stronger than that of the heterodimer ions. From Table 1 it can be seen that indeed the dissociation energies £Ό(Β ~) and £o(T^) are nearly the same and significantly higher than £Ό(ΒΤ+). In the case of F ~ the low-ering of the dissociation energy is attributed to a partial charge localization at the electronegative fluorine atom, which most likely leads to a decreased charge transfer resonance interaction.

It is interesting to study the contribution of the charge transfer resonance interaction to the bond-

ing in van der Waals dimer ions in more detail. For this reason we investigated the extreme situation of an heterodimer consisting of an (aromatic) benzene and the nonaromatic cyclohexane molecule. After two-photon ionization benzene-cyclohexane dimer ions (BC+) show a slow metastable dissociation in the drift region of the RETOF leading to benzene daughter ions (B+) and neutral cyclohexane mol-ecules (C). The first laser frequency was fixed to the 6o transition in benzene and the energy of the second photon was varied. The appearance energy of the above mentioned decay was found to be 9.32 eV and the ionization energy 9.12 eV. Using these values, the binding energy of the neutral dimer and the dimer ion are calculated from Eqs. (1) and (2), respectively, to Z)0(BC) = 80meV and £o(BC) =200meV (see Table 1). The bind-ing energy in the neutral heterodimer Z)0(BC) = 80meV is equal or nearly the same as in B2, BF and F2 dimers. Since all of these molecules (B, F and C) have no dipole moment and their polarizabilities do not differ very much (cyclohexane: a = 10.9 À [27]), this result is in line with simple theoretical arguments. The binding energy in the heterodimer ion (BC+) is E0(BC+) = 200 meV. If we suppose that no charge transfer resonance interaction contributes to the bonding in BC+ (because of a missing delocalized π-electron system in cyclohexane) the bonding in BC+ is purely electrostatic. By comparing the data of the benzene dimer Bj with charge transfer resonance interaction (E0 = 660 meV) and BC+

without charge transfer resonance interaction (E0 = 200 meV), we conclude that the contribu-tion of charge transfer resonance to the binding energies is as much as 400 meV in sandwich-like dimers of aromatic components.

2.4. Larger benzene complexes ( 2 ^ n ^ 5 )

Measurements of the breakdown of the meta-stable decay efficiency have also been performed for larger benzene clusters up to the pentamer [29]. For these clusters the intermediate state spec-tra are sufficiently separated so that a selective

HJ. Neusser and H. Krausejlnt. J. Mass Spectrom. Ion Processes 131 (1994) 211-232 219

excitation and ionization of the cluster species under consideration is possible.

All results for the ionization and appearance energies and the deduced neutral and ionic disso-ciation energies are listed in Table 2. In addition, the total binding energies are listed. They are obtained from the sum of the dissociation energies for stepwise evaporation of a monomer. A striking result is the large dissociation energy of the neutral trimer of 200 meV as compared to that of the dimer. This suggests a particularly stable structure of the trimer. Assuming that only pair interactions contribute to the binding of the trimer, the maxi-mum binding energy of the trimer should be three times the dimer binding energy, and the measured value of Z>bind(3) = 270meV appears to be too high compared to the value for the dimer. It is hardly possible to decide whether an additional three-body interaction contributes to the binding in the trimer or whether the deviation is caused by the experimental error. A fourth benzene molecule is bound to the trimer with a somewhat smaller energy of lOOmeV.

Several theoretical methods have been applied to calculate the binding energy of the benzene dimer. Van de Waal [30] assumed a nearly parallel

arrangement of the two benzene molecules and calculated a binding energy of 114meV employing potential-energy minimization with empirical atom-atom potential functions. Similar results were obtained by de Meijere and Huisken [31]. This structure, however, is at variance with the T-structure in the benzene crystal. In ab initio calcu-lations a T-structure was found as the most stable structure of the benzene dimer. In an early work of Cârsky et al. [32] a binding energy of 65 meV was calculated and in a more recent study by Hobza et al. a higher binding energy of 117meV was obtained [33]. When comparing the theoretical results with the experimental values presented here, it must be recognized that the calculated bind-ing energies yield the energy of the potential mini-mum De, whereas the experimental ones represent the lowest vibrational energy level D0 (see e.g. Fig. 4 in our previous work [17]). The results differ by the zero-point energy of the multidimensional van der Waals potential.

For the larger clusters, ab initio calculations are not available. Binding energies have been calcu-lated from empirical atom-atom potentials employ-ing a potential energy minimization program [30,31] assuming a triangular structure of the

Table 2 Dissociation energies of small neutral and ionic benzene clusters. Measured values for the ionization energies (IEs) and the appearance energies (AEs) of the monomer evaporation of benzene clusters (C6H6)„(« = 1 - 5 ) . From these values the dissociation energies of the neutral clusters (Z)0) and the cluster ions (E0) are deduced; the errors represent the reproducibility of the measured values and do not include a systematic error in the values of IE and AE; calculated values Z)e from the literature, which represent the potential minimum, are also listed

Parameter

AE(eV) IE (eV) D0(meV) E0 (meV) A>ind (meV)

£bind (meV) De (meV)

(Benzene),,

1

_ 9.243

-----

2

9.31 ± 0.01 8.65 ± 0.01

70 ± 1 0 660 ± 20

70 660 117a

114bc

3

8.85 ± 0.02 8.58 ± 0.02

200 ± 30 270 ± 40 270 930 219b

192c

4

8.68 ± 0.02 8.55 ± 0.02

100 ± 4 0 130 ± 4 0 370

1060 243b

147c

5

<8.61 8.50 ± 0.02 < 60 ^ 110

--

243b

For comparison of D0 with De see text. a Reference 33. b Reference 30. c Reference 31.

220 HJ. Neusser and H. Krause)Int. J. Mass Spectrom. Ion Processes 131 (1994) 211-232

trimer. Results for the trimer are in surprisingly good agreement with our experimental value. This is a strong corroboration of the triangular structure assumed in both theoretical approaches [30,31]· New spectroscopic measurements of isoto-pically substituted benzene trimers also show evi-dence for a triangular structure [34]. The agreement is less conclusive for the tetramer. Here, the value of van de Waal (243 meV) [30] for a tetrahedral structure is more than twice the experimental value. The experimental value is close to the disso-ciation energy calculated by de Meijere and Huis-ken [31] for a different structure. Here it is assumed that the fourth benzene molecule is added to the strongly bound trimer in such a way that the angle between the plane of the attached monomer and one neighboring molecule of the trimer ring is close to the dimer angle. Our experimental results seem to favor this structure for energetic reasons. For the pentamer, only one theoretical value exists which has been found for the ditetrahedral struc-ture. The much smaller experimental value does not support this structure.

Contrary to the neutral binding energies, the ionized cluster dissociation energies show a mono-tonic decrease with increasing cluster size. As already mentioned above, the binding energy of the dimer cation is nearly one order of magnitude larger than that of the neutral dimer. This was attributed to a strong charge transfer resonance interaction in the dimer ion, typical for a sandwich structure of the benzene dimer ion [17,35-38]. The binding energy of the third benzene molecule in the trimer ion is less than half the binding energy of the dimer cation. Badger and Brocklehurst [36] pre-dicted that charge transfer interaction should be responsible for the binding in the trimer ion, if one assumes a triple sandwich structure. A simple Hückel molecular orbital calculation, assuming a delocalized charge in the trimer and the tetramer ion, yields relative binding energies that are in excellent agreement with the experimentally observed increase of the total binding energy [29]. From this we conclude that charge transfer reso-nance interaction is the dominating contribution to

the binding energy in benzene cluster ions up to the tetramer. The main assumption in the Hückel calculation is a delocalization of the charge that points to a non-cyclic sandwich structure of the cluster ions. It is interesting to compare the sand-wich structure of the trimer ion with the proposed cyclic structure of the neutral trimer. Whereas for the charged trimer the strongest stabilization is achieved by a charge transfer resonance (£bind(3) = 930 meV) in a non-cyclic structure, for the neutral trimer the largest dispersive binding energy (A)ind(3) = 270 meV) is achieved in a cyclic configuration.

Experimental values for the binding energy have been obtained from high-pressure mass spectrome-try (HPMS) experiments [35,39]. The most recent value of Meot-Ner et al. is E0(2) = 740 ± 65meV [35]. This is in reasonable agreement with our result. New results of Hiraoka et al. [39] point to a somewhat higher binding energy for the dimer of EQ(2) = 894 ± 43 meV and a binding energy of the trimer ion OÎEQ(3) — 338 ± 22meV. The latter is in reasonable agreement with our result of £0(3) = 270±40meV.

3. Production and decay of state-selected cluster ions

3.1. Introductory remarks

A common characteristic of the clusters studied in the preceding section is that they undergo strong structural changes after the ionization process. For example, in the benzene dimer the T-shape struc-ture of the neutral complex changes to a sandwich structure in the ion. As a consequence, the vertical ionization process leads to the excitation of many intermolecular modes in the ion and the adiabatic ionization transition is weak. Some of the six van der Waals modes have low frequencies and produce a smoothly increasing ionization efficiency curve with densely packed ion states rather than a step-like behavior of the ion current with each new channel separated from the preceding one. For this reason there is little chance to selectively excite defined vibrational states in the dimer ion.


In this section we focus our attention on aro-matic molecule-noble gas complexes with three van der Waals modes and no strong structural change during the ionization process. We will demonstrate that, in these cases, the newly developed technique of pulsed field threshold

ionization leads to the production of state-selected complex ions.

3.2. Pulsed field threshold ionization

Delayed pulsed field ionization is a powerful

hv| hv2

Vjet

^jet

2V

0 V 0 V

Θ

►1000 V

©

2 V 0 V

=□"· 0 V

acceleration drift reflector Fig. 5. Principle of mass selected pulsed field ionization in a reflectron time-of-flight mass spectrometer, (a) t — 0: molecules are excited by resonance enhanced two-color two-photon excitation. A weak deceleration field leads to the separation of instantaneously ionized molecules (+) and the neutral Rydberg molecules (R). (b) t « 5 /is: a strong electric field is applied. It ionizes the still existing Rydberg molecules and accelerates all ions. The directly ionized molecules are now separated from the newly produced ions and accelerated by a higher potential difference, (c) t « 25 /is: due to their larger kinetic energy they penetrate through the reflector, whereas the ions

produced by pulsed field ionization are reflected and monitored separately.


technique when combined with narrow bandwidth laser excitation. It allows the separation of mol-ecules in a narrow energy range close to the ioniza-tion threshold from simultaneously excited molecules in lower-lying Rydberg states and from non-energy-selected ions produced by direct ioni-zation above a lower ionization potential. This is probably due to the strongly increasing lifetime of the Rydberg states approaching the ionization energy. The lifetime enlarges due to an increase of the radiative and the non-radiative lifetime, both increasing with n3, where n is the principal quantum number. For states with n > 100, this lifetime exceeds several microseconds in many molecules. Thus, delay times in the microsecond range can be used to field ionize the molecules from their Rydberg states. During this long delay time, the simultaneously-produced, undesired, non-energy-selected ions can be spatially sepa-rated from the still-neutral molecules in Rydberg states simply by applying a weak field insufficient to field ionize all of the latter but sufficient to decele-rate the ions.

First, pulsed field ionization has been used to measure electrons with nearly zero kinetic energy [8]. The technique developed by Reiser et al. [8] and Müller-Dethlefs and Schlag [40] displays a consid-erably increased resolution in photoelectron spec-troscopy of the electronic ground state ions. Recently, Zhu and Johnson presented a method to combine pulsed field ionization with ion detec-tion [9]. In their technique they separated the threshold ions from ions below this threshold that are responsible for the background in photoioniza-tion efficiency spectra by a special technique. This includes three different acceleration regions and a time-of-flight analysis in the mass spectrometer. A similar method was presented by Jouvet et al. [41]. In our recent work we introduced another versatile, easily applicable method for separation of thresh-old and non-threshold ions. It is based on the energy selection in the reflecting field of a linear reflectron mass spectrometer and provides high mass resolution which is particularly important for cluster investigations [18,19].

The principle of our method to separate the ions produced by pulsed field ionization (PFI) from directly ionized molecules is shown in Fig. 5. The molecules enter the mass spectrometer with the velocity ν·]ξΧ of the molecular beam, which depends on the noble gas used (He or Ar). At this time

a

u G O

hv, + hv7 = 74900 cm'1 C6H6 a

^v. -5n6

J L Û _

hv. + hv9 = 74700 cm"1

JL

44 44.5

Time-of-Flight [με]

45

Fig. 6. Time-of-flight mass spectrum of benzene obtained for two different two-photon ionization energies and fixed hvx

tuned to the 61 intermediate state, (a) The two-photon energy hv\ + hv2 = 74900 cm"1 is in resonance with the ionization threshold of the 61 (3/2) ion state (see right arrow in Fig. 9). Two different ion peaks are observed: threshold ions produced by delayed pulsed field ionization according to the method de-scribed in Fig. 5 (shaded peaks) and promptly produced ions due to the lower lying adiabatic ionization energy. The same peak pattern is observed for the "light" benzene i2C6H6 and the "heavy" l3C12C5H6 isotope. The time separation of threshold ions (shaded) and promptly produced ions is due to the mechan-ism explained in Fig. 5. Note the high mass resolution which is sufficient to separate all different peaks, (b) Two-photon energy hv\ + hv2 = 74700 cm- 1 is not in resonance with an ionization threshold (see left arrow in Fig. 9). No threshold ions appear in the time-of-flight mass spectrum.


(t = 0) a weak electric field of Ex « -0 .2 to -0.6 V cm- 1 is present (Fig. 5(a)). Its direction is chosen such that the ions instantaneously produced by the laser pulse are decelerated. In the present experiment the pulsed acceleration voltage is delayed by several microseconds after the exciting laser pulse. This is the main difference between the

a

u a o

hv, + hv, = 74900 cm"1

Uref=847V

C6H6

13CC5H6

Uref=840V

_A_ yL_JL

Uref = 820V

± 44 44.5 45

Time-of-Flight [ is]

Fig. 7. Time-of-flight mass spectra of benzene obtained for a two-photon energy hv\ +hv2 = 74900 cm-1 (right arrow in Fig. 9) in resonance with the ionization threshold of the 6'(3/2) ion state. The reflecting voltage of the reflectron mass spectrometer is decreased from (a) to (c). (a) For UTe{ = 847 V, threshold ions (shaded) and directly produced ions are reflected and monitored. Same situation as in Fig. 6(a). (b) For i/ref = 840 V, directly produced ions are reflected with smaller efficiency due to their higher kinetic energy, (c) For i/ref = 820 V, the directly produced ions are no longer reflected due to their higher kinetic energy. Only threshold ions with low kinetic energy are reflected and exclusively monitored (see Fig. 5(c). Note that under these conditions the resulting mass spec-trum consists of threshold ions (shaded) exclusively without any background from directly produced ions.

conventional mass spectrometric operation of the reflectron and the technique described here.

After a delay of 3-100 μβ the highly excited neu-tral molecules in long-lived Rydberg states and the ions originally produced at the same place are spa-tially separated — by about 1 mm in the case of a 5 s delay (Fig. 5(b)). At this time a pulsed positive voltage of 1000 V is applied to the repeller plate. This leads to an electric field of +333 Vcm- 1 , caus-ing the field ionization of the still existing long lived Rydberg states and the rapid acceleration of all ions. Due to the spatial separation the instanta-neously produced ions are accelerated by a higher potential difference than the ions produced by delayed PFI. For a typical spatial separation of 1 mm, this energy difference is about 33 eV. It is sufficient to separate both species within the reflec-tor of the mass spectrometer. The reflecting voltage £/ref is chosen so that only PFI ions are reflected, while the instantaneously produced ions penetrate through the reflector (Fig. 5(c)). In this way the PFI ions are detected after reflection and after they have passed the second drift region.

The procedure of separating the directly pro-duced from field ionized ions is demonstrated in Figs. 6 and 7 for the benzene cation. In these spec-tra a delay time between the laser pulse and the pulsed field ionization of 22 μβ was used. During this time a permanent field of 0.5 Vcm - 1 was pre-sent to decelerate the directly produced ions. After 22 ßs an electric field of about 500 Vcm - 1 was turned on in order to ionize the Rydberg states and to accelerate all ions into the mass spectro-meter. A reflector voltage Ure{ = 847 V is high enough to guarantee the detection of all, i.e. directly ionized and field ionized, ions. The reflect-ing conditions were chosen in such a way that both species are well separated in time. Figure 6(a) shows a part of the mass spectrum in the region between 77 and 79 u, corresponding in flight times between 44 and 45 ßs. In Fig. 6(a) the wavelength of the second laser was chosen so that Rydberg states converging to the ionization energy of the 6!(3/2) state are excited, as marked by the right arrow in Fig. 9. In this case, additional ion peaks


(shaded) with a 100-ns-longer time of flight than the directly produced ions are observed in the mass spectrum and can be clearly distinguished in the mass spectrum due to the high mass resolution achieved in the reflectron mass spectrometer. Both types of ions are observed for the same excitation energy. Direct ionization corresponds to the adia-batic IE(0°) whereas threshold ions correspond to the 61 (3/2) state.

The small mass peak at longer flight times results from the 6% 13C12C5H6 which is present in the natural isotopic benzene sample. For the photon energy hvx = 38 609 cm- 1 resonance enhancement of both isotopic species, 12C6H6 and 13Cl2C5H6

occurs (see Fig. 10). The adiabatic ionization ener-gies differ only slightly so that threshold ions of both isotopes are observed for the same two-photon energy. The smaller peak at shorter flight times results from the fragmentation of the benzene cation at the four-photon level at the high laser intensity in this experiment. Two additional photons with energy hv2 are absorbed and lead to the ejection of an H atom from the benzene cation since the total internal energy of the benzene cation after four-photon absorption is more than 8.9 eV. Thus, a rapid dissociation takes place in the ionization region [42,43] and the fragment ion peak appears at the position of the daughter ion.

In the mass spectrum of Fig. 6(b) the total two-photon excitation energy was decreased by 200 cm- 1 (see left arrow in Fig. 9). It is no longer in resonance with long lived Rydberg states close to an ionization potential. For this reason only the directly produced ions corresponding to the 0° adiabatic ionization energy appear in the mass spectrum, whereas the threshold ion peak has dis-appeared from the spectrum. For this excitation energy no threshold ions exist since the two-photon energy is not resonant with an ionic state.

In Fig. 7 we demonstrate the energy analysis in the reflecting field for suppression of the ions pro-duced by direct ionization. These have a somewhat higher kinetic energy, as demonstrated in Fig. 5. Figure 7 shows a series of mass spectra that have been obtained for different reflection potentials.

The exciting two-photon energy is 74900 cm" and marked by the right arrow in Fig. 9. For this two-photon energy both directly produced and threshold ions (shaded peaks) are produced. In Fig. 7(a) the reflectron voltage is largest, C/ref = 847 V, and both types of ions are reflected and detected by the channel plates. In Fig. 7(b) £/ref is decreased to 840 V leading to a smaller signal of the directly produced ions, since only a small part of them is reflected under these conditions. Finally in Fig. 7(c), at i/ref = 820 V, the directly produced ions are no longer reflected and do not appear in the mass spectrum, i.e. the observed ions are exclu-sively produced by pulsed field ionization of long lived Rydberg states. (It should be noted that in the spectra of Fig. 7 the mass resolution is not as high as usual, since the reflectron is operated with incomplete correction so that both ion species are separated in time of flight and can be clearly dis-tinguished from each other.) As a result we obtain ion signals originating exclusively from threshold ions with no background from non-energy-selected directly produced ions.

Figure 8 demonstrates the high mass resolution of our experimental setup. Now the reflectron is operated in the complete correction mode. The spectrum displays two sharp PFI peaks which

C6H6 (78 u)

,3CC5H6 (79 u)

1 k 1 1 i i i ■ 1 ■ ■ ■ ■ 1 I i i i i I ■ ■ ■ ■ I

41 42 43

Time-of-Flight [μ8]

Fig. 8. Mass spectrum of the threshold ions (shaded) of natural isotopic benzene. The voltages of the reflecting field were opti-mized to achieve a high mass resolution. Note the large spacing between the two benzene isotopic peaks with one mass unit difference.

HJ. Neusser and H. Krausejlnt. J. Mass Spectrom. Ion Processes 131 (1994) 211-232 225

belong to the benzene cation Q H ^ ^ u ) and the isotope 13CC5Hj(79u) resulting from an excita-tion to the ionic ground state 0°. A mass resolution of about m /Am = 1000 is deduced from the peak widths and the distance between both peaks. In another reflectron apparatus with a longer drift region we were able to realize a mass resolution up to m/ Am = 3000 for the threshold ions. The vibrational spectra resulting from a scan of the second laser frequency and the monitoring of the threshold ion current will be discussed in section 3.3.

An important difference to the technique of Zhu and Johnson [9] is the high field of some 300 V cm-1

used for delayed field ionization in our experiment. Assuming the well-known field dependence of the ionization threshold [44,45] this field is sufficient to ionize Rydberg states down to 100 cm"1 below threshold. If all Rydberg states within the energy range of 100cm"1 below threshold survived the delay of several microseconds, the spectral resolu-tion and thus the energy selection would be limited to 100 cm"1. As will be shown below our achieved energy resolution is a few reciprocal centimeters and thus much better than expected for the high ionizing field. This means that Rydberg states within a narrower energy range are ionized and the lifetime of the lower Rydberg states accessible to field ionization is too short to survive the long delay time of several microseconds. From this result we expect a further improvement of spectral resolution when longer delay times up to 100 /is are used. It should be mentioned that the high ioniza-tion field is not principally necessary in our tech-nique, but facilitates the separation of threshold ions from directly produced ions. Up to now, for all investigated molecules, a resolution of several reciprocal centimeters was found and we have not been able to find an example where the high field of 300 V cm"1 led to a lower resolution. Alternatively, a pulsed ionization field comparable to that used in zero kinetic energy electron (ZEKE) experiments (some 1 Vcm"1) [40] can be applied. An additional delayed high field is then used for the acceleration of the so-produced ions.

3.3. Threshold ion spectra

Figure 9 shows a spectrum of the ionic ground state of benzene (X, 2Elg) from zero to 1000 cm"1

excess energy. The total ion current spectrum (middle trace) is compared to a conventional time-of-flight photoelectron (TOF-PE) spectrum (upper trace) and the PFI spectrum (bottom trace). The total ion efficiency curve displays a step-like increase of the ion current whenever a new ionization threshold is reached. The two arrows mark the excitation energy leading to the mass spectra shown in Figs. 6-8. The TOF-PE spec-trum has been measured in our laboratory but also previously with a somewhat better resolution by Long et al. [46]. All three experiments have been performed using a two-photon ionization process

74400 74600 74800 75000 75200

Two-Photon Energy [cm-1]

75400

Fig. 9. Three spectra of the benzene cation C6H£ obtained with three different techniques. Top: time-of-flight photoelectron spectrum after ionization of benzene with a two-photon energy of 77210 cm- 1. The spectrum reflects the ground state and the lowest vibrational states of the benzene cation. Middle: total ion current as a function of two-photon energy. For each new ioni-zation threshold a step is observed due to the additional ions produced. Bottom: threshold ion (PFI) signal as a function of two-photon energy. Peaks are observed when a new ionization threshold is reached. For two-photon energies not in resonance with an ionization threshold the threshold ion signal is zero (see Fig. 6). Note the increased resolution of this spectrum leading to additional features not resolved in the upper two spectra.


with the 61 state acting as an intermediate state. The TOF-PE spectrum was recorded in a one-color two-photon ionization experiment leading to an excess energy of 2655 cm- 1 in the ion. In the other experiments with ion detection, the wave-length of the second photon was scanned from the adiabatic ionization potential up to 1000 cm- 1

excess energy. Obviously, the resolution in the PFI spectrum is much better than in the TOF-PE spectrum and new vibrational features are resolved. Due to the peaked structure and flat base-line of the PFI spectrum it obviously contains more spectral information than the total ion efficiency curve. Several peaks show a previously unresolved splitting and additional peaks with low intensity are observed. The relative integrated peak intensi-ties in the PFI spectrum are similar to that in the TOF-PE spectrum. This means that in the PFI spectrum, as well as in the TOF-PE spectrum, the intensities are mainly determined by the Franck-Condon factors of the respective transition.

The peak width of 9 cm- 1 in the PFI spectrum (bottom) is smaller by a factor of 10 than that in

the TOF-PE spectrum (top). This still does not represent the experimental resolution but is caused by the rotational structure of the vibronic transi-tion; the relatively broad linewidth (about 0.8 cm- 1) of the laser providing the first photon leads to the excitation of many rotational states in the 61 vibronic intermediate state and conse-quently to many rotational states in the ion. The detailed assignment of all vibrational bands has been discussed in our previous work [10]. Briefly, the splitting of the 6!(3/2) band is due to a pre-viously unresolved quadratic Jahn-Teller splitting.

The mass selectivity of pulsed field threshold ionization with ion detection is demonstrated in Fig. 10. Here the threshold ion current (PFI) spec-trum at different masses (78 and 79 u) is shown. The frequency of the first laser is tuned to the maximum of the 60 band of 13C12C5H6 present with an abun-dance of 6% in the natural isotopic mixture. Since the 6o intermediate state spectrum of 13Cl2C5H6

overlaps with the band of 12C6H6, a separation of both spectra solely by resonance enhancement in the intermediate state is not possible. Here, the

74600 74800 75000 75200 75400 Two-Photon Energy [cm"1]

75600

Fig. 10. Vibrational spectra of the ionic ground state of the two benzene isotopes, I2C6H6 (upper trace) and l3C12C5H6 (lower trace) in natural abundance. The frequency vx of the first laser is the same in both spectra. It is in resonance with the maximum of the 6Q band of I3C12C5H6 and favors the ionization of this isotope. For the chosen frequency vx, high rotational /states of 12C6H6 are excited leading to broad peaks in the upper spectrum. Both isotopic spectra can be measured without interference due to the high mass resolution achieved

for the threshold ions in the reflectron mass spectrometer (see Fig. 8).


mass selectivity of ion detection is needed. As shown in Fig. 8, the mass resolution of the reflec-tron mass spectrometer for the threshold ions is sufficient to separate completely these neighboring isotopic mass peaks in the mass spectrum. In Fig. 10 the frequency of the second laser is scanned across the ionization thresholds. The lower trace represents the threshold ion current at mass 79 u (13C12C5H6) whereas the upper spectrum was obtained for 78 u (12C6H6). The spectrum of l3C12C5Hj resembles the spectrum of ' l ight" ben-zene (78 u) shown in Fig. 9. However, there are two small differences: (i) The 0° peak of 13C12C5Hj

.-1

12 4 cm * higher in appears at 74 559cm~ , i.e

energy than the 0° transition in i 2 C 6 Hj . (ii) There is a small peak on the red wing of the 0° transition of 13C12C5Hj which can be tentatively explained by a splitting due to the reduced sym-metry that has been theoretically predicted for iso-topically substituted benzene cations [47].

The striking difference between these spectra is the broad structure of the vibrational peaks in the "light" benzene (78 u) spectrum in Fig. 10 (upper trace) which also differs from that in the spectrum of Fig. 9. This is probably caused by different sets of excited J, K rotational levels in the intermediate state. We have to bear in mind that in Fig. 10 the first laser frequency was scanned to the maximum of the 13C12C5H6 absorption. This frequency posi-tion coincides with the red wings of the rotational contour of the 12C6H6 band containing preferen-tially high J levels. High J levels in the intermediate state lead to rotational transitions in the vibra-tional band of the ion which are widely spaced.

To summarize, the main result of this section (see Fig. 9) is that molecular ions can be prepared in various defined vibrational states when the second laser frequency is tuned to the respective transition and threshold ions are exclusively observed.

3.4. Cluster ion dissociation

After having demonstrated the virtue of pulsed field threshold ionization for studies of molecular

ions, we apply this technique to weakly bound molecular complexes. The frequency of the first laser is now tuned to the intermediate state of the benzene-Ar complex that is produced in the cooled molecular beam. The 6o transition is red-shifted by 21 cm- 1 from the corresponding transition in bare benzene. Under these excitation conditions prefer-entially benzene-Ar complexes are ionized due to the selectivity of the intermediate state. A further selectivity is achieved by the mass selective detec-tion of the threshold ions that is possible in our experiment.

Figure 11 shows the PFI spectrum of the

Ö ) 3

£S2

C6H6+-C6H6.Ar^

\616\±3I2)

0"

'U-^

6'(±3/2) C6H6-Ar+

41

172 cm·' M H

QtV

-+J LA

74400 74600 74800 75000 75200 75400

Two-Photon Energy [cm1]

Fig. 11. Three threshold ion spectra obtained for a molecular beam with benzene seeded in Ar under high pressure. Bottom: threshold ion signal at 78 u for a photon energy hvx in resonance with the 6l state of C6H6; the vibrational spectrum of the ben-zene cation is measured. Middle: threshold ion signal at 118 u for a photon energy hvx in resonance with the 61 intermediate state of the neutral benzene-Ar van der Waals dimer. The resulting spectrum is shifted by 172 cm- 1 to the red of the benzene spec-trum (lower trace) due to the increased binding energy in the ion. All peaks with a larger energy than the 41 state disappeared. Top: threshold ion signal at 78 u for a photon energy hv\ in resonance with the 6l intermediate state of the benzene-Ar dimer. The signal at the benzene mass (78 u) results from a decay of the complex ion and can be observed at the 16V (±3/2) peak and for peaks with higher energy. For discus-sions of the dissociation threshold see text.

228 HJ. Neusser and H. KrausejInt. J. Mass Spectrom. Ion Processes 131 (1994) 211-232

benzene-Ar cation (118u) (middle trace). For comparison, in the lower trace the spectrum of the bare benzene cation (Fig. 8) is given. It was obtained for a frequency of the first laser tuned to the 6Q transition of bare benzene. The 0° tran-sition of the benzene-Ar cation is red-shifted by 172 cm"1 due to the stronger bonding of the benzene-Ar cation compared to the neutral com-plex in the electronic ground state. This represents the shift of the adiabatic ionization potential after complexation. The ionization energy of benzene-Ar is found to be IE = 74383 ± 2 cm"1 in good agreement with recent ZEKE results [48]. On the blue side of the 0° transition small features indicate the additional excitation of van der Waals modes.

The striking result is the absence of any vibra-tional structure above the 41 band in the spectrum of the benzene-Ar cation. The position of the split 61(3/2) band relative to the origin is the same as in the benzene cation. However, in the benzene cation spectrum the 6!(3/2) mode is the strongest band, while in the benzene-Ar cation the 0° band is the strongest. From the missing peaks at higher energy we conclude that at excess energies of more than 418 cm"1 (position of the 41 band) fragmentation of the (benzene-Ar)+ complex occurs. If this con-clusion is true, a signal should be observed at the daughter ion mass (i.e., the mass 78 u of benzene). The PFI signal monitored at 78 u for the same frequency of the first photon v\ is given in Fig. 11

c

U a o

55 56 57

55 56 57

C6H6-Ar*

C6H6+-j

6!(l/2)-Band

6!(3/2)-Band

0°-Band

1 I 1 1 M 1 1 1 1 1

68 68.5 69

J-(C6H6)$

40 50 60 70 80 Time-of-Flight [μ5]

90 100

Fig. 12. Three threshold ion mass spectra obtained after excitation of three different final states of the benzene-Ar cation. Bottom: final state is the vibrational ground 0° state of the ion. Only parent ion peaks (C6H6 · Ar+) appear. Middle: final state is the 61 (3/2) state of the ion (Emt = 368 cm-1). In addition to the parent ion peak, weak features around the daughter (C6H^) mass are observed and shown on a magnified scale in the inset. For explanation see text. Top: final state is the 61 (1/2) state of the ion (£int = 677 cm-1). The parent ion

peak has disappeared and a strong daughter ion peak appears. For explanation of the inset see text.


(upper trace) and compared with the spectrum of (benzene-Ar)+ monitored at mass 118 u measured at the same excitation conditions (middle trace). Obviously, the missing bands in the spectrum observed at the parent ion mass (118u) appear when monitoring the daughter ion signal at 78 u. The quality of the mass resolution and of the sup-pression of the ion signal of non-energy-selected directly produced ions that is achieved with the reflectron mass spectrometer is demonstrated in Fig. 12. The lowest trace displays the threshold mass spectrum of the 0° band. Here the (benzene-Ar)+ peak at 118u dominates the spectrum. The achieved mass resolution is demonstrated in the inset of Fig. 12 with an expanded horizontal scale. Only very weak signals at the time of flight of the benzene monomer of 55.75 /xs, the benzene dimer of 78 ^s, and the trimer of 96 /xs are seen. These are produced by direct ionization. Basically the same situation is observed for the 6!(3/2) band with the benzene-Ar peak representing the strongest signal (middle trace). Finally, for the highest excited 61 ( 1 /2) ion state at an excess energy of 677 cm-1 the C6H6 · Ar+ peak disappears and instead a strong benzene peak grows in. Since nothing was changed in the intermediate state these benzene cations are clearly the product of the (benzene-Ar)+ dissocia-tion. The additional small features at somewhat (0.2 and 0.6 /xs) longer flight times (see inset) result from a slow metastable dissociation of non-energy-selected benzene-Ar and benzene dimer ions, respectively. Thus we conclude that the dissociation of C6H6 · Ar complexes in long-lived high Rydberg states occurs before they are ionized by the pulsed field. This implies that the dissociation process does not disturb the Rydberg electron. Since the dissocia-tion process is the same for the electron in a high Rydberg orbit or removed from the core, this clearly demonstrates that the benzene-Ar cation dissoci-ates on a time scale of less than lOO^s by evapora-tion of the Ar atom at an internal energy smaller than 629 cm"1 (position of the 16161(3/2) band). In con-clusion, the first observed peak in the daughter ion spectrum at 629 cm-1 yields an upper limit for the dissociation energy of the benzene-Ar cation, i.e.,

E0 < 629cm"1 (78meV), {\6l6x{$/2) level). In our recent publication we found daughter ions already at the smaller internal energy of the 61 (3/2) peak [10]. At present we cannot exclude the possibility that this signal originated from a three-photon excitation process. To be on the safe side we take the energy of the 1616!(3/2) band as an upper limit. Experiments on this point will be discussed in more detail in a separate publication [49]. From the 172 cm-1 red-shift of the ionization energy of benzene-Ar, an upper limit of D0 < 457 cm"1 (56meV) results for the dissociation energy of the neutral benzene-Ar complex using the relation D0 = E0- ΔΙΕ. These upper limits can be reduced considerably by com-paring the results of the benzene-Ar complex with those of benzene-Kr or other benzene-noble gas clusters, as will be presented in a forthcoming paper [49].

It is interesting to compare the results for the upper limits of the dissociation energies of the neu-tral and ionic benzene-Ar complex with the results for heterogeneous benzene-molecule dimers in Table 1. In the ionic benzene-Ar complex, charge transfer resonance interaction is not expected to contribute to the binding energy. The ionization energy of Ar (15.76eV [50]) is much higher than that of benzene (9.243 eV [48,51]). Thus, the charge is located on the benzene part of the com-plex. This is the same situation as e.g. in the ben-zene-cyclohexane dimer where the charge is also located on the benzene side of the complex. The smaller dissociation energy of the ionic benzene-Ar complex is then qualitatively explained by the smaller polarizability ( 1.64 À [27]) of Ar compared to cyclohexane (11.0 À [27]). The same argument also holds for the neutral benzene-Ar complex: the dissociation energy of benzene-Ar is expected to be smaller than the dissociation energy of the other benzene-molecule dimers listed in Table 1 because of the smaller polarizability of Ar.

The upper limit for the dissociation energy of the neutral benzene-Ar complex found in this work from the PFI spectra in Fig. 11 is in formal agree-ment with the binding energies found with different theoretical methods. Force field calculations

230 HJ. Neusser and H. Krausejint. J. Mass Spectrom. Ion Processes 131 (1994) 211-232

yielded a value of 393 cm- 1 [52], and recent ab initio calculations a value of 380 cm"1 [53]. How-ever, we have to bear in mind that the experimental upper limit of this work represents a value that is most likely above the real value. The reason for this is the sparse vibrational level structure of ihe benzene-Ar cation around the dissociation thresh-old. A dissociation cannot be observed for the 41

level at 418 cm- 1 excess energy but it is observed at the 16161(3/2) level at 629cm"1. For a critical comparison with theoretical predictions the gap between the highest non-dissociating level and the lowest dissociating level should be decreased. This will be shown in a separate publication on hand of measurements of benzene-Kr dissociation thresh-old [49].

4. Summary and conclusion

In this work, various new aspects of time-of-flight mass spectrometry are discussed which are important for the investigation of weakly bound van der Waals clusters. Two different methods of studying the energetics of molecular clusters have been presented. Both techniques employ a cooled supersonic beam combined with a time-of-flight mass spectrometer. The special instrument used here is a linear reflectron mass spectrometer with a molecular beam collinear to the ion flight paths. For ionization, resonance-enhanced two-photon ionization is particularly suitable as the resonance enhancement in the first absorption step assists the selection of a single cluster species from the mixture of clusters produced in the molecular beam. For the determination of their dissociation threshold the investigated ions must be produced with defined internal energy distributions. In break-down measurements there exists a sharp upper limit for this energy distribution given by the total photon energy. In this limiting case the excess energy above the adiabatic ionization energy is present as internal energy of the ion, i.e. the kinetic energy of the electron is zero. When the decay of an ion is observed as a function of this photon energy the fragment ion intensity breaks down for a

photon energy below threshold. The resulting "breakdown graphs" yield appearance energies for the fragments. This breakdown is close to the dissociation threshold with a negligible kinetic shift when slow metastable dissociation is observed. With this technique the dissociation energies of several molecular clusters containing an aromatic molecule as a chromophor have been obtained.

The second method permits the production of cluster ions with sharply defined internal energy rather than a distribution of energy. This is a novel result for clusters achieved with a special technique based on pulsed field ionization of long-lived Rydberg states close to a higher ioniza-tion threshold. Threshold ions appear when the increasing two-photon energy reaches a new ioni-zation threshold. The separation of the energy-selected threshold ions from non-energy-selected ones is performed by the energy-analyzing proper-ties of the reflecting field in a reflectron mass spec-trometer [10]. As a result the threshold ion spectrum contains sharp peaks at the different vibrational levels of the ions on a flat baseline rather than steps as in the conventional ionization efficiency curves. Although the separation of ions can be achieved with other techniques [9,41], the energy analysis in a reflectron time-of-flight mass spectrometer provides in addition a high mass reso-lution due to its flight time correction properties. This is particularly important for cluster investiga-tions (as their mass spectra contain a great number of peaks and the spacing between neighboring peaks is small). Several features of the mass-selective pulsed field ionization technique used in this work are described.

First experiments with this powerful technique demonstrate that the production of state-selected benzene-noble gas dimer ions is possible and that their decay can be observed. Future applications will include other dimer ions, larger ionic clusters and spectroscopic studies of their ground states. Good candidates for these investigations are clus-ters that do not undergo large structural changes after ionization and have a small number of van der Waals modes. This behavior is expected for systems

H.J. Neusser and H. Krause/Int. J. Mass Spectrom. Ion Processes 131 (1994) 211-232 231

with weak contributions from charge transfer reso-nance interaction in the ionic complex, such as clusters composed of constituents with strongly differing ionization energy. For the spectral resolu-tion demonstrated in this work the investigation of van der Waals modes, even in weakly bound ionic complexes, is possible [54]. This is important infor-mation complementary to the spectral information available for the respective neutral systems. At present, the spectral resolution is limited to about 1.6 cm- 1 for delay times of some 20/is [55]. One expects that the lifetimes of higher Rydberg states are even longer. Thus, experiments with longer delay times of some lOO^s at low densities and small electric fields will further increase the spec-tral resolution and suppress the linewidth below the 1 cm"1 level. First experiments in our laboratory have indeed shown that a sufficient number of Rydberg states survives a 200//s delay and a threshold ion signal can be detected [49]. This high spectral resolution, together with the high mass resolution of the reflectron mass spectro-meter, makes this technique valuable for the spectroscopic investigation of not only clusters, but also radicals or transient species that are pres-ent at low concentrations within a molecular beam sample.

Not only spectroscopic but also kinetic investi-gations of ions become accessible with this new technique. When a metastable decay of the energy-selected threshold ions occurs in the drift region of the reflectron time-of-flight mass spectro-meter, the resulting drift peak can be analyzed and decay constants can be deduced. In this way, uni-molecular kinetic models and the validity of statis-tical assumptions can be experimentally tested. This is of particular importance for molecular clus-ters since, in these systems, the intramolecular energy redistribution process is expected to be dif-ferent due to the existence of two types of vibrations, intra- and intermolecular, with strongly differing frequencies.

In conclusion, we have shown that recent developments in pulsed field ionization demon-strate the importance of time-of-flight mass spec-

trometry for the investigation of the spectroscopy and decay kinetics of molecular and particularly cluster ions.

Acknowledgments

The authors thank Professor Schlag for the invi-tation to contribute to this special issue. They are grateful to Dr. Alice Smith for careful reading of the manuscript. Financial support from the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie is gratefully acknowledged.

References

1 R.G. Cooks, J.H. Beynon, R.M. Caprioli and G.R. Lester, Metastable Ions, Elsevier, Amsterdam, 1973.

2 W. Forst, Theory of Unimolecular Reactions, Academic Press, New York, 1973.

3 W.J. Chesnavich and M.T. Bowers, in M.T. Bowers (Ed.), Gas Phase Ion Chemistry, Vol. 1, Academic Press, New York, 1979.

4 C. Lifshitz, Int. J. Mass Spectrom. Ion Processes, 118/119 (1992)315.

5 H.J. Neusser, J. Phys. Chem., 93 (1989) 3897. 6 B. Brehm and E. von Puttkamer, Z. Naturforsch. Teil A, 22

(1967) 8. 7 T. Bear, in M.T. Bowers (Ed.), Gas Phase Ion Chemistry,

Vol. 1, Academic Press, New York, 1979. 8 G. Reiser, W. Habenicht, K. Müller-Dethlefs and E.W.

Schlag, Chem. Phys. Lett., 152 (1988) 119. 9 L. Zhu and P. Johnson, J. Chem Phys., 94 (1991) 5769.

10 H. Krause and H.J. Neusser, J. Chem. Phys., 97 (1992) 5923.

11 U. Boesl, H.J. Neusser and E.W. Schlag, Z. Naturforsch. Teil A, 33 (1978) 1546.

12 L. Zandee, R.B. Bernstein and D.A. Lieh tin, J. Chem. Phys., 69 (1978) 3247.

13 E.W. Schlag and H.J. Neusser, Ace. Chem. Res., 16 (1983) 355.

14 H.J. Neusser, Int. J. Mass Spectrom. Ion Processes, 79 (1987) 141.

15 C. Lifshitz and N. Ohnmichi, J. Phys. Chem., 93 (1989) 6329.

16 M.R. Nimlos, M.A. Young, E.R. Bernstein and D.F. Kel-ley, J. Chem. Phys., 91 (1989) 5268.

17 B. Ernstberger, H. Krause, A. Kiermeier and H.J. Neusser, J. Chem. Phys., 92 (1990) 5285.

18 A. Kiermeier, B. Ernstberger, H.J. Neusser and E.W. Schlag, J. Phys. Chem., 92 (1988) 3785.

19 H. Kühlewind, H.J. Neusser and E.W. Schlag, Int. J. Mass Spectrom. Ion Phys., 51 (1983) 255.

232 H.J. Neusser and H. Krause/Int. J. Mass Spectrom. Ion Processes 131 (1994) 211-232

20 H.L. Selzle, H.J. Neusser, B. Ernstberger, H. Krause and E.W. Schlag, J. Phys. Chem., 93 (1989) 7535.

21 P.M. Dehmer, J. Chem. Phys.,76 (1982) 1263. 22 E. Rühl, B. Brutschy and H. Baumgärtel, Chem. Phys.

Lett., 157 (1989) 379. 23 J.R. Grover, G. Hagenow and E.A. Walters, J. Chem.

Phys., 97(1992)628. 24 K.H. Fung, W.E. Henke, T.R. Hays, H.L. Selzle and E.W.

Schlag, J. Phys. Chem., 85 (1981) 3560. 25 B. Ernstberger, H. Krause and H.J. Neusser, Z. Phys. D, 20

(1991) 189. 26 W. Scherzer, O. Krätzschmar, H.L. Selzle and E.W. Schlag,

Z. Naturforsch. Teil A, 47 (1992) 1248. 27 R.C. Weast and M. J. Astle (Eds.), Handbook of Chemistry

and Physics, CRC Press, Boca Raton, FL, 1979. 28 A.L. McClellan, Tables of Experimental Dipole Moments,

Rahara Enterprises, El Cerrito, 1974. 29 H. Krause, B. Ernstberger and H.J. Neusser, Chem. Phys.

Lett., 184(1991)411. 30 B.W. van de Waal, Chem. Phys. Lett., 123 (1986) 69. 31 A. de Meijere and F. Huisken, J. Chem. Phys., 92 (1990)

5826. 32 P. Cârsky, H.L. Selzle and E.W. Schlag, Chem. Phys., 125

(1988) 165. 33 P. Hobza, H.L. Selzle and E.W. Schlag, J. Chem. Phys., 93

(1990) 5893. 34 O. Krätzschmar, H.L. Selzle and E.W. Schlag, in prepara-

tion. 35 M. Meot-Ner, P. Hamlet, E.P. Hunter and F.H. Field, J.

Am. Chem. Soc, 100 (1978) 5466. 36 B. Badger and B. Brocklehurst, Nature, 219 (1968)

263.

37 F.H. Field, P. Hamlet and W.F. Libby, J. Am. Chem. Soc, 91 (1969) 2839.

38 E.G. Jones, A.K. Bhattacharya and T.O. Tiernan, Int. J. Mass Spectrom. Ion Phys., 17 (1975) 147.

39 K. Hiraoka, S. Fujimaki, K. Aruga and S. Yamabe, J. Chem. Phys., 95 (1991) 8413.

40 K. Müller-Dethlefs and E.W. Schlag, Ann. Rev. Phys. Chem., 42 (1991) 109.

41 C. Jouvet, C. Dedonder-Lardeux, S. Martrenchard-Barra and D. Solgadi, Chem. Phys. Lett., 198 (1992) 419.

42 H. Kühlewind, A. Kiermeier and H.J. Neusser, J. Chem. Phys., 85 (1986) 4427.

43 H. Kühlewind, A. Kiermeier, H.J. Neusser and E.W. Schlag, J. Chem. Phys., 87 (1987) 6488.

44 W.E. Cooke and T.F. Gallagher, Phys. Rev. A, 17 (1978) 1226.

45 K.H. Fung, H.L. Selzle and E.W. Schlag, Z. Naturforsch. Teil A. 36 (1981) 1257.

46 S.R. Long, J.T. Meek and J.P. Reilly, J. Chem. Phys., 79 (1983) 3206.

47 J. Eiding and W. Domcke, Chem. Phys., 163 (1992) 133. 48 L.A. Chewier, K. Müller-Dethlefs and E.W. Schlag, Chem.

Phys. Lett., 135 (1987) 219. 49 H. Krause and H.J. Neusser, J. Chem. Phys., in press. 50 S.G. Lias, J.E. Bartmess, J.F. Leibman, J.L. Holmes, R.D.

Levin and W.G. Mallard, J. Phys. Chem. Ref. Data, 17 (suppl. 1) (1988).

51 S.G. Grubb, R.L. Whetten, A.C. Albrecht and E.R. Grant, Chem. Phys. Lett., 108 (1984) 420.

52 N.B. Horin, U. Even and J. Jortner, J. Chem. Phys., 91 (1989)331.

53 P. Hobza, H.L. Selzle and E.W. Schlag, J. Chem. Phys., 95 (1991)391.

International journal of Mass Spectrometry and Ion Processes 131 (1994) 233-264 0168-1176/94/S07.00 © 1994 - Elsevier Science B.V. All rights reserved

233

Using reflectron time-of-flight mass spectrometer techniques to investigate cluster dynamics and bonding

Shiqing Wei, A.W. Castleman, Jr.* Department of Chemistry, Pennsylvania State University, University Park, PA 16802, USA

Laser based time-of-flight mass spectrometer systems affixed with reflectrons are valuable tools for investigating cluster dynamics and reactions, spectroscopy and structures. Utilizing the reflectron time-of-flight mass spectrometer techniques, both decay fractions and kinetic energy releases of metastable cluster ions can be measured with high precision. By applying related theoretical models, the desired thermochemical values of metastable species can be deduced, which are otherwise very difficult to obtain. Several examples are discussed with attention focused on ammonia as a test case for hydrogen bond systems, and xenon for weaker van der Waals clusters. A brief overview of applications to investigating solvation effects on reactions and structures, delayed electron transfer and ionization through intracluster Penning ionization, is also given.

Key words: Reflectron; Clusters; Penning ionization; Hydrogen bonding

(Received 23 June 1993; accepted 6 August 1993)

Abstract

1. Introduction

Studies of the dynamics of formation and dissociation, and the changing properties of clusters at successively higher degrees of aggregation, enable an investigation of the basic mechanisms of nucleation and the continuous transformation of matter from the gas to con-densed phase to be probed at the molecular level [1]. In this context, the progressive clustering of a molecule involves energy transfer and redistri-bution within the molecular system, with atten-dant processes of unimolecular dissociation taking place between growth steps [2]. Related processes of energy transfer and dissociation are also operative during the reorientation of mole-cules about ions produced during the primary

ionization event required in detecting clusters via mass spectrometry [3], providing further motivation for studies of the dissociation dynamics of clusters [4-9].

The cooling of a body which accompanies evaporation from its surface is a familiar phe-nomenon, and readily explicable in terms of a phenomenological approach. Recent progress in the field of cluster beam research now enables the details of these evaporation processes to be quantitatively investigated at the molecular level [8]. Even though the method often used to prepare clusters, e.g. supersonic expansion of gases through a nozzle, will yield "cold" neutral clusters, ionization of these neutral species makes detailed investigations of the evaporation processes possible. This is due to the fact that clusters "warm-up" upon ionization, and the ionic clusters are normally formed with enough energy to undergo a number of first-order processes, in * Corresponding author.

SSDI0168-1176(93)03886-Q

234 S. Wei and A.W. Castleman, Jr./Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264

particular those of evaporative metastable evaporation. The ionized clusters can be mass selected and the evaporation processes can be probed quantitatively [8].

As the body of knowledge on cluster dissociation continues to increase, the complex issue of interpreting the processes arises. Several models [10-13] have been presented in order to account for these unimolecular evaporative processes. Klots [12,13] in particular has elucidated various important factors regarding the formation and energetics of dissociating clusters or, as they are now termed, evaporative ensembles. From the theoretical models the relative binding energies of the evaporating cluster ions can be derived from studies of the extent of the evaporation processes over a given period of time, but accurate measure-ments of decay fractions together with kinetic energy release values are required in order to obtain absolute energies. The application of the measurements of decay fractions in obtaining relative energies is discussed first, with application to ammonia and xenon cluster cations. Thereafter consideration is given to methods for determining absolute bond energies.

Until now, most studies of dissociation dynamics of metastable cluster ions have been made using a double focusing mass spectrometry method [7,11]. In general, it is difficult to determine the kinetic energy release of metastable cluster ions, and the measurements are often limited to small cluster ion systems. As discussed herein, the technique of reflectron time-of-flight (TOF) mass spectrometry is a valuable alternative approach to more standard methods. With carefully designed experiments it is possible to derive both kinetic energy releases and decay fractions for cluster ions undergoing unimolecular decomposition following multi-photon ionization. The precision which can be obtained with the method is demonstrated for studies of cluster ions (NH3)„ · H + , n — 4-17. By applying both Klots' evaporative ensemble model and Engelking's modified RRK/QET theory, it is demonstrated that binding energies of ammonia cluster ions can be derived.

Other areas of current research interest in the cluster field to which reflectron TOF finds particular application include the study of cluster reactions, structure, and processes of ionization. Drawing largely from work in our own labora-tory, we discuss one typical example pertaining to the influence of solvation on dehydration reactions for alcohol clusters, which display a delay or induction time for the reaction, a fact that is easily revealed using the reflectron technique. Then, we discuss the structures of hydrogen-bonded cluster ions, particularly a titration technique enabling insight into cluster structures, as applied to the case of protonated water clusters. The first experi-mental evidence for the influence of isomeric structures on dissociation dynamics of hydrogen-bonded clusters was found by using the reflectron TOF mass spectrometer. Finally, we discuss the observation of delay electron transfer and ioniza-tion revealed through the laser excitation and resulting intracluster Penning ionization in mixed clusters.

2. Experimental

2.7. General considerations

A brief description of the apparatus and tech-nique is given here, with details available elsewhere [14,15]. A schematic of the apparatus is shown in Fig. 1. Neutral clusters are formed in a supersonic expansion of desired gases from a pulsed nozzle (diameter = 150 //m), and thereafter ionized using laser techniques. For most of the work discussed herein, 355 nm light from a frequency-tripled Nd:YAG laser is employed. The laser system typically produces pulses of approximately 6 ns duration with fluxes of 1016

photons per pulse. Ions formed by multiphoton ionization process are accelerated in a double electrostatic field [16] to several kilo-electronvolts of energy and directed through a 130 cm (lx) long field-free region toward a reflectron. Ions are then reflected at an overall angle of about 3° and there-after travel 80 cm (/2) through another field-free

S. Wei and A.W. Castleman, Jr./Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264 235

Pulsed Nozzle

Skimmer ^ Ionization Laser

TOF Lenses

Gas Inlet

First Field Free Region

Metastable Cluster Ions

dl·*----''*'

Reflectron

MCP Detector

t$ \)8^f>.

?* xet* I \<>*e

l ^ _ Second Field _ ^ J Free Region

Fig. 1. Reflectron TOF mass spectrometer.

region toward a chevron microchannel plate detector. The signal received by the detector is fed into a 100 MHz transient recorder coupled to an IBM PC/AT computer. The experiments typically operate at 10 Hz and TOF spectra are usually accumulated for at least 1000 laser shots.

A reflectron is employed to separate daughter and parent ions in order to measure kinetic energy releases (KERs) and decay fractions of dissociating cluster ions in the field-free region of the apparatus. The initial parent ion energy (£/0), established by potentials applied to the TOF lens elements (Ux

and C/2), is measured by using the reflectron as an energy analyzer. Details of this approach are dis-cussed in a later section. When a metastable parent ion decomposes to a daughter ion and a neutral species, the daughter ion has an energy of Ud = (Aid/Afp) i/0> where Afd and Afp are masses of daughter and parent ions, respectively. Studies of metastable decomposition processes are done by setting proper voltages to the second grid (Ut) and the last grid (£/k) of the reflectron unit [8,15].

There are three operating modes for the reflectron: (1) when Ut > U0 and t/k = 0, both parent and daughter ions are reflected back in the first reflective field, the recorded mass spectrum is called a "hard reflection" TOF spectrum (e.g. Fig. 2(a)); (2) when Ud < Ut < U0 and Uk = 0,

only daughter ions are reflected back and detected; this gives rise to what is termed as a "daughter only" TOF mass spectrum (e.g. Fig. 2(b)); (3) when Ut < Ud and Uk > U0 both parent and daughter ions are reflected in the second reflective field, the recorded spectrum is then called a "reflectron" TOF mass spectrum (e.g. Fig. 3). Measurement of the kinetic energy release of metastable ions is based on a peak shape analysis of the TOF spectra. Decay frac-tions of dissociating cluster ions are determined from the analysis of integrated peak intensities of parent and daughter ions. In order to compensate errors resulting from instrumental artifacts and different ion trajectories, the reflectron is operated under varying conditions (as discussed below).

2.2. Measurement of decay fractions

2.2.1. Hard reflection TOF spectrum. A typical "hard reflection" TOF spectrum is recorded under the following conditions: Ut > U0 and Uk = 0 so that both parent and daughter ions are reflected back in the first reflective field. Figure 2(a) displays a typical "hard reflection" TOF spectrum of the ammonia cluster ions (NH3)„ · H + , n = 2-20, taken at Ut = 2400 V > U0 = 1650 V and t/k = 0. These cluster ions are produced via multi-photon ionization of neutral ammonia clusters,

236 S. Wei and A.W. Castleman, Jr.j Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264

10

-

(a)

(b)

t 1

1

ί J

J 1 |

\ \

r~

1 I J 1 J i ί i i

- γ -

X

r~

20

30 50

Flight Time (microsecond)

70

44

-\ (c)

H (d)

J

111 u

1

1 — i 1 —

111 L A

1

1 i J

1 1

V

L

x 2

— I

46 48


50

S. Wei and A.W. Castleman, Jr. j Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264 237

which typically leads mainly to protonated cation clusters [8]. These generally readily evaporate ammonia molecules in the microsecond time domain for reasons detailed later in this paper. In the case that a metastable decomposition in the field-free region is involved, at Ut > U0 both parent and daughter ions can be observed in the same TOF spectrum. The daughter ions have less energy than the corresponding parent ions ((Md/Mp)U0 versus t/0, and Md < Mp) and hence do not penetrate into the reflective field as deeply as do the corresponding parent ions. Therefore, the daughter ions arrive at the detector earlier than the corresponding parent ions. Figure 2(c) is a portion of the "hard reflection" TOF spectrum (Fig. 2(a)) in the region showing the parent ions (ΝΗ3)„·Η+, η = 11-13, and their corresponding daughter ions (the peaks on the right side of each peak pair are the parent ions and the corresponding daughter peaks are on the left side). The time difference for each set of parent and daughter ions is typically 0.10-0.30 //s, depending on the initial parent ion energy and the mass. It is important to verify that those peaks on the left side of the peaks corresponding to the parent ions (ΝΗ3)„·Η+ in Fig. 2(c) are the daughter ions (NH3)„_! · H+ and not other parent ions of differ-ent masses. This is done by varying Ut in the range of U0 < Ut <2U0 and observing the relative arri-val times of those peaks. If the peaks were attribu-table to parent ions of other masses, their flight times would correlate to the exact mass assign-ment. However, the experimental results show that the relative arrival times of those peaks change with different Ut settings and they do not correlate to mass assignment as a parent.

2.2.2. "Daughter only" TOF spectrum. When a metastable parent ion decomposes into a daughter ion and a neutral species, the daughter ion carries an energy of Ud = (Md/Mp)U0, where Md and Mp

are the masses of the daughter and parent ions, respectively. Figure 2(b) is a daughter only TOF spectrum showing peaks of (NI^^H"1" corre-sponding to metastable dissociation of the parent cluster ions (ΝΗ3)Λ·Η+, « = 2-20. The peak assignments are simply done by comparing the ion arrival times in the daughter ion spectrum with those in the hard reflection spectrum. Besides, the peak assignments can be further confirmed by a daughter ion cutoff potential method [15] (detailed discussion of the cutoff method is given in section 2.4). Figure 2(d) is a portion of Fig. 2(b) in the region showing the daughter ions (ΝΗ3)Λ_ιΗ+, n = 11-13, to be com-pared to Fig. 2(c) which displays both parent ions (NH3)„ · H+ and their daughter ions (ΝΗ3)Λ_!Η+. Note that the arrival times of daughter ions in Fig. 2(d) are larger than those in Fig. 2(c). This difference in ion flight time results from the lower Ut setting (whereupon the ion spent more time in the reflectron) when the daughter ion spectrum was taken.

2.2.3. "Reflectron" TOF mass spectrum and measurement of decay fractions. The decay frac-tions of metastable cluster ions can be derived from the integrated intensities of the parent and daughter ion peaks in a TOF spectrum. However, there are several problems which cause inaccuracy in the measurement of ion decay fractions. First, the signal distortion in the microchannel plate (MCP) ion detector may affect the peak intensity measurement. Each channel (typical diameter « 100/im) of the MCP usually [17,18] has a dead time of about 8 ms. This slow recovery time can cause some degree of signal distortion in the ion peak intensity, although the presence of 105—106

channels in a MCP plate may partially compen-sate this problem [17,18]. Since the decay fraction measurements are based on peak intensity analysis, the dead time problem in the miniature channels of

Fig. 2 (opposite). TOF mass spectrum of (NH3)nH+, n = 2-20, U0 = 1625 V. (a) A hard reflection TOF spectrum at Ut = 2400 V and Uk = 0. (b) A daughter only TOF spectrum at Ut = 1625 V and Uk = 0. (c) Parent (NH3)„H+ (right) and daughter ( N H ^ ^ H * (left),

n = 11-13. (d) Daughter peaks of ( Ν Η ^ Η + , η = 11-13.


an MCP must be taken into consideration. This is done by controlling the ion trajectories and dis-playing the parent and daughter peaks in the same TOF spectrum. Second, the correction of the ion trajectory in the reflective field (as discussed earlier) can account for a 20-30% difference in the peak width and a 50% deviation in the peak intensity measurements. The third factor which may cause inaccurate measurement of the relative parent and daughter ion intensities is the peak overlap in the TOF spectrum. In order to control ion trajectories and avoid peak over-lapping, the parent and daughter ions are directed and reflected in the second reflective field of the reflectron unit; this is called the "reflectron" TOF mass spectrum. Potential settings on the second and last grids of the reflectron are varied to control parent and daughter ions to follow the same flight paths. To force the daughter ions to follow the same flight path of their parents, the potential settings on the reflectron grids are varied (based on the consideration of parent and daughter ion energies). Different combinations of the new t/t/ and t/k/ settings may also lead to the same arrival time of daughter ions. However, the same arrival time of daughter ions results from the fact that the daughter ions follow the same trajectory of their parents at £/t/ = (Md/Mp)Ut

and Uv = (Md/Mp)Uk.

2.3. Determination of KER values

2.3.1. General considerations. The intensity and width of the ion peaks carry information on the parent ion internal energy as reflected in the KER measured when the metastable parent ion decom-poses. In order to obtain an accurate measurement of the ion peak shapes (including intensity and width), it is necessary to vary the Ut setting until the parent and daughter ions have the same flight paths. The ion trajectory correction can be done by making the parent ion and its corresponding daughter ion follow the same paths in the reflec-tive field of the reflectron unit. This can be accom-plished as described in what follows.

At Ux = Ut , the flight time that the parent ion spends in the reflectron, fp, is expressed as follows,

/p = C[U0/MpY'2Mp/Utp

where C is a proportionality constant, Mp is the parent mass and Ut is the voltage setting on the reflectron for the parent ion TOF spectrum. At t/t = Ut , the flight time of the daughter ion td is given by

td = C[U0/Md]l/2Md/Utd

where Ut is the voltage on the reflectron for the daughter ion TOF spectrum. Note that the pro-portionality constant C — 2.$SLq~1'2 with tp and ta in microseconds, Mp and Md are in atomic mass units, UQ, Ut and Ut in volts and q is the charge unit. By comparing these two equations, one can conclude that the trajectories of parent and daughter ions will be the same at Ut/Utd= Mp/Md. Since the ratio of Μρ/Μά

varies with the cluster size, the different values of Ut and Uta must be set for each group of parent and daughter cluster ions. Based on our experi-mental results, this correction in the ion trajectory can account for a 20-30% difference in the peak width measurement and a 50% deviation in the peak height measurement.

2.3.2. Measurement of average values of KER. Measurement of the average values of KER of metastable ions is based on a peak shape analysis of the TOF spectra. This method was originally proposed by Berry [19] and later developed by Franklin et al. [20] for use in a TOF mass spectro-meter. During the decomposition of the metastable parent ion in the field-free drift region, the internal energy of the parent ion is converted to the transla-tional energy of the daughter ion. As a result, the translational energy distribution of the daughter ion is expected to be broader than that of the parent ion due to the KER in the decomposition process.

Baldwin et al. [21] pointed out that the broadening width Wu resulting from the KER,


can be obtained from the relationship W} — Wd- Wp, where Wd is the width of the daughter ion at 22% of the peak height, and Wp is the width of the parent ion peak in the mass-selected ion kinetic energy spectra. The formula w\ = W\ - Wl is exact only if both parent and daughter ion peak shapes are gaussian.

It can be shown [8] that the relationship Wl = W\ - W\, (where Wd is the width of the daughter ion at 22% of the peak height, and Wp

is the width of the parent ion peak) is also applicable in obtaining the KER from the parent and daughter ion TOF spectra.

2.4. Mass assignment via TOF and cut-off potentials

In general, there is a small difference in the flight times of the daughter ion and its parent (a few tenths of a microsecond) due to different ion flight trajectories in the reflection. This allows one to assign the metastable ion peaks by simply compar-ing the arrival times of daughter and parent ions in the TOF spectra. In order to complete the meta-stable decomposition dynamics studies, we applied a daughter ion cut-off potential method. This method has been introduced previously for studies of cluster ion dissociation dynamics using our reflectron TOF mass spectrometer [15]. Briefly, the technique involves observing the presence/absence of daughter ion peaks upon changing the voltage settings on the middle plate of the reflec-tron unit (Ut).

In the general case, there are two possible meta-stable decomposition channels for the mixed cluster ion (A)„(B)+:

(AMBU-CA^CBJi + A (1) ( A ) n ( B ) ^ ( A ) „ ( B ) ; _ , + B (2)

For a parent cluster ion of mass Afp, the corre-sponding daughter ion masses are (Mp — A) for process (1) and (Mp - B) for process (2), res-pectively. Here, we apply the daughter ion cut-off

potential method for investigation of the three possible metastable decomposition mechanisms: (a) both channels coexist; (b) only channel (1) is open; (c) only channel (2) is open.

First of all, if two channels open simultaneously, two daughter ion peaks will appear in the daughter ion spectrum for a corresponding parent ion. The separation of these two peaks in the daughter ion TOF spectrum is determined by the following equation:

Ai = C(t/0/Mp)1/2(Md, -Md2)/Ut

where Md\ and Mdi are the masses of the two daughter ions from the same parent ion as indicated in processes (1) and (2). The constant C is equal to 2.SSLq~l^2(L = 1.5 cm, is the distance of the first grid and the middle grid of the reflectron unit) with At in microseconds, A/p, Mdi, and Mdi in atomic mass units, UQ and Ut are in volts and q is the unit of charge. As an example, for parent ion NH3(TMA)4 .H+ (Mp = 254u) with an energy of 2600 ± 10 eV, the daughter ion energies are 2426 and 1996eV for channels (1) and (2), respectively. If the two channels are open simul-taneously, two daughter ion peaks for parent ion NH3(TMA)4H+ will be separated by 0.23 ^s at £/t = 2550 V. This can be easily identified with our apparatus which has a TOF resolution better than 0.02 /is (mass resolution mj Am = 1200) in the mass range of 200.

Secondly, if only one of the two dissociation channels is open, only one daughter ion peak will be seen in the daughter ion spectrum for a corre-sponding parent ion. If the metastable decomposi-tion only follows channel (1), the daughter ion peak will disappear when Ut is set between 1996 and 2426 V. However, if the daughter ion peak still exists at 1996 < Ut < 2426, it may be considered that channel (2) is the only possible metastable decomposition process. Whether channel (2) is the only possible metastable decomposition process can be confirmed by observing the dis-appearance of the daughter ion peak when Ut is lowered to a value less than 1996 V.


3. Studies of metastable dissociation and relative cluster ion bond energies

3.1. Quantifying the phenomenon

3.1.1. General considerations. Following ioniza-tion, cluster ions typically undergo extensive evaporative dissociation in time windows of sev-eral of tens of microseconds. The excess energy available for this process is generally available from several sources including rearrangement of the ions following the instantaneous ionization of the stable neutral structure, as well as photoabsorption into the ion state during multiphoton ionization, and possible thermal fluctuations among the many modes of clusters which relay internal energy during the formation process.

The general procedure for measuring the decay fractions is given in section 2.2. Consider the general reaction for cluster A comprised of n units undergoing evaporative dissociation:

A„+-+A„+_1 +A (3)

The decay fractions, D = Ιά/(Ιά + Ip) where /d and /p are the daughter and parent ion intensities, can be measured with high precision by integrating the parent and daughter ion peaks.

3.1.2. Evaporative ensemble model for decay fractions. The evaporative ensemble [12,13] model assumes that each cluster ion has suffered at least one evaporation before entering the field-free region of the TOF mass spectrometer. For each cluster ion, t0 is defined as the flight time that the parent ion spends from the ionization region to the last TOF lens, whereas / is the flight time in which the parent ion reaches the first grid of the reflectron unit. The value of t0 can be calculated using a formula given by Wiley and McLaren [16] and / can be obtained from the TOF spectrum. The evaporative ensemble predicts that at time / the normalized population of daughter ions is given [12] by

D= (C„/72) /«{i / [ / 0+(i- io)exp(-72/C„)]} (4)

7 2 = 7 2 / [ l - ( 7 / 2 C „ ) 2 ] (5)

where Cn is the heat capacity of the cluster ion (in units of Boltzmann constant &B), 7 is the Gspann parameter. Studies of clusters containing many thousands of atoms by using the electron diffraction method [22-24] suggest that 7 is « 25, usually independent of cluster size [25].

Using Eqs. (4) and (5) with experimental values of /, t0, and D, the Gspann parameter 7 and heat capacity Cn may be determined. Since the cluster (intermolecular) modes are much more important than the internal molecular motions in the evaporative dissociation process of a cluster containing n molecules, Cn is chosen to be propor-tional to n — 1.

When one considers that the cluster ions of sizes n and n + 1 have different binding energies as denoted by AEn and ΔΕη+Ϊ9 the daughter ion population can be calculated from the following equation [13]:

D = 1 - (aWnyl 1η{1 + [βχρ (α^ ) - \}t0/t} (6)

where

aWn = S - W**> 2 -2 (7)

Wn = AEÎ{\ + [(dE/dAEn)];]}

χ(Δ£„-Δ£„_,) /Δ£„} (8)

(dE/dAEn)k = ( C / 7 )

x [ l + 7 / 2 C w + (7/C„)2/12...] (9)

By fitting the calculated decay fractions to the measured ones, the values of (AEn - AEn+{)/AEn

are readily obtained.

3.2. Considering the ammonia system

Ammonia clusters have been studied extensively and provide an excellent test case for these considerations. As discussed later in this paper, comparison of the deduced thermochemical values


is possible by employing data available in the literature obtained by other methods.

Upon multiphoton ionization of the neutral ammonia clusters, protonated ions are formed via the following internal ion/molecule reactions [6,8]:

NH^ + NH3 -► NHj- + NH2

(A//° = -0.74eV) (10)

The excess energy within the cluster ion owing to the above exothermic reaction and the relaxation around the newly formed ion, as well as possible further multiphoton excitations, contributes to heating of the cluster ions and concomitant evaporative dissociation. As the cluster ions evaporatively cool, the dissociation extends to longer times, and the present study is directed to an investigation of the metastable dissociation

processes of cluster ions (ΝΗ3)Λ · Η + in the field-free region (where the time window is about one to a few tens of microseconds). In the case of the ammonia cluster ion system, the dissociation process can be expressed as

(NH3)„ - H + - (ΝΗ3)„_Χ - H + + xNH 3 ( i i ;

where x ^ 1. Under our usual experimental con-ditions, and in the absence of collision induced dissociation processes, we have found (as dis-cussed later) that the dissociation process involves losing only one molecule (i.e. x = 1 in Eq. (11)) for cluster ions of all sizes (n = 4-22).

3.2.1. Measurement of decay fractions. Figure 3 displays "reflectron" TOF spectra taken at (a) Ut = 390 V and Uk = 3901 V and (b) Ut = 420 V and Uk = 4200 V (the parent ion peaks are


Fig. 3. Reflectron TOF spectra of (NH3)„H+, « = 1 1 - 1 5 , (P = Parent, D = Daughter), (a) Ut = 390 V and t/k = 3901V. (b) Ut= 420 V and Uk = 4200 V.


labelled as "P" and the corresponding daughter peaks are labelled as "D"). Here, the dissociation process with n = 14 in Eq. (11) is used as an exam-ple. First, the U{ and Uk potentials are set 420 and 4200 V, where the overall ion intensity is optimized. As shown in Fig. 3(b), the arrival time of the parent ion (ΝΗ3)14Η+ is 52.22/is. Since the daughter ions (NH3)13H+ have an energy of Ud = (Md/Mp) U0 as a result of metastable decomposition, different tra-jectories for parent and daughter ions in the reflec-tron are expected. Corrections for this effect are made as discussed earlier. It is clear that the arrival time (52.22 ßs) of the daughter ions (ΝΗ3)13Η+ in Fig. 3(a) is equal to that of their parent ions (NH3)14H+ in Fig. 3(b). The integrated intensities of these two peaks are then used to compute the decay fraction, D = Id/(Ip + Id) where Id and Ip

are the daughter and remaining parent ion inten-sity, of (NH3)„H+, n = 14. Similar procedure fol-lows for all cluster species of various size n. The

experimentally measured values of the metastable cluster ions (ΝΗ3)„Η+, n — 4-23 are plotted as a function of cluster size, as shown in Fig. 4 (the data points are plotted as open squares). The experi-mental uncertainties for each measured point are less than 5%.

The solid line in Fig. 4 represents the decay frac-tion of metastable cluster ions (NH3)rtH

+, n = 4 -23, calculated using Eqs. (4) and (5) with 7 = 24.5 and Cn = 6(n - 1). An alternative way [26] involves (a) using the bulk heat of evaporation of liquid ammonia and a first estimate of 7 = 25 to estimate the Gspann temperature, Tg = AEn/ (7&Β), (b) obtaining the bulk heat capacity of liquid ammonia at Tg, (c) applying this value as Cn in Eq. (4) to get 7, and (d) repeating processes (a)-(c) if necessary. It was found that values of 7 = 24 and Cn = 5.5(n - 1) give a line which can be superimposed with the solid line in Fig. 4. The fact that these two methods are consistent indicates

Fig. 4.

2 6 10 14 18

Number of Ammonia, n

A plot of measured decay fractions as a function of cluster size, n: □, measured points; ( ) Eq. (4).

S. Wei and A.W. Castleman, Jr.j Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264 243

10 14


Fig. 5. Relative binding energies plotted as a function of«: +, Eq. (6); □, ref. 27.

both sets of estimated Cn and 7 values are quite plausible.

The general trend agrees well with the experi-mental measurements; however, it is evident that the predictions from Eq. (4) are in discrepancy with the experiments for small cluster sizes. As discussed in the following section, this discrepancy results from different binding energies of ammonia cluster ions at various sizes; however, the binding energy difference was neglected in Eq. (4).

Using Eqs. (6-9) it is possible to determine relative bond energies for the clusters from measurements of the decay fractions. The deduced values of {ΔΕη/ΔΕη+\) for ammonia cluster ions (NH3)„H+, n = 4-22, are shown in Fig. 5. The points designated as crosses are obtained by assum-ing Cn = 6(/i - 1) and 7 = 24.5. It should be noted that the values of (ΔΕη - ΔΕη+χ)/ΔΕη are very

sensitive to the choice of 7 for small cluster sizes; however, the values are insensitive to 7 for large ones.

By using the best known literature value of binding energy of (NH3)5H+ as a starting point, binding energies of (NH3)nH+, n = 4-22, are calculated and shown in Fig. 6 (labelled as open triangles) with 7 = 24.5 and Cn = 6(n - 1). The literature values (designated as open squares) are also shown in the figure for comparison. It is seen that values from the present approach are in good agreement with the available thermo-chemical data for n = 4-7. An abrupt decrease in the binding energy of (NH3)„H+ from n = 5 to n = 6 is observed. This indicates that (NH^NHJ" is a particularly stable ion and can be pictured as a complete solvation shell formed by four NH3 molecules bound to a central ammonium ion.


900


Fig. 6. A plot of binding energies as a function of cluster size n: □ , Literature values [27]; Δ» Klots' model, decay fractions; X, Klots' model, KER; + , Engelking's model, KER.

3.3. The metastability and thermochemistry of xenon cluster cations

The ion spectrum Xe,J~, n — 6-31, after multi-photon ionization of xenon neutral clusters is shown in Fig. 7. In agreement with earlier observa-tions [28-30], magic numbers at 13, 16, 19, 23, 25 are evident. In the time window of a few tens of microseconds, the observed dissociation processes are:

Xe+ -► Xei_! + Xe, n = 5-40 (12)

A daughter ion spectrum is demonstrated in Fig. 8. By using the reflectron, the parent and the corre-sponding daughter ions can be readily separated [8]. As an example, a typical reflectron TOF spec-trum is shown in Fig. 9. The decay fractions, D — Ιά/(Ιά + Ip) where /d and Ip are the daughter and parent ion intensities, are determined by integrating the parent and daughter ion peaks.

The measured values are presented in Fig. 10, along with the error bars from 12 independent measurements. The trends and the variations with size are in excellent agreement with studies of xenon cluster ion distributions reported previously [29].

The method only produces the ratio of AEn/AEn+x and for very small cluster ions, n < 10, the values are very sensitive to the choices of other parameters [31]. For larger cluster ions, n > 10, however, the values are found to be relatively insensitive to the choices of the heat capacity and Gspann parameter. In applying the above equations, for n > 10 7 is found [13] to be « 25, Cn is estimated from the bulk heat capacity, and t0/t is calculated from the experimental conditions. By fitting the calculated decay fractions to the measured ones, the values of (ΔΕη - ΔΕη+\)/ΔΕη, or ΔΕη/ΔΕη+χ are readily obtained.


120 140 160

Flight Time (microseconds)

Fig. 7. A hard reflection TOF spectrum of Xe+, n = 6-31: t/0 = 2300 V; Ut = 2800 V; Uk = 0.

200

In the temperature range 70-100 K, which is that expected for the evaporating aggregates of xenon clusters [25], the molar bulk heat capacity (in units of Boltzmann constant) was measured to be 2.77 ± 0.03 [32], the heat capacity was found to be a weak function of temperature in that range [32]. Considering that the heat capacity of cluster ions Xe+ mainly results from the cluster modes, a function of the form Cn = (2 Jin - 6) is used. By fitting to the measured decay fractions, the values of AEn/AEn+x are plotted in Fig. 10. Since the analysis applied to small clusters is very sensitive to the choices of other parameters, only the values of large (n ^ 12) cluster ions are shown in the figure. Typically, a 5% uncertainty of 7 results in about 25% error of AEn/AEn+x for cluster ion XeJ~2, a 10% error for XeJ , and 4% for XeJ .

By comparing Figs. 7 and 11, it is evident that every magic number observed in the mass spectra matches exactly the local maxima of the relative binding energies. The qualitative relationship

between the intensity anomalies and stability has been often assumed, but without proof. The pre-sent results provide evidence for the existence of the correlation between the relative binding energies AEn/AEn+\ and the intensity anomalies of the mass spectrum for xenon cluster ions. It is important to note that there are two possible origins for the local maxima of the relative binding energy: (i) when the cluster size n is a particularly stable ion, a substantial drop between AEn and AEn+\ will lead to the local maximum at n\ (ii) when there is a substantial binding energy increase after AEn+x (i.e. for cluster size n + 2), this will also result in a local minimum at n + 1 or a magic number at n in the mass spectrum. In the first case, the intensity maximum does reflect the stability of the cluster ions as in the present situa-tion. In the second case, however, the intensity drop between n and n + 1 does not reflect the stability of cluster ion n since the local minimum at n + 1 results only because of the increasing

246 S. Wei and A. W. Castleman, Jr./Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264

13

Λ/ 1 \A \ä tiL V JL

19

80 100 120 140 160


180 200

Fig. 8. A daughter only TOF spectrum of Xe+_,, n = 6-31: t/0 = 2300 V; U{ = 2250 V; Uk = 0.

stability after n + 1. This suggests that the intensity anomalies of mass spectra alone do not conclu-sively lead to knowledge of the stabilities of the cluster ions.

4. Application of KER measurements; bonding: ammonia as a paradigm

4.L Experimental measurements

The TOF-Reflectron technique has applications far beyond mass analysis and determination of metastable dissociation fraction. In particular it provides a very valuable approach to determining KER during evaporative dissociation. As shown in this section, these data also find application in determining absolute values of cluster bond energies for systems without a barrier to evaporative metastable dissociation. As discussed in section 2.3., the technique depends on accurately measuring the temporal profile of the parent and daughter clusters.

In studies of ammonia clusters, all parent and cluster ion peaks are observed to display a gaussian shape in the TOF spectra. As an example, the experimental data points (10 ns apart from each other) of the normalized peak of the parent ion (NH 3 ) n H + (indicated as crosses) and daughter ion (NH3)10H+ (indicated as open circles) are fitted to the pure gaussian curves (indicated as solid lines) as shown in Fig 12. Clearly, the daughter ion peak is broader than the parent ion peak due to the KER involved in the decomposition process. A small tail on the later time side of the parent cluster ion peak is observed, resulting from rapid ion fragmentation in the TOF acceleration zone. This tail does not affect our KER measurements for cluster size up to 17 because the width used to compute the average KER values is measured at 22% of the peak height. However, it was found that the tail became more noticeable for peaks of larger parent cluster ions due to more dissociation in the TOF lens region.

S. Wei and A.W. Castleman, Jr.jInt. J. Mass Spectrom. Ion Processes 131 (1994) 233-264 247

13

u

11 d ' a X 0

UUuUXjilÂiiA — i 1 1 ι 1 1 1 1 1 1 —

80 100 120 140 160 180


200 220 240

Fig. 9. A reflectron TOF spectrum of Xe+, n = 6-36: U0 = 2400 V; Ut = 500 V; t/k = 3100 V.

With a measured parent ion birth potential U0, an estimated daughter ion traveling distance in the field-free drift region L (distance from the position where the daughter ion is born to the ion detector), and a broadening width Wt observed in the TOF spectra, the average KER is given [8] by

[13] using the following equation:

AEn = Ί(Ετ)/[\ - (7/2Ç,)] (14)

(Er) = l/(L2){[U0Wt}2[Md/(2MpM)}} (13)

where Mp, Md and M are the masses of the parent ion, daughter ion, and neutral fragment, respect-ively. The values deduced by this method are shown in Fig. 13.

4.2. Evaporative ensemble model for KER

The evaporative ensemble approach suggests that the evaporation energy can also be calculated from the measured kinetic energy release with a properly chosen heat capacity and Gspann parameter. It follows that the binding energy of a molecule in a cluster ion of size n can be calculated

Figure 6 displays the binding energies of (ΝΗ3)ΛΗ+, n = 4-17, as a function of cluster size n (labelled as "x") using the measured KER and 7 = 24.5, Cn = 6(n - 1) and compares the data with prior measurements based on high pressure mass spectrometry and values deduced from decay fractions.

4.3. QET/RRK theory and approach

Engelking has also proposed a modified QET/ RRK statistical model [10] to determine the bind-ing energy of a molecule within a metastable cluster ion and suggested that it can be determined from the evaporative lifetime and the average KER. In the present work we adopt this simple theoretical model and by employing the measured average KER values, we obtain the binding energies of the protonated ammonia cluster ions (NH3)„H+,


0 10 20 30

Number of Xenon, n

Fig. 10. A plot of measured decay fractions as a function of cluster size n.

40

n = 4-17. The binding energy of (ΝΗ3)„Η+, can be calculated using the following equation:

E: = A(s- l)[Cl/{s~l\ETYs-2)/{s-l) - (Er)] (15)

where, (Er) = the measured average kinetic energy release, s = cluster mode (number of internal motions), A — 0.5 (model scaling parameter, given by Engelking [10], and C = l6nu3ß^yS/Tn. In order to calculate the constant C, one needs to obtain information on v (vibrational frequency of cluster mode), μ (reduced mass), g (remaining channel degeneracy of the ejected neutral fragment), S (geometrical cross section for form-ing cluster ion) and Tn (unimolecular dissociation rate). From Eq. (15) it is known that the dissoci-ation rate and the KER are the only required experimental values for obtaining the binding energy of a metastable cluster ion. However, the metastable process can be observed only if the life-times of the dissociating cluster ions fall within the experimental time window. Since the decay rate

depends upon a high power (s - 2) of the internal energy, even a very broad range of experimental lifetimes can imply a narrow range of internal energies of the metastable ions. This point can be further visualized in Eq. (15). The lifetime is included in the constant C which is taken to a large root. Therefore, the variation of dissociation rate constant does not greatly affect the determin-ation of the binding energy.

In computing the binding energies of (NH3)„H+, the cluster modes including rotations and transla-tions of each monomer are treated active, providing 6(n-l) degrees of freedom. The channel degeneracy g is set to n - 1 and the vibrational frequencies are set to 100 cm"1 for (NH3)„H+, and 50 cm"1 for (NH3)„H+, n = 6-17 considering that the cluster ion (NH3)5H+ forms a closed solvation shell. The geometrical cross section S is chosen to be 100 À . The lifetimes (1/Γ„) of the metastable cluster ions are estimated from the observed ion arrival times assuming the


22 26

Number of Xenon, n

Fig. 11. The relative binding energies calculated from Eq. (6) plotted as a function of cluster size n.

ion dissociates in the region midway between the last TOF lens and the first plate of the reflectron unit in our TOF mass spectrometer. They are in the range of 8-20/xs for (NH3)17H+, n = 4-17.

It is necessary t< choose a single proper model scaling parameter (A) such that the computed binding energies of ammonia cluster ions match the literature values. Figure 13 shows the deter-mined binding energies of (NH3)WH+, n = 4-17, as a function of cluster size n for A = 1 (shown as a cross in the figure). All formula parameters remain the same for cluster ions of all sizes except the reduced mass, lifetime, channel degeneracy, cluster mode, and measured average KER. The deduced binding energy values are scaled to match the best determined value for (NH3)5H

+ in the literature [27]. The determination of binding energies of the protonated ammonia cluster ions using Eq. (15) are relatively insensitive to the choices of the vibrational frequency and lifetime. A 50% variation in the chosen vibrational fre-

quencies gives rise to about 10-15% change in the calculated binding energy. In addition, the binding energies change about 5-10% when lifetime is varied from the earliest to the latest time window, 0.6 to 18 /is for the tetramer and 1.2 to 36 μ$ for the 17-mer.

As seen from Fig. 6, it is evident that binding energies determined from different methods are in good agreement with the known thermochemical data for n = 4-7. However, more importantly, the present approach enables binding energies to be determined for larger clusters, which are very difficult to obtain using the more traditional high pressure mass spectrometry technique.

5. Reactions, structure, and delayed ionization

The reflectron TOF technique is also useful for investigating processes which require an induction time such as one which has been observed for reactions in alcohol clusters, solvation dependent dehydration reactions in acetone, and delayed



Fig. 12. Peak shapes of parent (NH 3 ) n H + ( + ) and daughter (o) fitted by gaussian functions ( ).

ionization through intracluster Penning ionization type phenomena in mixed clusters.

5.7. Reactions influenced by solvation

The application of the reflectron technique in

o

(U °0

E

1-Lü to

■<J-

-

' I

I -

I 1

H T I T lU i Ξ i I x j

-

3 5 7 9 11 13 15 17 19

Number of Ammonia , n

Fig. 13. A plot of the measured average KER as a function of«.

studying reactions in clusters is evident from dis-cussions in the foregoing sections. It has been widely applied to a number of systems where it has been found of value in unraveling reaction mechanisms. Two especially interesting examples involve acetone clusters [15], where the reaction mechanism is strongly dependent on the degree of solvation, and alcohol clusters [33-35] where evidence suggests that there is a substantial delay/induction time associated with a solvation dependent hydration process for the protonated dimer.

5.1.1. Methanol clusters: size specific intracluster reactions. Other solvation dependent reactions have been observed in methanol. For example, upon ionization, the protonated cluster ions of methanol H+(CH3OH)„, are also found to undergo several other intracluster reaction pathways which display a dependence on the degree of aggregation. A conventional TOF mass


H

Z ω H z

5-0 15.0

TIME OF FLIGHT

25.0

Fig. 14. A conventional TOF spectrum of H+(CH3OH)n; laser wavelength, 266 nm.

spectrum is shown in Fig. 14. Two distinct distri-butions of cluster ions are observed in the figure along with other ions of interest. The sequence of ions labelled a consists of protonated methanol clusters of form H+(CH3OH)„. A second sequence labelled b is clearly evident after the n = 7 peak in the a series; these ions are of form H+(H20)(CH3OH)„. A peak corresponding in stoichiometry to C H 3 0 + (ionization potential 11.55eV from CH3OH), labelled c, is also observed and apparently arises from three-photon, non-resonant ionization of CH3OH at 266 nm. The undissociated monomer ion, CH3OH+, is not observed in the mass spectrum. No intensity anomalies are observed in the smooth distribution of protonated methanol cluster ions. The prominence at « = 1 3 and the shoulder on the right of n = 11 result from contributions of background peaks.

The origin of the sequence a, corresponding to protonated methanol peaks, is identified as arising from a rapid intracluster proton-transfer reaction

following ionization of the neutral clusters. This reaction has a well known bimolecular counter-part that proceeds at near collision rate [36].

(CH3OH)+ - Η+(€Η3ΟΗ)Λ_! + CH 3 0 (16)

The formation of sequence b, H+(H20) (CH3OH)„_3 is envisioned to occur via

[H+(CH3OH),]* - H+(H20)(CH3OH)„_3

+ (CH3)20 + CH3OH (17)

Incorporation of a reflectron in the TOF mass spectrometer enabled investigation of the dissocia-tion processes, and resolution of the operative mechanisms. Daughter ions are easily identified under conditions where parent ions are excluded from the spectrum as seen in Fig. 15.

Five types of peaks (labelled d-h) are seen clearly in the spectrum. Peaks labelled d correspond to a mass loss of 32 u from a parent ion cluster that enters the drift region as H+(CH3OH)rt (these are analogous to the a peaks in Fig. 14) as shown


>-

W

z ÜJ

z d n-l«6

d n-l=2

L_L I » I I » « » » i I t I » i I I t I 10.0 20.0 30.0 40.0

TIME OF FLIGHT 50.0

Fig. 15. A daughter-ion TOF spectrum of Η+(€Η3ΟΗ)π_!; laser wavelength, 266 nm.

below

H+(CH3OH)M -> H+(CH3OH)n_, + CH3OH

(18)

Mass losses of more than one monomer unit, not readily apparent in Fig. 15, appear as unresolved shoulders on the early arrival side of the d peaks in the figure. The loss of up to five methanol mono-mers from the protonated octamer is observed.

The peak designated e in Fig. 15, corresponds [34] to loss of water from the protonated methanol dimer ion:

H+(CH3OH)2 -> (CH3)2OH+ + H 2 0 (19)

Experimentally, it has been found to be a process requiring an induction time of at least several tenths of a microsecond.

The reflected ions which do not dissociate in the field-free region account for the peaks labelled f. Of particular interest are the peaks labelled g and h. Peaks labelled h correspond to loss of 78 u from H+(CH3OH)n, indicating

loss of both a methanol monomer and C2H60 (dimethyl ether) for parent ions n = 4-9 in the drift region. Peaks designated as g in Fig. 15 represent the loss of one methanol moiety from the mixed cluster species H+(H20)(CH3OH)„ (peaks labelled b in Fig. 14). If these clusters are also formed by reaction (17), they initially contained nine or more methanol molecules. It is observed that the protonated methanol dimer ion eliminates H 2 0 while retain-ing dimethyl ether as described by reaction (19). This has recently been found in thermal reaction experiments [36] and is clearly due to solvation effects. Analogous water-elimination reactions are not observed for the parent cluster ions larger than

In the case of these larger clusters, H 3 0 + is solvated more strongly by methanol than is protonated dimethyl ether, (CH3)2OH+ [37]. Hence for these clusters water retention and dimethyl ether elimination lead to production of mixed clusters to form H+(H20(CH3OH)„.


Prompt (immediately following ionization) loss of dimethyl ether in the TOF ion lens, results in the mixed cluster sequence labelled b in Fig. 14, the conventional TOF mass spectrum. These mixed clusters formed by rapid decay processes in the ion lens subsequently dissociate in the field-free region via loss of one methanol monomer to form the sequence g peaks in Fig. 15. In contrast a "slower" elimination of dimethyl ether in the drift region (along with CH3OH) from cluster sizes H+(CH3OH)„ « = 4-9 results in the peak sequence labelled h in Fig. 15. Thus, a significant difference in reactivity with cluster size is observed; the small clusters lose dimethyl ether on a longer (field-free region) time scale while the larger clusters undergo comparatively rapid loss of C2H60 in the ion lens.

5.2. Hydrogen-bonded cluster ions: compositional and structural changes due to solvation

5.2.1. Structures of (A)n(M)mH+: A = NH3 or H20, M = Proton acceptor. Studies of ion solvation, structure, and stability are also possible using the reflectron TOF mass spectrometer tech-nique. In small hydrogen-bonded cluster ions it is known that stability arises as a result of the nature and location of ligand molecules that bond to the central proton or protonated species; for example (CH3COCH3)2H+ [38], (CH3OCH3)3H30+ [36], and (NH3)4NHj [14,37]. It is interesting to investigate the location of the proton in the general case. Consider, for example, one situation in which the proton affinity of molecule X in a mixed cluster ion (ΝΗ3)„(Χ)^Η+ is less than that of another such as ammonia, and compare it to the alternative case where the proton affinity of molecule X is larger than that of ammonia.

The proton affinity of ammonia is 204.0 cal mol"1 and those of CH3COCH3, CH3CN, and CH3CHO are 196.7, 188.4, and 186.6calmol_1, respectively [15]. In all three mixed cluster ion systems, the intensity distributions show that there is a maximum [39] at n + m = 5. Results of metastable decomposition studies [40] of the mixed

cluster ions were determined to be

(NH3)„(X)WH+ - (ΝΗ3)Α(Χ)„-!Η+ + Χ

for n = 1 (20)

(NH3)M(X)WH+ -> (ΝΗ3)Λ.,(Χ)ΜΗ+ + ΝΗ3

torn ^2. (21)

The results clearly indicate that N H j is the core ion; this is in accordance with expectation in view of their relative proton affinities. The four available hydrogen-bonding sites are occupied to complete the first solvation shell. The four ligands can include any combination of ammonia and molecule X. The loss of an ammonia molecule resulting from the metastable decomposition for n ^ 2 indicates that X is more strongly bonded to the NH4" than is another NH3. This is not surprising since all molecules considered here have higher dipole moments and polarizabilities than ammonia, which leads to them having a greater ion/dipole and ion-induced-dipole inter-action.

It is instructive to next consider the case of pyridine (C5H5N) and trimethylamine (TMA = (CH3)3N) [41] whose proton affinities are 220.8 and 225.1 cal mol"1, respectively, which are larger than that of ammonia [15]. Metastable decomposition studies of NH3(C5H5N)wH+(w = 1-5) yield the following results:

NH3(C5H5N)„,H+ - (C5H5N)WH+ + N H 3

for m < 4, (22)

NH3(C5H5N) + m + H+ - NH3(C5H5N)m_1+H+

+ C5H5N for m ^ 4 (23)

For m < 4, the higher proton affinity of pyridine leads to its stronger bonding to the proton, whereby the ammonia molecule is lost upon meta-stable dissociation. The loss of pyridine in the case of m ^ 4 can be accounted for by the fact that a central NH4" core ion is formed that provides four hydrogen bonding sites for the ligands, hence a structure dictated by the net energetics of bond-ing. This finding is consistent with the previous


proposed stable structure for the cluster ion (NH3)„(X)mH+.

Extensive studies conducted on mixed proto-nated clusters of ammonia and trimethylamine show that the ion intensity distributions of (NH3)„(TMA)mH+ [41,42] display local maxima at (n9m) = (1,4), (2,3), (2,6), (3,2), and (3,8). As before, the fact that the maximum intensity occurs at (n,m) = (1,4), (2,3), and (3,2) indicates that a solvation shell is formed around NH4" ion with four ligands of any combination of ammonia and TMA molecules. In the situation where the maximum of the ion intensity occurs at (n,m) = (2,6) and (3,8), the experimental ions [H 3 N-H-NH 3 ] + (with six available hydrogen bonding sites) and [H3N-H(NH2)H-NH3]+ (with eight available hydrogen bonding sites). The observed metastable unimolecular decomposition processes [41] support the above solvation model. Through studies of the ternary systems comprised of the foregoing bound with water, additional observations revealed the general validity of these concepts. Some structures pertinent to the present considerations are shown in Fig. 16. Of course, account must be taken of the fact that these systems are very floppy and one should not expect them to display a rigid structure.

Strong evidence has also been obtained [42] for ringlike structures for mixed neutral clusters (A)„-(M)W (A is both the proton donor and accep-tor such as ammonia and water; M is only a proton acceptor such as acetone, pyridine, and

TMA

H

TMA H — N — H O /

A. H H

TMA

\

TMA

TMA

TMA

Fig. 16. Structure compatible with the observed magic number corresponding to H+(NH3)(H20)2[(CH3)3N]6.

M-. /

M

I H

M

H

\ /

O-H~

M

*f /

Fig. 17. Structure compatible with the observed magic number corresponding to (H20)5(M)6H+.

trimethylamine). For example, under a wide vari-ety of experimental conditions, mixed cluster ions display a maximum intensity at m = 2{n + 1) when n^ 5 for (NH3)„-(M)mH+, and m = n + 2 when n ^ 4 for (Η20)η·(Μ)^Η+. These findings reveal that the cluster ions with these compositions have stable closed shell structures as discussed above. However, a breakdown of the pattern occurs at n > 5 for the ammonia system and n > 4 for the water system, with the most intense peaks occur-ring for species with one molecule less than the expected pattern, i.e. m = 2(n + 1) - 1 when n = 6 for (NH3)rt.(M)wH+ and m = (n + 2) - 1 when n = 5 for (H20)„-(M)wH+. These results are compatible with suggestions that hydrogen-bonded ring structures form such as the one for (H20)5(M)+ shown in Fig. 17.

In like manner, we have investigated [43,44] the structure of pure water clusters. Particularly inter-esting species are H+(H20)2o and H+(H20)2 1 , the latter being the very prominent magic number in the water system that was mentioned earlier whose structure has been the subject of much debate. First, direct experimental evidence for clathrate structures of (H20)„H+ (n = 20 and 21) was obtained based on a technique similar to the one above that allowed the number of non-hydrogen-bonded surface hydrogens to be counted. Neutral clusters (H20)„*((CH3)3N)m, prepared in a pulsed nozzle supersonic expansion, were ionized by


Fig. 18. Water clusters with H 3 0 + or metal ions (depicted by a shaded circle) encaged inside the clathrate of (Η20)2ο·

multiphoton ionization and investigated with a reflectron TOF mass spectrometry technique. The magic numbers (n, m) in the ion intensity distribu-tions of (H20)W-[(CH3)3N]/W-H+ served to reveal the stable hydrogen-bonding structures. For the mixed cluster ion (H2O)20-[(CH3)3N]w-H+, the intensity distribution was found to display an abrupt intensity drop after the magic number at (20,11), while for (H20)2I-[(CH3)3N]W-H+ the magic number appeared at (21,10). The findings gave experimental evidence for a stable clathrate structure of (H2O)20H+, being a pentagonal dode-cahedron with the proton residing on the surface, while for (H20)2 1H+, the H 3 0 + ion is encaged inside the clathrate structure of (H2O)20; see Fig. 18. The latter structure provides a total of ten hydrogen-bonding sites for (CH3)3N. More recent studies have revealed that other cations can be trapped in the clathrate structure in an analogous fashion.

5.2.2. Influence of isomeric structures on dissoci-ation dynamics. Isomeric forms have been proposed to explain the observed differences in the spectral properties for several neutral van der Waals systems [45-48]. Recent developments in IR and

microwave spectrometry techniques provide sufficiently high-resolution rotational spectra to establish the existence of conformers in a hydro-gen bonded neutral with more than two individual entities [49-51]. Although isomeric forms of some ions and ionic dimers, such as HCO+/HOC+

[52,53], HCN+/HNC+ [54], and c-C3H^-OH2

[55], are well known, currently there is a paucity of evidence for isomeric cluster ion systems [56,57]. To the best of our knowledge there is no prior evidence for their existence in the case of hydrogen-bonded complexes of small cluster ions until our recent experimental proof [58] for the existence of isomers in hydrogen-bonded cluster ions, and their influence on the dynamics of cluster ion dissociation for the cluster ion NH3[(C2H5)3N]3H+.

A hard reflectron TOF spectrum of ionized mixed ammonia-triethylamine clusters is shown in Fig. 19(a). The three parent ion peaks which appear at 53.47, 54.94 and 56.39 /is are [(C2H5)3N]3H+, NH3[(C2H5)3N]3H+, and (NH3)2[(C2H5)3N]3H+, respectively. When a metastable decomposition pro-cess occurs in the field-free region, the daughter ion has an energy of (Md/Mp)£/0, where Md and Mp

are the daughter ion and parent ion mass. The daughter ion spectrum can be obtained when the reflecting potential, t/t, is set lower than U0 but higher than (Md/Mp)U0. In general, there is a dif-ference of a few tenths of a microsecond in the flight times of the daughter ion and its parent due to different ion flight trajectories in the reflectron as discussed in section 2.2. Moreover, the heavier the molecule lost in the dissociation process, the larger the difference between the flight times of the daughter and its parent ion. This difference allows an assignment of the metastable ion peaks by simply comparing the arrival times of daughter and parent ions in TOF spectra; but cut-off potential measurements described below are also made for confirmation.

Figure 19(b) displays a daughter ion spectrum taken at Ut = 2570 V. The four observed peaks arrive at 53.00, 54.47, 55.03, and 56.47/xs. By com-paring Fig. 19(a) with Fig. 19(b) it can be seen that

256 S. Wei and A.W. Castleman, Jr.I Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264


Fig. 19. TOF spectra of mixed ammonia-triethylamine clusters: U0 = 2600 V. (a) Hard reflection at Ut = 2840 V. Γ3 = (TEA)3H+; A„TW - (NH3)„(TEA)WH+. (b) Daughter ion spectrum at Ut = 2570 V. (c) Ut = 1860 V.

the daughter ion arrives earlier than its parent ion [(C2H5)3N]3H+ by 0.47 /xs, resulting from a dissociation process involving a loss of a TEA moiety (101 u). It is evident that there are two daughter ion peaks (second and third peaks from the left in Fig. 19(b)) from one common parent ion NH3[(C2H5)3N]3H+. The wider peak shifts toward earlier time by 0.47 μβ, and the narrower one shifts toward later time by 0.09 μ8; this enables differentiating the two processes which involves a loss of a TEA (101 u) and an ammonia (17 u) molecule, respectively. The parent ion (ΝΗ3)2[(€2Η5)3Ν]3Η+ loses one ammonia mole-cule in the TOF field-free region and gives rise to a daughter ion peak (the fourth peak from the left in Fig. 19(b)) which shows up at a time 0.08 ^s later.

In order to further confirm the metastable decomposition dynamics, we apply another complementary technique termed the daughter

ion cut-off potential method discussed in section 2.4. This is done by observing the presence/ absence of daughter ion peaks while the voltage settings on the middle grid of the reflectron unit (C/t) are varied. When C/t is lowered to 1860 V, only the two peaks corresponding to processes involving loss of a triethylamine molecule are observed, as shown in Fig. 19(c). This can be easily understood since daughter ions which result from loss of TEA have energies of [(Afp - 101)/Afp]t/0. Therefore, the cut-off potential measurements establish that the four peaks (from left to right) in Fig. 19(b) correspond to the following dissoci-ation processes (TEA=(C2H5)3N):

(TEA)3H+

NH3(TEA)3H+

NH3(TEA)3H+

(NH3)2(TEA)3H+

-> (TEA)2H+ + TEA -> (TEA)2H+ + TEA -► (TEA)3H+ + NH3

-+ (TEA)3H+ + NH3

(24) (25) (26) (27)

S. Wei and A.W. Castleman, Jr. I Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264 257

Dissociation of a metastable cluster ion is usually accompanied with some kinetic energy release during the process as discussed in Section 2.3. To further investigate the dissociation dynamics of metastable cluster ions of the general form of (ΝΗ3)„[(€2Η5)3Ν]^Η+ was investigated during dissociation [58]. Here, we concentrate only on the dissociation dynamics of the metastable ΝΗ3[(€2Η5)3Ν]3Η+ which displays evidence of the existence of isomeric structures. Measurement of the average KER is based on peak shape analysis of the TOF spectra of parent and daughter ions as described in section 2.3. A crucial requirement of this method is that the shape of both peaks of parent and daughter ions must be gaussian. Experimental data points of the peaks of daughter ions NH3[(C2H5)3N]2H

+ and [(C2H5)3N]3H+ (labelled as (2) and (3) in Fig. 19(b)), which come from the same parent ion ΝΗ3[(€2Η5)3Ν]3Η

+, are

fitted to the pure gaussian curve as shown in Fig. 20. The fit is seen to be excellent. The broadening width Wt9 resulting from kinetic energy release during the metastable dissociation can be obtained from Eq. (13), as discussed earlier. The average KER values for Processes (25) and (26) are measured [58] to be 21.7 and 5.3 meV, respectively.

By itself, the observation that two metastable decomposition channels open simultaneously for the mixed cluster ions does not necessarily prove the existence of an isomer. However, in con-junction with the dramatic difference of the KER of these two channels, the new findings serve as clear evidence for the existence of isomeric structures of NH3[(C2H5)3N]3H+ and their influ-ence on the dynamics of dissociation. The reasons for this are as follows. First, if one assumes as a premise that there is only one stable structure

54.8 54.9 55


55.1 55.2

Fig. 20. The experimental data points of the daughter ions NH3(TEA)2H+ (+) and (TEA)3H+ (O) are fitted by a gaussian function ( ).

258 5. Wei and A.W. Castleman, Jr./Int. J. Mass Spectrom. Ion Processes 131 (1994) 233-264

NH3((C2H5)3N)2H+

+ (C2H5)3N \

((C2H5)3N)3H+

+ NH3

NH3" ((C2H5)3N)3H+

(Π)

(C2H5)3N " H - N - H ·· N(C2H5)3

H

N(C2H5)3

(I)

Fig. 21. Model potential energy surface for dissociation of NH3(TEA)3H+.

for NH3[(C2H5)3N]3H+, the observation of two channels corresponding to the loss of (C2H5)3N and NH3 proceeding at the comparable decompo-sition rates in the same metastable decomposition time window implies that the dissociation energies of these two channels are about the same. Other-wise, only one channel with the lower dissociation energy would be observed. Second, consider the cluster ion metastable decomposition process for which the KER is well accounted for by the modi-fied RRK theory as proposed by Engelking [10] and the evaporative ensemble [8] by Klots [12,13]. In both models, the KER is found to be directly proportional to the cluster ion dissociation (bond) energy. The KER is determined from Eq. (14) as discussed earlier. For most clusters as well as the particular cluster ion in question, ΝΗ3[(€2Η5)3Ν]3Η+, the Gspann parameter is a constant (^ 25) and the heat capacity Cn should be the same for the parallel metastable pathways [12,13]. Therefore, on the premise of a single structure the KERs of these parallel channels should be about the same based on both Engel-king's and Klots' models. This follows from the fact that the dissociation energies of these two

channels would have to be about the same since the two channels proceed at the same decom-position rate at the same metastable time window (about 10/xs).

However, the measured KER for the triethyl-amine loss channel (21.7 ± 0.04 meV) is much larger than the ammonia loss channel (5.3 ± 0.7 meV) which definitely rules out the assumption of only one stable structure for cluster ion NH3[(C2H5)]3H+. Consequently, the dynami-cal studies prove the existence of more than one structure.

Two possible structures for NH3[(C2H5)3N]3H+

are: (1) the proton is bonded to NH3 forming NH^ and three other (C2H5)3N molecules are hydrogen-bonded to N H j (Structure I in Fig. 21); (2) the proton is solvated by three (C2H5)3N molecules forming [(C2H5)3N]3H+ to which the ammonia molecule is loosely bonded (Structure II). The exis-tence of a barrier between them is not surprising since substantial molecular Fig. 21 rearrangement is involved. The metastable decomposition could proceed with or without a structural rearrange-ment between these two isomers. In the case that the rate of the rearrangement is very slow com-


pared to molecular evaporation, the metastable dissociation will take place from two different energy surfaces of two distinct isomers (Struc-tures I and II) for the species NH3[(C2H5)3N]H+ as shown in Fig. 21. These two isomers must come from the different structures of the neutral species due to the Franck-Condon principle. Therefore, the memory of the neutral is main-tained and the dramatically different KER is due to the bonding energy difference of two distinct isomers. In the case that structural rearrangement would be possible, the observation of different KERs between the two channels shows that the two isomers must be individual long-lived species following energy randomization from which the memory of the original structure is lost; other-wise, two different KER values would not be found.

5.3. Spectroscopy of mixed clusters and delayed ionization through intracluster Penning ionization

5.3.1. Spectral shifts upon solvation and influence on ionization. The influence of clustering on the ionization of molecules has been a problem of long standing interest [59]. Resonance enhanced multiphoton ionization through the specific exci-tation of an electronic state of a chromophore contained within a cluster is a powerful method [60] of ascertaining the properties of clusters in relating these to their counterparts in the con-densed and isolated gas phases. Generally, the clustering of atoms or molecules onto a chromo-phore result in a perturbation of the electronic states of that chromophore. The spectral shift of a given electronic transition from that of the isolated chromophore is a measure of the relative differences between the lower and upper states of the energetic perturbation induced by clustering. This is analogous to the spectral shift of electronic transitions observed for molecules in solutions or matrices from their gas-phase transitions. A red shift implies that complex formation has reduced the energy difference between the two states, whereas a blue shift indicates an increase in the

difference. The magnitude and direction of the shift are due to a combination of effects, including dispersive and repulsive interactions, hydrogen bonding, and electrostatic forces involving such processes as dipole-induced-dipole or dipole-dipole interactions. Details of such studies are presented elsewhere in this issue.

We have employed one- and two-color reso-nance enhanced multiphoton ionization to investi-gate the S\ <— S0 π-electron transition in several substituted benzene systems such as phenylacetyl-ene and p-xylene clustered with various solvents. Single color multiphoton ionization studies of the perturbed Lb (!B2) states of phenylacetylene (PA) bound with Ne, Ar, Kr, and Xe were all found to induce a lowering of the Si resonance with respect to the ground state. These observed red shifts have been attributed to dispersive interactions by three factors: (i) short-range electronic repulsive interac-tions which result in a blue shift; (ii) electronic dispersive interactions which result in a red shift, and (iii) differences of zero point energies between the excited and ground states. Results from our

T

E u

I CO

_J

< cr H U u CL (/)

o ö -CD

o Ö -

o ö -C\J

Xe

/ / / / / / / /

/ Kr / / /

/ ?ΆΓ /

/ / / / / / N e

/

/ ,o '

— i 1

0.0 2.0 4.0

POLARIZABILITIES a (A )

Fig. 22. Spectral shift of PA-R (relative to the nascent PA) versus the polarizability of the rare-gas atom: a — 0.40 (Ne); 1.63 (Ar); 2.48 (Kr); 4.01 Â3 (Xe). The spectral shift of the two atom complexes of argon and krypton with phenylacetylene are almost exactly twice that of the respective single atom complexes.


laboratory, as well as others, support the finding [61] that in aromatic molecule-rare gas atom sys-tems the spectral shifts are dictated by atom polar-izability, i.e., the important role of dispersive forces in the perturbation of the Sj excited state. Figure 22 displays the results of our study which showed a direct linear dependence of the spectral shift on the electrostatic polarizability of the rare gas atom. The results conform to the Onsager model, but on a microscopic level [47].

Investigations with large-ringed systems have generally shown an approximate additivity of spec-tral shifts based on the number of rare gas atoms clustered on to the aromatic [61]. In our own work, we find that this additivity is apparently applicable only up to the clustering of two atoms per aromatic ring [60]. Since the additivity is nearly exact for the two-atom case, the spectral shifts shown in Fig. 22 are identical on a spectral shift per atom basis for the phenylacetylene system in the case of the two-atom containing rare gas complex. Interesting trends are also seen for larger clusters, a subject also discussed elsewhere in this issue.

Interesting spectral shifts have been observed for the clustering of other molecules such as N2, 0 2 , N 2 0 , NH3, H 2 0 , CC14, and CH4 to phenylacetyl-ene. In most cases the main resonance is also red-shifted, although in a few a substantial blue shift is observed, most notably for H 2 0 . The striking dif-ference seen between the isoelectronic molecules H 2 0 and NH3 can be rationalized in terms of the excitation of the π system leading to a repulsive interaction with the two long-pair electrons of the H 2 0 molecule [57,60].

Investigations of the shifts in the ionization potential of molecules with degree of aggregation is another subject of related interest. For example, studies have been made in which the ionization potentials of p-xylene (PX) bound with argon (ΡΧ·ΑΓΛ) were determined through studies in which the energy of one photon was fixed at the Lb resonance and the wavelength of the second laser was scanned. Resonance-enhanced ionization with a single-color laser results in significant frag-mentation due to the fact that the absorbed energy

is substantially above the ionization threshold since the S2 state lies more than halfway to the ionization continuum. Cluster fragmentation was found to be suppressed to a negligible amount in the two-color experiments, enabling a detailed investigation of the variation in ionization potential with degree of aggregation to be definitively established.

It is well-known that the Stark effect leads to a shift in ionization potential when measured in an electric field and correction is necessary to account for shifts in the order of 50 cm"1. The ionization potentials are found to vary with the square root of the electric field present in the region of ionization in accordance with expectations and findings of others [62]. Extrapolation to zero field is readily accomplished in view of the linear dependence and the fact that various cluster systems display lines of identical slopes in these weakly perturbed rare-gas aggregates.

The shifts in ionization potential of p-xylene in the rare gas aggregates is shown in Fig. 23 for clusters with one to six argon atoms. The shift in relative ionization potential is observed to display a

-800

_ - 6 0 0 + X OL

a. < I - 4 0 0

< x a. - 2 0 0

PX-Arr

2 NUMBER

4 6 OF ARGON ATOMS n

Fig. 23. Relative appearance potentials. Field ionization of PX-Arn (« = 0-6) in a 150Vcm-1 d.c. field. AP(PX-Ar+) increases with the coordination number n.


broadly linear dependence on the number of argon atoms. The largest deviations from this trend are observed for the dimer and pentamer. The observed total shift of about 750 cm"1 for the hexamer is to be contrasted with the matrix isolated value which is about 6000 cm"1 for a simi-lar molecule (benzene) in an argon matrix [63]. Evidently, the "local environment" with which a molecule interacts is relatively large in such a matrix and the observed shift is far from the expected bulk value.

5.3.2. Intracluster "Penning ionization" and evidence for slow ionization processes in clusters. Clusters provide interesting systems for comparing ionization and concomitant electron transfer processes for bimolecular processes in the gas phase, including Penning ionization [64-66], with analogous ones in the condensed phase. Toward this goal, ionization of clusters comprised by p-xylene bound to NH3 and N(CH3)3 were studied [64] following the absorption of photons through the perturbed S\ state of p-xylene. The findings are of significance in that they also bear on a problem of considerable current interest regarding the extent of ionization attainable in large molecules, and the significance of Rydberg states in delayed ionization processes in molecules and clusters.

An interesting finding is provided by results of studies involving adducts of p-xylene bound to NH3 and trimethylamine since the ionization of p-xylene is less than that of ammonia but greater than that of trimethylamine. In the case of ΡΧ·ΝΗ3, ionization by absorption of a second photon which is absorbed by the perturbed S\ state of p-xylene begins near the ionization thresh-old of p-xylene and leads to the expected cluster ion PX-NH^. Two other channels are possible at higher photon energies, namely the formation of NH4" at 0.1 eV above the ionization potential of p-xylene and NH^ at 1.8 eV above; N H j is observed in the two-color experiments at high fluence of the ionizing laser where two photons are absorbed by the Si state.

By contrast, absorption into high Rydberg states

of p-xylene below its ionization potential in PX*N(CH3)3 leads to the production of predomi-nantly N(CH3)J with H+N(CH3)3 as a minor product. No PX-N(CH3)J ion is detectable. One conclusion is that photoexcitation of p-xylene leads to an intercluster ionization process bearing analogy to Penning ionization where the perturbed high Rydberg states of p-xylene interact with the partner molecule N(CH3)3. A second, and more startling observation, was the finding of a slow ionization process as evident in the TOF peak shapes shown in Fig. 24. Since the laser interacts with the molecules in the first of a two-field accel-eration region, a long tail is only possible when the ionization process is slow. Fragmentation leads to a knee in the peak shape and not a long tail as observed in Fig. 24. Interestingly, the process is substantially slower with a decrease in the energy of the ionizing photon. Questions arise whether the slow step is associated with the proton transfer channel (i.e., the (CH3)3NH+ product) or an

100 torr

11.0 12.0 13.0 14.0

TIME OF FLIGHT U s e c )

Fig. 24. Ion mass peaks at different two-photon energies. Broad-enings of TMA+ ion peaks as a function of the ionization energy. A: hv2 > 3.875eV; B: hv2 = 3.688 eV; C: hv2 = 3.607 eV. Ionization energy of PX(S,) = 3.90eV. The broaden-ings in B and C correspond to time constants of 160 ± 20 and 200 ± 20 ns, respectively. The peaks corresponding to ΤΜΑ·Η+

are also observable.


electron transfer process (i.e., the (CH3)3N product). Careful measurements with deuterated species reveal that the tail is largely associated with the electron transfer process. Interestingly, Wada et al. [67] have found that orientational effects in the liquid phase, where motion is restricted, can lead to a significant reduction in the rate of Penning ionization. A plausible explanation for the present findings is that a large geometry change is involved in the formation of the trimethylamine ion, and/or, the long lifetime of Rydberg states near the threshold for ionization, followed by field ionization [68-70] is responsible.

6. Perspectives for the future

The importance of using a reflectron TOF mass spectrometer in the studies of metastable uni-molecular dissociation dynamics is that it enables simultaneous determination of the decay fractions and the kinetic energy release. Here, we present a precise method for measuring both kinetic energy releases and decay fractions of metastable cluster ions with corrections concerning instrumental arti-facts and the ion trajectory of the parents and daughters. The experimental data are used to derive the Gspann parameter and heat capacity of clusters as described in Klots' evaporative ensemble model. With these two parameters, we apply the evaporative ensemble model to obtain binding energies of ammonia cluster ions (NH3)WH+. The deduced binding energy values are in very good agreement with each other and with the thermochemical data.

For small cluster sizes, the values of relative binding energies from the approach using decay fractions are very sensitive to choices of Gspann parameter and heat capacity. It is therefore evi-dent that the method based on kinetic energy release measurement is the preferred one for small cluster sizes. By contrast, the method using decay fractions is more applicable to the larger cluster sizes, where it is comparably insensitive to the choice of parameters which must be employed

in deducing the values from theoretical models. Most importantly, the present approach enables binding energies to be determined for large clusters, which are very difficult to be obtained using traditional high pressure mass spectrometry technique. In addition to the above, the reflectron TOF technique enables studies of other processes such as delayed reaction, delayed ionization and the existence of isomeric structures in cluster ions. It can be expected that these processes will be the subject of continuing investigations in view of their fundamental importance, and the prospects for their study offered by recent experimental advances. Femtosecond laser techniques [71-73], combined with the above, hold promise of even more detailed elucidation of the dynamics of phenomena associated with the ionization of clusters.

Acknowledgments

Financial support from the U.S. Department of Energy, Grant No. DE-FGO1-88ER60658, and the National Science Foundation, Grant No. ATM-9015855, is gratefully acknowledged.

References

1 A.W. Castleman, Jr. and R.G. Keesee, Ace. Chem. Res., 19 (1986) 413; Science, 241 (1988) 36.

2 B.D. Kay and A.W. Castleman, Jr., J. Chem. Phys., 78 (1983) 4297.

3 A.W. Castleman, Jr. and R.G. Keesee, Chem. Rev., 86 (1986) 589.

4 A.J. Stace and C. Moore, Chem. Phys. Lett., 96 (1983) 80. 5 (a) A.W. Castleman, Jr. and R.G. Keesee, in P. Ausloos

and S.G. Lias (Eds.), Structure/Reactivity and Thermo-chemistry of Ions, NATO ASI Series, Reidel, Dordrecht, 1987, pp. 185-217. (b) T.D. Mark and A.W. Castleman, Jr., Adv. At. Mol. Phys., 20 (1984) 65. (c) T.D. Mark, Int. J. Mass Spectrom. Ion Processes, 79 (1987) 1. (d) U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spectrom. Ion Processes, 112 (1992) 121.

6 (a) O. Echt, P.D. Dao, S. Morgan and A.W. Castleman, Jr., J. Chem. Phys., 82 (1985) 4076. (b) S. Morgan and A.W. Castleman, Jr., J. Phys. Chem., 93 (1989)4544.


7 (a) C. Lifshitz and F. Louage, J. Phys. Chem., 93 (1989) 5633. (b) C. Lifshitz and F. Louage, Int. J. Mass Spectrom. Ion Processes, 101 (1990) 101. (c) C. Lifshitz, M. Iraqi and T. Peres, Rapid Commun. Mass Spectrom., 4 (1990) 485.

8 S. Wei, W.B. Tzeng and A.W. Castleman, Jr., J. Chem. Phys., 92 (1990) 332; 93 (1990) 2506.

9 (a) R.D. Levine and R.B. Bernstein, Molecular Reaction Dynamics and Chemical Reactivity, Oxford University Press, Oxford, 1987. (b) W. Begemann, K.H. Meiwes-Broer and H.O. Lutz, Phys. Rev. Lett., 56 (1986) 2248. (c) T.D. Mark, P. Scheier, K. Leiter, W. Ritter, K. Stephan and A. Stamatovic, Int. J. Mass Spectrom. Ion Processes, 4 (1986)281.

10 P.C. Engelking, J. Chem. Phys., 85 (1986) 3103; 87 (1987) 936.

11 A.J. Stace, J. Chem. Phys., 85 (1986) 5774. 12 CE. Klots, J. Chem. Phys., 83 (1985) 5854; Z. Phys. D., 5

(1987) 83. 13 CE. Klots, Z. Phys. D, 21 (1991) 335; 20 (1991) 105. 14 J.J. Breen, W.B. Tzeng, K. Kilgore, S. Wei, R.G. Keesee

and A.W. Castleman, Jr., J. Chem. Phys., 90 (1989) 19. 15 W.B. Tzeng, S. Wei and A.W. Castleman, Jr., J. Am. Chem.

Soc, 111 (1989)6035. 16 W.C Wiley and I.H. McLaren, Rev. Sei. Instrum., 26

(1955) 1150. 17 Galileo Electro-Optics Corporation, Sturbridge, MA.

J.L. Wiza, Nucl. Instrum. Methods, 162 (1979) 587. 18 A. Seko and H. Kobayashi, Rev. Sei. Instrum., 44 (1973)

400. 19 CE. Berry, Phys. Rev., 78 (1950) 597. 20 J.L. Franklin, P.M. Hierl and D.A. Whan, J. Chem. Phys.,

47(1967)3148. 21 M.A. Baldwin, P.J. Derrick and R.P. Morgan, Org. Mass

Spectrom., 11 (1976)440. 22 J. Farges, M.F. de Feraudy, B. Raoult and G. Torchet,

Surf. Sei., 106 (1981) 95. 23 G. Torchet, P. Schwartz, J. Farges, M.F. de Feraudy and B.

Raoult, J. Chem. Phys., 79 (1983) 6196. 24 E.J. Valente and L.S. Barteil, J. Chem. Phys., 79 (1983)

2683; 80 (1984) 1451, 1458. 25 CE. Klots, Nature, 327 (1987) 222. 26 CE. Klots, personal communication, (1989). 27 R.G. Keesee and A.W. Castleman, Jr., J. Phys. Chem. Ref.

Data, 15(1986) 1011. A.W. Castleman, Jr. and I.N. Tang, J. Chem. Phys., 62 (1975) 4576.

28 O. Echt, K. Sattler and E. Recknagel, Phys. Rev. Lett., 47 (1981) 1121.

29 D. Kreisle, O. Echt, M. Knapp and E. Recknagel, Phys. Rev. A, 33 (1986) 768.

30 O. Echt, M.C. Cook and A.W. Castleman, Jr., Chem. Phys. Lett., 135(1987)229.

31 S. Wei, K. Kilgore, W.B. Tzeng and A.W. Castleman, Jr., J. Phys. Chem., 95 (1991) 8306.

32 V.A. Rabinovich, A. Vasserman, V.l. Nedostup, L.S.

Veksler and T.B. Selover, Jr., Thermophysical Properties of Neon, Argon, Krypton, and Xenon, Springer Verlag, Berlin, 1988, p. 510.

33 S. Morgan and A.W. Castleman, Jr., J. Am. Chem. Soc, 109 (1987) 2867.

34 S. Morgan, R.G. Keesee and A.W. Castleman, Jr., J. Am. Chem. Soc, 111 (1989)3841.

35 S. Morgan, R.G. Keesee and A.W. Castleman, Jr., Proc 35th ASMS Conference on Mass Spectrometry and Allied Topics, May 24-29, 1987, Denver, CO, pp. 187-188.

36 M.T. Bowers, T. Su and V.G. Anicich, J. Chem. Phys. (1973)5175. L.M. Bass, R.D. Cates, M.F. Jarrold, N.J. Kirchner and M.T. Bowers, J. Am. Chem. Soc, 105 (1983) 7024.

37 R.G. Keesee and A.W. Castleman, Jr., J. Phys. Chem. Ref. Data, 15(1986) 1011.

38 S. Morgan and A.W. Castleman, Jr., J. Am. Chem. Soc, 109 (1987) 2867.

39 L.W. Sieck and P. Ausloos, Radiât. Res., 52 (1972) 47. 40 K.A.G. MacNeil and J.H. Futrell, J. Phys. Chem., 76

(1972) 409. 41 Z. Luczynski and H. Wincel, Int. J. Mass Spectrom. Ion

Phys., 23 (1977) 37. 42 A.J. Stace and A.K. Shukla, J. Phys. Chem., 86 (1982) 865. 43 X. Yang and A.W. Castleman, Jr., J. Am. Chem. Soc, 111

(1989) 6845. 44 S. Wei, Z. Shi and A.W. Castleman, Jr., J. Chem. Phys., 94

(1991) 3268. 45 N. Halberstadt and B. Soep, J. Chem. Phys., 80 (1984)

2340. 46 M. Castella, A. Tramer and F. Piuzzi, Chem. Phys. Lett.,

129(1984) 105; 129(1986) 112. 47 P.D. Dao, S. Morgan and A.W. Castleman, Jr., Chem.

Phys. Lett., I l l (1984)38. 48 D.H. Levy, C.A. Haynam and D.V. Brumbaugh, Faraday

Discuss. Chem. Soc, 73 (1982) 137. 49 K.W. Jucks and R.E. Miller, J. Chem. Phys., 88 (1988)

2196. 50 T.D. Klots, R.S. Ruoff and H.S. Gutowsky, J. Chem.

Phys., 90 (1989) 4216. 51 C.N. Lovejoy and D.J. Nesbitt, J. Chem. Phys., 90 (1989)

4671. 52 R.C Woods, CS. Gudeman, R.L. Dickman, P.F. Gold-

smith, G.R. Huguenim, W.M. Irvine, A. Hjalmarson, L.A. Nyman and H. Olofsson, Astrophys. J., 270 (1983) 583.

53 W. Wagner-Redeker, P.R. Kemper, M.F. Jarrold and M.T. Bowers, J. Chem. Phys., 83 (1985) 1121.

54 F.W. McLafferty and D.C McGilvery, J. Am. Chem. Soc, 102 (1980)6189.

55 J.D. Shao, T. Baer, J.C Morrow and M.L. Fraser-Monteiro, J. Chem. Phys., 87 (1987) 5242.

56 C.A. Deakyne, M. Meot-Ner, CL. Campbell, M.G. Hughes and S.P. Murphy, J. Chem. Phys., 84 (1986) 4958.

57 S.T. Graul and R.R. Squires, Int. J. Mass Spectrom. Ion Processes, 94(1989)41.

58 S. Wei, W.B. Tzeng and A.W. Castleman, Jr., J. Phys. Chem., 94 (1990) 6927.


59 P.D. Dao, S. Morgan and A.W. Castleman, Jr., Chem. Phys. Lett., 113 (1985) 219 (and references cited therein).

60 R.G. Keesee and A.W. Castleman, Jr., in E.R. Bernstein (Ed.), Atomic and Molecular Clusters, Elsevier, Amster-dam, 1990, pp. 507-550.

61 S. Leutwyler, U. Even and J. Jortner, J. Chem. Phys., 79 (1983) 5769.

62 K.H. Fung, H.L. Selzle and E.W. Schlag, Z. Naturforsch Teil A, 36(1981) 1257.

63 J. Jortner, in E.E. Koch, R. Haensel and C. Kunz (Eds.), Vacuum Ultraviolet Radiation Physics, Pergamon, Oxford, 1974, p. 291.

64 P.D. Dao and A.W. Castleman, Jr., J. Chem. Phys., 84 (1986) 1435.

65 (a) W. Kamke, B. Kamke, H.U. Kiefl and I.V. Hertel, Chem. Phys. Lett., 122 (1985) 356. (b) W. Kamke, B. Kamke, H.U. Kiefl, Z. Wang and I.V. Hertel, Chem. Phys. Lett., 128 (1986) 399.

66 B. Brutschy, P. Bisling, E. Rühl and H. Baumgärtel, Z. Phys. D., 5 (1987) 217.

67 T. Wada, K. Shinsaka, H. Namba and Y. Hatano, Can. J. Chem., 55 (1977) 2144.


69 E.W. Schlag, J. Grotemeyer and R.D. Levine, Chem. Phys. Lett., 190 (1992) 521.

70 D. Bahatt, U. Even and R.D. Levine, J. Chem. Phys., 98 (1993) 1.

71 S. Wei, J. Purnell, S.A. Buzza and A.W. Castleman, Jr., J. Chem. Phys., submitted.

72 J. Purnell, S. Wei, S.A. Buzza and A.W. Castleman, Jr., J. Phys. Chem., submitted.

73 S. Wei, J. Purnell, S.A. Buzza, R.J. Stanley and A.W. Castleman, Jr., J. Chem. Phys., 94 (1992) 9480.


The one dimensional photofragment translational spectroscopic technique: intramolecular clocking of energy redistribution for molecules falling apart l

Hyun Jin Hwang1^ Jennifer Griffiths, M.A. El-Sayed* Department of Chemistry and Biochemistry, University of California, Los Angeles, CA, 90024-1569, USA

(Received 1 July 1993; accepted 29 July 1993)

Abstract

The details of a simple modified form of a one dimensional photofragment translational spectroscopic technique is described which is used to determine the rate of energy distribution in a molecule falling apart. The apparatus utilizes lasers for photodissociation of the parent ion and the ionization of one of the fragments and a one dimensional time of flight mass spectrometer with a pulsed extraction electric field for data collection. This technique uses molecular rotation as an internal clock and the recoil velocity of the fragments as a monitor of the excess energy distribution prior to photodissociation. The method is applied to study the photodissociation dynamics of iodobenzene. The rate of energy distribution in the predissociative π,π* state at 304.67 nm is found to be « 23 kcal mol"1 ps"1.

Key words: One dimensional photofragment translational spectroscopy; Multiphoton dissociation-ionization; Intra-molecular clocking

1. Introduction

The combined use of high power pulsed laser as an ionization source and mass spectrometers to study molecules and ions dates back to the early 1970s. In 1970, Berezheskaya et al. [1] reported on a technique in which a Nd : glass laser was used to ionize and then dissociate diatomic hydrogen. It was stated that at the laser intensities used, diatomic hydrogen absorbed fourteen photons {hv = 1.18 eV) which led to the formation of its radical cation. Subsequent absorption of several

* Corresponding author. * Current address: Department of Chemistry, Kyung Hee University, Seoul, Korea. 1 Work done in partial fullfilment of the Ph.D. requirement of H.J. Hwang, University of California, Los Angeles, (1992).

more photons resulted in the dissociation of H^ into H + and H. Almost simultaneously Chin [2] reported multiphoton ionization-dissociation (MPID) of I2, D 2 0 , and CC14 using a pulsed ruby laser {hv = 1.785 eV). Both studies used time of flight mass spectrometers to analyze the charged fragments which were generated in the MPID process. Several other research groups com-menced study of the MPID of large molecules with high power lasers used in conjunction with mass spectrometers. Early contributions in the field include Boesl et al. [3], Letokhov and co-workers [4], Zandee and Bernstein [5], Fisanick et al. [6], and Lubman et al. [7].

Understanding intramolecular energy distri-bution in molecules and ions is vital to our under-standing of reaction dynamics. During the past few

266 H.J. Hwang et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 265-282

decades, at least three different groups have been involved in research aimed at this understanding. One group tried to describe unimolecular decom-position of ground state neutral molecules (by thermal activation). The second group was involved in understanding nonradiative processes of electronically excited neutral molecules. The third group was trying to describe the mass spectra of molecules by examining the unimolecular decomposition of initially electronically excited ions formed upon ionization by using high energy electrons or photons.

In studying thermal unimolecular decom-position, the coupling of internal motion is assumed to lead to very rapid intramolecular vibrational energy randomization before decom-position takes place (the basic assumption in the RRKM theory [8]). The dependence of the non-radiative [9] processes from electronically excited states on parameters such as the energy difference between the coupled states, the Franck-Condon factors, the type of the electronic states coupled by spin-orbit operations in intersystem crossing, and the vibrational or rotational quantum numbers have been [10] and are now being studied.

As a result of the efforts of the third group, statistical theories, such as the quasiequilibrium theory [11] (QET), constitute the framework of our understanding of dissociation rates of polyatomic ions. At the cornerstone of the QET stands the quasiequilibrium (or ergodic) hypothesis. This states that the rate of energy redis-tribution among all the internal degrees of freedom (electronic and vibration) is faster than the rate of ionic dissociation. Ionization with high energy electrons or photons (e.g. in photoionization) is believed to give rise to different electronically excited states which rapidly internally convert to produce the lowest electronic (ground) state of the ion. As has been postulated by the RRKM theory [8] in the case of unimolecular thermal reac-tions of neutral molecules, the coupling of internal motions is also assumed to be responsible for rapid intramolecular vibrational energy random-

ization within the ground electronic state of the decomposing parent ion.

The search for nonrandom dissociation and the limits of the applicability of statistical behavior is at the forefront of research on ionic and neutral systems alike. It is the statistical behavior that has tarnished the hope for accomplishing mode selective chemistry by using multiphoton vibrational excitation of ground state of neutral molecules with C0 2 and similar IR lasers.

For thermal reactions, the test of the statistical behavior is made by comparing experimental results with RRKM calculations of the A factors as well as theoretically fitting the pressure falloff dependence of the rate constant. Such methods cannot be applied to ionic decomposition, as their studies require low pressures for mass spectro-metric detection and to prevent ion/molecule inter-actions. However, comparisons [8,12] of the observed breakdown curves with those calculated using the QET and the RRKM theories have been used to test random or nonrandom behavior. Furthermore, by making this comparison at different temperatures one could test the validity of energy randomization (rapid intramolecular vibrational relaxation). For photochemical neutral unimolecular transformation, e.g. iso-merization, a laser was used to follow the change in real time of the isomerization-fluorescence quenching efficiency on the nano- and picosecond time scale in cold beams [13]. The effect of increas-ing internal energy (by changing the laser wave-length) on the rate of energy redistribution was studied [14] and the results were compared with the RRKM predictions. Unfortunately, in these calculations as well as in ionic fragmentation of the ions, assumptions were made concerning the structure and the vibration frequencies of the transition state that are not usually known.

7.7. Determination of energy distribution rates by mass spectrometry

Electron impact (El) excitation has been the most widely applied method in ionization-

HJ. Hwang et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 265-282 267

dissociation mass spectrometry in the past. The disadvantage of this method as it is used today lies in the fact that the residence time of the ions in the ion source of the mass spectrometer is on the microsecond time scale. Thus the observed mass spectrum is the result of processes occurring over the period between 10~l4-10~6s. Only if the ion dissociation is in the microsecond time range (the ion extraction times) will metastable peaks appear in the mass spectra and a knowledge of the dissociation rates of the precursor ion can be inferred. By modifying the ionization source, the study of metastables on the nanosecond time scale was made possible [15]. Even in this case, no knowledge of the rates of the preceding energy distribution processes can be obtained, only a limit can be set. These constraints are also present if the ionization is carried out by conventional photoionization laser techniques. It is the mass spectrometer detection that sets up this limit, whether it is magnetic sector or time of flight (TOF) with El or laser multiphoton techniques. Field ionization techniques extend the time of observing rapid ion rearrangement into the picosecond time domain [16]. This technique, however, cannot be used with multiphoton laser ionization sources and does not yield quantitative results. However, by use [17] of two pico-second lasers for sequential ionization, the change in the mass spectra with the time delay in between them gave indirect information of the time scale of energy redistribution within the parent ion.

1.2. Rapid molecular photodissociation and molecular beams

A great deal of understanding of dissociation dynamics has been achieved from studies that combined molecular beams with lasers in the photodissociation of molecules with repulsive excited state potentials. Since the pioneering work of Wilson and co-workers [18] on TOF photo-fragment spectroscopy, much attention has been paid to this technique as a method of obtaining

detailed knowledge on the photodissociation dynamics of isolated molecules [19]. Measure-ments of photofragment recoil speed distributions allows the determination of internal energy distributions [20-28] of photofragments (and often vibrational [29-31] or rotational [32] state distributions with sufficiently high resolution) via energy conservation. In addition, the vector properties of photofragment recoil, i.e., the angular distribution of the photofragments with respect to the electric vector of the photolysis light, give insight into the symmetry of the excited state of the excited molecule as well as the time scale of photodissociation [33-37].

Wilson's original scheme of the TOF method has been widely employed due to its universal applic-ability and superior resolution [18-30,36,37]. In this method, the recoil speed and angular dis-tribution are obtained by measuring arrival time distributions of neutral photofragments over a calibrated distance with angle resolved detection. Several recent variations of the TOF method have been developed which employ a simple TOF spectrometer with state-selective detection scheme using multi-photon ionization (MPI) [38-42]. In these "state selective photofragment spectro-scopy" techniques, photofragments are selectively ionized immediately after photodissociation and allowed to drift in field free conditions before being accelerated to the detector. Although this is a rather low resolution method which makes it difficult to accurately determine the translational properties of photofragments, their state selec-tivity to probe individual quantum states gives advantages of measuring various vector cor-relations between properties such as recoil speed and the quantum state of the fragment, between the recoil speed of a state-selected fragment and its direction, or between the recoil direction of a diatomic fragment and its rotation vector. Several groups [42-48] have recently examined such cor-relations and have been able to extract more detailed information on the nature of the photo-dissociation dynamics such as the internal energy (or internal state) correlation between two


fragments, the geometry or structure of the dissociating molecule, involvement of multiple potential surfaces, etc.

There are three important characteristic times that determine the observed results in the above experiments; the dissociation time (rdiss), the energy distribution time (rdist), and the rotation time (rrot). Three interesting types of limits can be identified. (1) rdis > rdist » rrot gives the statistical dissociation limit. This produces fragments of relatively small translational energy having a spatially isotropic distribution. (2) rdiss <C rdist <C rrot which is the case for rapid dissociation in which the excess energy goes into the kinetic energy of the fragments, which have a very aniso-tropic spatial distribution that is determined by the laser polarization direction and the direction of the absorption transition moment relative to that of the dissociation direction within the molecule. In this limit, the recoil velocity of the fragments is independent of its spacial distribution. (3) Tdiss ~ Tdist ~ rroti m this case, the recoil velocity of the fragment depends on its spacial distri-bution. This is the most useful limit in determining the rate of energy distribution. If a molecule is excited along the molecular axis and the x-lab direction and dissociates along the same direction (i.e. μ is parallel to the dissociation direction), those molecules with long rdiss will give fragments along an axis rotated from the x axis (i.e. will have small anisotropy parameter β) and will also have smaller recoil velocities since they would have more time for more effective energy distribution.

In the present paper we give the details of a modified simple technique of state-selective photo-fragment translational spectroscopy that has been developed in order to determine the rate of energy distribution for molecules whose dissociation time is comparable to energy distribution and molecular rotation times. The photodissociation dynamics of aromatic iodides are found to satisfy this criteria. We show that when the observed correlation between the spatial anisotropy and the recoil velo-city is transformed into a relation between time and internal energy via the time dependent rotation

correlation function of the molecule, the energy distribution rate can be determined. In section 2 the details of the experimental apparatus and data analysis are given. In section 3, the application of the technique to the photodissociation dynamics of iodobenzene is discussed and the rate of energy distribution prior to photodissociation at 304.67 nm is determined.

2. The technique

2.1. Basic idea of the experiment

The basic idea of the method is summarized as follows. An organic iodide is dissociated with a polarized laser and the iodine atoms formed are selectively ionized by a two-photon resonance-one-photon ionization by using the same or a different laser. After a short time following the dissociation ionization laser pulse(s) an extraction pulsed elec-tric field is applied and the TOF of I+ is deter-mined. The ionizing laser is fixed at either 304.67 or 304.02 nm for detecting the iodine atom in the ground or the excited 2P\/2 state. Since the iodine ions collected could be moving away from or towards the detector when the extraction field is applied, two distributions arriving at the detector at different times are observed. The separation between these two distributions is correlated to the recoil velocity of the iodine photofragments. In addition to determining the recoil velocity (for which one can determine the translational energy released), the spatial anisotropy factor β can be calculated from the intensity of the signal at any velocity within either distribution when the electric field direction of the photolysis laser is parallel and perpendicular to the detection axis.

2.2. Experimental apparatus

2.2.1. Vacuum chamber. Figure 1 shows a diagram of the experimental apparatus which employs a single stage acceleration TOF spectro-meter and an effusive sample source. The TOF spectrometer is designed to operate in two modes:


L,

Fig. 1. Diagram of the experimental apparatus. Shown are: A, the back electrode; B, the extraction acceleration electrode; C and D, mounting plates; E, the Einzel lens; F, the discrimination electrode; G, the microchannel plate detector. Lasers are focused in the center of the region between A and B.

(1) the translational spectrometer mode, and (2) the mass spectrometer mode, by biasing the back electrode, the acceleration electrode, and the center electrode of the Einzel lens to different voltages. The discrimination electrode and the two wing electrodes of the Einzel lens are held at ground potential in both operation modes via electrical connections to the vacuum chamber. Further details of the two operational modes will be given later.

The acceleration electrode and the back elec-trode are mounted to the mounting plate C via

ceramic spacers. The acceleration electrode consists of a cylindrical shape polished stainless steel electrode with a grid attached to each side. The grids are made from 90% transmissive stain-less mesh. The Einzel lens assembly is mounted in between the mounting places C and D. The Einzel lens is made of 1.5 in o.d. stainless tubing and consists of one center electrode and two wing electrodes. The discrimination electrode and the detector are mounted to the top flange. The dis-crimination electrode is a polished circular stain-less steel plate with a 6 mm diameter hole at the center. To prevent intrusion of the electric field from the detector through the discrimination pinhole, a grid is attached to the detector side of the discrimination electrode. The detector is a two stage microchannel plate detector. At the experimental conditions, the front microchannel plate is held at -1800 V.

The steel cross at the bottom of the vacuum chamber (see Fig. 1) is equipped with a sample vapor inlet line and two 1-in quartz windows for laser beam access. The entire assembly is pumped by a 4 in liquid nitrogen cooled diffusion pump. The pressure in the chamber, measured about 30 cm upstream from the sample inlet, is lower than 5 x 10~8Torr without the sample and rises to 1 x 10_6Torr when the sample is introduced.

The sample inlet consists of a foreline with two separate inlet lines equipped with a bellows valve and an adjustable leak valve. This allows for two samples to be used without contamination i.e. I2

for calibration and the sample of interest. The sample inlet inside of the chamber consists of a stainless steel needle covered with a 1 mm Teflon tubing tip to prevent arcing between the needle and the electrodes.

222. Optical layout. The optical layout in Fig. 2 is designed to control two pulsed laser beams, one at 304 nm and the other at a variable wavelength. The 304 nm laser pulse is generated by pumping a pulsed dye laser (Quanta-Ray Model PDL-1) with the second harmonic of a Nd : YAG (Quanta-Ray Model DCR-2A) and frequency doubling the

270 HJ. Hwang et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 265-282

Transient Digitizer

Digitized

Ion Signal IBM/PC Computer

High Voltage Pulse

Generator

Fused Dye Laser &

2nd Harmonic

Generator

2nd & 4th

Harmonic

Generator

Trigger Pulse

Nd:YAG User

Fig. 2. Block diagram of the optical and triggering experimental scheme. Pi and P2 are Wollaston prism polarizers, Rl is a 355 nm half-wave plate, R2 a 266 nm half wave plate, LI a 12.5 cm focal length plano-convex lens, L2 a 20 cm focal length plano-convex lens,

11-14 are irises.

output of the pulsed dye laser. The second laser pulse is either 266 nm from the fourth harmonic of the Nd:YAG or a wavelength obtained by Raman shifting the 266 nm light. Both pulses have a FWHM of about 20 ns. Each laser beam passes through a retardation plate and a polarizer, and is then focused into the center of the ionization region of the TOF spectrometer in between the back electrode and the first grid of the acceleration electrode. The intensities and the polarization direction are controlled by rotating the retardation plate and the polarizers, respec-tively, around the axis of the laser beam propaga-tion. Irises are used to both reduce the laser beam size (to about 4 mm) and to align the laser beam. The travel distances of the two laser beams is adjusted so that the 266 nm laser pulse arrives at the center of the ionization region 5 ns later than the 304 nm pulse.

The optical setup described above can be used to study the photodissociation of iodide molecules in

the 304 nm region as well as at 266 nm with the state selective capability for the iodine photo-fragments in either the ground I(2P3/2) or spin orbit excited Γ(2Ρι / 2) state. Two wavelengths, 304.02 nm and 304.67 nm with pulse energies from 20 to 50 uJ per pulse, are used to state selec-tively ionize the iodine atoms in the ground and excited states, respectively, via two photon reso-nance to the 4Ό°ι/2 and 2Ό°5/2 states respectively plus one photon ionization.

2.2.3. Electronics and timing. The TOF spectro-meter can be operated in two modes depending on the voltages biased to the back electrode, the acceleration electrode, and the Einzel lens. The first one is the translational spectrometer mode in which the recoil velocity distribution of the iodine atom photofragment can be determined by measur-ing the TOF distribution of I+. The second one is the mass spectrometer mode that is used to obtain the mass spectrum. Details of the experimental


conditions and procedure for each mode are described below.

In the translational spectrometer mode, the back electrode and all three electrodes of the Einzel lens are held at ground potential and a pulsed acceleration voltage of -1500 V with a pulse duration of 1.0 μ$ and a risetime of 40 ns is applied to the acceleration electrode at a variable delay time rd with respect to the timing of the laser pulse. A time zero, a focused laser pulse (304.02 nm or 304.67 nm) interacts with neat vapor of the iodide molecules at room temperature at the center of the ionization region in between the back electrode and the front grid of the acceleration electrode. As a result, the iodide molecules are photodissociated with one photon, and the resulting iodine atoms are state selectively ionized within the same pulse depending on the wavelength. During the delay time rd the ions spread out from their initial position. At time rd

the acceleration voltage is applied and the ions are accelerated towards the detector. When all the ions are inside of the acceleration electrode (between the front and rear grid) the acceleration voltage is turned oif to ensure field free drifting of the ions up to the discrimination electrode. After passing through the discrimination electrode, the ions are accelerated further and detected by the microchannel plate detector.

In mass spectrometer mode, the photoions are accelerated right after formation by the use of a CW acceleration field generated by biasing the back electrode to 1500 V and holding the acceleration electrodes at ground. The Einzel lens is used to focus the ions into the discrimination pinhole, thus avoiding the discrimination effects sought in the TOF mode.

Figure 2 also shows the scheme of triggering and signal processing. The voltage generator is triggered by a trigger pulse from the Nd:Yag laser. The timing of the trigger pulse is adjusted to make the timing of the laser pulse and the onset of the acceleration pulse identical when the delay time in the high voltage pulse generator is set to zero. The high voltage pulse generator then

sends an acceleration pulse to the acceleration electrode of the TOF spectrometer and a trigger pulse to the transient digitizer (Gould, Biomation 8100) at the same time. The ion signal from the microchannel plate detector is amplified by a lOOx preamplifier (Pacific Instruments, Model 2A50) and digitized as a function of time with a sampling time of 10 ns by using the transient digitizer. The digitized signal is transferred to an IBM/PC through a digital input/output board (Data Translation Inc., Model DT2817).

2.3. Presentation of the results

2.3.1. Overview of the methodology. We start with a brief overview in order to familiarize the reader with the basic physical concepts and methods utilized in the translational energy spectroscopy of photodissociation. The experimental apparatus and data manipulation will be described in greater detail later. Figure 3 shows a schematic representa-tion of the ion packet spatial distribution at dif-ferent times during the experiment. At time equal to zero, a focused linearly polarized laser pulse propagating along the x axis dissociates gaseous iodinated molecules in the ionization region. The resulting iodine photofragments are ionized within the same laser pulse via 2 + 1 resonance enhanced multi-photon ionization (REMPI), thus creating a small ion packet at the focal point of the laser. The photoions are allowed to spread out from their initial position under field free conditions for a delay time rd. The spatial distribution of the photoions at time rd is thus dependent on the recoil velocity imparted to the ions during the dis-sociation. At time rd the ion packet is accelerated by a pulsed electric field towards the detector. As a result, the photoions acquire different accelerations in the z direction that depend on their one dimensional positions along the direc-tion of the acceleration field (i.e. along the z axis) at time rd. The accelerated ions travel under field free conditions to the detector. Field free con-ditions are ensured by timing the acceleration pulse width such that it is turned off only after all


L 600

(d) vxy discrimination & detection time s rd ♦ TOF

(c) Acceleration off Y times τά -Μ.Ομβ

Vê = -1500 -> 0 V

(b) Acceleration on time s τά

VÊ = 0-> -1500 V

(a) Photolysis & ionlzation time s 0

Fig. 3. A schematic representation of the ion packet distribution and TOF during the state-selective photofragment spectroscopic experiment. A laser beam propagates along the x axis with polarization that makes an angle a with respect to the detection (z) axis. At point (a) ionization of the gaseous sample molecules occurs, (b) After a delay time rd the ions are accelerated to point (c) at which time the acceleration field is turned off (the acceleration field was on for 1 ^s) and the ions drift under field free conditions towards the detector, (d) At the detector a core of the ion distribution is sampled and detected.

the photoions are inside a field switching electrode. At the detector, a 6 mm pinhole selects out a core of the ion packet whose distribution is recorded as a function of arrival time.

The central point in this detection method is that it breaks down the initial recoil velocity of the photofragments v into two components vz

and vxy. The discrimination pinhole allows one to monitor the vz component directly. While the acceleration field converts the component vz to TOF, no external field alters vxy. Thus, the ions continue to spread out according to their vxy in the field free drift region before the detector. At the pinhole, photoions with high vxy are not detected. Typically, only those ions with vxy < 120-130ms-1 will be detected, which reduces the effort and increases the accuracy in separation of the angular and speed parts of the recoil distribution.

-600-400-200 0 200 400 600

\)v m/s l.<h

Kb)

•2

* 0 . 0 -

? He) 0^0.5-

0.0--600 -400 -200 0 200 400 600

vv m/s Fig. 4. Effect of the vxy discrimination for an isotropic (β = 0) recoil velocity distribution of iodine atoms at three different velocity components of equal probability, (a) A two-dimensional {vz,vxy) depiction of the vxy discrimination for three recoil velocities (v = 100, 250, and 500 ms - 1 ) given by the circles. The discrimination velocity νά (which discriminates against photoions such that those with vxy^vd are not detected) is also shown; (b) and (c) show simulated recoil velocity distributions of the iodine atoms in the vz domain with-out and with the vxy discrimination. As shown in (a) and (c), the higher the recoil speed, the stronger the vxy discrimination (i.e. the sharper the observed peak). When the recoil speed is lower than vd, (in our instrument υά = 100ms-1), there is no discrimination effect.

The effect of the vxy discrimination in the detection of the photoions is well illustrated in Figs. 4 and 5 for the two cases of an isotropic recoil distribution (no angular dependence) and an anisotropic recoil distribution (including angular dependence), respectively. As shown, the three dimensional recoil distribution is monitored through the vz component of the distribution. Figure 4(a) shows the two dimensional view of an isotropic distribution of ions at three different


-600-400-200 0 200 400 600 vv m/s

Fig. 5. The effect of the vxy discrimination on the observed value of ß for anisotropic recoil velocity distributions of iodine atoms with a single recoil speed (v = 500ms"1); (—) and (· · ·) corespond to simulated recoil velocity distributions in the vz

domain at laser polarizations of a = 0° and 90° with respect to the detection axis, respectively. The discrimination criteria is shown with vertical lines marked with arrows, specifying the onset of the discrimination at two values of vz. Photofragments with their vz between these two values of vz are not detected in the presence of the vxy discrimination.

recoil speeds (v = 100, 150, and 500ms"1). With-out the vxy discrimination, the one dimensional projection of this distribution results in a broad distribution in the vz domain, as illustrated by Fig. 4(b). In reality, the dissociation of molecules from linearly polarized light is often anisotropic. Thus, the recoil velocity distribution depends on the angle of observation with respect to the direction of the electric vector of the dissociating light. The observed velocity distribution is a function of both the degree of anisotropy and the angle of observation, and without the vxy

discrimination is not easily separated into its constituent angular and speed components. With the discrimination pinhole it is possible to independently monitor the speed and angular parts of the distribution. As shown in Fig. 4(c), use of the discrimination pinhole (Fig. 4(a) shows the cutoff for the vxy discrimination) sharpens the distribution in the vz domain as photoions with low vxy and thus small θζ can only be detected. The

observed peak position in the vz domain is thus very close to the values of v. In addition, as shown in Fig. 5, the angular dependence can be monitored by simply examining the polarization dependence of the intensity of the photoion signal rather than the broad lineshape resulting from the full one dimensional projection, especially when v is high.

Using the time delayed method and the dis-crimination pinhole outlined above, the recoil velocity component of vz of the dissociating molecules can be monitored by measuring the TOF of the photoions. As mentioned earlier, the acceleration, and thus kinetic energy, accessed by the photoions during acceleration depends on their one dimensional position along the z axis at the time of acceleration, which in turn is directly related to their displacement from the initial ionization position by νζτά due to the recoil velocity of the photofragments. Thus, the measured TOF of the ions is a direct probe of their recoil velocities vz. The details of the TOF calculation and the conversion of TOF to vz are given below. In addition, the con-version from laboratory to center of mass (CM) frame is described, as well as the method of deconvolution of the recoil and speed parts of the velocity distribution. The conversion from TOF to the energy domain in order to investigate the energy partitioning between internal and translational modes of the fragments is discussed.

2.3.2. Calculation of the TOF distribution. In general, the TOF in the laboratory frame of a photoion is dependent on the component of the initial velocity of the photofragment (i.e. the recoil velocity) along the detection axis as well as the acceleration due to the electric field, £a . For a photoion of mass m and charge q, the kinetic energy after acceleration U is given by

U= U0- qE9ravz + Ur (i;

where UT is the kinetic energy of the photofragment prior to acceleration which is dependent on the


recoil velocity vz and is given by

UT = mv2/2 (2)

and i/0 is the kinetic energy accessed by a photoion

with vz = 0. The term -qE2Lravz takes into account the direction of recoil along the detection axis i.e. if vz is negative (the detected fragment recoils away from the detector), the photoion spends more time in the acceleration field and thus acquires more kinetic energy than those ions with positive vz

(recoil towards the detector). The kinetic energy after acceleration U can be

related to the TOF of the photoion. The TOF of the photoion with respect to the onset of the acceleration voltage is given by the sum of the flight times spent for acceleration ( r a ) , field free drift (r f) , and the detector preacceleration (Τά). Thus,

TOF(*;z, r d£a , i/o) = T, + Tr+Td (3)

where

ra = (2m)1/2(â)-,(t/I/2±t/r,/2) (4)

7}=(m/2)1/2Z)fC/-1/2 (5)

Td = (2mY'2Dd(qVd)-l[(U+qVdy/2 - Uxl2)] (6)

Dd is the detector preacceleration distance, Z)f is the field free drift distance (from the first grid of the acceleration electrode to the shielding grid of the discrimination pinhole), and Va is the elec-tric potential change from the shielding grid of the discrimination pinhole to the first plate of the microchannel plate detector. The plus and minus signs in Ta correspond to recoil directions away from and towards the detector, respectively.

23.3. Calibration of the instrument. One needs to measure the TOF of the photofragment with respect to the onset of the acceleration voltage pulse. Although triggering of the transient digitizer is synchronized with the acceleration voltage pulse, the TOF measured by the transient digitizer is found to be shorter due to an intrinsic delay Δ Γ in triggering the timing circuit of the

transient digitizer. The value of Δ Γ can be accurately determined by taking the TOF mass spectrum in the mass spectrometer mode. At this condition, the TOF with respect to the onset of the acceleration voltage pulse must be proportional to the square root of the mass m of the ion: TOF(ra) = Araly/2. Thus, the measured time of flight (TOFm) must be shorter than the TOF defined in Eq. (3) by ΔΓ.

TOFm = Am1/2 - Δ Γ (7)

Equation (3) contains many apparatus constants which must be accurately known to determine the lab. velocity distribution from the TOF distribu-tion. While Dr, Z)d, and Va are measured directly, £a, i/o, and a small systematic error associated with the delay time measurements (eT, on the order of 0.01 /xs) are determined by studying the delay time dependence of the TOF peak position of the iodine ion produced from photodissociation of I2 at 304.02 nm and a laser polarization angle a — 90°. At this wavelength, I2 dissociated into I and Γ by a perpendicular transition and only I* atoms are ionized by REMPI and detected. Since I2 does not have any internal modes once dissociated, all avail-able energy is released in translation. Thus, the center of mass recoil velocity of the dissociated iodine atom is accurately known (1114ms-1) by the energy conservation expression.

Ε{ = Ην + Ε[η1(ρ)-Ο°0(1-1)-Ε$ο (8)

where Et is the translational energy release (37.6 kcal mol-1), hv is the photon energy, E[nt(p) is the initial internal energy of the parent I2

(0.9 kcal moP1), Z>(j(I-I) is the dissociation energy of the ground state I into two ground state iodine atoms at OK (35.6 kcal mol"1) [49], and Eso is the spin-orbit excitation energy of the iodine atom (21.7 kcal mol-1). For the value of £int(p) w e u s e d the average thermal internal energy at 298 K calcu-lated by using the known moment of inertia and vibrational frequency of I2 [49] assuming a Boltzmann distribution.

Figure 6 shows a TOF spectra of I2 at various delay times taken in the translational spectrometer


555 S 2 I f&udih

TOF, \i%

Fig. 6. Delay time dependence of the TOF spectra of resonantly ionized Γ atoms produced from the photodissociation of I2 at 304.02 nm (30//j per pulse) and laser polarization of a = 90°. The numbers shown are the measured delay time in micro-seconds. From this dependence and the knowledge of the recoil velocity of P , the different parameters used to calibrate the TOF instrument are calculated. I2 is then replaced by iodobenzene and the same parameters are used to determine the time of flight distribution of its dissociated iodine atoms.

mode. The two sharp peaks observed at non-zero delay time correspond to iodine ions with their recoil direction towards (the peak at longer TOF corresponding to positive v2) and away from (the peak at shorter TOF corresponding to negative vz) the detector. While these two peaks collapse into each other at zero delay, they move away from each other at longer delays. This delay time dependence of the TOF peak position is fitted to the TOF Eq. (3) by nonlinear least squares method to determine 2sa, t/0, and eT. In this fitting, it is assumed that the position of the TOF peaks correspond to vz equal to plus and minus CM recoil velocity of the iodine atom (1114ms-1). The parameter eT takes into account a small systematic error in the delay time rm and is defined as

Ta = rm + er

Since the value of U0 changes when the laser is realigned due to small changes in the laser focus position, it is independently determined for each measurement.

2.3.4. Conversion of TOF distribution to the velocity distribution in the vz domain. The TOF distribution obtained by the preceding analysis

is a function of the recoil velocity vz in the lab. frame. In order to determine the velocity distribution in the v2 domain, fz

a{vz) from the measured TOF distribution ha(JO¥) the following relation is used:

fza(vz) = /*Q(TOF)[dTOFK)/d*,z (9)

where a is the laser polarization angle with respect to the z detection axis. As will be shown below, measuring ha(TOF) and thus fz{vz) at two polarization angles, a = 0° and 90° is sufficient to uniquely determine the photo-fragment recoil velocity distribution in the lab. frame including both the speed and angular parts.

We first derive a velocity distribution in the vz

domain fza(vz) resulting from a monoenergetic

velocity distribution with a single speed v (a similar treatment in the absence of the vxy dis-crimination can be found in ref. 50). The velocity along the z axis vz is given by (see Fig. 7 for the coordinate system)

V7 — 1>COS07 (10)

where Θζ is the lab. recoil angle with respect to the z detection axis. Thus, dvz = -vsm6zd9z. We need to find the probability of detecting photofragments at a lab. recoil angle of 9Z regardless of the azi-muthal angle φζ when the laser polarization angle is a. The probability of recoiling into this solid

I—^k * *>υ

VI

Fig. 7. Coordinate system for the photofragment recoil velocity distribution resulting from photodissociation via a linearly polarized light. E is the electric vector of the dissociating light and v is the recoil vector. The experiment monitors the com-ponent of v along the z axis (vz = UCOS0Z) via the TOF of the photofragments.

276 HJ. Hwang et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 265-282

angle Ω is given by

Ρα(η)άΩ = sin0zd0z [ * άφζΙ(θ) (11)

where θ is the lab. recoil angle with respect to the electric vector of the dissociating light. By using the relation cos Θ = sin a sin θζ sin φζ + cos a cos θζ and integrating, we obtain

Pa(ü)aÜ = ^ [1 + ßP2(cos a) P2(cosθζ)\ sin θζ άθζ

(12)

Substituting in for άθζ and θζ gives the following relation.

fza(vz)dvz = (2^)-1[l +ßP2{cosa)P2(cosvz/v)]dvz

(13)

Without the ^ discrimination, the value of vz in the above equation can range from - v to D (see Figs. 2 and 3). In our method, however, photofragments with height vxy are not detected. The criteria for discrimination is given by

vxy = νύηθζ = {v2 - v2z)

x/2 < vd(vz)

= Rd/[rd + TOF'(vz)] (14)

where vd is the discrimination velocity (120-130ms"1 for I+ depending on the value of vz) determined by the radius of the discrimination pinhole Ra and the time lapse to reach the discrimination pinhole after the photodis-sociation. The flight of time TOF' of the photoion to reach the discrimination pinhole after the acceleration electric field pulse is 21-25 μδ for the I+ in the typical experiment. This flight time is slightly shorter than the flight time to reach the detector and depends on vz. From the relation shown in Eq. (14), the range of vz when the vxy

discrimination is present is given by the following relations:

-v^vz<-{v2-vl)xl2

and

(v2-v2d)

l/2<vz*:v (15)

Equation (13) can be generalized to an arbitrary recoil velocity distribution by integrating from ^min t o max a n d weighting with the lab. speed distribution g(v):

(Umax

[g(v)/2v][\+ß(v)P2(cosa)

x P2{cosv/vz)] (16)

The integration ranges are determined using Eqs. (10) and (15) and are as follows:

Vmin = ( c ^ J m i n= |cos<?z|max

= N ( 1 7 )

and

v =(-^-) = N max \cosezJmax |cos0z|min

= pt=(v2 + vl)^ (18)

To observe a value of vZ9 the recoil speed v must be equal to or higher than \vz\. The lower limit of the integration range vmin = \vz\ corresponds to the case where the photofragments recoil along the z detection axis, giving |cos0z|max = 1. For the case where v > \vz\, the lab. recoil angle θζ must deviate from 0° or 180° according to Equation (10) but it cannot deviate too much due to the vxy discrimination as is shown in Equation (18). Thus there should be a mini-mum value of |cos0z| which is given by

cmin = iî^2 + ^ ) - 1 / 2 .

2.3.5. Determination of the anistotropy parameter β and the relation between the lab. and CM frames. The method for determination of the anisotropy parameter in photodissociation from experimental photofragment angular distributions has been well discussed in the literature. The basics of the ideas are, however, outlined here. For photodissociation resulting from a one-photon electric dipole transition induced by a linearly polarized laser, the normalized CM angular distribution of the photofragments is given by the

HJ. Hwang et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 265-282 277

following relation [16,28-30]:

/CM (0CM) = W 1 ! 1 + / W > 2 ( C O S 0 C M ) ] (19)

where 0CM is the angle between the electric vector of the dissociating light and the CM recoil direction of the photofragments and P 2 ( C O S ^ C M ) =

(3cos20CM - l) /2 is the second order Legendre polynomial. If the dissociation is prompt (short compared to the rotational period of the molecule), the angular distribution reduces to cos2 9CM and sin2 0CM dependence for parallel and perpendicular, respectively. Thus the anisotropy parameter can range from 2 for a prompt parallel dissociation to - 1 for a prompt perpendicular dissociation. However, ßCM can be reduced from either of these limits depending on the angle X between the dissociation bond axis and the transition dipole moment of the parent molecule μ, as well the lifetime of the excited state. As the dissociation time becomes longer, the absolute value of /5CM is reduced while retaining its sign because the initial alignment of the dissociation bond axis with respect to the electric vector of the dissociating light, and thus that of the photo-fragment recoil direction, is lost due to the vibrational and rotational motions of the excited parent molecule during the finite dis-sociation time. More details of the relation between depolarization of the anisotropy parameter and the dissociation lifetime will be presented later.

In order to fully characterize the photo-fragment recoil velocity distribution in the CM frame the CM speed distribution £CM(^CM) must be included with the angular distribution as follows:

/CM(^CM^CM) = ( 4 π ) - 1 [1 + ßcM^cu)

x P2(coseCM)]gCM(vCM) (20)

Note that the anisotropy parameter is expressed as a function of vCM. If all the photofragments are produced from a single transition and with the same dissociation time or if photodissociation occurs with different dissociation times that result

in the same speed distribution, the anisotropy parameter would be constant and not dependent on vCM. However, in reality photodissociation occurs with a range of dissociation times and dif-ferent speed distributions. As such, Eq. (20) is thus a more general expression and takes into account possibilities such as photodissociation from a mixed transition or predissociation from a bound state resulting in a broad range of dissociation times.

It is necessary to convert the measured recoil velocity distributions in the vz domain (Eq. (20)) to the CM frame. We assume that the photo-fragment recoil velocity distribution in the lab. frame can be expressed as analogous to that in the CM frame:

AvJ)=fr[\+ß(v)P2{cös0)]g(v) (21)

where Θ is the lab. recoil angle with respect to the electric vector of the dissociating light and β(υ) and g(v) represent the anisotropy para-meter and speed distribution measured in the lab. frame, respectively. This equation was found to be valid within the experimental parameters. Note that the angle Θ will coincide with the laser polarization angle a with respect to the z detection axis if the photofragments are detected with an infinitely small detection solid angle. However, due to the finite size of the detector as determined by the diameter of the pinhole, photofragments will be detected with a range of Θ values.

2.3.6. Deconvolution o/g(v) and β(ν) from f"(vz). Using the relations derived earlier for fz

a(vz), g(v), and ß(v) (Eq. (16)) the observed distribution fz

a{vz) can be deconvoluted to g(v) and β(ν). With-out knowing the shape of g(v) and β(ν) we solve Eq. (16) assuming that the integrand is linear within the integration ranges given by Eqs. (17) and (18), which gives the following result:

fzaM = [g(v)/2v][l+ß(v)P2(cosa)P2(vz/v)}Av

(22)


where the values of v and Av are related to the observed value of vz according to Eq. (22):

(«min + «max) | « z | ( l + C m i n ) V =

Δν = ümax - vmin =

IC

c

(23)

(24)

Obviously, this linear approximation is valid only if the integration range Av is narrow, which is the case when v is high enough compared to the discrimination velocity νά so that the vxy

discrimination is strong. Substituting for v and Av into Eq. (22) we get

J?M =fzM(vz)[\ + ßz(vz)P2(cosa)P2(vz/v)]

(25)

where

f!AM = g(v)(\-cMn)/(\ + cnia)

and

ßzM=ß(v)l6CZia/(\ + C1Bin)2-{]

(26)

(27)

Note that the value of v in Eqs. (26) and (27) is related to the observed value of vz according to Eq. (23).

Equations (26) and (27) are the key equations. If one determines fz

M{vz) and ßz{vz), g(v) and ß(v) can be automatically determined using these equations along with Eqs. (17), (18) and (23). Note that the term ßz{vz)P2(cosa) vanishes when a is the magic angle 54.74°, giving a magic angle distribution of fz

M{vz). Thus the speed distribution g(v) can be determined by measuring the TOF distribution at the magic angle and using these equations. Instead of measuring fz

u(vz) directly, we measure fza(vz) at a = 0° and

90° and use the following relation to obtain

/ z > z ) H [ / . ° > z ) + / Z9 0 > z ) ] (28)

This is convenient because ßz(vz) and thus ß(v) can be determined according to the following relation:

/ z > z ) - Λ 9 ° > ζ ) Pz\vz) j -o"/ \ , /·90°/ \ (29)

In summary, the speed distribution g(v) and the velocity dependence of the anisotropy parameter β(ν) can be determined by measuring the velocity distribution in the vz domain (i.e. the TOF distribution) at two laser polarization angles a = 0° and 90°.

2.3 J. Transformation of the photofragment recoil velocity distribution to the energy domain. Since we are interested in the energy partitioning between the internal and translational modes of the photo-fragments, we need to examine the photofragment distribution as a function of translational energy release Et. Here, Et is the translational energy of both the I and radical R photofragments. The g(v) and β(ν) distributions can be transformed into the energy domain as G(Et) and ß{Et) according to the following relations:

E{ = [mi(mi + mR)v2]/2mR

G(Et) = [mR/mi(mi+mR)]\g{v)/v]

(30)

(31)

where mj and mR are the mass of the iodine atom and the radical, respectively. Once the translational energy release distribution is determined for each electronic state of the iodine (I(2P3/2) or 7*(2P1/2)), the internal energy distribution of the radical Eini is calculated using the following energy conservation relations:

£avi = hv - D°0 + £int(p) =Et + Eini

E1* τρ r? 17* 17*

(32)

(33)

where £avl and £^1 are the energies available to be partitioned into internal and translational modes for the I and Γ dissociation channels respectively, £int(p) is the internal energy of the parent molecule at room temperature, DQ is the dissociation energy of the ground state parent molecule to form the ground state radical and iodine at room temperature, and Eso is the spin-orbit excitation energy of the iodine atom (£so = 21.7 kcal mol"1).


3. Clocking of intramolecular energy transfer

3.1. Converting ß into time

Two of the central questions in the study of photodissociation processes are what is the life-time of the excited state of the molecule in the absence of collision? and how, during this lifetime, is the electronic energy of excitation redistributed to vibrational, rotational, and translational energy? One handle on trying to answer these questions is the anisotropy parameter. As mentioned earlier, ß can be diminished from its limiting values for a prompt dissociation by either a mixed polarization character of the absorption transition or by rotation of the molecule during dissociation. It is on this latter relationship between the anisotropy parameter and rotational period of the molecule which we will focus in order to relate ß to the dissociation lifetime of the molecule. Many authors [33-36] have treated the time-anisotropy parameter correlation assuming a purely rota-tional depolarization. Yang and Berhson [36] have outlined for arbitrary molecules the dependence of ß on the orientation of the transition dipole with relation to the dissociation direction in the molecule framework as well as the rotational correlation functions for rotating molecules. We consider here the specific case of a symmetric top molecule. In this model, the vibrational motion is ignored although excitation of the vibrational modes such as in-plane or out-of-plane C-I bending modes could cause reduction of ß. The value of the anisotropy parameter at time /, /?(/), can be related to the rotational correlation function for a symmetric top molecule with the transition dipole along the symmetry axis as follows [36]:

ß(t)= ß0(Dlo(t)) (34)

(Z>o,o(0) is ^ e rotational correlation function for which the bracket denotes the average over the rotational motion of the molecule. The ensemble average depends on the temperature T and the

asymmetry parameter b = (I-Iz)/Iz, where / is the perpendicular moment of inertia and Iz is the moment of inertia around the figure axis of the symmetric top. Yang and Berhson [36,37] add a second average of the correlation function over the lifetimes of the excited state. However, we exclude this average as our results suggest that the anisotropy parameter is dependent on the translational energy release, implying that dissocia-tion events take place at different dissociation times which results in a range of translational energy release. Thus, the value of ß observed at each v is more closely related to the dissociation time t than the average dissociation lifetime r. ß0 is the limiting value of ß in the case of instantaneous dissociation and is given by [34-36]:

ß0 = 2P2(cosX) (35)

where χ is the angle between the direction of dissociation and the transition dipole moment.

3.2. The system: iodobenzene

The A band of the alkyl iodide is an absorption continuum in the 200-300 nm [51] range arising from the transition of a nonbonding electron of iodine to a a* molecular orbital [52]. As observed in various studies, the σ* <— n transition is followed by a very fast dissociation along the repulsive state [37,53], and results in production of a ground state alkyl radical and a ground state I(2P3/2) or spin-orbit excited state Γ ^ Ρ ^ ) iodine with the majority formed in the Γ state [22,54] and a large fraction of the available energy released in translation [22,25-27]. According to the orbital theory of Mulliken [52], the σ* <— n transition is composed of three overlapping transitions from the ground N state to the repulsive 3 Q b 3Q0, and lQ] states in order of increasing energy. Mulliken predicted that the 3Q0 <— N transition is polarized parallel to the C-I bond axis and dissociates to Γ while the 3Q! <— N and XQ\ <— N transitions are polarized perpen-dicular to the C-I bond axis and dissociate into I.


-1200 -800 -400 Ô 400 800 1200 υ,, m/s

0.0 J - l 400 , 800

υ, m/s 1200

Fig. 8. (a) The velocity distribution Aa(TOF) measured for resonantly ionized I atoms produced from the photodissocia-tion of iodobenzene at 304.76 nm at two laser polarization angles a = 0° and 90° with respect to the detection axis. The distribution clearly shows two overlapping recoil distributions; a sharp distribution at approximately ±770ms"1, and a broader distribution in the ±200-600 m s- 1 range. The small peaks at vz = ± 1200 m s - 1 are from photodissociation of I2 which was used for determining the instrumental parameters, (b) The velocity dependence of the anisotropy parameter β(ν) and g(v) showing the deconvolution of g(v) and β(ν) into high velocity (—) and low velocity ( · · · ) distributons. This shows that while β is independent of vz for the high velocity sharp distribution, it decreases with vz for the broad lower velocity distribution. In the latter case, the determination of the rate of energy distribution prior to dissociation becomes possible.

Since alkyl iodides photodissociate in a time faster than rotation time, this technique is able but to set a limit on the energy distribution rate. In order to slow down the photodissociation process, we used molecules in which the «,σ* repulsive state is mixed with its stable non-dissociative states. In aromatic iodides, the «,σ* absorption of the carbon iodine system is observed in the same energy range as the different π,π* (Lb and La) states of the aromatic system. Excitation to these states is expected to dissociate

Fig. 9. The time dependent rotation correlation function: from a knowledge of the moment of inertia of iodobenzene and the temperature, the time dependent rotational correlation function can be calculated [50] as a function of time scale /* = t/rrot = t(I/kT)~^2, where / is the moment of inertia around the axis perpendicular to the figure axis of the symmetric top molecule, k is the Boltzman constant, and T is the temperature of the parent molecule.

the C-I bond with a slower rate than excitation to the alkyl iodide type of n, σ* state.

Figure 8(a) shows the velocity distribution in the vz domain for the I(2P3/2) from the photo-dissociation of iodobenzene at 304.67 nm with a = 0° (top) and 90° (bottom). Figure 8(b) gives the deconvoluted recoil velocity g(v) for the high velocity (sharp) and the lower velocity (broad) distributions. In the same figure, the value of β for each velocity is calculated and plotted. The sharp distribution at high velocity is very similar to that observed for alkyl iodides and is assigned [55] to iodine atoms produced in the photo-dissociation of molecules absorbing to the η,σ* repulsive state of the C-I system. Consistent with rapid dissociation behavior, this distribution has a large average translational energy release (high recoil velocity) and a β value that is independent °f recoil (s e e ßh(v) in Fig· 8(b)). The broad dis-tribution in Fig. 8(a), however, seems to be for atoms having smaller recoil velocities and with ß values that are observed to depend on their value of the velocity (see ß\ (v) in Fig. 8(b)). Atoms with small ß value (i.e. for iodine atoms dissociated from molecules that have under-gone more rotation prior to dissociation) also have small recoil velocity (since they are

H.J. Hwang et al.lint. J. Mass Spectrom. Ion Processes 131 (1:

\ I(2/>3/2)/C6H5I

20 - j 304.67 nm 'S \ Low E. channel

^ J <-i 1 •«ion

5 I "**': -I '"'-'<

O H — i — i — i — · — i — i — i — i — i — i — « — i — i — i — I

0.0 0.5 1.0 1.5

i, ps

Fig. 10. The time dependence of the translational energy released during the dissociation of iodobenzene at 304.67 nm vs. time. The conversion of v vs. β into translational energy (Et) vs. time is carried out from the results shown in Fig. 8(b) and Fig. 9. Since Et reflects the internal energy of the molecule prior to dissociation (due to energy conservation), this figure shows the time dependence of the internal energy of iodoben-zene prior to its photodissociation at 304.67 nm. This depen-dence is almost linear, with a slope of 23kcalmor1 ps~l, which is the excess energy distribution rate at this wavelength.

produced from molecules that have had more time to distribute their excess energy prior to dissociation). These broad distributions are assigned [55] to iodine atoms dissociating from molecules that absorb to the predissociative π, π* states.

Figure 9 shows the plot of the time dependent rotational correlation function (which is equal to ß/ßo) as a function of time (scaled by rrot) for iodobenzene. From this figure and the different values of ß for different i>recoil (Fig. 8(b)) a plot of time vs. translational energy is obtained and shown in Fig. 10. From the slope of this plot, the rate of energy distribution is determined to be ISkcalmoP1 ps"1. This is indeed rapid as assumed in the RRKM. Thus for molecules or ions with not too large excess energy and transition states having a very slight minimum on the top of their activation barrier, complete randomization of the excess energy is expected to take place.

Acknowledgments

The authors would like to thank the National Science Foundation for its support.

265-282 281

References

1 N.K. Berezhetskaya, G.S. Vornov, G.A. Delone, N.B. Delone and G.K. Piskova, Sov. Phys. JETP, 31 (1970) 403.

2 S.L. Chin, Phys. Rev. A, 4 (1971) 992. 3 U. Boesl, H.J. Neusser and E.W. Schlag, J. Chem. Phys., 72

(1980) 4327. 4 (a) V.S. Antonov and V.S. Letokhov, Appl. Phys., 24

(1980) 89. (b) V.S. Antonov, V.S. Letokhov and A.N. Shibanov, Appl. Phys., 22 (1980) 293.

5 L. Zandee and R.B. Bernstein, J. Chem. Phys., 71 (1979) 1359.

6 G.J. Fisanick, T.S. Eichelberger IV, B.A. Heath and M.B. Robin, J. Chem. Phys., 72 (1980) 5571.

7 D.M. Lubman, R. Naaman and R.N. Zare, J. Chem. Phys., 72 (1980) 3034.

8 For review see: (a) L. Oref and B.S. Rabinovitch, Ace. Chem. Res., 12 (1979) 166. (b) W. Forst, Theory of Unimolecular Reactions, Academic Press, New York, 1973.

9 For a review see: P. Avouris, W.M. Gelbart and M.A. El-Sayed, Chem. Rev., 77 (1977) 793.

10 E. Riedle, H.J. Neusser and E. Schlag, J. Chem. Phys., 75(9) (1981)4231.

11 H.M. Rosenstock, M.B. Wallenstein, A.L. Wahrhaftig and H. Eyring, Proc. Natl. Acad. Sei. U.S.A., 38 (1952) 667.

12 PJ . Robinson and K.A. Holbrook, Unimolecular Reactions, Wiley-Interscience, London, 1972.

13 J.W. Perry, N.F. Scherer and A.H. Zewail, Chem. Phys. Lett, 103(1983) 1.

14 L.R. Khundkar, R.A. Marcus and A.H. Zewail, J. Phys. Chem., 87 (1983) 2473.

15 O. Osberghaus and C. Ottinger, Phys. Lett., 16 (1965) 121.

16 (a) H.D. Beckey, H. Hey, K. Levsen and G. Tenschert, Int. J. Mass Spectrom., Ion. Phys., 2 (1969) 101. (b) P.J. Derrick in A Maccoll (Ed.), Mass Spectrometry, International Reviews in Science, Physical Chemistry, Series II, Vol. 5, Butterworth, London, 1975, pp. 1-46. (c) K. Levsen, Fundamental Aspects of Organic Mass Spectrometry, Verlag Chemie, Weinheim, West Germany, 1978, p. 312.

17 D.A. Gobeli, J. Morgan, R. St. Pierre and M.A. El-Sayed, J. Phys. Chem., 88 (1984) 178.

18 (a) G.E. Busch, R.T. Mahoney, R.I. Morse and K.R. Wilson, J. Chem. Phys., 51 (1969) 449. (b) 51 (1969) 837. (c) G.E. Busch, J.F. Cornelius, R.T. Mahoney, R.I. Morse, D.M. Schlosser and K.R. Wilson, Rev. Sei. Instrum., 41 (1970) 1066.

19 For reviews see: (a) R. Bersohn, IEEE J. Quantum Electron., 16(1980) 1208. (b) S.R. Leone, Adv. Chem. Phys., 50 (1982) 255. (c) R. Bersohn, J. Phys. Chem., 88 (1984) 5145.

20 A.M. Wodtke and Y.T. Lee, in M.N.R. Ashforld and J.E. Baggott (Eds.), Molecular Photodissociation Dynamics, Royal Society of Chemistry, London, 1987, pp. 31-59.


21 (a) G.E. Busch and K.R. Wilson, J. Chem. Phys., 56 (1972) 3626. (b) 56 (1972) 3838. (c) 56 (1972) 3655.

22 S.J. Riley and K.R. Wilson, Faraday Discuss. Chem. Soc, 53 (1972) 132.

23 (a) J.H. Ling and K.R. Wilson, J. Chem. Phys., 63 (1975) 1010. (b) R.D. Clear, S.J. Riley and K.R. Wilson, J. Chem. Phys, 63 (1975) 1340.

24 (a) M. Kawasaki, S.J. Lee and R. Bersohn, J. Chem. Phys., 66 (1977) 2647. (b) A. Freedman, S.C. Yang, M. Kawasaki and R. Bersohn, J. Chem. Phys., 72 (1980) 1028.

25 R.K. Sparks, K. Shobatake, L.R. Carlson and Y.T. Lee, J. Chem. Phys., 75 (1981) 3838.

26 (a) G.N.A. van Veen, T. Bailer, A.E. de Vries and N.J.A. van Veen, Chem. Phys., 87 (1984) 405. (b) G.N.A. van Veen, T. Baller, A.E. de Vries and M. Shapiro, Chem. Phys., 93 (1985) 277.

27 (a) M.D. Barry and P.A. Gorry, Mol. Phys., 52 (1984) 461. (b) C. Paterson, F.G. Godwin and P.A. Gorry, Mol. Phys., 60 (1987) 729. (c) F.G. Paterson, F.G. Godwin and P.A. Gorry, Mol. Phys., 60 (1987) 729.

28 (a) T.K. Minton, P. Felder, R.J. Brudzynski and Y.T. Lee, J. Chem. Phys., 81 (1984) 1759. (b) D. Kranjovich, L.J. Butler and Y.T. Lee, J. Chem. Phys., 81 (1984)3031. (c) T.K. Minton, G.M. Nathanson and Y.T. Lee, J. Chem. Phys., 86 (1987) 1991. (d) G.M. Nathanson, T.K. Minton, S.F. Shane and Y.T. Lee, J. Chem. Phys., 90 (1989) 6157.

29 (a) R.K. Sparks, L.R. Carlson, K. Shobatake, M.L. Kowalczyk and Y.T. Lee, J. Chem. Phys., 72 (1980) 1401. (b) A.M. Wodtke and Y.T. Lee, J. Phys. Chem., 89 (1985) 4744. (c) R.E. Continetti, B.A. Balko and Y.T. Lee, J. Chem. Phys., 89 (1988) 3383.

30 G.N.A. van Veen, K.A. Mohamed, T. Baller and A.E. de Vries, Chem. Phys., 74 (1983) 261.

31 Z. Xu, B. Koplitz and C. Wittig, J. Chem. Phys., 90 ( 1989) 2692.

32 H.J. Krautwalk, L. Schneider, K.H. Welge and M.N.R. Ashfold, Faraday Discus. Chem. Soc, 82 (1986) 99.

33 G.E. Busch and K.R. Wilson, J. Chem. Phys., 56 (1972) 3638.

34 C. Jonah, J. Chem. Phys., 55 (1971) 1915. 35 R.N. Zare, Mol. Photochem., 4 (1972) 1. 36 S. Yang and R. Bersohn, J. Chem. Phys., 61 (1974) 4400. 37 M. Dzvonik, S. Yang and R. Bersohn, J. Chem. Phys., 61

(1974)4408. 38 S.R. Gandhi, T.J. Curtiss and R.B. Bernstein, Phys. Rev.

Lett., 59(1987)2951. 39 S.M. Penn, C.C. Hayden, K.J. Carlson Muykens and F.F.

Crim, J. Chem. Phys., 89 (1988) 2909. 40 J.F. Black and 1. Powis, Chem. Phys., 25 (1988) 375. 41 T.A. Spiglanin and D.W. Chandler, Chem. Phys. Lett., 141,

(1987) 428. 42 (a) R. Ogorzalek Loo, G.E. Hall, H.-P. Haerri and P.L.

Houston, J. Phys. Chem., 92 (1988) 5. (b) R. Ogorzalek Loo, H.-P. Haerri, G.E. Hall and P.L. Houston, J. Chem. Phys., 90 (1989) 4222.

43 R. Schmiedl, H. Dugan, W. Meier and K.H. Welge, Z. Phys. A., 304(1982) 137.

44 (a) R. Vasudev, R.N. Zare and R.N. Dixon, Chem. Phys. Lett., 96 (1983) 399. (b) J. Chem. Phys., 80 (1984) 4863.

45 I. Nadler, D. Hahgerefteh, H. Reisler and C. Wittig, J. Chem. Phys., 82 (1985) 3885.

46 M.P. Docker, A. Hodgson and J.P. Simons, Faraday Dis-cuss. Chem. Soc, 82 (1986) 25.

47 G.E. Hall, N. Sivakumar, R. Ogorzalek, G. Chawla, H.-P. Haerri, P.L. Houston, I. Burak and J.W. Hepburn, Faraday Discuss. Chem. Soc, 82 (1986) 13.

48 For reviews see: (a) M.P. Docker, A. Hodgeson and J.P. Simons, in M.N.R. Ashfold and J.E. Baggott (Eds.), Molecular Photodissociation Dynamics, Royal Society of Chemistry, London, 1987, pp. 115-137. (b) P.L. Houston, J. Phys. Chem., 91 (1987) 5388.

49 K.P. Huber and G. Herzberg, Molecular Spectra and Molecular Structure IV. Constants of Diatomic Molecules, Van Nostran, Princeton, NJ, 1979, p. 330.

50 S. Yang and R. Bersohn, J. Chem. Phys., 61 (1974) 4400. 51 D. Porret and C.F. Goodeve, Proc R. Soc, London Ser. A,

165(1983)31. 52 R.S. Mulliken, J. Chem. Phys., 8 (1940) 382. 53 J.L. Knee, L.R. Khundakar and A.H. Zewail, J. Chem.

Phys., 83 (1985) 1996. 54 J.V.V. Kasper and G.C. Pimentai, Appl. Phys. Lett., 5

(1964)231. 55 H.J. Hwang and M.A. El-Sayed, J. Chem. Phys., 96(1)

(1992) 856.


283

Quantitative determination of kinetic energy releases from metastable decompositions of sputtered organic ions using a time-of-flight mass spectrometer with a single-stage ion mirror

D.R Barofskya'*, G. Brinkmalmb, P. Hâkanssonb, B.U.R. Sundqvistb

^Department of Agricultural Chemistry, ALS 1007, Oregon State University, Corvallis, OR 97331-7301, USA bDivision of Ion Physics, Department of Radiation Sciences, Uppsala University, Box 535, S-751 21 Uppsala, Sweden

A method has been developed for determining the kinetic energy released when metastable organic ions, produced by particle-induced desorption-ionization, decompose in a time-of-flight mass spectrometer having a single-stage ion mirror. To the best of our knowledge, this is the first report of a fully developed, quantitative procedure for this particular combination of ionization method and mass analysis.

In order to obtain the kinetic energy released in a specific metastable decay, the rate constant for the unimolecular reaction has to be estimated, and the widths of the precursor and charged fragment ion peaks have to be measured. The rate constant for a specific decomposition reaction is determined by deflecting all ions away from the optic axis at positions of increasing distance along the flight path through the first field-free region of the spectrometer and by counting the neutral fragments that reach the detector located at the back end of the mirror. The widths of the precursor ion peak and the charged fragment ion peak are measured respectively after both types of ion species have independently followed precisely the same flight path in space and time. With a single-stage mirror, the latter condition is met by reflecting the precursor and fragment ions with mirror potentials that are in proportion to the respective masses of the ions. Theoretical, experimental, and error analyses are described and illustrated with examples.

Key words: Kinetic energy release; Sputtered organic ions; Single-stage ion mirror

(Received 3 June 1993; accepted 15 September 1993)

Abstract

Introduction

It is well documented that repeated particle bombardment or pulsed photon irradiation of an organic sample in vacuum ejects a large fraction of ions that contain excess internal energy. In a mass spectrometer these ions decompose unimolecularly en route to the instrument's detector. Even though the vibrational energy acquired during the sputter-


ing or ablation process may exceed that required for decomposition via a given reaction path, the fragmentation does not occur immediately because the excess energy must first become suitably distributed within the molecule. In mass spectrometric parlance, these delayed reactions are referred to as metastable decays.

The study of unimolecular decompositions via mass spectrometry has played a major role in advancing our understanding of molecular structure, ionic structures produced by various

SSDI0168-1176(93)03889-T

284 D.F. Barofsky et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 283-294

ionization processes employed in mass spectro-metry, ion dissociation energetics and kinetics, transition states, and thermochemical quantities [1,2]. The behavior of the charged and neutral products of organic ions that have been sputtered or ablated from condensed phases and have sub-sequently undergone metastable decomposition in time-of-flight (TOF) mass spectrometers has been investigated for kilo-electronvolt ion bombard-ment [3,4], mega-electronvolt particle bombard-ment [5-7], and pulsed laser irradiation [8]. With few exceptions, these past studies were qualitative in nature; they elegantly revealed the character-istics exhibited by metastable organic ions pro-duced by particle-induced processes and, in so doing, established the possibilities and, in some instances, the feasibility for using TOF techniques to study metastable decompositions.

In order to study the dissociation dynamics of ammonia cluster ions formed by irradiating super-sonically expanded ammonia gas with intense pulses of 355 nm light, Wei et al. recently demon-strated how to take advantage of a double-stage ion mirror in a TOF mass spectrometer to make precise measurements of the intensities and widths of the peaks produced respectively by the meta-stable reactant ions and product ions [9]. By adapt-ing their technique to an instrument with a single-stage mirror and combining it with several of the techniques referenced in the previous paragraph, we have developed a general protocol for precisely and quantitatively estimating the average kinetic energy released when organic ions, produced by nuclear or electronic sputtering processes, undergo a specific metastable decomposition.

Experimental

All experiments were conducted on a TOF mass spectrometer designed and constructed at Uppsala University, see Fig. 1. Samples can be bombarded with mega-electronvolt ions from the Uppsala EN-tandem accelerator, with kilo-electronvolt ions from a commercial liquid metal ion column (FEI Company), or with UV photons (not indicated in

the figure) from an excimer-pumped dye laser. The reader is referred elsewhere for a complete descrip-tion of this instrument [10,11].

The mega-electronvolt primary ions pass through a thin (5 /ig cm2) carbon foil before reach-ing the target. Transit through this foil forces the primary ions into an equilibrium charge state [12,13]. The secondary electrons emitted from the carbon foil after passage of a single primary ion are drawn to a microchannel plate detector to produce a signal that starts the timing electronics. The mega-electronvolt primary ions impinge on the front of the sample at an angle of 45° from the surface normal. 72.3 MeV 127i13+ primaries were used in the high energy component of this study.

The kilo-electronvolt ion column is equipped with a set of blanking plates that can be pulsed to sweep the continuously generated ion beam across a timing aperture and thereby periodically produce bursts of primary ions for TOF-SIMS measure-ments. The length of the ion packets can be varied from 5 to 100 ns. The pulse generator also delivers the start signal to the timing electronics. Since the primary ions have to pass through the 4.5 mm long acceleration zone separating the sample from the grounded grid that defines the beginning of the field-free flight region (Fig. 1), the angle at which the primaries impinge on the sample depends on their kinetic energy and on the voltage used to accelerate the secondary ions into the flight tube. In this study, 20keV (at impact) 69Ga+ and 115In+

primaries were used, and samples were maintained at a potential Ka of +5154 V. Under these con-ditions, the primary ions strike the sample at an angle of about 50° from the surface normal.

MicroChannel plate detectors are located both behind and in front of the ion mirror (Fig. 1) so that secondary ions and neutrals passing straight through the mirror when it is at ground potential as well as secondary ions reflected by the mirror when it is at some active potential can be detected. Flight times of individual ions are registered on a multistop (up to 255 stops per start) time-to-digital converter (CTN-M2, IPN, Orsay, France) with 0.5 ns time resolution per

D.F. Barofsky et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 283-294

I y^ z

285

Stop detector 2 (reflected mode)

Electrostatic mirror Vr Stop detector 1

(straight mode)

Sample

Start detector

Fig. 1. A schematic of the TOF mass spectrometer.

channel. Spectra are generated by accumulating data on a desorption event by desorption event basis in an ATARI Mega computer.

An einzel lens and two sets of deflection plates are located along the secondary ions' flight path between the ion source and the ion mirror. Together with the mirror these devices were used for ion deflection in the decomposition rate con-stant measurements (see Average squared lifetime section).

Valine and tri-alanine were obtained in powder form from Sigma Chemical Co. Samples were pre-pared by dissolving the molecules in trifluoroacetic acid (10 /ig /iL"1) and then spin-coating « 10 /xL of the solution on a silicon wafer backing [14,15]. Film thicknesses, which were monitored by ellipso-metry, were between 30 and 60 nm.

Average kinetic energy release

An ion is ejected from the sample probe in the ion source with some initial velocity δν^, and it acquires an additional component of velocity v^ ( > δν^) due to its acceleration out of the ion

source. If a set of ions of a given mass do not decompose before striking a detector, they will generate a signal whose width reflects, for the most part, the distribution of initial velocities among the given ion set.

The unimolecular decomposition of an ion P+

into a charged fragment C+ and a neutral frag-ment N is represented by the reaction

C + + N (1)

If this reaction takes place in the field-free region preceding an ion mirror in a TOF mass spectro-meter, the charged fragments register in a spec-trum of the reflected ions as diffuse (metastable) peaks whose centroids generally do not coincide with those of the normal mass peaks (Fig. 2). As is well known from studies of unimolecular decompo-sitions with sector instruments, the diffuse shape is due to the conversion of a fraction of the precursor ion's internal energy into kinetic energy that separates the fragments. Elementary consideration of the laws of conservation of momentum and energy leads to the following expression for the ionic fragment's velocity u, relative to the


60 70 Mass/charge [a.u.]

Fig. 2. Section from a TOF mass spectrum of tri-alanine recorded in the reflected ion mode. Four metastable peaks at m/z 53.55, 57.35, 68.98, and 74.01 are prominent among the stable-ion mass peaks.

decomposed molecule's center of mass:

/m N 2r r ^ u= \ ιυ

V mc mP

(2)

where raP, rac, and raN are the masses of the precursor ion, the charged fragment, and the neutral fragment respectively, Tr is the amount of internal energy converted into translational energy during fragmentation of P + , and ζ is a unit vector along the direction of if [1], Forming the scalar product of u with itself, solving for Tr, and averaging give the following expression for the average kinetic energy released in reaction (1):

mpmç 2 (3)

The velocity of the charged fragment in the fixed coordinate system of a mass spectrometer v* is obtained by superimposing its velocity due to scattering u on the velocity of the decomposed molecule's center of mass ϊ^ + 6v*0 i.e.

V*=~VQ + 6V*Q +~W (4)

If the instrument's flight axis is taken as the z-direction

- ßqV& r vo = v0iz = J iz V Hip

(5)

acceleration potential (Fig. 1). Hence, the axial velocity component of the ionic fragment can be expressed as

= i>o ± 6v0z ± uz = o ± 6vz (6)

where q is the charge on the ion and Fa is the

Equations (2)-(6) are generally valid for any type of mass spectrometer in which the ions enter a field free region after exiting the ion source. However, the manner in which a fragment ion's velocity of separation contributes to the broaden-ing of its signal is specific to the type of instrument used to record the mass spectrum. The relationship between the width of a metastable peak produced by a TOF mass spectrometer equipped with a single-stage ion mirror and the average kinetic energy released in a particular unimolecular decomposition can be derived by considering the average flight path of a fragment ion through the mirror. The total length L of field-free drift space in the instrument is equal to the sum of the lengths of the field-free drift space between the ion source and the ion mirror Lx and the field-free drift space between the ion mirror and the detector for reflected ions L2. Consider a metastable ion that decays at a time r (= ßL/v0, 0 < ß<Lx/L or 0 < T<LX/VQ) after it exits the ion source. Start-ing with an equation derived by Tang et al. [4], the time of the charged fragment's arrival at the detector for reflected ions can be expressed to the

D.F. Barofsky et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 283-294 287

first order by

2mcv0 'c

. u

L + — v0J

δν, Όζ '2mcv0 L

2mcv0 L i — v + r

v0 [ qE v0

±... (7)

where E is the electric field strength generated by the mirror. The electric field strength near the axis of a single-stage electrostatic mirror is directly pro-portional to the voltage applied to its exit grid. When the mirror voltage is set to a value KP for which

£ P = 2mpvl

(8)

then the flight time of every precursor ion, regard-less of its mass, is minimized to 2L/v0 and, to the first order, is independent of variations in axial velocity [4]. When the mirror voltage is lowered from V? to a new setting Vc given by

Vc = ^V* Wp

then

EQ — Ep mP

2mcv0

qL

(9)

(10)

With this mirror setting the second term in Eq. (7) vanishes, and the remaining expression simplifies to

('c)min« ± — r (11)

Thus, fragment ions originating from a given meta-stable decomposition in Lx will, on the average, travel the same minimized flight path in time and space as their precursors would have, had the latter survived their flight to the detector for reflected ions. The flight time of the fragment ions will fluctuate about the average of their precursors by the factor

/ X 2 ^ , UZ fo= (fc)min « ± — T

^0 ^0 (12)

Fragment ions having the largest or smallest axial

velocities possible will be those that are scattered forward or backward along the flight axis. For these ions, uz — w, and

^max = ± — T (13)

Except for a small degree of tailing due to instru-mental effects, the shapes of precursor ion and fragment ion mass peaks produced on TOF mass spectrometers equipped with a single-stage ion mirror are well approximated by gaussian curves (Fig. 3); this has also been reported for the case of instruments having a double-stage ion mirror [9]. Hence, 6t appears to be a normal variate. Since the distribution of <5i, i.e. the mass peak itself, is defined primarily by those ions scattered forward or back-ward along the flight axis and since (u) = 0, the variance of the distribution should have the form

(14)

The proportionality constant in Eq. (14) is set by the fact that the width of a normally shaped meta-stable peak measured at 22% of peak height (5 ^0.22 = 1-74σ) corresponds to (TT) [16,17], i.e.

(1.74σ,)2=^(τ*> (15)

Eliminating (u2) between Eqs. (3) and (15) and using Eq. (5) to substitute for VQ give

(16)

Thus, obtaining an experimental value for the aver-age kinetic energy released in a given decompo-sition reaction reduces to determining the average squared lifetime of the metastable precursor ion and the variance in the charged fragment's mass peak attributable to the kinetic energy the frag-ment acquires when its precursor decomposes.

Average squared lifetime

If the rate constant of a given decomposition in the field-free region preceding the ion mirror is well


300

250

4->

e D O o G O

200

150

100

50

50 F-

a 40

30 t-

c s o o

I 20

10

0

(a) ~ i — i — I — i — i — i — i — i — i — i — i — i — I — i — i — i — i — i — i — i — i — r

■t—M-»^

(b) I °°|a<*|°°Tt0y>TT°tT»"ta9°tr f?tw"

ΐ ♦.-Τ, Ι ' , - Ϊ , Λ ^ ?

24850 24870 24890

Flight time [ns] 24910

Fig. 3. The molecular ion of valine undergoes the metastable decomposition process [M + H]+ —> [M — CH02]+ + CH202. Shown are:

(a) the region around the [M + H]+ precursor peak in a TOF mass spectrum recorded in reflected mode with mirror potential Vm = VP

and (b) the region around the [M-CH02]^ fragment peak in a TOF mass spectrum recorded in reflected mode with mirror potential Vm= Vc. The fitted curves are gaussians on top of constant backgrounds.

represented by a single value k, the average squared lifetime of the metastable ion can be calculated in the usual manner from

\'Ll t2Q-k,kàt

1 (17)

t~ktkàt

where tL = L\/v0. Evaluation of Eq. (17) leads to

/ 2\ .2

(ktLly + [ l + _ ) ( l - ^ ) .*/#.. \ - i

(18)

In general, the three main factors that govern the kinetics of the metastable decomposition of a large organic ion favor describing the reaction with a single, average rate constant [1,18]. First, the range

of rate constants over which metastable reactions can usually be detected is relatively narrow; the upper limit, « 106s_1, is roughly determined by the length of time it takes the ions to accelerate from their point of origin into L\9 and the lower limit, « 104 s"1, is roughly determined by the transit time tL]. Second, k for a given reaction is a relatively slow function of the internal energy E of an ion having a large number of effective oscillators (Fig. 4). Third, the range of internal energies over which metastable ions exist is fairly narrow (Fig. 4).

In the particular case of 252Cf plasma desorption-ionization of organic molecules in the mass range 900-6000 u, a body of work (for a review, see Ref. 5) has shown that a very high proportion of the ionized molecules undergo a succession of fragmen-tations in times less than 10~8s after the ionizing events and that, of those surviving acceleration, only a small fraction arrive intact at the detector


E(eV)

Fig. 4. The rate constant k for the unimolecular reaction of an organic ion, P+ —► C+ + N, as a function of the internal energy £of P+. The curve is generated from the semiquantitative equation, k(E)= v{\ - E0/E)s~\ which is based on the classical harmonic oscillator model for unimolecular fragmentation [1,18]. The frequency factor v and the activation energy E0 were taken respectively to be 1013 s"1

and 2.5 eV; these values are typical for the cleavage of a single bond in an organic ion [18]. The effective number of oscillators s was estimated simply from (37V - 6)/3 where N was taken to be 33, the number of atoms in tri-alanine. The shaded region shows the relationship between the range of rate constants and the range of internal energies over which metastable decompositions can

typically be observed.

some 200 /xs later. The ions that disintegrate after acceleration have rate constants ranging from about 104 to 106s_1. In light of the generalizations noted above, plasma desorption processes apparently produce a population of ions possessing a broad distribution of internal energies with a high mean. After the chain of prompt decompositions take place during acceleration, a small subpopulation, characterized by a narrower distribution of internal energies with a smaller mean, remains.

As with any chemical reaction, determining the rate of a given metastable decomposition requires that some measure of either the precursor ion population or the population of one of the frag-ment species be plotted over time. Wahrhaftig, who was apparently both the first to recognize the importance of measuring the rates of unimol-ecular dissociations in mass spectrometers and the first to attempt to make such measurements, obtained ion signals as a function of time by using delayed drawout from a pulsed electron impact ion source to vary the time between ion-ization and mass analysis on a TOF instrument [19]. Hunt et al. pointed out that the reaction

time for unimolecular processes occurring in the drift tube of a TOF analyzer could be varied either by changing the length of the flight path to the detector or by changing the velocity of the pre-cursor ions, and they devised and demonstrated a method for the latter approach [20]. More recently, a slightly modified version of Hunt et al.'s tech-nique was employed to estimate the rates by which electronically sputtered ions metastably decompose [5]. Also recently, decay rates for metastable ions produced by nuclear sputtering processes were determined by using a movable detector to vary the length of the flight path and, thus, the time allowed for decomposition [3,4]. We achieve the same end as in the latter case by letting an einzel lens, a set of deflection plates, and the exit grid of the ion mirror serve as ion-valves at different positions along the drift region Lx preceding the ion mirror and by monitoring the signals produced by neutral rather than charged fragments (Fig. 5). Since these devices, which pre-existed in our instrument for other purposes, do not stop ion flow at accurately known positions, they introduce


0 4000 Flight time [ns]

8000

Fig. 5. A schematic of the mass spectrometer showing the different ion deflection devices used as ion-valves in the decomposition rate experiments: · , the number of neutral fragments detected (when the deflection device depicted above a given data point is activated) from the decomposition process [M + H]+ -> [M-CH0 2 ] + + CH 2 0 2 that occurs when valine is bombarded by 20keV 69Ga; , the

fit from Eq. (19) to the data points.

small systematic errors that will be discussed in the final section of this paper.

The required reaction rate data for a given system of analyte, matrix, and primary beam is obtained by recording a neutral particle spectrum for each of the three possible combinations of one ion-valve on and two off. All spectra are recorded with the same number of starting pulses, usually between 106 and 107, to ensure consistent stat-istics. Exponentially modified gaussian curves are fitted to the neutral fragment peaks using PeakFit (Version 3.1, Jandal Scientific), and the areas of these curves are used as measures of the particle counts. Exponential modification to the curves is used to obtain a better measure of the tailing

displayed on the high-time sides of the neutral fragment peaks.

The particle counts are plotted versus time. Assuming that a single reaction pathway and first order kinetics prevail, the general equation

N=P0{\-e-M) (19)

is fitted to the three data points and the origin (Fig. 5) to obtain estimates of the average reaction rate constant k and the population of precursor ions P0 that survived acceleration. Reaction rate data for two major metastable fragmentations of valine, including the computed values of (r2), are given in Table 1. We have also been able to obtain excellent fits (r2 ^ 0.99) for several of the

D.F. Barofsky et al.I Int. J. Mass Spectrom. Ion Processes 131 (1994) 283-294 291

Table 1 Decomposition rate parameters, including the correlation coefficient r2, determined from fitting Eq. (19) to the neutral fragment peak data for two metastable decomposition reactions of valine bombarded by different primary ions

Decomposition process k(ßs (τ2)(μ*2)

Primary ion : 20keV69Ga

[M + H]+ -> [M - CH0 2 ] + + CH 2 0 2

[2M + H]+ -► [M + H]+ + M

Primary ion :

[M + H]+ -» [M - CH0 2 ] + + CH 2 0 2

[2M + H]+ -> [M + H]+ + M

Primary ion: 72 Me V1271

[M + H]+ -► [M - CH0 2 ] + + CH 2 0 2

[2M + H] + ->[M + H]+ + M

720 ±140 3 1 4 ± 6

0400 ± 400 6000 ± 300

2060 ± 60 620 ± 30

0.61 ±0.03 0.47 ± 0.02

0.67 ± 0.07 0.47 ± 0.06

0.76 ± 0.06 0.64 ± 0.08

0.9995 0.9994

0.997 0.998

0.998 0.997

3.3 4.2

3.0 4.3

2.6 3.1

metastable decompositions that behenic acid and tri-alanine undergo [21]. We believe these results provide independent support for the expectation, which derives from the general behavior of metastable ions, that particle-induced desorption-ionization processes will in most instances spawn narrow distributions of rate constants for observ-able decomposition reactions.

When two major, parallel decomposition path-ways exist for a given precursor ion, neutral fragments contribute to the same peak in a neutral particle spectrum and, thus, do not provide sufficient information to determine the rate constants for the two competing reactions. This is the same situation that would exist were it possible to follow the decline in the precursor ion popu-lation for successively longer reaction times. Fortunately, it is only necessary to determine the ratio of the ionic fragments for any reaction time within the range studied in order to deconvolute the total reaction rate curve for the combined neutral fragment species into two curves corre-sponding to the individual fragmentation reac-tions [22]. An example of such a deconvolution is shown in Fig. 6.

Peak variances

Ferguson et al. [23] were the first to point out the possibility of using peak broadening in TOF mass

spectra to obtain information about the kinetic energy released in metastable decompositions, and Franklin and co-workers [24,25] were the first to show how, with TOF peaks, the contri-bution of the precursor ions' thermal movements could be subtracted from the overall motion of the ionic fragments. In the independent development of related approaches for sector mass spectro-meters [26,27], it was realized that correction for the inherent energy distribution of the reactant ion should be made by subtracting the square of its peak's width from the square of the metastable peak's width [27]. Also from work performed on

400

2 10° 4 10° 6 10° 8 10'° Flight time [s]

Fig. 6. The prompt fragment ion [C 5 H n N 2 0] + from a tri-alanine sample decomposes metastably either through the neutral loss of CO or C3H5NO. Deconvolution of the total decomposition curve into the decomposition rate curves for the two decay channels is shown: # , measured data points.

292 D.F. Barofsky et aL/Int. J. Mass Spectrom. Ion Processes 131 (1994) 283-294

sector instruments, it was ultimately found for strictly gaussian peak shapes that the best measure of peak width is at 22% of peak height [16] and that the addition of peak widths in quadrature is exact [28].

The shape of a metastable peak is the result of a statistical convolution of the distributions of the kinetic energies released in the center of mass system of the reactants and of the trans-lational energies of the population of precur-sor ions, from whence the fragments came, in the fixed frame of reference of the mass spectro-meter. The latter distribution, which originates from both ionizing events and factors specific to the particular mass analyzer used, ultimately determines the shape of the precursor ion's mass peak. In general, it is difficult to deconvolute the metastable peak shape with the precursor ion peak shape to obtain information about the distribution of kinetic energies released; however, use of a TOF mass analyzer with an ion mirror reduces the deconvolution problem significantly.

Experimentally, the reactant and product ion populations for a given decomposition reaction are made to follow essentially identical flight paths through the instrument and, as a result, to be subjected to essentially the same instrumental factors that affect peak shape. This is accom-plished by successively recording the precursor ion's and fragment ion's signals subject respec-tively to the conditions expressed in Eqs. (8) and (9). It can be ascertained that condition (8) is satisfied by varying the source potential over a small range and finding the voltage that exactly minimizes the precursor ion's flight time. To ensure that conditions (9) and (10) are satisfied, a value for Vc is calculated from Eq. (10), and then starting from this calculated value, the mirror vol-tage is varied until a setting is found at which the centroid, i.e. mean flight time, of the recorded frag-ment ion peak closely matches the centroid of a precursor ion peak recorded at a VP that satisfies condition (8). This entire procedure, which is much more precise than relying on high voltage readings

from meters, assures that the mean flight times of the two ion populations agree to better than 1 part in (2-3)xlO4 and that, therefore, any difference between their variations about that mean, i.e. between the widths of the two peaks, will be due solely to the kinetic energy releases from the decompositions [9].

The shapes of precursor ion and fragment ion peaks in TOF mass analyzers are fundamentally normal. This is borne out by fitting gaussian curves (PeakFit, Version 3.1, Jandal Scientific) to the shapes of recorded peaks (Fig. 3). Systematic tailing is observed on the high-time side of all of the precursor ion peaks because the ions continue to decompose during the deceleration phase in the ion mirror. We found by trial and error that the most robust curve fits to these peaks are obtained by using all of the data on the low-time side of the peaks and only that data above 0.55 of the highest channel count on the high-time side of the peak. The variance of a fitted gaussian curve is taken as the best estimate of the variance of the correspond-ing ion population. Since it is well established that the variances of normal populations add in quadrature (a fact that accounts for the finding of Baldwin et al. [28]), the fragment ion's extra variance in flight time σ], which is strictly due to the distribution of translational energy acquired by the charged product of the metastable reaction, is simply obtained by subtracting the estimated variance of the precursor ion peak σ\ from the estimated variance of the metastable peak σ^, i.e.

a2tâ2

m-al (20)

The values of σ] obtained from Eq. (20) are used in Eq. (16) to compute the average kinetic energy releases from the metastable decomposition reactions.

Evaluation

We have developed a method for quantitatively determining the kinetic energy released during a metastable decomposition in a TOF mass spectro-meter with a single-stage ion mirror. To obtain the


kinetic energy released in a specific metastable decay, the rate constant for the unimolecular reaction as well as the widths of the precursor and charged fragment ion peaks have to be measured. The average rate constant for a specific decomposi-tion reaction is determined by deflecting all ions out of the flight path at positions of increasing length along the first field-free region of the spec-trometer and measuring the concomitant increase in the number of neutral fragments detected after the mirror. The widths of the precursor and the charged fragment ions are measured after both types of ions have independently followed pre-cisely the same flight path in space and time. With a single-stage mirror, the latter condition is met in the reflected mode by using mirror voltages that, in accordance with Eq. (9), are in direct proportion to the masses of the fragment and precursor ions respectively.

Table 2 lists our experimental estimates of the average kinetic energies released in the two valine reactions

[M + H]+ -> [M - CH0 2 ] + + CH 2 0 2 (21)

and

[2M + H]+ -► [M + H]+ + M (22)

under three different ion bombardment conditions. These values are in the range expected for such reactions [1], but a detailed understanding of their differences requires further study. One obvious dif-ference is that, regardless of the bombardment con-ditions, the kinetic energy released from cleavage of the covalently bound CH 2 0 2 in reaction (21) is consistently greater than the kinetic energy released from dissociation of the noncovalently bound valine molecules in reaction (22).

The overall precision of the technique is compar-able to that obtained by TOF with a double-stage

ion mirror [9]. The accuracy of the technique is difficult to assess without comparable data obtained by independent methods. Based on the fits to the reaction rate data (Fig. 5), the systematic error introduced by using imprecise ion-valves does not appear to be significant. Decompositions that take place in the deceleration regions of the einzel lens and the ion mirror (particularly the latter) produce tails on the longer flight time sides of the neutral fragment peaks and, thus, are a source of systematic error in the measured areas of peaks; this error also does not appear to be appreciable. Probably the largest errors result from decomposition reactions that produce low ion counts; the resulting peaks are not statistically well defined and estimates of statistics from fitted curves must carry a high degree of inaccuracy.

Investigations of the dynamics of metastable decompositions of organic ions produced by ion bombardment should provide useful insights into the mechanisms by which electronic and nuclear sputtering processes impart internal energy into ejected molecules. Moreover, the techniques described in this paper should make it possible to extend the study of unimolecular metastable decompositions to moderately large biomolecules for the traditional purposes of determining binding energies and molecular structure.

Acknowledgement

This work was supported by the Swedish Natural Sciences Research Council (NFR) and the Wallenberg Foundation. Additional support for D.F.B. was provided by US National Institute of Environmental Health Science (NIEHS) Grant ES00210.

Table 2 Kinetic energy releases, (TT) (meV), from decompositions of two types of metastable secondary ions ejected from solid valine bombarded by three different primary ions

Decomposition process/primary ion 20keV69Ga 20keV115In 72MeVI27I

[M + H]+ -+ [M - CH0 2 ] + + CH 2 0 2

[2M + H]+ -+ [M + H]+ + M 71 ± 5 1 8 ± 1

63 ± 1 0 8 ± 2

34 ± 5 22 ± 5


References

1 R.G. Cooks, J.H. Beynon, R.M. Caprioli and G.R. Lester, Metastable Ions, Elsevier, Amsterdam, 1973.

2 A. Fraefel and J. Seibl, Mass Spectrom. Rev., 4 (1985) 151. 3 B. Schueler, R. Beavis, G. Bolbach, W. Ens, D.E. Main and

K.G. Standing, Chem. Phys., 44 (1986) 57. 4 X. Tang, R. Beavis, W. Ens, F. LaFortune, B. Schueler and

K.G. Standing, Int. J. Mass Spectrom. Ion Processes, 85 (1988) 43.

5 B.T. Chait, Int. J. Mass Spectrom. Ion Processes, 92 (1989) 297.

6 S. Della-Negra and Y. Le Beyec, Anal. Chem., 57 (1985) 2035.

7 P. Demirev, J.K. Olthoff, C. Fenselau and R.J. Cotter, Anal. Chem., 59 (1987) 1951.

8 B. Spengler, D. Kirsch and R. Kaufmann, Rapid Commun. Mass Spectrom., 5 (1991) 198.

9 S. Wei, W.B. Tzeng and A.W. Castleman, Jr., J. Chem. Phys., 92 (1990) 332.

10 G. Brinkmalm, P. Hâkansson, J. Kjellberg, P. Demirev, B.U.R. Sundqvist and W. Ens, Int. J. Mass Spectrom. Ion Processes, 114 (1992) 183.

11 G. Brinkmalm, P. Hâkansson, J. Kjellberg and B.U.R. Sundqvist, to be published.

12 P. Hâkansson, E. Jayasinghe, A. Johansson, I. Kamensky and B. Sundqvist, Phys. Rev. Lett., 47 (1981) 1227.

13 A. Brunelle, S. Della-Negra, J. Depauw, Y. Le Beyec and K. Wien, Nucl. Instrum. Methods Phys. Res. B, 43 (1989) 484.

14 G. Save and B.U.R. Sundqvist, TLU Report 149/87, Tandem Accelerator Laboratory, Uppsala University, Uppsala, Sweden, 1987.

15 G. Brinkmalm, G. Jonsson, B.U.R. Sundqvist, A. Hedin and P. Hâkansson, in A. Hedin, B.U.R. Sundqvist and A. Beninghoven (Eds.), Ion Formation from Organic Solids (Proc. IFOS V), Wiley, Chichester, 1990, p. 55.

16 D.T. Terwilliger, J.H. Beynon F.R.S. and R.G. Cooks, Proc. R. Soc. London, Ser. A, 341 (1974) 135.

17 J.F. Elder Jr, J.H. Beynon and R.G. Cooks, Org. Mass Spectrom., 10(1975)273.

18 F.W. McLafferty and F. Turecek, Interpretation of Mass Spectra, 4th edn., University Science Books, Mill Valley, CA, 1993.

19 A.L. Wahrhaftig, in D. Waldron (Ed.) Advances in Mass Spectrometry, Vol. 1, Pergamon, New York, 1959, p. 274.

20 W.W. Hunt Jr., R.E. Huffman, J. Saari, G. Wassel, J.F. Betts, E.H. Paufve, W. Wyess and R.A. Fluegge, Rev. Sei. Instrum., 35 (1964) 88.

21 G. Brinkmalm, P. Hâkansson, B.U.R. Sundqvist and D.F. Barofsky, to be published.

22 R.A. Alberty, Physical Chemistry, Wiley, New York, 1987.

23 R.E. Ferguson, K.E. McCulloh and H.M. Rosenstock, J. Chem. Phys., 42 (1965) 100.

24 J.L. Franklin, P.M. Hierl and D.A. Whan, J. Chem. Phys., 47(1967)3148.

25 M.A. Haney and J.L. Franklin, J. Chem. Phys., 48 (1968) 4093.

26 J.H. Beynon, R.A. Saunders and A.E. Williams, Z. Natur-forsch. Teil A, 20(1965) 180.

27 C. Ottinger, Phys. Lett., 17 (1965) 269. 28 M.A. Baldwin, P.J. Derrick and R.P. Morgan, Org. Mass

Spectrom., 11 (1976)440.


Kinetic energy analysis in time of flight mass spectrometry: application of time of flight methods to clusters and pyrolysis studies in supersonic expansions

John S. Riley, Tomas Baer* Department of Chemistry, University of North Carolina, Chapel Hill, NC 27599-3290, USA


Abstract

A new experimental technique, based on high resolution time of flight analysis of ions in a molecular beam, is described with the use of several examples. Although the approach used here is based on threshold photoelectron photoion coincidence (TPEPICO), the technique can also be used with pulsed laser photoionization, albeit without the benefit of ion energy selection. The approach is based on the fact that the TOF distribution of parent ions formed from cold neutral molecules are narrow, while product ion TOF distributions are broad due to kinetic energy release. This distinction permits differentiating cluster ions formed by "simple" ionization of the corresponding neutral clusters from similar mass cluster ions formed by dissociative ionization. Results so far obtained indicate that most clusters ionize, even at threshold, via dissociative ionization. The technique is also suitable for obtaining the threshold photoelectron spectra (TPES) of mass selected cold species in the presence of a mixture of warm and cold species as might be encountered in a pyrolysis experiment.

Key words: Cluster ions; Threshold photoelectron photoion coincidence; Threshold photoelectron spectra; Supersonic expansions

Introduction

The use of pulsed laser initiated time of flight (TOF) mass spectrometry has revolutionized this method of mass analysis by improving its reso-lution to levels that approach those of double focusing electric/magnetic sector instruments [1,2]. This resolution, combined with the large mass range and the capability of collecting all mass peaks simultaneously, makes this the method of choice in a growing number of applications. However, the desire for high mass resolution has encouraged the use of large electric draw-out fields

which minimize the effects of initial translational energy distributions in the sample. This prevents the utilization of a unique TOF capability, namely the measurement of the neutral sample or product ion kinetic energy. In this paper we demonstrate, using several examples, the various advantages that accrue from the use of low draw-out fields which result in broadened TOF peaks. The ideas pre-sented here apply to experiments in which the ions are produced by pulsed laser excitation, as well as by continuous VUV photoionization. How-ever, all of the data presented are obtained by the method of photoelectron photoion coincidence in which the photoelectron not only energy selects the ion of interest, but also provides a precise start signal for measuring the ion TOF. * Corresponding author.

SSDI0168-1176(93)03880-U

296 J.S. Riley and T. Baer/Int. J. Mass Spectrom. Ion Processes 131 (1994) 295-305

Consider the production of a given mass peak (A+) by the following two reactions:

A + hu -»■ A+ + e

AB + hu -► A+ + B + e"

(1)

(2)

where (AB) may be a covalent molecule or a van der Waals bound cluster. Standard mass spectro-metric approaches such as mass analysis with a quadrupole mass filter or TOF mass spectrometry with the use of high draw-out fields cannot distin-guish A+ formed by reaction (1) from reaction (2). However, via multiphoton ionization it is possible to select the cluster size optically in the first photon absorption to the resonance level so that the second photon ionizes just size selected clusters [3,4].

When space focusing conditions are met in a TOF mass spectrometer, the TOF peak width is a function only of the ion's initial kinetic energy [5]. This kinetic energy can come either from the mole-cule's thermal energy, or from kinetic energy release during a dissociation process. The TOF dispersion due to the forward and backward ejec-tion of energetic fragment ions (turn around time) remains and cannot be compensated for by such devices as a reflectron. The only way to minimize their effects is by delayed pulse extraction [5], by apply-ing a high electric field to the ionization region, or dynamic focusing later in the ion's flight [6].

The basis of distinguishing between reactions (1) and (2) lies in the kinetic energy of ion A+ imme-diately following ionization. For simple ionization reaction (1), A+ is formed with whatever kinetic energy distribution the ensemble of neutral A had prior to photon absorption, which in a supersonic beam based experiment is extremely narrow, espe-cially if the ion is extracted in a direction perpendi-cular to the molecular beam. However, for the production of the A+ ion by a dissociative ioniza-tion reaction (2), the A+ ion gains kinetic energy from the dissociation process which can be orders of magnitude greater than the perpendicular beam energy. The benefits of analyzing kinetic energy broadened TOF peaks have recently been illus-trated for several cluster ions [7-10].

A Maxwell-Boltzmann kinetic energy distri-bution produces a gaussian velocity distribution along the time of flight axis. For a TOF spectro-meter with space focusing, the ATOF = T(v) — T(v = 0) oc v so that the resulting TOF distri-bution for a gaussian velocity distribution is itself a gaussian [11,12]. For the simple ionization reac-tion (1), the full width at half maximum (FWHM) of the A+ TOF peak is related to the average per-pendicular beam energy by

FWHM[A+(1)] = (KE)beamMA

0.261e2 (3)

where e is the electric field in the ionization region in Vcm"1, (KE) b e a m is the average perpendicular beam energy, FWHM is in microseconds, and MA

is the mass of A in atomic mass units [11,13]. How-ever, when A4" is formed from a dissociative ioniza-tion, as in reaction (2), its kinetic energy is

K E [ A + ( 2 ) ] = - A ( K E ) A B MAB

+ K E R ( M A B Z M A ) (4)

where (KE)AB is the average perpendicular beam energy for the AB molecule (from Eq. (3)), and KER is the total kinetic energy release for the dis-sociation reaction [11]. In a molecular beam, the KER is generally much greater than the transla-tional energy of the parent ions ((KE)AB), so that the second term dominates. Dropping the small first term yields the approximate expression for the FWHM for the A+ fragment ion

FWHM[A+(2)] = f (KER)(M A B -M A )M A

0.261e2MAB

(5)

Equations (3) and (5) show that when MAB and MA

are sufficiently different and when the extraction field 6 is small, the different ionization reactions (1) and (2) will produce A+ TOF peaks of drama-tically different widths.

In this paper we illustrate the usefulness of

J.S. Riley and T. Baerjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 295-305 297

the distinction between narrow and broad TOF peaks in several areas, including threshold photo-electron spectra (TPES) of cold molecules and the differentiation between dissociative and non-disso-ciative photoionization processes in clusters and pyrolysis products.

Experimental details

The threshold photoelectron photoion coinci-dence (TPEPICO) set up is shown schematically in Fig. 1. A skimmed, continuous molecular beam intersects the photon beam at right angles. The nozzle is located about 1.5 cm from the 1 mm skimmer thereby providing a beam with a divergence of / = 15. The ionization region is about 5 cm from the nozzle so that the beam width is about 0.4 cm wide. The light source is a low pressure H2 continuous discharge, and the quasi-continuum emission is dispersed by a 1 m normal incidence monochromator with a 2 À band-pass. As the light leaves the monochromator exit slit it diverges in both the horizontal ( / = 10) and vertical ( / = 15) directions. At the ionization

1 1600 1/s

<H IONS

DRIFT

e-H> 1

km 111|nnq mill a—iri Γ; ~v

NOZZLE

5300 1/s

Fig. 1. The experimental arrangement of the molecular beam based TPEPICO experiment. The VUV photon beam enters the cell at right angles to the ion and electron extraction fields as well as at right angles to the molecular beam. Two stages of ion acceleration (2 and 3) are followed by a 26 cm drift region. The mild focusing lens (4) is used to increase the ion collection efficiency. Zero energy electrons are energy selected by a stera-diancy analyzer (1) followed by a hemispherical energy analyzer (6).

region, which is located 2.5 cm from the exit slit, the photon beam is 0.25 cm wide. Therefore, the photon beam width of 2.5 mm is the limiting factor in determining the size of the ioniza-tion region.

Photoelectrons are extracted from the ionization region by a continuous 20 Vcm-1 electric field into a three-element lens (which for optimum resolution is operated as a drift tube with V\ = V2 = V3). This steradiancy analyzer [14] with a \\d ratio of 40 discriminates against energetic electrons with per-pendicular velocity components and yields a total electron resolution (photon plus electron) of 30 meV. The electron lens/drift tube is terminated by a 180° hemispherical energy analyzer operated with a resolution of 0.3 eV, whose major function is to eliminate stray electrons and to discriminate against energetic electrons whose initial velocity is in the direction of the energy analyzer. Better reso-lution of the hemispherical analyzer would not improve the apparatus because the electron ener-gies are dispersed by the voltage drop across the 2.5 mm photon width. After energy analysis, the electrons are detected by a spiraltron secondary electron multiplier. Each photoelectron signal pro-vides a start pulse for measuring the TOF of its associated ion.

Photoions are extracted by the same 20 Vcm"1

field which extends for 4.5 cm. This is followed by a 0.2 cm field of 390 Vcm-1 which accelerates the ions to their final drift energy of 200 eV. The ions are collected at the end of the 25 cm drift tube by a set of microchannel plates. Because the ionization region is 2.5 mm wide, space focusing is necessary to ensure the optimum mass resolution [5].

The experiments were carried out with commer-cially available samples. In order to promote vibra-tional cooling, most samples were diluted (10%) in an Ar seeded beam. Various nozzles were used in these experiments, among them a stainless steel nozzle with a ΙΟΟμιη orifice. The nozzle for the pentane study was a quartz free-jet nozzle with an orifice of 60 /xm, while the acetylene and acetone experiments were carried out with the same nozzle opened up to 120/im. These nozzles produced


1 r

σ c

00

0

Γ

* F "

7] ■ / '

n 1 1

É

i >

fe

>

\ ■\

1 *■

k ' . ,' ^TmqnnVp-- - ,— ,

17.40 17.60 17.80 18.00 TOF (/xs)

18.20 18.40

Fig. 2. A comparison of the TOF distribution of room tempera-ture and molecular beam cooled pentane ions: (o), seeded mo-lecular beam; (■), effusive nozzle; ( — ) (I = 5 K) and (θ θ Θ) (I = 298 K) are gaussian functions. The symmetric tails on the cold TOF distributions are due to warm background gas in the chamber.

terminal Mach numbers [15,16] ranging from 10 to 20. About 8 cm of the quartz nozzle was wrapped with 0.010 in tantalum wire and was covered with ceramic epoxy (Aremco Products Cermabond 569) to provide mechanical support and heat shielding. This nozzle assembly could be heated by passing up to 20 W through the tantalum wire which yielded temperatures as high as 900 K as measured with an Alumel-Chromel thermocouple. This basic design has been successfully used to produce free jets of pyrolysis reaction products for spectroscopic stu-dies [17,18].

Examples of TPEPICO ion TOF distributions are shown in Fig. 2 for the case of pentane ions [19]. The broad peak was obtained with a thermal 298 K sample. The broken line is a gaussian curve simulated for a Maxwell-Boltzmann distribution of 298 K. The FWHM of this gaussian is obtained from Eq. (3) in which the (KE) = 3/2 RT. It is evident that the full width of this distribution is accounted for by the initial velocity distribution of the pentane molecules in the source. The nar-row peak is obtained when the pentane is seeded in a supersonic Ar expansion. The gaussian simula-tion yields a temperature of 5 K. However, note the broad tail under the narrow gaussian. This is due to thermal background gas in the ioni-zation cell.


Warm and cold acetylene

As shown in previous publications [9,20], as well as in Fig. 2, the TOF distributions obtained with the cold molecular beam sample are contaminated by room temperature background gas in the ioni-zation region. Figure 3 shows TOF distributions near the ionization energy (IE) for a neat C2H2

beam. In the top spectrum, taken 30meV below the IE of acetylene, the broad component of the spectrum is considerably enhanced relative to the spectrum at the IE (bottom), which indicates that the warm portion of the beam ionizes at a lower energy than the cold beam. This signal below the IE is a result of hot bands. However, since the lowest vibrational frequency of acetylene is i/4 = 611.7cm"1 [21], the observed hot bands, which are within 300 cm"1 of the origin, are rota-tional hot bands.

Much of this signal below the acetylene IE can probably be attributed to autoionization of rota-tional hot bands [22-24]. Rotational autoioni-zation results when rotationally excited Rydberg states with total energy above the IE, exchange rotational for electronic energy yielding a rota-tionally relaxed ion. Figure 4 shows a plot of the

m/z 26

1 1 . 4 1 eV

1 0 . 5 0 1 0 . 6 0 1 0 . 7 0 1 0 . 8 0 1 0 . 9 0 11 .00 TOF ( μ δ )

Fig. 3. The acetylene ion TOF distributions at and below the acetylene ionization energy. The broad portion of the peak is due to the ionization of rotationally warm acetylene sample. Note the 13C peak at 10.97//s.

J.S. Riley and T. Baer/Int. J. Mass Spectrom. Ion Processes 131 (1994) 295-305 299

0.4 ςο CM

O

o CJ

0.1

C2H2 IP

11.35 11.37 11.39 11.41 11.43

P h o t o n E n e r g y (eV)

11.45

Fig. 4. The fractional abundance of cold acetylene signal in TPEPICO TOF distributions such as the ones shown in Fig. 3. The fraction of cold sample in the beam is 40%. This fraction dips below 40% just below the IE where the warm sample can ionize by virtue of rotational hot bands, while the cold acetylene cannot ionize.

ratio of cold/total mass 26 signal versus ionizing energy in the region of the IE of acetylene. At energies well below the IE, the signal is due entirely to stray light. The fraction of cold signal at these energies should reflect the actual sample composition. Similarly, at high energies, well above the IE, the fractional abundance of cold acetylene signal should reflect the sample composi-tion. Thus the similarity between the very low

energy and high energy fractions is expected and so is the constancy of the fraction at energies above the IE.

The data in Figs. 3 and 4 demonstrate how read-ily the cold signal can be distinguished from the warm signal when the low electric draw-out field permits the TOF peak widths to reflect the sample translational energy distribution. Further, such thermal effects are traditionally difficult to observe in polyatomic molecules, but due to the dramatic differences in TOF peak widths, the presence of such a thermal effect is unquestionable.

Narrow parent and broad daughter ion TOF peaks

In Fig. 5 is shown the TOF distribution resulting from the photoionization of isobutane at 11.04eV [20]. This energy is very close to the dissociation limit so that both parent and daughter ions are evident in the TPEPICO TOF distribution. Most of the parent peak is narrow because its width is given by Eq. (3) and the beam energy is very small. The broad part of this peak is a result of thermal background gas and the dissociative photoioni-zation of neutral dimers. In contrast to the parent

en

m <

< z CD M if)

z o

.9

.7

.5

.3

. 1

1

1 42

,

1 1 1 1

ISO-BUTANE

MOLECULAR BEAM

11.04 eV

1

58

J

1

W*d

A

J

-J

A

-\

1 3 . 5 1 4 . 5

TIME OF FLIGHT 1 5 . 5 1 6 . 5

/ MICROSEC.

Fig. 5. The TOF distributions of parent and daughter ions from the TPEPICO study of isobutane. The daughter ion peak is broadened by kinetic energy release, while the parent ion remains narrow except for the broad features due to thermal gas in the cell and the

dissociative ionization of isobutane dimers. (Taken with permission from ref. 20).


C2H2 +

hu = 10 .85eV

CAWA '

hi/ = 1 0.68 eV

C6H6 +

hv = 1 0.68 eV

1 3 nsec.

-JL·

1 45 nsec.

^ W i P i i !

1 90 nsec.

10.25 10.75 11.25 14.75 15.25 15.75 18.20 18.70 19.20

Time of Flight, ^sec .

Fig. 6. The TOF distributions of acetylene monomers, and "dimer" and "trimer" ions. The latter two are formed by dissociative ionization of the next higher neutral cluster, and their structures as ions are very different from their neutral counterparts [10]. Each of the photon energies corresponds approximately to the appearance energy of the particular species. (Taken with permission from ref. 9.)

peak, all of the daughter ion peak is broad. Its width is determined by the kinetic energy released during the dissociation process.

The broad daughter ion peak is the major reason that molecular beam based TOF mass spectrometers operate at high draw-out fields. Only high fields are capable of narrowing the daughter ion peaks because the width is due to the turn around time in the source. A reflectron cannot compensate for this width. However, in the examples that follow, we make use of this distinction between narrow parent and broad daughter ion peaks to investigate the photoioni-zation process.

The dissociative ionization of clusters

Acetylene clusters. Figure 6 shows TOF mass spectra of acetylene monomers, dimers and trimers [9,10]. Each spectrum was collected at the lowest possible energy at which the peaks appeared. Note that the acetylene monomer peak at the C2H2 IE is very narrow. The beam was adjusted in order to minimize the warm back-ground signal. Well below the IE, we observe "dimer" and "trimer" signal. However, these peaks are broad and have near perfect gaussian shapes. There is no evidence of any narrow component. We thus draw the conclusion that

both of these peaks arise from a dissociative ioni-zation process. Detailed analysis of the peak widths shows that C4H4

+ signal comes from (C2H2)3 and the C 6 H 6

+ signal comes from (C2H2)4, each with the loss of a single acetylene unit [10]. Apparently the dimer and trimer ion and neutral structures are very different so that the Franck-Condon factors for direct ionization are negligibly small. Ab initio calculations of acetylene cluster ions support this conclusion [10]. In fact, they show that the C4H4

+ ion structure is totally different from the loosely bound neutral dimer. In effect, the acetylene dimer has no adiabatic ioni-zation potential.

Methylchloride dimers. Another interesting example of cluster photoionization is that of methylchloride clusters, whose dimer ion TOF distribution is shown in Fig. 7 [9]. The sharp compo-nent is a result of the simple ionization of CH3CI dimers (Eq. (1)), while the broad component is a result of a dissociative photoionization (Eq. (2)). It is interesting that the major contribution to the dimer ion signal is from the latter process. The clear distinc-tion between simple and dissociative ionization, now permits the collection of, for instance, the photoioni-zation efficiency curves or threshold photoelectron spectra of each separate reaction which produces the methylchloride dimer ion.


C2H635CP7CI +

C2H637CI2 +

20.75 21.25 21.75

Time of Fl ight, /^sec.

Fig. 7. The TOF distribution of methylchloride dimers. The sharp peaks are due to the direct ionization of methyl chloride neutral dimers, while the broad, and major fraction is due to dissociative ionization. (Taken with permission from ref. 9.)

Methanol and methyl mercaptan dimers (CH3XH)2 +

The TOF distribution for methanol and methyl mercaptan dimers are shown in Fig. 8. The metha-nol dimer spectrum is dominated by the protonated dimer peak, which has been formed by the disso-ciative ionization of the neutral methanol trimer. The methanol dimer peak is narrow and very small

Y\v = 10.20eV

(CH3OH)2-H+ (CH5SH)2 +

(CH3OH);

7.0 I 8.0 21.0 21.5 22.0

Time of F l ight , /^sec.

Fig. 8. The methanol and methyl mercaptan dimer TOF spectra. Note that the broader methyl mercaptan dimer ion peak is at lower photon energy.

which demonstrates that these dimer ions are pro-duced by the "simple" ionization process of reac-tion (1). In contrast, there is no evidence for the protonated dimer in the methyl mercaptan TOF distribution. However, the dimer peak is rather broad, but with a multi-component shape. An interesting feature of the mercaptan spectra, which at first seems counter intuitive, is that the peak width at 9.19 eV photon energy is narrower than the corresponding peak width at the lower energy of 9.02 eV.

We propose that these peaks consist of a narrow component on top of a broad peak. The two com-ponents arise from the following two reactions:

M2 + Ai>-> M 2+ +e" (narrow peak) (6)

M3 + hv —► M2 + + M + e (broad peak) (7)

However, why is the narrow component stronger in the higher energy spectrum? Figure 9 explains this feature in terms of the Franck-Condon factors for ionization. Apparently, the dimer ion can only be produced from the dimer neutral at higher energies, most likely producing a vibrationally excited dimer ion because the Franck-Condon factors near the adiabatic ionization potential are negligibly small. Spectra with higher TOF resolution and colder beam temperatures would permit the separation of these two processes more clearly.

It is interesting to compare the methyl mercap-tan with that of the methanol. First, there is no protonated dimer formed in the methyl mercaptan

M2+ + M

hvi

\ M 3

hV2

M2 + M

Fig. 9. A schematic potential energy diagram for the methyl mercaptan ionization. At low energies, only the dissociative photoionization is possible. However, at higher energies, the direct ionization to highly vibrationally excited states becomes possible.

302 J.S. Riley and T. Baerjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 295-305

photoionization. Secondly, the mercaptan trimer neutrals evidently dissociate by the loss of a mono-mer unit, a reaction not found in the methanol cluster ions. These differences can perhaps be explained in terms of the different thermo-chemistry. The relevant heats of formation of the clusters (neutral trimers and protonated dimers) are not known. However, we can obtain an approximate idea of the energetics by analyzing the similar reaction for the dimers and mono-mers, for which the thermochemistry is known [25].

The proton transfer reaction can be divided into two half reactions, one of which involves the pro-ton affinity of CH3XH:

CH3OH + H+ -► CH3OH2+ PA = 761 k J m o P 1

CH3SH + H + -+ CH3SH2+ PA = 782kJmor 1

This indicates that the mercaptan is stabilized slightly more than methanol by the attachment of a proton. The other reaction is the dissociation of the CH3XH+ ion through the loss of H + . Here the thermochemistry is:

CH3OH+ -+ CH 3 0 + H + AH°r = 691 kJmol"1

CH3OH+ -♦ CH2OH + H+ ΔΗ°τ = 649 kJ mol"1

CH3SH+ -+ CH3S + H+ AH°T = 756 kJ mol"1

It is clear from the above energetics that proton transfer in the methanol dimer is exothermic by 112kJmol~l, while it is only 26kJmol_ 1 exother-mic for the methyl mercaptan dimer. The heat of formation of the CH2SH radical is not known, a fact which may suggest that it is less stable than the CH3S isomer.

Acetone clusters. Figure 10 shows three spectra near the IP of the acetone monomer for a seeded beam of acetone. The top spectrum at hv = 9.65 eV, contains a narrow signal at 22.75 μϊ {m/z 116) due to ionization of acetone dimer, and a small peak at m/z 58. At 9.66 eV, a new narrow peak appears at m/z 98, which corresponds to the mass of the Ar · · · Acetone heterocluster ion. The bottom spectrum, taken above the IE of acetone

(9.71 eV) [25], contains the expected large and narrow TOF peak corresponding to the mono-meric acetone parent ion. The bottom spectrum also serves to show the relative concentrations of the various beam components. If we assume the ionization cross sections are nearly equal, the monomer:dimer:heterocluster ratio is about 100:10:1.

That the m/z 98 signal is due to the heterocluster can be inferred from two observations in the spec-trum. First, its narrowness means that it does not arise from dissociation of the dimer (e.g. via H 2 0 loss), even though such intra-cluster chemical reac-tions have been observed in acetone clusters [26]. Second, this signal has no broad component which is consistent with it being a weakly bound cluster, since sufficient thermal energy that would broaden the TOF peak would also cause dissociation of the cluster. The presence of this cluster in the molecular beam is interesting since it was formed at rather modest expansion conditions.

A small portion of the coincidence TPES of the monomer and the Ar · · · Acetone cluster is shown in Fig. 11. The first striking conclusion drawn from this picture is that the ionization onset for the Ar · · · Acetone mixed cluster is very sudden, suggesting that this onset results from a favorable

m/z 116

m/z 58

u l L · » , 9.65 eV m/z 98

9.66 eV Λ 9.72 eV -X-

15.00 17.00 19.00 21.00 TOF (μδ)

23.00

Fig. 10. The TPEPICO TOF spectra of acetone seeded in Ar at three photon energies at and below the monomer ionization energy. Note the sharp peak at around 20.5//s TOF in the 9.66 eV spectrum which is due to an Ar ■ · · Acetone heteroclus-ter ion.


~^N-

/ / /

Φ / i

t

J 1 /

1

J A -

\ (Ar — C3H60)+

Λ (C3H60) +

Gl J \

J "" "° 9.60 9.65 9.70 9.75

Photon Energy (eV) 9.80

Fig. 11. A portion of the Ar · · · Acetone heterocluster and acetone monomer coincidence TPES collected in a single experi-ment. The arrows at 9.66 and 9.71 eV locate the heterocluster and monomer IEs, respectively.

Franck-Condon factor for adiabatic ionization. Such adiabatic ionization of small clusters has been elusive [27], since the electrostatic properties responsible for the primary bonding in clusters change dramatically upon ionization [28(a)]. This results in large minimum energy geometry differ-ences between the neutral and ionic clusters. Appar-ently, in the Ar · · ■ Acetone complex, a significant geometry change does not occur upon ionization, perhaps in part due to the large Ar polarizability.

This experimental conclusion is supported by ab initio calculations of the Ar · · · Acetone complex structure. The starting point for these calculations was a geometry in which the Ar atom was placed nearest the acetone oxygen atom co-linear with the C = O bond. Molecular orbital calculations with the GAUSSIAN 90 program using an MP2/3-21G* basis set, show the O-Ar bond distances, in the fully optimized structures, to be 3.1 and 3.3 À in the neutral and ion, respectively. Furthermore, the rest of the acetone structure did not change signifi-cantly. In view of the modest sized basis set used in these calculations, the precise values of the calcu-lated Ar-Acetone bond lengths are less interesting than the fact that the Ar-Acetone ion appears to have a stable minimum with the same geometry as the neutral hetero dimer. We did not explore other ionic dimer geometries. It is possible that a more stable structure exists with the Ar on the other side of the acetone ion. However, such a structure would not be accessible in a one photon ionization

because of poor Franck-Condon factors. The TPES of the heterocluster and acetone in Fig. 11 have several features in common. They have simi-larly sharp onsets, the same vibrational frequency (up to v = 1) is excited, and they have a similarly sharp first PES band.

With the adiabatic ionization energy of the heterocluster in hand, we can determine the differ-ence in the neutral and ion cluster binding energies. This is given by

BE+ - BEC = IEm - IEC = 0.050 eV

where IEm is the monomer IE of 9.71 eV, IEC is the cluster IE of 9.66 eV obtained from Fig. 11, and BEC

and BE* are the neutral and ion cluster binding energies, respectively. While the IEs provide infor-mation about the relative binding energies, the clus-ter ion dissociation limit contains information about the neutral cluster binding energy. The cluster ion dissociation limit should be shifted from the acetone ion IE by the neutral binding energy. It is evident in Fig. 11 that the TPE spectrum for the cluster drops to zero above about 9.78 eV. However, we do not know if this decrease in the TPE spectrum is a result of the cluster ion dissociation or the decreasing transition probability due to the onset of the Franck-Condon gap in the PE spectrum of ace-tone. Furthermore, in this energy region, we cannot distinguish the product acetone ion from the large acetone monomer ionization signal. For these rea-sons, we can derive only a lower limit for the neutral binding energy. The observed shift (9.78-9.71 eV) provides us with a B E n ^ 0.070 meV (560 cm"1 or 6.7kJmol"1), which is a reasonable value.

The acetone dimer signal at m/z\\6 is narrow as well, indicating that this cluster ion also arises from a simple ionization process. A similar conclu-sion has recently been reached by Furuya et al. [28(b)]. Again, this suggests an ability of the acet-one dimer to hold itself together in spite of the significant electro-static changes imparted by ioni-zation. This can be qualitatively understood in terms of two dipole-dipole interactions in the acetone dimer. Ionization by removal of a non-bonding elec-tron from one of the oxygen atoms certainly disrupts

304 J.S. Riley and T. Baerjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 295-305

the binding at this site, but it leaves the other dipole interaction intact, allowing the dimer in this particular geometry to remain bound. This geometry may not correspond to the global energy minimum of the acetone dimer ion.

Pyrolysis of pentane

We now extend the ideas developed above to a system that contains several species produced in a pyrolysis source. Figure 12 shows the PEPICO TOF distribution obtained from 10% pentane seeded in 90% Ar and expanded from a nozzle heated to 730 K, which is sufficiently hot to pyro-lyze pentane [29]. The mass spectrum obtained with the room temperature nozzle at the same photon energy contains only the parent ion because the photon energy of 10.68 eV is below the pentane ion dissociation limit of 10.90 eV [25,30]. The TOF spectrum from the heated nozzle contains three mass peaks. The two peaks near 13.6^s (m/z 42 and 43) are broad, which indicates that they are produced by dissociative ionization. The broken line at mass 42 is a simulated spectrum showing the expected width if a mass 42 neutral were ionized according to reaction (1). The broad peaks are therefore not produced by "simple" ionization of pentane pyrolysis.

In contrast to the broad m/z 42 and 43 ion TOF distributions, the signal at 10.75 s (m/z 28) is nar-row, indicating that this ion is formed simply by

Pentane Pyrolysis Products

m/z 28

m/z 42

L_ 11.00 11.50 12.00 12.50 13.00 13.50 14.00

TOF (/zs) Fig. 12. The TPEPICO TOF distribution at 10.68 eV obtained from a supersonic expansion of pentane and Ar from a quartz nozzle heated to 730 K. The corresponding room temperature spectrum shows only parent pentane ions at m/z 72.

OH E-0) o Ö CD

Ί3

O Ö

o o

I

"*!r 1 2

i 1

. 1

^ 4 '

10.40 10.50 10.60 10.70

Photon Energy (eV) 10.80

Fig. 13. The TPES of the cold ethylene sample obtained as a product of the pentane pyrolysis.

ionization of a neutral mass 28 molecule. The most likely candidate for this ion is ethylene which has an IE of 10.51 eV [25] and is a known pentane pyrolysis product [29]. The narrow TOF signal for m/z 28 shows that the TOF width is inde-pendent of nozzle temperature which means that the pyrolysis products have been cooled by the free jet expansion. This is an important conclusion which can be directly inferred from the narrow TOF peak.

As was done for the case of the Ar · · · Acetone complex, we can collect the sharp signal at 10.75 μ$ as a function of the photon energy and obtain a coincidence TPE spectrum, which is shown in Fig. 13. The band assignments indicated in the figure are taken from Stockbauer's TPE spectrum of ethylene [31]. This spectrum illustrates several important features. First, the assignment of the spectrum further verifies this species as ethylene. Second, the threshold TPES technique is indeed useful for obtaining spectra of species from a multi-component beam, and that the spectra are identical to those obtained from a pure beam.

Conclusion

TOF mass spectrometry with low draw out fields is shown to be highly useful for extracting infor-mation about the kinetic energy of the mass peaks. This information is vital in unraveling the complex series of reactions which take place in the

J.S. Riley and T. Baerjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 295-305 305

photoionization of cluster beams of stable mole-cules as well as beams obtained from a pyrolysis source. This technique is particularly advantageous when combined with photoelectron photoion coin-cidence (PEPICO) because it is then possible to collect coincidence TPES of mass selected samples as has previously been reported also by Ng and co-workers [32,33] and Furuya et al. [28(b)]. How-ever the approach is also useful in non-coincidence TOF experiments where the distinction between "simple" and dissociative ionization processes could be distinguished.

Acknowledgement

We thank the Department of Energy for support of this work under Grant No. DE-FG05-88ER13950. We are grateful to the North Carolina Supercomputing Center for time on the CRAY Y-MP 8/464 supercomputer, and the University of North Carolina Computing Center for time on the CONVEX system. Finally, we thank Dr. Jon Booze for valuable contributions to this research project.

References

1 (a) R.B. Opsal, K.G. Owens and J.P. Reilly, Anal. Chem., 57(1985) 1884. (b) R. Grix, R. Kutscher, G. Li, U. Grüner and H. Wollnik, Rapid Commun. Mass Spectrom., 2 (1988) 83.


3 M. Mons, J. LeCalve, F. Piuzzi and I. Dimicoli, J. Chem. Phys., 92(1990)2155.

4 Ch. Riehn, Ch. Lahmann, B. Wassermann and B. Brutschy, Chem. Phys. Lett., 197 (1992) 443.


6 G.E. Yefchak, CG. Enke and J.F. Holland, Int. J. Mass Spectrom. Ion. Processes, 87 (1989) 313.

7 E. Holub-Krappe, G. Gantefor, G. Broker and A. Ding, Z. Phys. D, 10(1988)319.

8 K.M. Weitzel, J.A. Booze and T. Baer, Z. Phys. D, 18 (1991) 383.

9 (a) J.A. Booze and T. Baer, J. Chem. Phys., 96 (1992) 5541. (b) K. Furuya and K. Kimura, J. Chem. Phys., 97 (1992) 1022.

10 J.A. Booze and T. Baer, J. Chem. Phys., 98 (1993) 186.

11 R. Stockbauer, Int. J. Mass Spectrom. Ion. Phys., 25 (1977) 89.

12 J.L. Franklin, P.M. Hierl and D.A. Whan, J. Chem. Phys., 47(1967)3148.

13 T. Baer, J.A. Booze and K.M. Weitzel, in C.Y. Ng (Ed.), Vaccum ultraviolet photoionization and photodissociation of molecules and clusters, World Scientific, Singapore, 1991, pp. 259-298.

14 T. Baer, W.B. Peatman and E.W. Schlag, Chem. Phys. Lett., 4 (1969) 243.

15 J.B. Anderson, in P.P. Wegener. (Ed.), Molecules, Beams and Low Density Gas Dynamics, Marcel Dekker, New York, 1974.

16 D.R. Miller, in G. Scoles (Ed.), Atomic and Molecular Beam Methods, Oxford University Press, New York, 1988, pp. 14-53.

17 J.R. Donlop, J. Karolcyak and D.J. Clouthier, Chem. Phys. Lett., 151 (1988) 362.

18 D.C. Moule, H.A. Bascal, Y.G. Smeyers, DJ. Clouthier, J. Karolcyak and A. Nino, J. Chem. Phys., 97 (1992) 3964.

19 T. Baer, J.A. Booze and J.S. Riley, in C.Y. Ng. (Ed.), Laser techniques for state selected and state to state chemistry, Proc. SPIE, Vol. 1858, 1993, pp. 158-163.

20 K.M. Weitzel, J.A. Booze and T. Baer, Chem. Phys., 150 (1991)263.

21 G. Herzberg, Molecular Structure and Molecular Spectro-scopy III. Electronic Spectra and Electronic Structure of Polyatomic Molecules, Van Nostrand, Princeton, NJ, 1967.

22 W.A. Chupka, P.J. Miller and E.E. Eyler, J. Chem. Phys., 88 (1988) 3032.

23 S.T. Pratt, J.L. Dehmer and P.M. Dehmer, J. Chem. Phys., 90(1989)2201.

24 M.A. O'Halloran, S.T. Pratt, F.S. Tomkins, J.L. Dehmer and P.M. Dehmer, Chem. Phys. Lett., 146 (1988) 291.

25 S.G. Lias, J.E. Bartmess, J.F. Liebman, J.L. Holmes, R.D. Levin and W.G. Mallard, Gas Phase Ion and Neutral Ther-mochemistry, 17th edn., NSRD-NBS, Washington, DC, 1988.

26 A.J. Stace, Org. Mass Spectrom., 28 (1993) 3. 27 C.Y. Ng, Adv. Chem. Phys., 52 (1983) 263. 28 (a) S. Tomoda and K. Kimura, in C.Y. Ng (Ed.), Vacuum

ultraviolet photoionization and photodissociation of mole-cular clusters, World Scientific, Singapore, 1991, pp. 101-168. (b) K. Furuya, S. Katsumata and K. Kimura, J. Electron. Spectrosc. Relat. Phenom., 62 (1993) 237.

29 D.L. Allara and D. Edelson, Int. J. Chem. Kinet., 7 (1975) 479.

30 K.M. Weitzel, J.A. Booze and T. Baer, Int. J. Mass Spec-trom. Ion. Processes, 107 (1991) 301.

31 R. Stockbauer and M.G. Inghram, J. Electron Spectrosc, 7 (1975) 492.

32 K. Norwood and C.Y. Ng, Chem. Phys. Lett., 156 (1989) 145.

33 K. Norwood, J.H. Guo, G. Luo and C.Y. Ng, Chem. Phys., 129 (1989) 109.


Photodissociation of magnesium ion/molecule complexes in a reflectron time-of-flight mass spectrometer

C.S. Yeh, K.F. Willey, D.L. Robbins, M.A. Duncan* Department ofChemistry, University of Georgia, Athens, G A 30602, USA


Abstract

Ion/molecule cluster complexes containing magnesium (e.g., Mg+-(C02)x, Mg+-(H20)x) are generated in a pulsed nozzle cluster source. Specific ions are size selected from the source distribution with a reflectron time-of-flight mass spectrometer for studies of their photodissociation dynamics. Photoexcitation of these complexes near the Mg+ (2S —>2P) resonance line causes a variety of novel photochemistry, ranging from simple ligand ejection, to metal insertion, to metal-ligand charge exchange. These reactions are first observed in the single molecule complexes, and they persist in larger aggregates with more extensive solvation. Excitation spectra probe the energy dependence of the photochemistry and they provide information on the structures of the ion/molecule complexes.

Key words: Ion/molecule complexes; Ion spectroscopy; Photochemistry; Magnesium complexes; Reflectron time of flight mass spectrometer

Introduction

As reflected in this special issue, time-of-flight (TOF) spectrometers have many advantages for modern mass spectrometry. They are ideally suited for experiments employing pulsed laser ionization, ablation, or desorption. The multi-channel aspect of mass analysis and ion detection allows rapid data acquisition and signal averaging. While TOF systems have historically had poorer resolution than many other mass spectrometers, the introduction of the reflectron time-of-flight (RTOF) design has improved resolution significantly [1-7]. The limited resolution of TOF instruments is offset by their expanded mass range compared to other instruments.


A serious disadvantage of TOF and RTOF instruments until recently has been the difficulty in utilizing these instruments for tandem mass spectrometry (MS-MS) [8-16]. By dissociation of a selected "parent" ion, and the measurement of its fragmentation spectrum, tandem experiments reveal the molecular sub-units present. Tandem experiments are required in many environments in which there are a mixture of possible com-ponents and the identity of a particular mass cannot be established. They are also useful in identifying the structure of the parent ion. In the present report, we describe the application of a new configuration for the reflectron TOF instrument which makes tandem TOF measurements more convenient [17]. This design uses laser excitation for ion photodissociation. We give examples of studies carried out on magnesium containing

SSDI0168-1176(93)03872-J

308 C.S, Yeh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 307-317

ion/molecule complexes, which investigate their dissociation dynamics, ion spectroscopy and intra-complex reactions.

Weakly bound complexes have been the subject of several recent photochemistry investigations. Van der Waals complex photodissociation has been employed to obtain onented reaction partners for the study of elementary reaction dynamics [18]. Laser induced fluorescence and/or multiphoton ionization of cluster-containing molecular beams have been used to infer intra-cluster reactions [19-22]. Mass spectra or product fluorescence spectra provide evidence of reactions in these systems. However, it is difficult in these experiments to ascertain the reaction mechanism including which cluster parent produced the product species observed. In more recent experi-ments, ion/molecule cluster complexes have been mass selected prior to photoexcitation [23-30]. This procedure eliminates the uncertainty in the identity of the parent molecule from which fragments originate, and makes it possible to investigate the influence of system size on the chemistry. These studies have revealed interesting phenomena in clusters analogous to familiar phenomena in condensed phases. These processes include caging-recombination [24], proton transfer [28], and charge transfer [23,27]. With the recent popularity of metal cluster research, metal atoms and metal ions have been incorporated into weakly bound complexes [26-32]. The photochemistry of these metal-containing complexes is in its infancy, but fascinating processes have already been observed.

In the present report, we focus on complexes containing magnesium ions. Magnesium ions are interesting because of the strong absorption in the 2S —>2 P resonance line near 280 nm. If complexes contain a "solvated" magnesium ion, the perturbed atomic resonance line should provide a convenient chromophore for photoexcitation. Magnesium ion complexes are also of interest because the possible reactions with small molecules are generally endothermic in the ground electronic state. It is therefore possible to synthesize weakly bound

complexes which retain the separate identity of the metal ion and the solvent molecule. However, the energy provided by the atomic excitation is enough to overcome this endothermicity. It may therefore be possible to observe reactions taking place within cluster complexes following photo-excitation. As described below, weakly bound complexes of magnesium are formed which are composed of Mg+ surrounded by small molecules. Photodissociation spectra establish the degree of the interaction and in some cases the structure of the complex. Photoinduced reactions are observed for magnesium ion complexes with small molecules (C02, CH3OH, etc.). However, not all energetically allowed reactions are photo-induced. When photoinduced reactions occur, the known thermochemistry for the non-cluster species can be used to derive information about the energetics of the ion complex. These experi-ments are all made possible by the application of photodissociation in the reflectron TOF system.


Complex synthesis

The ions studied here are produced in a modified pulsed-nozzle laser vaporization cluster source described previously [27]. The molecular species for complexes with the metal is added to the expansion gas (helium or argon) as a dilute component. C0 2 complexes are also produced by using a pure C 0 2 expansion gas.

Mg+-(C02)x

UJLdLI 0 5 10 15

C l u s t e r Size (x)

Fig. 1. The distribution of clusters produced when a magnesium sample is vaporized in an expansion of C0 2 . Complexes of the form Mg+-(C02)x are produced.

C.S. Yeh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 307-317 309

Figure 1 shows the distribution of cluster com-plexes resulting from vaporization of magnesium in an expansion gas containing carbon dioxide. This mass spectrum was obtained with a specially designed reflectron TOF spectrometer, with switched acceleration plates to extract ions from the molecular beam. These complexes have therefore grown as ions, and they have been supersonically cooled after their formation. The predominant mass peaks throughout the spectrum correspond to ions of the form Mg+-(C02)x. There are virtually no complexes of the form MgO+-(C02)x, which would be expected if hot metal ions reacted with C02 in the vaporization plasma. As shown in Table 1, the ground state reaction of magnesium ion with carbon dioxide is endothermic by about 74 kcal mol-1, suggesting that reactions would be unlikely. Because of the simple mass spectra observed we conclude that these com-plexes are "solvated" magnesium ions with differ-ent numbers of complexed molecules. As indicated, the positive charge is expected to be localized on the metal atom because its ionization potential (7.65 eV) is significantly lower than that of C02

(13.77eV). Figure 2 shows a similar mass distri-bution for magnesium-acetone complexes. Again, simple spectra with molecular units added to the metal ion are observed as the primary intense peaks, with only slight evidence for frag-mentation. Similar spectra are obtained for a vari-

\hJ

Mg+— (Acetone)>

IjLJ lu 0 1 2 3 4 5 6 7 8

Cluster S ize ( x )

Fig. 2. The distribution of clusters produced when magnesium is vaporized and the helium expansion gas is seeded with acetone. Mg+-(acetone)x complexes are produced.

ety of small molecules (benzene, H20 and CH3OH).

To study the photodissociation of these cluster ions, we mass select the complexes one at a time from the initial distribution of species produced by the source. The configuration of the reflectron TOF instrument used for these experiments is shown in Fig. 3. Clusters are mass-selected by their TOF through the initial drift tube section. Undesired clusters are rejected and the selected size trans-mitted with pulsed deflection plates located just before the reflectron plate assembly. Pulsed laser excitation is accomplished with a tunable Nd:YAG pumped dye laser (Spectra Physics PDL-2) which intersects the trajectory of the selected ion packet as it reaches its turning point in the reflectron field. Parent ions and their

Table 1 Reaction schemes and energetics for magnesium ion/molecule cluster complexes

Reaction8 Δ//Γ

(kcal mol"

Mg+(2S) + C02( 'E;)

Mg+(2S) + H20('A1)

Mg+(2S)+CH3OH

Mg+(2S) + (CH3)2CO('Al)

MgO+(2II) + CO(1E+)

MgO+(2n) + H2('E+) MgOH+('E+) + H(:rS)

MgO+(2n) + CH4

MgOH+('E+) + CH3(2A2')

MgO(,E+)+CH3+(,A'1) + H(2S)

MgOH(2E+) + CHi('A',)

MgO+(2n) + C3H6 (propene) MgO+(2n) + C3H6 (cyclopropane)

+74.2

+64.2 +44.3

+36.8 + 15.8

+ 186.4 +76.3

+63.2 +71.1

As indicated, all these reactions are endothermic for ground state magnesium atoms.

310 C.S. Yeh et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 307-317

Reflection Grids

Fragmentation Laser

\

+100 V

0 V J£ Pulsed

Mass Gate

\W Einzel Lens

Vaporization Laser

\

Einzel j | [ Lens n = n

/ \ EMT

Metal Cluster Source ,1 Deflection

Plate

Acceleration Plates

rr - - +150

- I ^ 0 V

+1500 V

Fig. 3. The reflectron TOF mass spectrometer configuration used to produce cluster ions, mass select specific sized ions from the cluster distribution, photodissociate the selected ion, and mass analyze the resulting fragment ions.

photoproducts are mass analyzed by their flight time through the second drift tube section. Dissociation spectra are recorded by monitoring the fragment Mg^ ion intensity as the dissociation laser is scanned.

Mg~h(C02)x complexes

Figure 4 shows the photodissociation mass spectra obtained when the complex Mg+-C02 is mass selected and photodissociated at two dif-ferent wavelengths near the expected atomic resonance line. These spectra are accumulated with a computer difference technique (fragment laser on-fragment laser off). Depletion of the parent ion is therefore plotted as a negative peak and fragment ions are indicated by positive peaks. The upper spectrum shows that the only fragment produced at 330 nm is the Mg* ion. At 260 nm, however, both Mg+ and MgO+ ions are observed as fragments.

Mg^-COs

337.4 n m

25 50 75 Mass ( a m u )

Mg+

tm^fttow

MgO+ I

V*MW^%WV

260 nm

\ A M A M ^ J ^ΛΛΛΑΑΛΛΛΑΛΜ

25 50

Mass ( a m u )

75

fig. 4. The photodissociation mass spectra obtained when M g + - C 0 2 is excited at two different wavelengths.

The observation of MgO+ ions as photo-fragments suggests that some fraction of the ions in this experiment may have reacted rather than form weakly bound complexes. However, the fragmentation to form Mg+ is consistent with the ions existing as weakly bound complexes. To investigate the structure of Mg+-C02, we have performed photodissociation spectroscopy experi-ments on this ion [33]. These experiments are guided by ab initio calculations performed by Sodupe et al. [34].

According to the ab initio calculations, the lowest energy structure for Mg+-C02 is the weakly bound electrostatic complex, which is predicted to be linear in both the ground and excited electronic states [34]. The energetics of this system are shown in Fig. 5. The ground state, which has a magnesium ion with a [Ne] s

C.S. Yeh et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 307-317 311

102.0 Mg - C 0 2 337.5 nm

rfg+(2P) + C 0 2 \

0.0

Mg+(2S) + C02 ■

à

252.2 n m

Mg+

k

>

3 3 7 6 n m

- C 0 2 (

Ou<)<D - -o=c=o

2n k

V)

ô

- 1 6 . 4

MgO+ + CO

Fig. 5. The energy level diagram for the excited electronic states of the Mg + -C0 2 complex derived from the ab initio calcula-tions of Godupe et al. [34].

configuration bound on an oxygen of C0 2 , is a Σ state. There are two excited states corresponding to different orientations of the excited p orbital on magnesium ion with respect to the molecular axis. The perpendicular orbital arrangement gives rise to a 2Π complex state, while the linear orientation produces a 2 Σ + state. The excited 2Π state is more strongly bound than either of the ground or excited 2 Σ + states. The electronic spectrum is there-fore predicted to consist of a 2 Σ + —>2Π+ transition red-shifted from the atomic resonance, and a 2 Σ + —>2 Σ + transition blue-shifted from the atomic resonance line.

The photodissociation spectrum of this complex indicates that it absorbs at almost exactly the wavelengths predicted for the weakly bound structure. This spectrum is shown in Fig. 6 [33]. Two electronic transitions are observed. The lower energy transition has an origin near 330 nm, and appears as a series of sharp doublet bands. The doublets are separated by about 400 cm- 1, and the doublet spacing is about 50 cm- 1. The higher energy region exhibits con-tinuous absorption without any structure. We have examined the sharp structure in the lower state by obtaining spectra for each of the isotopic forms of magnesium (24Mg+, 25Mg+ and 26Mg+). These isotopic studies con-firm that the sharp structure is the result of

Mg+-C0 2

Mg + (2S - 2p)

279.8 nm

266 nm

-6000 -4000 -2000 0 2000 Relative Frequency ( c m - 1 )

Fig. 6. The photodissociation spectrum for Mg + -C0 2 obtained by mass selecting the parent ion, photodissociating it with a tunable dye laser, and measuring the wavelength dependence of the Mg+ fragment ion.

a vibrational progression in the metal-C02

stretching frequency. The doublet structure, however, is invariant with isotopic substitution, indicating that it is not the result of some low frequency vibrational mode. Instead, it is electronic in origin, arising from the spin-orbit splitting in the excited electronic state. This structure confirms that the complex is linear with the 2ü-excited state symmetry predicted by theory.

A fit of the vibrational progression determines that the vibrational frequency in the excited state is Jt = 381.8 cm- 1 . This vibrational motion also corresponds to the dissociation coordinate of the molecule. Therefore, it is possible to fit the vibrational progression observed to a Morse potential to obtain the dissociation energy in the excited state (11194cm-1; 32.0 kcal mol-1). While the Morse potential form is not expected to hold near the dissociation limit of the complex, it does describe the levels we observe with good accuracy. Since we do not have data near the dissociation limit, the simple Morse form is most appropriate. Using the excited state dissociation energy DQ, the known asymptote of the Mg+ atomic transition (2S —>2 P) and the electronic origin observed here (z/00 = 29 625 cm- 1) , a simple cycle can

312 C.S. Yeh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 307-317

be used to determine dissociation energy:

the ground state

This treatment reveals that the ground state dissociation energy for the complex is Do = 5 150 cm"1 (14.7 kcal mol"1). Therefore, this photodissociation study makes it possible to determine the structure of the complex, its electro-nic state energies, its binding energy, and its vibrational frequency.

One further piece of information is provided by this photodissociation spectroscopy. The two dissociation channels seen in Fig. 4 at 260 nm can now be seen to fall in the region of the 2 Σ + —>2 Σ +

electronic transition, while the single channel (Mg+) is observed in the 2 Σ + —>2 Π transition. The observation of the metal oxide does not result because the ions have previously reacted. It comes instead from a photo-induced reaction in the excited state of the complex [33(b)]. An induced reaction in this system is possible because the photon energy (280 nm; 4.42 eV; 102.1 kcal mol"1) exceeds the endothermicity (74.2 kcal mol"1) for the reaction of Mg+ with C0 2 . Excited state reactions have been observed previously for neutral Mg and H2 [31], in low pressure collisional environments, but this is the first example of a photoinduced reaction observed for magnesium ion within a weakly bound complex. In another recently studied system, metal oxidation has been observed upon excitation of the V + - C 0 2 ion/ molecule complex [29]. The reaction here is fascinating because it only occurs within the upper excited state. It is not clear whether this is merely an energetic effect, or whether the orienta-tion of the orbitals in this excited state are respon-sible for preferred reactivity.

The photoinduced reaction observed here is also found for larger aggregates with additional C 0 2

molecules present. For example, Fig. 7 shows that the products of Mg+(C02)7 are Mg+(C02)2 and MgO+(C02)2. We have been able to obtain similar spectra for complexes containing up to 12 C0 2

molecules. In every case, both solvated metal ion

0)

M g + ( C 0 2 ) 2 .

Mg+C02 1

1 MgO+(C02)2

XmàÊÊéu

Mg+ (C02)7

\ήφ^^ à m % ^

1 (

100 300 Mass

Fig. 7. The photodissociation mass spectrum of the Mg+-(C02)7 complex.

and solvated metal oxide ion channels are observed. Figure 8 shows a plot of the number of molecules lost as a function of the size of the complex. The broken line indicates the number of C0 2 molecules expected to be lost if the sample had the binding energy of bulk C0 2 . As indicated, the number of C 0 2 molecules "evaporated" is always less than the number expected for the bulk C0 2 . This is because the cluster binding includes metal ion-C0 2 interactions as well as C 0 2 - C 0 2 inter-actions. As the cluster size increases, the average number of C 0 2 species evaporated increases, indicating that the interactions are weaker on the average. This is expected as the metal ion becomes

Mg(C02)n

16 -|

14 -

12 -

"w 10 -O

8 -

O 6 -O

4 -

2 -

0 -

■

co2

■

1

b

■

ulk

■

1

■

■

1

■

■

1

■

1 1

0 2 4 6 8 10

P a r e n t C lus te r Size (n)

Fig. 8. A plot of the number of C0 2 molecules lost by dissociation of different sized Mg+-(C02)x complexes. ( ) The number expected to be lost if the C0 2 binding energy were the same as it is in bulk frozen C0 2 .


more "solvated" by C0 2 molecules, which eventually shield the outer layers from direct interaction with the metal ion. In other similar systems, a discontinuity is sometimes observed in this evaporation behavior, indicating the cluster size at which solvation shells are completed. However, there is no evidence for a solvent shell closing here. It seems unlikely that the first solvent shell closing would occur at a cluster size larger than 12. If this is not the case, it must be true that there is no abrupt difference in binding interactions at the early solvent shells in this system.

Mg+-H20

Magnesium-water complexes provide another fascinating example of metal complex spectro-scopy. According to Bauschlicher and co-workers, this complex is also bound electro-statically, but by charge-dipole interactions [35]. As in the C0 2 complex, the metal ion is expected to bind on the oxygen atom of water, which produces a complex with planar C2v symmetry. The orbital diagram for this system is shown in Fig. 9. Three excited electronic states are predicted corresponding to on-axis (2A{), perpendicular in-plane (2Bi), and perpendicular out-of-plane

102.2 Mg+-H20

M g + ( 2 P ) + H 2 0 \

0.0 Mg+(2S)+H20 ,

2 Λ ,

i

Mg

ι

i

265.6 nm

+ - H

2 R i Mg0++H2

i l

334 nm

20(

i ^D2 i MgOH++H

357.7

-32.2 2 A i )

64.2

_44.3

o8

2 2

\"\

1 1

ö 2* 1

r

-1 Mg" ■ - H 2 0

hw

[jj[yi' 5»2»

/ 51

Fig. 9. The energy level diagram showing the excited electronic states expected for the Mg+-H20 complex. The data are from ab initio calculations by Bauschlicher and co-workers [35].

0 1000 2000 3000 4000 Relative Frequency (cm-1)

Fig. 10. The photodissociation spectrum of Mg+-H20. The structure in this spectrum is assigned to vibrational structure and partially resolved rotational structure in two excited electronic states of the complex.

(2B2) orientations of the Mg+ p orbital with respect to the C2 axis. As in the C0 2 complex, the perpendicular orbital arrangements expose the positive ion core more effectively to the oxygen, resulting in stronger binding than the on-axis orientation or the ground state. The electronic transitions to these states are therefore expected to be significantly red-shifted from the atomic resonance line.

The photodissociation spectrum obtained for M g + - H 2 0 is shown in Fig. 10 [36]. It consists of a series of sharp resonances over a region of about 5000 cm- 1. The vibrational assignments are indicated in the figure, showing that we have assigned progressions in the metal-water stretch (mode 2), the metal water in-plane and out-of-plane bends, the water scissors mode, and the H 2 0 asymmetric stretch. The identity of all of these modes are confirmed by a similar study of M g + - D 2 0 . The most prominent mode is the metal-water stretch, which has a frequency of Je = 518 cm"1. As we described for M g + - C 0 2 , the observation of a progression in the metal-water stretching mode makes it possible to determine the dissociation energy of the ion complex. We have also done this here, obtaining a value of 25.0 kcal mol- 1 for the ground state dissociation energy. Additionally, the spectrum shown in Fig. 10 contains partially resolved rotational structure. Simulation of the band contour obtained for both isotopomers, in terms

314 C.S. Yeh et al /Int. J. Mass Spectrom. Ion Processes 131 (1994) 307-317

of the symmetric top structure suggested by theory, confirms that the complex has a planar C2v

structure.

Mg+-CH3OH

Magnesium-methanol complexes can also be formed by these laser vaporization methods, and they have also been studied with theory. We have not yet investigated their spectroscopy, but we have found that they dissociate at 280 nm. As shown in Table 1, several dissociation channels may be expected in the energy region accessible with the laser. Metal oxide and perhaps hydroxide formation are possible. Additionally, hydroxide formation with charge transfer may also occur, producing the methyl radical cation. As shown in Fig. 11, we observe essentially all of these photo-dissociation channels, and their branching ratios are dependent on the laser energy employed. Hydrogen increments on Mg+ or MgO^ ion channels are difficult to ascertain for sure because of the overlap with the 25 and 26 isotope channels of magnesium. However, these channels do appear to indicate hydride and hydroxide ion formation. At 355 nm, a multiphoton process leads to charge transfer, forming the CH3OH+ ion, as discussed below for metal-benzene ions. The ions in this experiment are formed under the same conditions as in the other experiments above, and so we do not believe they have undergone these reactions in the source region. Instead, it seems that there are also photo-induced intracomplex reactions in this system. Using the known energetics of the isolated species, and the energy thresholds of these photo-induced reactions, it should be possible to deduce energetic information on the ion/molecule com-plex dissociation energy. These experiments are presently ongoing in our lab.

Mg^-benzene

Figure 12 shows the photodissociation prod-ucts which result from decomposition of the Mg+-benzene complex [37]. As indicated, both the

Mg+-CH 3 0H

266 n m

25 35 45 55

Mass ( a m u )

Fig. 11. Photodissociation mass spectra of Mg+-methanol at various excitation energies.

Mg+ and C6H^ fragment ions are formed. The Mg+ channel is usually observed from dissociation of these metal ion complexes, as indicated in the other systems described here. However, the observation of the benzene cation channel is surprising because of the energetics required to produce this channel. The ionization potential of magnesium is 7.646 eV, while that of benzene is 9.24 eV. Dissociation to produce the


M g + — b e n z e n e

bz +

L, 1 1 1

0 50 100 150

Mass ( a m u )

Fig. 12. Photodissociation mass spectrum of Mg+-benzene, indicating evidence for dissociative charge transfer.

benzene cation fragment is a higher energy process by about 1.6eV. This channel therefore cannot result from dissociation on the ground electronic state.

We have actually observed this kind of dissocia-tion behavior for a variety of metal ion-benzene complexes [37]. Because of the low ionization potential of benzene compared to the other small molecules studied in complexes here, the electronic level structure which results in benzene com-plexes has a new feature. The ground state of M+-benzene complexes usually correlates to an asymptote of M + + benzene because of the lower ionization energy of M relative to benzene. However, there is also a low energy excited elec-tronic state correlating to the M + benzene* asymptote, which can be referred to as a "charge transfer" excited state. Charge transfer electronic transitions are well known in condensed-phase inorganic chemistry, but until recently they had not been observed for gas phase metal complexes. The electronic transition from the ground state in these complexes to the charge transfer excited state is strongly allowed, and therefore absorption into this state can compete with other chromophores in the system, such as the solvated 2S —>2 P atomic transition.

For absorption into the charge transfer state to produce the benzene cation fragment, however, the vertical electronic excitation must access the

excited potential surface above its dissociation limit. As we have described previously, this excited state dissociation process requires an energy greater than or equal to the ground state dissocia-tion energy of the complex plus the ionization potential difference between the metal and benzene. By measuring the threshold energy at which charge transfer dissociation first occurs, and using the known ionization potential dif-ference, it is possible to determine an upper limit on the ground state dissociation energy of the complex:

D'o^hvcj- ΔΙΡ

Binding energies of these metal ion complexes are in general not known and they are difficult to measure. This photochemical threshold method has now been used to obtain dissociation energies for several metal ion-benzene complexes. In the case of magnesium, the threshold for the appearance of the benzene cation channel is at 450 nm, which puts an upper limit on the dissociation energy of 26.9 kcal mol- 1 . This can be compared to the similar binding energy of Ag+-benzene (30.0 kcal mol"1) and the much larger value for Fe+-benzene (62.1 kcal moP1) determined by this same method [37]. The dissocia-tion processes in Mg+-benzene therefore seem to be determined by absorption into the two electro-nic systems of the charge transfer state and the solvated atomic resonance line.

Mg* -acetone

The photochemical dynamics in ion/molecule complexes such as those shown here is expected to become even more varied and complex as the size of the system increases. The Mg*-acetone system provides an initial example of the com-plexity which may be expected in larger systems. As shown in Fig. 13, the dissociation product from this ion is mainly Mg+, but there are minor channels corresponding to MgCH^ and CH3CO+. We believe that the CH3CO+ ion is the result of a charge transfer dissociation like that reported

316 C.S. Yeh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 307-317

Mg+

L UeCHÎ CHjCO

4>

Mg+-

+

(Acetone)

355 nm

7 25 50 75

M a s s ( a m u )

100

Fig. 13. Photodissociation mass spectrum for M g+-acetone.

above for Mg+-benzene. The acetone molecular ion would be the initial result of this process, but if there is excess energy (about 0.65 eV is required), the acetone ion can fragment to form the acetyl fragment ion observed. Charge transfer is possible in this system, as in the benzene system,because of the low ionization potential of acetone (9.7 eV). The MgCH^ ion is believed to result from photo-chemistry localized on the acetone moiety in the complex. Isolated acetone dissociates to form the acetyl radical and the methyl radical, and so the metal-methyl ion could result from this process occurring with the metal ion as a spectator. While these mechanisms are difficult to prove, they are energetically and chemically reasonable. More-over, they illustrate the fact that the photo-chemistry in larger complexes may have many possible branchings depending on the various chromophores in the system, their relative probability of absorption, and the dissociation paths most accessible.

Conclusion

We have shown here how a new reflectron mass spectrometer system can be used to study the photodissociation processes of a variety of magnesium ion complexes. In some systems, precise spectroscopic data has been obtained, making it possible for the first time to determine the structures of metal-containing ion complexes.

In other systems, the photochemistry is more qualitative, but thermochemical information on metal-molecular binding energies may be obtained. These experiments are still very new, and they serve to illustrate the many possible varieties of photochemistry which can be studied. The reflectron system, when combined with the pulsed molecular beam and laser photo-dissociation methods, provides a powerful experi-mental tool with which to investigate fundamental phenomena in ion/molecule cluster systems.

Acknowledgments

This research is supported by the National Science Foundation through grant CHE-9008246. This research is also supported by the Air Force Office of Scientific Research through grant no. AFOSR-91-0001.

References

1 B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A. Zagulin, Zh. Eksp. Teor. Fiz., 64 (1973) 82.

2 U. Boesl, H.J. Neusser, R. Weinkauf and E.W. Schlag, J. Phys. Chem., 86 (1982) 4857.

3 D.M. Lubman, W.E. Bell and M.N. Kronick, Anal. Chem., 55 (1983) 1437.

4 H. Kuehlewind, H.J. Neusser and E.W. Schlag, Int. J. Mass Spectrom. Ion Phys., 51 (1983) 25.

5 (a) H. Keuhlewind, H.J. Neusser and E.W. Schlag, J. Phys. Chem., 88 (1984) 6104. (b) H. Kuehlewind, H.J. Neusser and E.W. Schlag, J. Phys. Chem., 89 (1985) 5600.

6 T.P. Martin, T. Bergmann, H. Gröhlich and T. Lange, J. Phys. Chem., 95(1991)6421.

7 T. Bergman, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 60 (1989) 347; 60 (1989) 792.

8 (a) W.F. Hadden and F.W. McLafferty, J. Am. Chem. Soc, 90(1968)4745. (b) W.F. Hadden and F.W. McLafferty, Anal. Chem., 41 (1969)31.

9 J.T. Stults, C G . Enke and J.F. Holland, Anal. Chem., 55 (1983) 1323.

10 G.L. Glish and D.E. Goeringer, Anal. Chem., 56 (1984) 2291.

11 G.L. Cooks, l l th International Mass Spectrometry Con-ference, Bordeaux, 1988.

12 M.E. Geusic, M.F. Jarrold, T.J. Mcllrath, R.R. Freeman and W.L. Brown, J. Chem. Phys., 86 (1987) 3862.


13 P.J. Brucat, L.S. Zheng, F.K. Tittel, R.F. Curl and R.E. Smalley, J. Chem. Phys., 84 (1986) 3078.

14 Y. Liu, Q.L. Zhang, F.K. Tittel, R.F. Curl and R.E. Smalley, J. Chem. Phys., 85 (1986) 7434.

15 (a) M.L. Alexander, N.E. Levinger, M.A. Johnson, D. Ray and W.C. Lineberger, J. Chem. Phys., 88 (1988) 6200. (b) N.E. Levinger, D. Ray, M.L. Alexander and W.C. Lineberger, J. Chem. Phys., 89 (1988) 5654. (c) J. Papanikolas, J.R. Gord, N.E. Levinger, D. Ray, V. Vorsa and W.C. Lineberger, J. Phys. Chem., 95 (1991) 8028.

16 L.A. Posey and M.A. Johnson, J. Chem. Phys., 89 (1988) 4807.

17 (a) P.Y. Cheng, K.F. Willey, J.E. Salcido and M.A. Duncan, Int. J. Mass Spectrom. Ion Processes, 102 (1990) 67. (b) K. LaiHing, P.Y. Cheng, T.G. Taylor, K.F. Willey, M. Peschke and M.A. Duncan, Anal. Chem., 61 (1989) 1458. (c) D.S. Cornett, M. Peschke, K. LaiHing, P.Y. Cheng, K.F. Willey and M.A. Duncan, Rev. Sei. Instrum., 63 (1992) 2177.

18 (a) S.K. Shin, Y. Chen, S. Nickolaisen, S.W. Sharpe, R.A. Beudet and C. Wittig, Adv. Photochem, 16 (1991) 249. (b) S. Buelow, M. Noble, G. Radhakrishnan, H. Reisler, C. Wittig and G. Hancock, J. Chem. Phys., 90 (1986) 1015.

19 R.G. Wheeler and M.A. Duncan, J. Phys. Chem., 90 (1986) 3876.

20 B. Brutschy, J. Phys. Chem., 94 (1990) 8637. 21 T. Maeyama and N. Mikami, J. Phys. Chem., 95 (1991)

7197. 22 J.C. Loison, C. Dedonder-Lardeux, C. Jouvet and D.

Solgadi, J. Phys. Chem., 95 (1991) 9192. 23 (a) M.F. Jarrold, L. Misev and M.T. Bowers, J. Chem.

Phys., 81 (1984)4369. (b) H.S. Kim, C.H. Kuo and M.T. Bowers, J. Chem. Phys., 87 (1987) 2667. (c) C. M. Roehl, J.T. Snodgrass, C.A. Deakyne and M.T. Bowers, J. Chem. Phys., 94 (1991) 6546.

24 (a) D. Ray, N.E. Levinger, J.M. Papanikolas and W.C. Lineberger, J. Chem. Phys., 91 (1989) 6533. (b) M. L. Alexander, N.E. Levinger, M.A. Johnson, D. Ray and W.C. Lineberger, J. Chem. Phys., 88 (1988) 6200. (c) J.M. Papanikolas, J.R. Gord, N.E. Levinger, D. Ray, V. Vorsa and W.C. Lineberger, J. Phys. Chem., 95 (1991) 8028.

25 L.A. Posey and M.A. Johnson, J. Chem. Phys., 89 (1988) 4807.

26 M. Shen and J.M. Farrar, J. Phys. Chem., 93 (1989) 4386. 27 (a) K.F. Willey, P.Y. Cheng, K.D. Pearce and M.A.

Duncan, J. Phys. Chem., 94 (1990) 4769. (b) K.F. Willey, P.Y. Cheng, M.B. Bishop and M.A. Duncan, J. Am. Chem. Soc, 113 (1991) 4721.

28 (a) J.A. Syage and J. Steadman, Chem. Phys. Lett., 166 (1990) 159. (b) J. Steadman and J.A. Syage, J. Chem. Phys., 92 (1990) 4630.

29 D.E. Lessen, R.L. Asher and P.J. Brucat, J. Chem. Phys., 93 (1990) 6102; 95 (1991) 1414.

30 (a) L. Operti, E.C. Tews and B.S. Freiser, J. Am. Chem. Soc, 110(1988)3847. (b) L. Operti, E.C. Tews, T.J. MacMahon and B.S. Freiser, J. Am. Chem. Soc, 111 (1989) 9152.

31 (a) C. Jouvet and B. Soep, Chem. Phys. Lett., 96 (1982) 26. (b) C. Jouvet, M. Boivineau, M.C. Duval and B. Soep, J. Phys. Chem., 91 (1987) 5416.

32 (a) R.P. Blickensderfer, K.D. Jordan, N. Adams and W.H. Breckenridge, J. Phys. Chem., 86 (1982) 1930. (b) W.H. Breckenridge, C. Jouvet and B. Soep, J. Chem. Phys., 84 (1986) 1443.

33 (a) K.F. Willey, C.S. Yeh, D.L. Robbins and M.A. Duncan, Chem. Phys. Lett., 192 (1992) 179. (b) C.S. Yeh, K.F. Willey, D.L. Robbins and M.A. Duncan, J. Phys. Chem., 96 (1992) 7833. (c) C.S. Yeh, K.F. Willey, D.L. Robbins and M.A. Duncan, J. Chem. Phys., 98 (1993) 1867.

34 M. Sodupe, C.W. Bauschlicher and H. Partridge, Chem. Phys. Lett., 192 (1992) 185.

35 (a) C.W. Bauschlicher, Jr. and H. Partridge, J. Phys. Chem., 95 (1991) 9694. (b) C.W. Bauschlicher, M. Sodupe and H. Partridge, J. Chem. Phys., 96 (1992) 4453. (c) M. Sodupe and C.W. Bauschlicher, Jr., Chem. Phys. Lett., 195 (1992) 494.

36 (a) C.S. Yeh, K.F. Willey, D.L. Robbins, J.S. Pilgrim and M.A. Duncan, Chem. Phys. Lett., 196 (1992) 233. (b) K.F. Willey, C.S. Yeh, D.L. Robbins, J.S. Pilgrim and M.A. Duncan, J. Chem. Phys., 97 (1992) 8886.

37 (a) K.F. Willey, P.Y. Cheng, K.D. Pearce and M.A. Duncan, J. Phys. Chem., 94 (1990) 4769. (b) K.F. Willey, P.Y. Cheng, M.B. Bishop and M.A. Duncan, J. Am. Chem. Soc, 113 (1991) 4721. (c) K.F. Willey, C.S. Yeh, D.L. Robbins and M.A. Duncan, J. Phys. Chem., 96 (1992) 9106.


Resonance-enhanced two-photon ionization time-of-flight spectroscopy of cold perfluorinated polyethers and their external and internal van der Waals dimers

Deon S. Anex*, Mattanjah S. de Vries, Arno Knebelkamp^ Joachim Bargon*, H. Russell Wendt, Heinrich E. Hunziker

IBM Research Division, Almaden Research Center, 650 Harry Road, San Jose, CA 95120-6099, USA


Abstract

Perfluorinated polyethers of the type Ra-(OCF2CF2CF2)„-F, where Ra is the end group C 6 H 6 - 0 - C H 2 C H 2 - 0 -(C=0)CF2CF2 -, were laser-vaporized and entrained in a pulsed jet expansion. Two photon ionization of the jet-cooled polymers via the phenoxy chromophore was combined with time of flight (TOF) mass spectrometry. Mass spectra were obtained for polymer distributions extending to 7000 Da, with minimal fragmentation. Under the appropriate expansion and desorption conditions parent masses of van der Waals dimers of these polymers were also observed. By scanning the ionization laser and monitoring particular mass-to-charge ratios, resonance-enhanced two photon ionization (R2PI) spectra were obtained for the jet-cooled polymers and their dimers near the electronic origin. Polymers with two end groups, present as an impurity in the samples, were detected exclusively in an internally dimerized form. In both the internal and external cases, the dimerization occurs only at the phenoxy chromophore. The R2PI spectra of a series of model compounds were measured and used to characterize the evolution of the spectra from phenol toward the polymer. The model compound spectra revealed the role of multiple conformations and molecular size in the polymer spectra, which are ultimately broadened by low frequency motions of multiple conformers. The results are discussed relative to the general problem of the photoionization of large molecules.

Key words: Resonance enhanced two-photon ionization; Perfluorinated polyethers; Van der Waals dimers

Introduction

Laser desorption has demonstrated great poten-tial for the study of complex molecules through the introduction of nonvolatile material into the gas phase. Large organic molecules have successfully

* Corresponding author. f Current address: Th. Goldschmidt AG, Goldschmidtstrasse, 100, D-4300 Essen 1, Germany. * Current address: Institut für Physikalische Chemie, Universität Bonn, 5300 Bonn, Germany.

been desorbed from various substrates with no or minimal fragmentation. While in a simpler version, a single laser pulse can vaporize the sample and also perform ionization, it is advantageous to separate these processes. A particularly powerful approach for the postionization step is that of resonance-enhanced two photon ionization (R2PI). When combined with time-of-flight (TOF) mass selection, this technique can provide detailed spectroscopic information about a specific molecule in a mixture and is very sensitive in

320 D.S. Anex et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 319-334

favorable cases. To capitalize on the strengths of R2PI detection it is necessary that the desorbed molecules be cooled by effective entrainment in a supersonic expansion. One advantage of cooling the internal degrees of freedom is the reduction in fragmentation, since less internal energy is available. A second advantage is an increase in sensitivity and selectivity, resulting from the cooling of molecules to the lower rotational and vibrational states, causing narrowed spectral lines and allowing the exciting laser line to interact with a larger fraction of the molecular population.

A disadvantage of this technique is that the molecule of interest may not absorb light at a convenient wavelength. Perfluorinated polyethers (PFPEs) are an example of such a case. To overcome this problem, a chromophore attached to the polymer can serve as a tag for R2PI studies. This approach has been demonstrated for DNA with an anthracene tag [1] and for non-aromatic peptides derivatized with a phenyl-containing chromophore [2].

Whereas many groups have successfully desorbed extremely large molecules into the gas phase [1,3-5], fewer have used jet-cooling techni-ques. Schlag and co-workers have produced mass spectra of a variety of large, jet-cooled molecules [6-8], including peptides [9,10] and polyenes [11]. These mass spectra are often dominated by parent ions, with the degree of fragmentation controllable with the intensity of the ionizing laser. Lubman and co-workers have also used similar techniques to study a variety of derivatized peptides [2] and aromatic polymers [12]. However, the mass spectra obtained for the aromatic polymers were dominated by the monomer unit that makes up the polymer or by fragment ions, with large oligomers rarely seen.

Even fewer examples exist where the ionization wavelength was scanned to generate the R2PI spec-trum of jet cooled species. To date, this technique has been successfully applied to molecules of mod-erate size only, typically with masses of several hundred atomic mass units and below. Cable et al. have obtained spectra of cold dipeptides and

tripeptides containing the amino acid tryptophan, which served as the chromophore [13,14]. These spectra provide information regarding the mole-cular conformations in the vicinity of the chromo-phore and serve as a nice complement to fluorescence studies [15,16]. Li and Lubman have measured the spectra of tyrosine-containing molecules cooled in a supersonic expansion [17]. Meijer et al. have obtained the jet-cooled spectra of a number of large organic molecules and their dimers [18-21].

Several factors related to molecular size make application of these techniques more difficult as the target molecules become larger. With increas-ing number of atoms in a molecule comes an increase in the number of degrees of freedom and low frequency vibrational modes. As a result, sub-stantially more vibrational states remain populated after cooling in the molecular beam than for molecules of a smaller size cooled to the same temperature. Metestable conformations also appear [22]. These effects multiply and ultimately broaden the R2PI spectrum.

Schlag and co-workers have cited several factors expected to impede the photoionization of larger molecules [23,24]. The leading one is delayed ionization. In this picture, the electronic energy needed to eject the electron to form the ion is tied up in vibrational motion before ionization, allow-ing other processes such as bond dissociation to compete. Other factors that may impede ioniza-tion of large molecules include unfavorable Franck-Condon factors, attachment of the ejected electron to another part of the molecule and excessive internal energy that dissociates the parent ion. In the current work we have not found these factors to present a significant impedi-ment to photoionization of long-chain polymers with terminal chromophores.

In this paper we report the extension of the tech-nique of R2PI-TOF mass spectrometry with laser-vaporization and jet-cooling to perfluorinated polyethers (PFPEs). These polymers serve as important industrial lubricants because of their chemical inertness, thermal stability and very low

D.S. Anex et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 319-334 321

vapor pressure. We have studied polymers func-tionalized with an end group that contains a phenyl ring which serves as a chromophore:

Ra-[OCF2CF2CF2]„-F I

where Ra is the aromatic end group C 6 H 5 -0 -CH2CH2-0-(C=0)-CF2CF2\ These poly-mers are available as Demnum SP® from Daikin Industries and will be referred to as type I. We have obtained mass spectra that extend to 7000 Da and are dominated by the parent ions of these polymers using ionization wavelengths fixed at 193 nm and near 274 nm.

We have measured the R2PI wavelength spectra in the vicinity of the S0-o transition (near 274 nm) of unfragmented polymers with masses up to about 2000 Da. We also studied polymers with the same repeat unit, but with two aromatic end groups:

Ra-[OCF2CF2CF2]„-0-Ra II

which we will refer to as type II polymers, as well as some with branched repeat units. Under appropri-ate expansion and desorption conditions, spectra of van der Waals dimers of the polymers were obtained as well. In order to better understand the polymer spectra, we also studied the R2PI spectra of smaller model molecules representing various lengths of the phenoxy end of the poly-mers. Figure 1 shows the various compounds that we studied, starting with phenol and evolving in complexity toward the PFPE. From trends in observed spectral shifts conclusions can be drawn about the structure and the dynamics of PFPEs, which will be discussed below.

Experimental

The laser desorption jet-cooling apparatus was described in detail before [21,25] and is shown schematically in Fig. 2. Molecules are desorbed from a sample bar on the vacuum side of a pulsed supersonic nozzle. Typically the desorption spot is positioned about 1 mm in front of the nozzle open-ing, which corresponds to a position 2 nozzle dia-meters downstream in the expansion. In the current

à PHENOL

CH3 O

à ANISOLE

O CH2 VH

O à 2-PHENOXYETHANOL

O CH2 Cf [F

à » · ESTERS (n=1,2,3, AND 7)

CF CHg , O w C F r CF2 1 0 W C F 2

O CHo C |0 CF CFo CF,

BRANCHED ETHERS (n-1, 2, AND 3)

CHo ¼ CF2

O CH2 C CF

è i CF2 CF,,]

STRAIGHT CHAIN ETHERS (<n>=8 AND <n>=20)

Fig. 1. Model compounds studied in this work.

studies we used the frequency-doubled output (532 nm) of a pulsed Nd:YAG laser for desorp-tion. Most samples were applied to a substrate of activated charcoal. This provided a surface film, which was replenished between successive laser shots from the subsurface bulk of the porous car-bon. Using a gas pulse from the molecular beam source timed to entrain the desorbed material, the molecules were cooled by collisions with the carrier gas. Xenon, at a stagnation pressure of 8 atm, was found to give the best cooling and most effective entrainment of the heavier molecules.

This method of sample introduction was used for all the molecules studied, with the exception of the anisole dimers. In this case, the charcoal would not hold enough of the more volatile anisole near its surface to create a high enough density to form dimers by laser desorption. Instead, the anisole was seeded in 1.5 atm of xenon in the pulsed jet.

The molecules in the beam were interrogated downstream by a pulse from a second laser, which was timed to ionize them by two photon

322 D.S. Anex et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 319-334

Fig. 2. Experimental apparatus.

absorption in the source of a linear time of flight mass spectrometer. Mass spectra were obtained using the 193 nm output of an ArF excimer laser or the doubled output of a fixed wavelength from a tunable dye laser. One color, R2PI spectra were collected by scanning the frequency-doubled out-put of the dye laser while monitoring ion masses of interest. In our setup several masses can be monitored simultaneously during a single scan to conveniently collect spectra as a function of chain length in a polymer mixture.

In most cases the signal to noise ratio of the R2PI spectra was limited by the shot-to-shot variation in the intensity of the desorption laser. This translates into fluctuations in the amount of material vaporized with each laser pulse and directly into signal variations. Improvements in the signal-to-noise ratio can be made by indepen-dently monitoring the amount of material present in the beam for each shot and normalizing the signal to this quantity after each shot. Efforts are underway to incorporate this approach into the present apparatus.

The smaller model compounds (phenol, anisole

I I I I I I I I I I 1 I I I I I 1 I I I I ! I I I I

HJ4U JJLiii Àt*v»t>-<wjtm, |

I I I I I I I I I

1 000 7000 3000 4000

MASS 5000 6000 7000 8000

(AMU)

Fig. 3. Mass spectrum of the type I PFPE sample with (n) = 20 ionized at 193 nm. The major peaks are parent masses spaced by 166 Da and range from 1114 Da (/i = 5) to 6924 au (n = 40). The origin of the minor peaks is described in the text.

and 2-phenoxyethanol) were obtained com-mercially and the esters with the perfluoroalkyl and branched ether chains were synthesized for this study. The PFPE sample with the shorter chain length distribution was provided as a gift by Daikin Industries and the longer PFPE was a commercial sample from the same manu-facturer.

19F NMR analysis of the sample of the shorter chain PFPE showed it to be 26 ± 2% nonfunc-tionalized (type I with Ra replaced by CF3CF2CF2 ), 66 ± 2% monofunctionalized (type I) and 7 ± 2% bifunctionalized (type II). The composition of the sample of the longer chain PFPE was found by the same technique to be 57 ± 2% monofunctionalized and 43 ± 2% non-functionalized.

Results

Mass spectra

Figure 3 shows the mass spectrum obtained using 193 nm ionization of a sample of the type I PFPE consisting of a distribution of chain lengths. The predominant peaks correspond to parent ions and are spaced apart by 166 Da, the mass of a repeat unit. They range from ra/z—1114Da


º I I | I | I | | | | | | | I | |—I—|—I—|—|—Ã

.JLL JUL /WVXÛJUAJ ' JJ_J d

B

.JJÜJÜJ

o o

UJuJlUaJiu^^ *.** ...—+j*,X »M

I I 1 I 1 l l I l I I I I 1 I I I 1 I I 1 I

500 1000 1500 2000 2500 3000 3500 MASS (AMU)

Fig. 4. Mass spectra of the type I PFPE sample with (n) = 8. (A) Ionization at 273.5 nm yields predominantly parent ions of the type I PFPE, (#). (B) Ionization at 274.8 nm favors ionization of the type II polymers present as an impurity in the sample ( x ) and van der Waals dimers of the type I polymers (O)· Peaks due to these two components of the mass spectrum are not visible in panel A. The parent

masses of the type I polymer are again indicated by # .

(n = 5) to m/z = 6924 Da (n = 40). Minor peaks are due to several components. Polymers that con-tain one repeat unit or an end group that has two rather than three CF2 units are shifted down from the main peaks by 50 Da. Some polymers are functionalized at both ends, containing two aromatic end groups, and appear as peaks shifted up 262 Da from the main peaks. At the higher end of the mass range, peaks due to van der Waals dimers of the polymers are observed. Polymers with up to 40 repeat units and a mass of the order of 7000 Da are readily measured with no observable fragmentation. The intensity distri-bution does not necessarily reflect the exact mole-cular weight distribution of the sample since entrainment and detection efficiencies may not be mass-independent over such a long mass range. However, we note that the observed mass distribu-

tion resembles distributions obtained in this laboratory from similar samples by a laser desorption-cationization technique [26].

Figure 4 shows mass spectra of another sample, which contained a mixture of shorter chain PFPEs of type I. Here, the ionization laser was the doubled dye laser operating at a fixed wavelength. In this case, the first ionization step is the absorption of a photon near the electronic origin. As a result molecules are selected that have absorbtion bands coinciding with the laser wavelength. Panel A shows the mass spectrum when ionizing with 273.5 nm radiation. This wavelength is resonant with polymers of type I and the mass spectrum is again dominated by parent ions, reflecting the shorter and narrower mass distribution of this sample, ranging from five (m/z = 1114 Da) to 13 (m/z = 2442 Da) repeat units. Parent masses


of the type I polymers are indicated by the solid circles.

The mass spectrum in panel B was obtained with the ionizing laser tuned to 274.8 nm. Type I poly-mers no longer dominate the mass spectrum. Instead, a new set of peaks (indicated by the crosses) appears that was not apparent in panel A. The new peaks correspond to polymers with two aromatic end groups, the type II polymers. This ionizing wavelength favors these polymers, whose parent ions now dominate the spectrum. In addition, parent ions of the van der Waals dimers of polymer I, indicated by the open circles, appear at the higher mass end of the spectrum. Thus, the wavelength that favors ionization of the polymers with two end groups also favors ionization of the dimers of the single end group compound. This suggests that the chromophores in the two species are similar. The reason for this similarity will be discussed below. In none of the mass spectra do we find evidence of delayed ionization, which would be indicated by tailing peaks.

Wavelength spectra

By monitoring a particular mass-to-charge ratio in the mass spectrum while tuning the wavelength of the ionization laser, a R2PI spectrum is obtained. Figure 5 shows R2PI spectra of a series of molecules of increasing size and complexity which were chosen to model the chromophore end of the polymer and converge to the structure of the PFPEs. We assign the main features in these spectra to the respective S0_o transitions for two reasons. (1) In each case the peak is close to the wavelength of the known S0_o transitions of phe-nol and anisole and (2) no other spectral feature which can be assigned to the electronic origin is observed when scanning at least 1300 cm"1 to the red. Panel A shows the spectrum of phenol; panel B, anisole; and panel C, 2-phenoxyethanol. Panels D through G show 2-phenoxyethyl esters of per-fluorinated carboxylic acids, with increasing length of the perfluoroalkyl chain. In H through J a series of esters with branched perfluorinated polyether

chains is shown. Finally panels K through N show spectra of straight chain perfluorinated poly-ethers found in the PFPE sample with the shorter chain distribution. Generally, the spectra evolve smoothly from phenol to the polymers. The transition shifts to higher energy in going from phenol (A) to the first ester (D). In D, E, and F, structure is observed that may be related to several electronic origins corre-sponding to different conformations of the molecule. In proceeding from the esters to the branched ethers (H-J) the spectra shift back slightly toward lower energy. The straight chain polymers (K-N) show no further shift or broadening. It appears that the spectra have converged to a limiting value at three ether oxygens in the chain (J) and the branched and straight chain polyethers have similar spectra.

When material is desorbed with high enough density into the early part of the supersonic expan-sion it is possible to form small van der Waals clusters. Figure 6 shows spectra of the dimers of type I PFPEs, obtained from the sample with the lower average molecular weight. Since the sample contained a distribution of chain lengths, each peak in the mass spectrum is due to a mixture of dimers. Each dimer mass can only be assigned in terms of the sum of two monomer chain lengths. For example, a dimer mass peak corresponding to a chain length of 16 can contain contributions of monomers with n and m repeat units in any combination for which n + m=\6. When a dimer peak in the mass spectrum is monitored dur-ing the wavelength scan, the resulting spectrum is a superposition of mixed dimers with the same mole-cular weight. The panels in Fig. 6 are labelled accordingly. All the dimer spectra show a broad-ening and a characteristic redshift of about 110 cm"1 from the monomer wavelength. Although the sample contains 26% nonfunctionalized poly-mers in addition to 66% of the monofunctionalized ones, no mixed dimers involving a nonfunctionalized polymer were observed. Dimers of the nonfunction-alized polymers are, of course, not observable.

For comparison with the polymer clusters, we measured the spectrum of the anisole dimer. The spectrum showed a sharp origin shifted -215 cm"1


2750 2740 2730 2750 2740 2730 WAVELENGTH (ANGSTROMS) WAVELENGTH (ANGSTROMS)

Fig. 5. R2PI spectra of the series of model compounds and PFPEs illustrated in Fig. 1: (A) phenol; (B) anisole; (C) 2-phenoxyethanol; (D-G) 2-phenoxyethyl esters of perfluorinated carboxylic acids: perfluoroacetate, perfluorobutyrate, perfluoropropionate and perfluor-

ooctanoate; (H-J) 2-phenoxyethyl esters with branched ether chains: n = 1,2 and 3; (K-N) type I PFPE with n = 5, 6, 7 and 9.

from anisole monomer. The trimer absorbed in the Figure 7 shows the R2PI spectra of doubly same region, but gave a much broader signal functionalized PFPEs of type II for a number of (about 100 cm-1 full width at half maximum) chain lengths. These spectra exhibit broadening without any sharp structure. and redshifts remarkably similar to those of the


E l i I I | I I I I | 1 I I I | I 1 1 M

Fi i .i..i_.l i .I_I_L.L± i i i i i i i iH 2760 2750 2740 2730 2720

WAVELENGTH (ANGSTROMS)

Fig. 6. R2PI spectra of van der Waals dimers of type I PFPE. Panels A-D are for n + m — 11, 12, 14 and 16, respectively, where n + m represents the sum of the number of units in the two polymer chains.

dimer spectra. The similarity between the spectra of the dimers of the type I polymers and the isolated, doubly functionalized type II polymers leads to the conjecture that the chromophore environment in both cases is similar, which is the case if the type II polymer chromophores are intramolecularly dimerized. When comparing the wavelength spec-tra of monomers (Fig. 5) and dimers (Figs. 5 and 6) with the mass spectra (Fig. 4), it is important to bear in mind that the dimer absorption is much weaker than that of the monomer due to lower concentration. Also the ratio of the type I dimer to type II absorption depends on the density of mate-rial injected in the molecular beam, that is, on the desorption laser intensity and the height of the sam-ple relative to the pulsed jet orifice.

To explore the dimer spectra more fully, the dimers of the shorter chain branched ethers were studied. Figure 8 shows R2PI spectra of dimers of the branched polymers with varying numbers of repeat units. At n = 3 the spectrum is remarkably

similar to the dimers of the type I polymers (Fig. 6), but at lower values of n a double-peaked structure appears.

Discussion

Ionization of large molecules

The data reported here represent an extension of R2PI spectroscopy to heavier molecules, i.e. those with masses between 1000 and 2000 Da. There are a number of reasons why application of the laser desorption jet-cooling spectroscopy to larger mole-cules is more challenging. Foremost is the fact that the large increase in the number of internal degrees of freedom and low frequency modes with number of atoms may increase the number of very low frequency vibrational states populated in a jet-cooled molecule. This then broadens the spec-trum, which reduces sensitivity and specificity of detection.

There is current interest in possible fundamental limitations on the size of molecules that can be efficiently ionized [23,24]. Schlag and co-workers have proposed that the origin of this limitation is related to the number of vibrationally excited levels that are coupled to the ionizing level. For a large molecule there is a very large number of non-ionic, superexcited states that are isoenergetic with the initially excited ionic state. It has been proposed that the loss of electronic energy (needed to eject the electron) to vibrational degrees of freedom as a result of coupling to these states can interfere with ionization. This is the ionization analog to internal conversion in fluorescence experiments.

A molecule excited above the ionization thres-hold, but having its energy in the wrong degrees of freedom for ejecting the electron, may undergo other processes. First, the molecule may auto-ionize, resulting in delayed ionization. Second, the energy-rich molecule may undergo bond dissociation, which competes with ionization. Third, the electron may attach to a remote part of the molecule, forming a zwitterion.

We find no evidence for delayed ionization in the


i i i i | i i i i | : i i i | M i i

h i i i 1 i i i i 1 i i i i I i é é i]

ÃÇ-Ç-fiH ijf II I I II II] L / V \ D J

M I I I I I I I I I I I I 1 I I I I N

2760 7750 7740 7730 7770 WAVELENGTH (ANGSTROMS)

Fig. 7. R2PI spectra of the doubly functionalized type II PFPE. Panels A - D are for n = 4, 5, 7 and 9, respectively.

current results, which include ionization of molecules up to 7000 Da. Nor do we find evidence of a gross decline in ionization efficiency with molecular size.

Our results also show that other factors thought to impede R2PI of large molecules are not signifi-cant in the examples studied here. First, a decline in the ionization efficiency due to increasingly unfa-vorable Franck-Condon factors with increasing molecular size is not germane. The sharp, isolated electronic origin of the chromophore indicates a vertical transition. The chromophore in the mole-cules studied here is fairly isolated from the rest of the polymer and is insensitive to changes in poly-mer size beyond a certain point (see Panels H - N in Fig. 5). Second, we find no evidence for the attach-ment of the ejected electron to the rest of the mole-cule. Dissociative attachment of electrons to PFPEs is a favorable process [26]. The separation of a negatively charged fragment [27] would result in positively charged ions with the polymer chain severed. We find no evidence of this in the PFPE mass spectra. Although the examples studied here may be classified as a special case (terminal chro-mophores attached to a perfluorinated chain), their

M I I I I I I I I I I I I I I I I I H

F A y X / \ i

| Ï M { | f + H ^ f f - f | - i f i ï i F B r ^ ^ s ^ 1

F I i i i I i i i i I i i i i I i i i i | F. c ^ - A ^ j

F Ï I I 1 1 I I I I 1 I I I I 1 I I I Ð

7 / 6 0 ?/b0 ?740 7 Ë50 ?/?0 WAVELENGTH (ANGSTROMS)

Fig. 8. R2PI spectra of van der Waals dimers of the branched chain ethers. Panels A-C are for n = 1,2, and 3 in the polymer chain, respectively.

behavior should be considered in general theories of photoionization of large molecules.

Evolution of the PFPE R2PI spectrum

The sequence of compounds in Fig. 5 represents a systematic series of molecules containing the phenoxy chromophore, from phenol to the perfluorinated polyethers. The spectral shift in going from phenol to the PFPEs covers roughly 200 cm- 1. In the first step, from phenol (panel A) to anisole (panel B), the spectrum does not change appreciably, with the single line corresponding to the electronic origin of the first excited electronic state shifting 32 cm"1 to higher energy, as has been previously determined [28-30]. The single line in each case indicates a single molecular conforma-tion, as reported for anisole by Breen et al. [30]. In their study of ethoxybenzene, workers from this group also reported a single molecular confor-mation, whose electronic origin is shifted 9 cm- 1 to lower energy for anisole [31]. In panel C, we find that the spectral shift due to the addition of the OH group to ethoxybenzene, to form 2-phenoxyetha-nol, is 104 cm- 1 to higher energy (relative to ethox-ybenzene). Again, a single molecular conformation is observed.


The sequence of spectra in panels D - F is parti-cularly remarkable, suggesting that the preferred conformation in the jet-cooled molecule depends critically on the number of carbons attached to the ester linkage. In panel D, the spectrum for 2-phenoxyethyl perfluoroacetate shows the first evi-dence for multiple conformations. The predomi-nant conformer exhibits an additional 104 cm- 1

shift from 2-phenoxyethanol. Minor peaks are observed at shifts of -77 cm- 1, -34 cm- 1 and + 67 cm- 1 from the major peak. That the minor peak at the +67 cm- 1 shift is indeed another con-former is indicated by the striking change in the spectrum going from the perfluoroacetate in panel D to the perfluoropropionate in panel E. By simply increasing the length of the perfluor-oalkyl chain by one CF2 unit, the major confor-mer now corresponds to the one shifted + 67 cm- 1 from the major one in the perfluoroace-tate.

With an additional CF2 in the chain, the spec-trum of the perfluorobutyrate (panel F) now resem-bles a combination of panels D and E. The major conformation has changed back to the one of the perfluoroacetate, but a significant contribution still comes from the one predominating in the perfluor-opropionate. Additional peaks appear in panel F which we interpret as additional minor conforma-tions that are shifted slightly from the major ones. We also expect a contribution from transitions that arise from molecules that are not in their vibra-tional ground state. With increasing size, the increasingly lower frequency torsional and bend-ing modes remain populated at the low tempera-ture of the molecular beam. Transitions from these states tend to further congest the spectrum.

The appearance of the spectrum of the 2-phenox-yethyl perfluorooctanoate shown in panel G can be interpreted as arising from multiple conformations broadened by transitions from molecules left with vibrational energy remaining in the long perfluor-oalkyl chain. The predominant conformations seem to be built on the same geometry that resembles the important ones in panel D and F, which would explain the position of the center of the band. Pre-

sumably, a combination of a range of conformers and broadening due to thermal congestion pro-duces the spectral width in panel G. Note that the bandwidth of this spectrum (about 100 cm"1 full width at half maximum) roughly matches the range spanned by the features in panel F for the perfluoropropionate.

The preceding discussion of the spectra in panels A - G showed the development of the spectrum as the molecule is built up through the ester linkage. The simplest ester was shown in panel D, with E - F showing the effect of increasing chain length. Panels H - N in the second column of Fig. 5 explore the effect of including ether linkages and then building the chain towards the PFPEs. Note that the structure of the chromophore end of these molecules through the ester linkages are identical to the esters discussed above. The spectra in panels H-J are of branched ethers, with increasing num-ber of ether oxygens. In panels K - N the spectra are shown of straight chain PFPEs with 5, 6, 7 and 9 repeat units from the polymer mixture whose mass spectrum is shown in Fig. 4.

The trend in the spectral shift in the branched ethers is a slight shift back towards the lower energy as the number of ether oxygens increases in panels H-J . The spectral width is again presum-ably due to a combination of multiple conforma-tions, absorbing at different wavelengths, and congestion due to transitions from molecules not in their vibrational ground state. These effects do not change significantly with length in this series.

A comparison of the branched polyether in panel J with the straight chain polyether (with two more additional repeat units) in panel K suggests that there is no difference in the spectra of the branched and straight chain polymers. The progression through panels K-N, corresponding to increasing the number of repeat units, indicates that both the position and width of the spectra are insensitive to chain length in this range.

In general, the spectral shift has reached its limit at the perfluoroacetate (panel D). Although an anomalous shift does appear in the perfluoro-propionate (panel E), the larger molecules seem


to prefer a conformation closer to one that resembles the perfluoroacetate. The spectral width seems to have converged with the first branched ether (panel H). The PFPE spectrum (position and width) seems to have converged at three repeat units, assuming no difference in branched and straight chain spectra, as the data suggest. Apparently, additional repeat units beyond three are too remote from the chromophore to induce additional shifts. The additional units also do not significantly affect the ionization efficiency. With increasing length of the ether chain, vibrations involving low frequency torsional and bending motions that remain populated in the beam must continue to increase in number and decrease in vibrational frequency. Transitions originating from these vibrational levels certainly add additional congestion to the spectrum. However, they must have small shifts relative to those already present in the spectrum and add un-observable broadening to the already broadened peaks.

We note that the increasing spectral bandwidth with increasing molecular size observed in Fig. 5, should serve as a caution about comparing ioniza-tion efficiencies at fixed wavelength. An apparent decline in ionization efficiency may be due to sampling a smaller fraction of the molecular population as the molecules get larger and the broadening effects cited above become important.

Internal and external van der Waals dimers

As seen in Fig. 4, the PFPE sample that is nominally a mixture of the singly functionalized type I polymers contains a small fraction consist-ing of the doubly functionalized type II polymer. We expected the wavelength spectra of the singly and doubly functionalized polymers to be the same, because in an extended polymer chain the chromo-phore groups cannot interact. Since the data in Fig. 5 show that the spectral shift and width are essen-tially unaffected by additions to the molecule beyond the ester linkage and are certainly not sen-

sitive to the length of the PFPE chain, a chromo-phore at one end of a molecule with an extended chain would not be affected by the termination at the remote end of the chain. This expectation of independent chromophores at the ends of the type II polymer chains was not borne out by the experimental results.

The first indication that the two chromophores are not independent in the type II polymers is seen in Fig. 4. If the type II polymers in fact had inde-pendent chromophores, they would appear with the type I polymers in the mass spectrum in panel A, which was taken with the ionization laser tuned to 273.5 nm, near the peak of the type I polymer absorption. Instead, they are most pronounced in the mass spectrum shown in panel B of Fig. 4, taken at 274.8 nm, the ionization wavelength that favors the van der Waals dimers.

The similarity between the chromophores in the van der Waals dimers of the type I polymers and the isolated type II polymers is clearly seen in the wavelength spectra shown in Figs. 6 and 7. In Fig. 6, the spectra of the van der Waals dimers are seen as broad peaks (about 200 cm"1 wide) centered at 274.4 nm, shifted -100 cm"1 for the type I polymer. Recall that only the total number of repeat units is known for a given dimer mass, how they are distributed between the two chains is undeter-mined. In Fig. 7 the wavelength spectra for type II polymers of several chain lengths are seen. There is no apparent variation with chain length and the similarity of the spectra to those of the van der Waals dimers is unmistakable. The type II polymer spectra certainly resemble the van der Waals dimer spectra in Fig. 6 rather than the type I PFPE spectra in panels K-N of Fig. 5. (Note that all spectra are plotted with the same wavelength range.)

The spectra in Figs. 6 and 7 indicate that the chromophores in the isolated type II polymers and the van der Waals dimers of the type I poly-mers are similar. One way of achieving this similar-ity is to form an w/ramolecular complex in the type II polymers that resembles the wteraiolecular com-plex in the van der Waals dimer. This would

330 D.S. Anex et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 319-334

involve interactions between pairs of chromo-phores in each case.

To support this idea, first consider the structure of the external van der Waals dimer. This complex could form through (1) interaction of polymer tails on each member, interaction of polymer tails and chromophores such that (2a) one chromophore interacts with the chain on the other polymer or (2b) both chromophores interact with the polymer chain on the other member of the dimer, or (3) interaction between chromophores on each.

The last possibility, (3), is strongly supported by the dimer redshift. A useful benchmark for comparison is the shift in the anisole dimer. We have measured the dimer S0_o transition shifted -215 cm- 1 from the anisole monomer. This com-pares favorably with -100 cm"1 for the broad PFPE dimer peak, falling within the observed range of shifts and showing that interaction of the chromophores is a plausible explanation for the source of the spectral shifts in the PFPE dimer spectra. The two shifts do not match exactly because anisole only approximates the chromo-phore and the polymer chain may restrict the dimerization geometry from matching that in the anisole dimer.

We argue that the first two possibilities, (1) and (2a), can be immediately eliminated, since each would give rise to a significant feature in the spectrum near where the "free" chromophore absorbs in the type I polymer (273.5 nm), which is not observed.

In (2b), where there is strong interaction between the chromophore and the polymer chain, each chromophore would be bound to a chain, also giving rise to no "free" chromophore. The spectral shift for this interaction is unknown, but this possibility can be eliminated by considering chromophore-chain interactions in other species. While for type I polymers, this interaction could be with the other member of the dimer, for type II polymers it would require the chromophores to complex with the chain of their own molecule. If so, the chromophores of an isolated type I molecule should complex with its own chain just as easily,

but the spectrum shows that it does not. There is no reason to believe that type II polymers form intra-molecular chromophore-chain complexes while type I polymers do not.

The possibility of dimers forming due to chromophore-chain interactions can also be elimi-nated by the absence of dimers between type I polymers and nonfunctionalized polymers in the mass spectrum in Fig. 4. Since the polymer mixture contains 28% of the nonfunctionalized polymer, strong interaction between the chromo-phore and the chain would result in a significant portion of the dimers involving one functionalized and one nonfunctionalized member. These were not observed. We conclude, therefore, that the van der Waals complex must be formed by intimate interaction between the chromophores.

If the intramolecular and intermolecular complexes are formed in the same way, as their wavelength spectra suggest, we expect that they both involve interaction of the chromophore ends. This scenario has two prerequisites. First, the chains must be flexible enough so that during the jet expansion they can efficiently bend to bring the chromophores together. In other words, the barriers to internal rotation must be low enough for the polymers to explore many conformations while they are being cooled, since the experiments suggest that the two ends of the molecule find each other with high efficiency. Second, the interaction between chromophores must be strong enough to effectively form the intramolecular complex once the chromophores are brought together.

The barriers to internal rotation around the bonds in these perfluorinated ether chains are not known accurately. Investigations on the inter-conversion of conformers during a supersonic expansion showed that barriers to internal rotation greater than 400 cm"1 were not effectively relaxed by helium, neon, argon or krypton [22]. Unfortunately, barriers to internal rotation in perfluorinated ethers have not been measured. High quality ab initio calculations for a model ether compound, 1,2-dimethoxyethane, predict barriers in the range of 500 to 800 cm"1 [32]. Ana-


logous barriers for the perfluorinated compound are expected to be similar or lower [33]. Once reli-able barriers for the internal motions of these chains become available, better predictions related to polymer flexibility may be made.

The strength of the ground state interaction between chromophores, the second aspect of effi-cient dimer formation, has also not been measured. Here, the binding energy of van der Waals dimers of smaller molecules may be used as a guide. To our knowledge, the binding energy of the anisole dimer, a useful model of our case, has not been measured. However, binding energies of the benzene dimer and the benzene-anisole complex have been deter-mined. For the benzene dimer, the binding energy is 560 ± 80 cm- 1 [34]. In the benzene-anisole com-plex it increases to 1360-1520 cm- 1, a strong inter-action [35]. It is reasonable to expect that the binding energy for the anisole dimer, containing two polar molecules, would be even larger.

We believe it is significant for the formation of the intramolecular dimer of the type II polymer that the chromophore dimer binding energy is sub-stantially larger than the torsional barrier heights. When the polymers are laser-vaporized from the surface they must have an internal temperature equal to room temperature or higher. At this stage they have sufficient internal energy to surmount the barriers to internal rotation and freely explore many conformations. A certain fraction of these conformations bring the chromo-phore ends together, but at this point the molecules may be too energetic to form a complex.

As the internal energy of the molecules is reduced during the expansion, there comes a stage when there is still sufficient energy for the conformations to interconvert, but the chromo-phores begin to get trapped in the dimer geometry if they happen to come together. If the dimer bind-ing energy is larger then the critical barrier heights, there is still sufficient vibrational energy in the chain for conformational changes after the dimers have begun to form. As the internal energy con-tinues to drop, the chain conformations are even-tually frozen in, but only after almost all the

chromophores have formed intramolecular dimers. Since the torsional barriers are likely lar-ger than 400 cm"1, we expect to see metastable conformations, in agreement with our observa-tions for some of the model compounds.

If this scenario is correct, a chromophore with dimer binding energy equal to or less than the barrier heights should show incomplete or no intra-molecular dimer formation. For example, a type II polymer with 2-phenylethyl rather than 2-phenox-yethyl esters in the end groups, would be predicted to have a substantially lower binding energy, comparable with the barriers to internal rotation. In this case, conformations with free chromphore groups would be expected to freeze before being converted to dimers, resulting in a free chrom-phore spectrum.

Combinations of polymer chains of variable flexibility with different chromophores will pro-vide an interesting arena for making predictions based on barriers to internal rotation and binding energy. Gas phase spectroscopic measurements, similar to those reported here, of the interaction between chromophores at remote positions on polymer chains may be excellent tests of those predictions. It may be possible to gauge the inter-nal barrier heights with a sequence of chromo-phores of ranging dimer strength.

To further explore the energetics of the forma-tion of the internal dimer, we have performed molecular mechanics calculations [36]. This was done to investigate the energy required to bend the polymer chain to form the internal dimer. If this energy was found to be prohibitive, then the strength of the chromophore bond would be imma-terial in the dimer formation. Preliminary results show that for the type II polymer with n = 4 the energy of the molecule with the extended chain is similar to that of the internally dimerized molecule. In fact, the internal dimer may be several kilo-calories per mole lower in energy. Although we have not explored the low energy pathways between the extended and ring geometries, these results indicate that the internal dimer geometry is not unreasonable.


A separate but related issue is the dissipation of the energy released in the complex formation. The formation of dimers of small molecules is accom-panied by stabilizing third body collisions that carry away the energy released when the complex is formed. In the polymer, the abundant low frequency modes may facilitate complex forma-tion by providing a sink for the dissipation of energy released in complex formation. This may result in the formation of a metastable complex that is later stabilized by collisions.

Van der Waals dimers of the branched chain ethers

To better characterize the dimers of the polymers discussed above, we also studied the dimers of a series of branched ethers. These were the same ethers shown in panels H-J of Fig. 5. One advan-tage of using these samples to study the dimers was that they contain molecules of only a single chain length. There was therefore no ambiguity in the length of the chains in the dimer, as there was in the case of the dimers formed from the mixture of type I polymers. Unfortunately, due to the limited availability of starting materials for the synthesis we were forced to use branched ethers. The mono-mer spectra discussed above showed that the branched and straight chain ethers have similar spectroscopy of the chromophore. However, pre-vious studies of similar branched perfluorinated polyethers in this laboratory showed an enhanced propensity for forming complexes with gas phase metal cations compared to the straight chain counterparts [26]. This enhancement was attribu-ted to the dipole moment of the repeat units due to the polarity of the CF-CF 3 bond. Ab initio calculations showed that the repeat unit in the branched perfluorinated polyether has a dipole moment of 1.2 D, whereas the repeat unit in the straight chain (type I) polymer has a dipole moment near zero [37,38]. Different polarity may affect the formation of the van der Waals dimer relative to the straight chain species discussed above.

The wavelength spectra of the dimers of the three

branched ethers are shown in Fig. 8. In some of these spectra two peaks appear, unlike the spectra of the dimers of the type I straight chain polymers, which only have one. The first appears near the monomer absorption and a second is redshifted by about 130 cm"1 from the monomer. The two features are most pronounced in the dimer of the shortest ether {n — 1, where n is the number of ether oxygens), shown in panel A. Proceeding to the n — 2 ether (panel B) the two features are less prominent, yet the spectrum is broad enough to consist of both. For the n = 3 ether in panel C, the redshift continues and the spectrum is begin-ning to resemble that of the dimers of the type I polymers in Fig. 6. A single broad feature is observed, shifted -60 cm- 1 from the n = 3 mono-mer, with no apparent peak in the vicinity of the monomer of this ether.

We interpret the two features in the spectrum in panel A as arising from (1) chromophores that are interacting, giving rise to the redshifted feature, and (2) free chromophores that occur in dimers that are bound together by the branched ether part of the chain. There may also be some contri-bution in the spectrum from chromophores that are complexed by the polar branched ether part of the molecule. The observed trend in the spectrum towards chromophores that are complexed as the size of the ether gets larger, can be explained by arguments similar to those used to explain the increased thermal congestion with size in the monomer spectra. Again, the increasing number and decreasing vibrational frequencies of the low frequency modes of the larger branched ethers results in molecules that survive with vibrational energy in these modes. A significant fraction of this energy will be in the low frequency modes that involve torsional and bending motions. This energy may be available for motion within the cluster as it is being formed, in a sense annealing it to its lowest energy geometry. Presumably, this is the structure with the two chromophores inter-acting in the cluster. The presence of the longer chains in the cluster may also facilitate the inter-conversion to the lower energy geometry by


providing a low energy path with intermediate geometries that involve bonding with the chain. In this picture, the smaller ethers may form dimers that do not have the chromophores interacting and cannot rearrange to the other geometry.

With increasing length of the branched PFPE chain, the appearance of the spectrum converges towards that of the dimers of the longer chain polymers of type I. With the present data it is impossible to completely separate the effect of having branched units in the molecule from the effect of the chain length in the PFPE chain. This is because there is no example where the chain length is the same for both the straight and branched ethers.

The spectra of all the dimers studied, including both internal and external ones, exhibit band-widths considerably broader than the corres-ponding monomers. This broadening arises from two sources. First, the dimers have additional vibrations involving the van der Waals bond. These are extremely low frequency vibrations that may remain populated in the cooled molecules and broaden the spectrum. A second effect arises from different conformations that have varying chromo-phore dimer geometries and, therefore, different spectral shifts from the monomer that contribute to a broadened spectrum.

Summary

We have demonstrated that it is possible to generate parent mass spectra of distributions of functionalized perfluorinated polyethers extending to 7000 au using laser desorption, jet-cooling and two photon ionization via a terminal chromo-phore. That the ionization technique is gentle is reinforced by the observation of weakly bound complexes of the polymers. The sensitivity of this technique is sufficient for the measurement of R2PI spectra for the polymers ranging in size to 1778 au by scanning the ionization wavelength in the vicinity of the electronic origin of the first excited electronic state and monitoring the ion signal at a particular mass-to-charge ratio.

These results show that factors thought to limit the R2PI detection of molecules of this size are not significant for these polymers. We see no evidence of tailing peaks in the mass spectrum which would indicate delayed ionization due to coupling of the electronic and vibrational degrees of freedom. There is also no evidence that the ejected electron is attaching to the polymer chain. This would be plausible for the PFPE chain, but the expected dissociative attachment would result in chain scission, which was not observed. The problem of increasingly unfavorable Franck-Condon factors with increasing molecular size is probably not relevant for these molecules, since the transitions seem to be essentially vertical and the spectrum does not change beyond a certain molecular size.

To better understand the polymer R2PI spectra, we studied a series of model compounds that show the evolution of the spectrum from the simplest molecule, phenol, representing the chromophore in the functional end group of the polymer, through intermediate model compounds represent-ing portions of the polymer of increasing length, to the polymers themselves. In going from phenol towards the straight chain polymer the total wave-length shift is about 200 cm"1. The spectral shift is essentially insensitive to structural changes more than six atoms removed from the chromophore, with the bandwidth also saturating. For the poly-ethers, no further change is seen in the spectrum for polymers containing three or more repeat units and there seems to be no sensitivity to the structural difference between the branched and straight polymer chains. Conformations for the three smaller esters in this series show a very interesting sensitivity to the length of the perfluoroalkyl chain.

The R2PI spectra of the van der Waals dimers of the straight chain polymers and the spectra of the polymers with two phenoxy end groups show a striking similarity, suggesting an accompanying similarity in structure. We argue that in both cases van der Waals complexes are formed between the two chromophores, which are inter-molecular for the dimer and wiramolecular for the bifunctionalized chains. The formation of the


intramolecular complex is an indication of the flexibility of the polymer chains and the strength of the interaction between chromophores. The barriers to internal rotation in the polymer must be low enough to allow the two ends of the chain to hook together before the conformations are frozen. This requires a chromophore dimerization energy which is substantially larger than the critical barrier heights.

Acknowledgments

We thank Daikin Industries for their gift of the short Demnum SP® sample and S. Gahderi (IBM) for providing a sample of the commercial Demnum SP®. We especially thank our colleague R. Johnson (IBM) for the 19F NMR analysis.

References

1 R.J. Levis, L.J. Romano, J. Rajan, D. Schilke and M. DeWitt, personal communication, (1992).

2 L. Li and D.M. Lubman, Appl. Spectrosc, 42(3) (1988) 411.

3 F. Hillenkamp, M. Karas, R.C. Beavis and B.T. Chait, Anal. Chem. A, 63 (1991) 1193.

4 C. Köster, J.A. Castoro and C.L. Wilkens, J. Am. Chem. Soc, 114(1992)7572.

5 D.M. Schielz, C.W. Chou, C.W. Luo, R.M. Thomas and P. Williams, Rapid Commun. Mass Spectrom., 6(10) (1992) 631.

6 J. Grotemeyer and E.W. Schlag, Ace. Chem. Res., 22 (1989) 399.

7 U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spectrom. Ion Processes, 112 (1992) 121.

8 U. Boesl, J. Grotemeyer, K. Müller-Dethlefs, H J . Neusser and E.W. Schlag, Int. J. Mass Spectrom. Ion Processes, 118/119(1992) 191.

9 G.R. Kinsel, J. Lindner and J. Grotemeyer, J. Phys. Chem., 96 (1992) 3157; 3162.

10 G.R. Kinsel, J. Lindner, J. Grotemeyer and E.W. Schlag, J. Phys. Chem., 95 (1991) 824.

11 M. Dey, F. Moritz, G.H. Atkinson, J. Grotemeyer and E.W. Schlag, J. Chem. Phys., 95(6) (1991) 4584.

12 D.A. Lustig and D.M. Lubman, Int. J. Mass Spectrom. Ion Processes, 107 (1991) 265.

13 J.R. Cable, M.J. Tubergen and D.H. Levy, J. Am. Chem. Soc, 110(1988)7349.

14 J.R. Cable, M.J. Tubergen and D.H. Levy, Faraday Dis-cuss, Chem. Soc, 86 (1988) 143.

15 J.R. Cable, M.J. Tubergen and D.H. Levy, J. Am. Chem. Soc, 111 (1989)9032.

16 S.J. Martinez III, J.C. Alfano and D.H. Levy, J. Mol. Spectrosc., 156(1992)421.

17 L. Li and D.M. Lubman, Appl. Spectrosc, 42(3) (1988) 418.

18 G. Meijer, M.S. de Vries, H.E. Hunziker and H.R. Wendt, J. Chem. Phys., 92 (1990) 7625.

19 G. Meijer, M.S. de Vries, H.E. Hunziker and H.R. Wendt, J. Chem. Phys., 94 (1990) 4394.

20 G. Meijer, G. Berden, W.L. Meerts, H.E. Hunziker, M.S. de Vries and H.R. Wendt, Chem. Phys., 163(2) (1992) 209.

21 G. Meijer, M. S. de Vries, H.E. Hunziker and H.R. Wendt, Appl. Phys. B,51 (1990) 395.

22 R.D. Ruoff, T.D. Klots, T. Emilsson and H.S. Gutowsky, J. Chem. Phys., 93(5) (1990) 3142.

23 E.W. Schlag, J. Grotemeyer and R.D. Levine, Chem. Phys. Lett., 190(6) (1992) 521.


25 P. Arrowsmith, M.S. de Vries, H.E. Hunziker and H.R. Wendt, Appl. Phys. B,46 (1988) 165.

26 E.F. Cromwell, K. Reihs, M.S. de Vries, S. Ghaderi, H.R. Wendt and H.E. Hunziker, J. Phys. Chem., 97(18) (1993) 4720.

27 S.M. Spyrou, I. Sauers and L.G. Christophorou, J. Chem. Phys., 78 (1983) 7200.

28 S.J. Martinez III, J.C. Alfano and D.H. Levy, J. Mol. Spectrosc., 152(1992)80.

29 H. Abe, N. Mikami and M. Ito, J. Phys. Chem., 86 (1982) 1768.

30 P.J. Breen, E.R. Bernstein, H.V. Secor and J.I. Seeman, J. Am. Chem. Soc, 111 (1989) 1958.

31 E.R. Berstein, H-S Im and M.A. Young, J. Org. Chem., 56 (1991) 6059.

32 G.D. Smith, R.L. Jaffe and D.Y. Yoon, J. Phys. Chem., (1993) in press.

33 D.Y. Yoon, personal communication, (1993). 34 H. Krause, B. Ernstberger and H.J. Neusser, Chem. Phys.

Lett; 184(1991)411. 35 F. Lahmani, C. Lardeux-Dedonder, D. Solgadi and A.

Zehnacker, J. Phys. Chem., 93 (1989) 3984. 36 MMX calculations, PCMODEL-PI, Serena Software,

Bloomington, IN. 37 J. Pacansky, M. Miller, W. Hatton, B. Liu, A. Scheiner and

R. Waltman, J. Am. Chem. Soc, 113 (1991) 329. 38 R. Waltman and J. Pacansky, personal communication, (1992).


Time of flight mass spectrometry of DNA laser-ablated from frozen aqueous solutions: applications to the Human Genome Project

Peter Williams Department of Chemistry and Biochemistry, Arizona State University, Tempe, AZ 85287-1604, USA

(Received 3 August 1993; accepted 5 August 1993)

Abstract

Time of flight mass spectrometry offers an extremely rapid and accurate alternative to gel electrophoresis for sizing DNA fragments in the Sänger sequencing process, if large single-stranded DNA molecules can be volatilized and ionized without fragmentation. A process based on pulsed laser ablation of thin frozen films of DNA solutions has been shown to ablate intact DNA molecules up to «400 kDa in mass, and also has been shown to yield molecular ions of single-stranded DNA up to «18 500 Da. The theoretical basis and the progress to date in this approach are described and the potential impact of mass spectrometry on large-scale DNA sequencing is discussed.

Key words: DNA; Laser ablation; Sequencing

1. Introduction

The objective of the Human Genome Project is to determine the entire three billion base sequence of human DNA. This will require a DNA sequen-cing effort of unprecedented scale. To accomplish the goals of the project, dramatic improvements in sequencing technology are being sought, aimed at improving sequencing speed and accuracy and increasing the length of DNA strands that can be sequenced in a single pass. While several groups have proposed radically new sequencing strate-gies, mass spectrometry offers a relatively conser-vative approach, aiming to improve the speed and accuracy of just one step in the conventional DNA sequencing process, by reading out DNA sequence "ladders" using time-of-flight mass spec-trometry rather than gel electrophoresis. The widely-used Sänger sequencing process uses

enzyme-catalyzed synthesis to generate four sets of complementary copies of a target DNA tem-plate in which growth is randomly interrupted at each appearance along the strand of a specific DNA base: adenine (A), guanine (G), cytosine (C) and thymine (T). The result is four sets of DNA strands, copied from one end of the tem-plate, whose relative lengths signal the relative positions along the overall copy of the respective bases. Currently, these four fragment sets are ordered by size using gel electrophoresis in four separate lanes. This produces a sequence "ladder" from which the sequence can be read directly simply by associating each successively longer DNA strand with the terminating base correspond-ing to the lane in which the strand was detected. In modern automated sequencers, the strands termi-nating in a specific base are each labelled by differ-ent colored fluorescent dyes and analyzed in a

SSDI0168-1176(93)03887-R

336 P. Williams j Int. J. Mass Spectrom. Ion Processes 131 (1994) 335-344

single electrophoresis lane with laser-induced fluor-escence readout of the different fragments as they pass the detector near the end of the gel. Despite significant improvements in the technology of elec-trophoresis in recent years, it remains a technique with a time scale on the order of hours, and limita-tions of resolution and absolute mass (or length) calibration. Time of flight mass spectrometry (TOF-MS) is a conceptually identical method of ordering such sequence ladders but with the poten-tial for orders of magnitude higher speed. Thus, several research groups have devoted considerable effort over the past few years to achieving a TOF-MS analytical capability for DNA mixtures.

2. Volatilization of intact DNA molecules

The idea of using mass spectrometry in DNA sequencing would have been unthinkable only a decade ago. Mass spectrometry requires gas-phase ions, and, despite remarkable progress by the early 1980s in volatilizing protein molecules with molecular weights as high as 30 000 Da by energetic ion impact, very little success was reported for DNA molecules of any size. Even if DNA in the 30 000 Da range could have been vola-tilized, this did not begin to compete with electro-phoresis which routinely handles 400-nucleotide DNA lengths, corresponding to molecular weights over 100 000 Da. It became clear to me in the mid 1980s that atomic ion impact was unlikely ever to produce intact gas-phase ions of large DNA mole-cules, because the linear dimensions of the area in a target surface excited by such impacts — perhaps 50 Ä radius — were much smaller than the lengths of the target DNA molecules which could range up to several hundred angstroms [1]. However, pulsed laser irradiation, which could excite much larger areas and so had no such dimensional limits, had been signally unsuccessful in desorbing intact molecular ions with molecular weights much beyond 2000-3000 Da.

For our work, the key breakthrough came when Bo Sundqvist and I realized that models of the ion impact desorption process had been unsuccessful

because they ignored the necessity to account for the sizeable momentum a large molecule must acquire in order to move away from a surface. A model detailing the origin of the required momen-tum in a recoil triggered by the rapid expansion of vibrationally excited protein molecules began to explain protein desorption from surfaces by fast ion impact [2], and the reason for the lack of prior success of laser irradiation became apparent: photons cannot directly transfer any significant amount of momentum to a large molecule. Thus, prior attempts to avoid thermal degradation in desorption of large molecules by heating the target molecules rapidly with short pulsed lasers simply resulted in frying the molecules faster. However, a focussed laser pulse excites an area which is much larger than molecular dimensions and so is almost the only type of excitation which has any possibility of desorbing intact massive DNA molecules. Photons supply heat: we needed a way to convert this heat into motion. Once the problem was phrased in this way it was apparent that the answer was several hundred years old: heat is converted to motion through the expansion of a working fluid, as in the steam engine, and, much earlier, the cannon. Our objective then became to create a laser-powered molecular cannon, in which an explosion in a suitable matrix would act to expel embedded large molecules into the vacuum. An anticipated further benefit was that the expand-ing matrix vapor should act like an expanding supersonic jet, not only entraining but also cooling the DNA molecules and so helping to ensure their survival. We briefly considered true explosive matrices, but rejected them because the explosions are exothermic chain reactions pro-ducing extremely hot atoms and free radicals which would be detrimental to the survival of intact bio-molecules. We then realized that lasers should have sufficient power to explosively ablate any material. Thus, constraints on our choice of matrix were relaxed and we could look for substances with optimum compatibility with the target molecules and with biochemical processes, and, in particu-lar, create endothermic explosions, in which all

P. Williams/Int. J. Mass Spectrom. Ion Processes 131 (1994) 335-344 337

processes following the absorption of the laser pulse act to cool the system.

3. Laser ablation

The natural first choice for a matrix was water, and we began to investigate pulsed laser ablation of aqueous solutions of DNA, frozen and cooled to liquid nitrogen temperatures for vacuum com-patibility. We realized that a chromophore was needed to couple the laser energy into the system and that IR lasers might couple efficiently to vibrational-rotational modes in the target water molecules, but no such lasers were available to us initially. The only laser available at the time was a dye laser operating in the visible at 581 nm. Some-what to our surprise, we found empirically that even such a visible laser, given sufficient power, would ablate thin frozen water films from a copper substrate. We now believe that the mechanism involved is one of shock heating of the thin ice film, driven by the rapid generation of very high pressures at the copper-ice interface by absorp-tion of up to 109Wcm~2 from the focussed laser beam: the copper substrate is the chromophore in this approach.

In initial studies beginning in 1987 we avoided the then unknown difficulties of ionization and detection of extremely large molecules by simply

Fig. 1. Contact autoradiogram of collector deposit of 32P-labeled DNA, ablated from a thin frozen film on an oxi-dized copper substrate. The small discrete spots correspond to spallation of small ice crystals from thicker regions of the film. The central diffuse deposit is the result of a supersonic jet expansion following shock vaporization of thin film regions.

collecting the ablated material and characterizing the transported DNA by gel electrophoresis [3]. Two characteristic results are shown in Fig. 1 and Fig. 2. First, using radiolabelled DNA, the angular distribution of the ablated material could be deter-mined from a simple contact autoradiogram. This revealed distributions characteristic of two distinct processes (Fig. 1). Discrete "hot spots" on the film

622 527 404 309

242

— 160

123 110

BP

90

76

67

— 34

— 26

Fig. 2. Gel electrophoresis assay of double-stranded DNA collected from a diffuse deposit similar to that of Fig. 1. Right: assay of starting mixture (pbR 322 digests of ë-phase DNA). The uppermost band (622 base pairs) corresponds to a mole-cular weight « 410 kDa. Left: ablated mixture. Laser power was « 2 x 108 Wem"2; wavelength, 581 nm.


with a broad angular distribution signalled the spallation of intact ice chunks; we believe these chunks were blown off from regions of the film too thick to allow complete vaporization by the high pressures generated at the base. Electro-phoresis of the DNA collected from these spallation spots showed intact DNA with no evi-dence that it had been exposed to elevated tempera-tures. When efforts were made to ensure that the target films were thin, we observed in the center of the collector a diffuse but strongly forward-peaked deposit which was the signature of the supersonic jet expansion we expected. By monitoring the pressure excursions in the vacuum system, it was estimated that film thicknesses ablated to produce such diffuse deposits were on the order of a few microns. Figure 2 shows an electrophoresis assay of a mixture of double-stranded DNA ablated at the highest power density used at that time ( ^ 2 x 108Wcm"2 with a corroded copper sub-strate acting as a good absorber of visible light). Comparison of the ablated DNA collected from the central ablation deposit (left) with the starting material (right) shows that the DNA was ablated substantially intact; in addition, there is signal between the bands for the ablated mixture show-ing that a portion of the DNA was fragmented in the ablation process. This is evidence that the DNA has been transiently exposed to high temperatures, and was not simply transported across the vacuum gap in small particles of ice. Surprisingly, fragmen-tation decreased as the laser power density was increased, and at the highest power densities a sizeable fraction of DNA survived intact, even up to molecular weights over 400 000 Da (Fig. 2).

By the time the results of ref. 3 were published, Tanaka et al. in Japan [4] and Karas and Hillen-kamp in Münster had begun to report striking success in laser ablation of large proteins using a different matrix-assisted technique [5]. These authors focussed on the problem of coupling energy into the system, rather than momentum, and chose their matrices to have a high extinction coefficient at the UV wavelengths they were using. Tanaka et al. used colloidal metal as a chromo-

phore in a glycerol matrix; Karas and Hillenkamp worked with aromatic acids which were both chromophore and ablation matrix and have since become the materials of choice. Both groups found that very large proteins could be ablated intact from UV-absorbing matrices; we believe the mechanism to be similar to ours, i.e. expulsion of the embedded biomolecules due to an explosion of the matrix material. In addition they found that a fraction of the ablated molecules were ionized so that mass spectrometry was simple and direct. Mass spectra from organic chromophores typi-cally contained parent molecular ions, together with some adducts of the matrix and a small com-ponent of fragments, multiply-charged ions and multimers (dimers, trimers, etc). Detection of mole-cular ions up to at least 300 000 Da was shown to be possible simply by accelerating the ions to suffi-ciently high energies before they struck an electron multiplier [5]. Protein characterization using the simple and direct approach pioneered by Karas and Hillenkamp, has come into widespread use. However, although occasional success was reported in ablating intact RNA molecular ions, DNA has turned out to be a much more difficult proposition.

4· DNA mass spectrometry

Our first mass spectrometric results, obtained with a very crude TOF-MS were reported in 1991 [6]. We found that ablation of frozen aqueous solutions directly produced ions of small DNA analytes (up to ^ 18,000 Da), as well as proteins, and that the mass spectra were simple, being domi-nated by the singly-charged molecular ion. Although spectra were not obtained routinely — typically only a small region of any given sample could be found which produced recognizable mole-cular ion signals — there was evidence that ioniza-tion improved if the dye laser used was tuned to the sodium D-line resonance at 589 nm. We suspected that ionization was occurring through adduction of sodium ions formed following resonant excitation and ionization of sodium atoms in the ablated

P. WilliamsjInt. J. Mass Spectrom. Ion Processes 131 (1994) 335-344 339

vapor plume (the frozen buffer solution contained sodium salts, and there was undoubtedly sodium contamination also on the substrate). At that point we set out to develop a TOF capability which could begin to impact DNA sequencing and the Human Genome Project.

The TOF-MS used in our current studies is a simple linear system. A thin frozen film of a DNA solution is produced on an oxidized copper substrate which can be floated to ±30 kV and maintained at liquid nitrogen temperature in vacuum. The sample manipulator allows the sample to be moved in the x and y directions so that different areas of the film can be ablated (each laser shot completely ablates a small area of the film). The mass spectrometer flight tube is approximately 1.4 m long. A Q-switched, frequency-doubled Nd-YAG laser (Lumonics HY-400) pumps a dye laser (Lumonics HD-300) which is operated with Rho-damine 6G dye in the wavelength range « 580 to 595 nm, and typically is tuned to 589 nm. The dye laser pulses are deflected and focussed through a 100 mm focal length quartz lens and a 60° prism to ablate a spot ^ 1 0 0 ÷ 2 0 0 ì é ç 2 on the sample surface. The tripled output of the Nd-YAG laser at 355 nm can also be focussed onto the sample surface through the same lens, giving a somewhat larger spot (« 300 x 400 μιή) for UV ablation studies.

DNA mass spectrometry for sequencing is a mixture analysis problem. Readout of sequence mixtures for a 400-base length of DNA will involve a mixture of DNA segments for each of the four types of base termination sites, each mix-ture involving on the order of 100 DNA segments. This places severe constraints on the mass spectral quality. If the mass spectrum of such a mixture contained a number of mass peaks for each indi-vidual component, for example fragments, clusters, multiple charge states and matrix adducts, deci-phering the sequence in the presence of multiple overlapping peaks from each of the four channels would be impossible. Ideally, the mass spectrum should contain no more than a single peak for each component (as is the case for gel electro-

phoresis). Resolving power should be sufficient to resolve two segments differing by one nucleotide over the entire read length — i.e. a resolving power of 400 is required for a 400-nucleotide read length. In molecular weight determinations of single proteins, rather poor resolution and broad tails on the peaks are tolerable because only the centroid of an isolated peak is required. In con-trast, in a DNA mixture analysis, the specified resolving power should be attained at the base of the peaks. This is particularly crucial in DNA sequence readout, because the enzyme reactions that produce the sequence ladder mixtures are not uniformly efficient, and some components may be present in low abundance; errors can arise from the submergence of small peaks in the tails of adjacent intense peaks. A final constraint on DNA sequence readout is that the technique must work with single-stranded DNA containing a mixture of bases. Our initial work used double-stranded DNA which might be expected to be somewhat more resistant to fragmentation. Recent work with UV ablation of DNA using organic acid chromophores has shown that intact ions of long chain polythy-midylic acid (poly-T) can be ablated intact from a number of matrices, but that the other homo-polymers (poly-A, -G or -C) or mixed-base single-stranded DNA fragment extensively under the same conditions and only very short molecules can be ablated intact.

To evaluate the capabilities of our approach for sequence mixture analysis, we synthesized a test mixture of mixed-base, single-stranded DNA (ss-DNA) lengths ranging from an eight-mer (molecular weight 2500 Da) to a 60-mer (18 500 Da). The sequence of each component was care-fully specified to avoid either interstrand or self complementarity, so that the tendency of any two strands to hydrogen-bond and anneal to each other was minimized. In this way we could evaluate the intrinsic tendency of the ablation process to produce clusters. Detection capability for clusters and possible multiply-charged ions was optimized in the ice studies by avoiding strand sizes which

340 P. Williams/Int. J. Mass Spectrom. Ion Processes 131 (1994) 335-344

differed by a factor of 2, so that any such artifact would be resolved from the peaks due to the individual strands.

For comparison with techniques involving UV-absorbing matrices we have also evaluated mass spectra obtained by ablating DNA mixtures from an anthranilic acid matrix (2-amino benzoic acid) using the 355 nm output of the YAG laser. These samples were prepared by mixing the DNA analyte solution with a 500 to 1500-fold molar excess of a saturated aqueous anthranilic acid solution and allowing 10-20 ì ß of the solution to dry on a stain-less steel sample holder. We show first a mass spec-trum from this matrix (Fig. 3) [7]. The mixture used here did not contain a 60-mer, but a 10-mer was added; independent studies of single-component samples showed that cluster formation was not a problem in this approach at the dilutions used. The spectrum is rather typical of results obtained by ourselves and others using UV laser ablation of mixed-base ss-DNA. Although molecular ion peaks are detectable for the lighter components in the mixture, the peaks show much evidence of fragmentation, and the higher molecular weight DNA components give simply a step in the spectrum corresponding to the onset of a range of

100

20 40 60

TIME (MICROSECONDS)

Fig. 3. Mass spectrum of six-component mixture of single-stranded DNA (8, 10, 14, 20, 26, 32 nucleotides) ablated from an anthranilic acid matrix. Laser wavelength 355 nm, pulse energy « O.lJcnT2. Adapted from ref. 7.

fragment peaks rather than giving a distinct mole-cular ion peak. This fragmentation must be occur-ring in the acceleration gap, implying that the ablated DNA is heated or excited to such an extent that the half-life for unimolecular fragmentation is on the order of 1 ßs or less. However, such spectra are obtainable quite routinely, given simple pre-cautions such as using freshly-made anthranilic acid solution. Because the extinction coefficient of anthranilic acid is high at 355 nm (e « 104) many laser shots can be made on the same area of sample before it is completely eroded.

In contrast to UV ablation, spectra of frozen aqueous films using visible laser irradiation are far more difficult to obtain. Our samples are prepared by smearing a thin film of an aqueous buffer solution of the DNA mixture onto a copper substrate which has been given a black hydrophilic surface coating by oxidation with an acid solution of potassium permanganate. The substrate is at a temperature of « -20°C so that the film freezes when it is applied. The frozen film is then allowed to sublime in vacuum until the films are judged to be appropriately thin (estimates from our initial work were that successful ablation required frozen films a few microns thick) and the sample is then plunged into liquid nitrogen to cool it before load-ing into the ion source of the mass spectrometer. Substantial areas of each sample film must typi-cally be sampled to find an area that emits mole-cular ions of the DNA analyte, and in many samples no such area can be found (in contrast each laser pulse produces ions characteristic of the substrate because we operate in the high power regime « 10 Jem - 2 where a plasma is pro-duced at the substrate surface). It appears that good spectra are obtained most readily when the laser impinges near an edge of the frozen film. Figure 4 shows one of the best mass spectra obtained to date of the eight-mer-60-mer six com-ponent mixture [7]. The spectrum is characterized by a single peak for each component, with no evi-dence of fragmentation or cluster formation. Two small peaks on the low-mass side of the eight-mer were at first suspected to be due to doubly-charged

P. Williams/Int. J. Mass Spectrom. Ion Processes 131 (1994) 335-344 341

150

100

50

8 14 20 26 32

LU yJiEiMMiiieii ™*fr4MMn*w*t ►* ^m

20 40 60

TIME (MICROSECONDS)

80

Fig. 4. Mass spectrum of six-component mixture of single stranded DNA (8, 14, 20, 26, 32, 60 nucleotides) ablated from a frozen solution film on an oxidized copper substrate. Laser wavelength 589nm, pulse energy « 10 Jem"2.

ions, but the absence of such peaks for the heavier components of the mixture (doubly-charged ion formation would be expected to be more prevalent for larger molecules) and discussions with DNA synthesizer staff lead us now to believe that these are impurities in the synthetic mixture.

Collector experiments show that DNA is ablated intact from substantial areas of routinely prepared ice films, but only rarely do we observe ionized DNA in the mass spectrometer. Calculations indi-cate that the initial temperature of the ablated water vapor is no higher than « 1000K, and that expansion rapidly cools the water, and presumably the entrained DNA, to room temperature or below [8]. Under these conditions it is not surprising that no intrinsic ionization processes occur. We believe that the observed ions are formed when metal ions from the substrate surface attach to ablated DNA molecules in the gas phase. However, ions formed at the base of a few-micron-thick ice film cannot immediately access DNA molecules which prob-ably accumulate near the surface of the film when it freezes. By the time the mean free path through the ablated water vapor plume is comparable to the plume dimensions, the leading edge of the plume is some millimeters from the copper surface.

Ions from the surface are accelerated by the high extraction field and typically will be too energetic to attach to the DNA molecules by the time they encounter them. Only if ablation occurs at a film edge, so that ions from the metal surface can imme-diately access the outer surface of the ice film, is ionization expected to be efficient. This is con-sistent with our experience, in that mass spectra such as that of Fig. 4 are most usually obtained when the laser impinges at or near an edge of the ice film. Cromwell et al. have recently demon-strated the efficacy of such a process in ionization of laser-ablated fluoropolymers [9]. However, as yet it has been difficult for us to control this process since visualization of the few-micron-thick ice film on the black copper surface is not good.

The mass spectrum of Fig. 4 has the key charac-teristic of simplicity, needed for successful mass spectrometric analysis of complex DNA sequence mixtures. The mass range sampled was limited by synthetic considerations, rather than by intrinsic limitations of the process, so that we believe con-siderably larger DNA strands could have been observed under these conditions. An improvement of less than a factor of 10 in mass range would make the technique competitive with elec-trophoresis (maximum read length « 400 bases); a corresponding improvement in mass resolving power is needed (resolving power in Fig. 4 is « 40 for full resolution at the base of the 60-mer peak, but we have not yet worked to optimize this in any way). Both capabilities appear within reach: UV laser ablation of 300 kDa proteins has been achieved, showing that there is no intrinsic barrier to detection of such large molecules (300 kDa DNA would correspond to « 1000-base lengths), and there is no intrinsic barrier to resolving powers well in excess of 1000 for TOF mass spectro-meters. Anticipating such progress we can begin to assess the possible impact of a successful TOF mass spectrometric capability on DNA sequencing.

5. Impact of TOF-MS on DNA sequencing

The most obvious advantage to be derived by


replacing electrophoresis with TOF-MS is that of speed. Multi-shot spectra such as that of Fig. 4 should in principle be obtainable in a few seconds, so it seems reasonable to assume a cycle time no longer than « 10 s to read out a complete sequence set. If read lengths are on the order of 400 bases per spectrum (up to « 120 kDa molecular weight), then throughput on the order of 100000 bases per hour appears possible, or « 106 bases per day (compared with « 104 bases per day for current electrophoresis-based automated sequen-cers). If the mass range achieved with proteins is attainable with DNA, then read lengths up to « 1000 bases come within reach. The most widely used sequencing method at present is the so-called "shotgun" approach, in which short (300-400 base) lengths of a long DNA template are sequenced randomly by generating "primers" (16-18 base lengths of DNA complementary to specific regions on the template strand). These primers initiate the enzyme transcription process at points where the primer anneals (hydrogen-bonds) to its complementary region on the template. The result is a set of (hopefully) overlapping lengths of sequence which must be assembled into the sequence of the entire template strand by aligning overlapping regions. Improving the read length has a disproportionate accelerating effect on overall sequencing speed in the shotgun approach, because assembly of a finished length of sequence from 1000-base rather than 400-base long pieces is considerably more than a factor of 2.5 faster.

An alternative to shotgun sequencing is the "primer walking" approach, which moves system-atically along a template strand, synthesizing primers specifically complementary to a chosen region at the end of each read increment so that the next set of sequence fragments begins where the last set terminated. Primer walking has no assembly step — the sequence is read linearly along the strand — but has not been widely applied up to now because the synthesis of a spe-cific 18-nucleotide primer for each increment is too time-consuming and expensive. However, recent

breakthroughs have shown that the desired primers can be assembled directly on the template strand by adding a set of three DNA hexamers which, when aligned end to end, have the desired primer sequence [10,11]. These hexamers bind to the complementary sites on the template and appear to initiate the polymerase reaction as effi-ciently as a single 18-mer with the same base sequence. The advantage of priming with hexamer sets is that there are only 4096 different possible hexamer combinations of the four DNA bases so that it is possible to assemble a library of the hexamers from which the appropriate combina-tions can be chosen rapidly for any priming site. An automated primer walking strategy then becomes conceivable which could march rapidly along a DNA template strand producing a "finished" sequence at each step: no assembly is required. However, in such a process, the sequence ladder readout becomes a rate-limiting step, because the information required to assemble the next set of hexamer primers arrives at the end of the ladder readout. Thus, mass spectrometry may well be essential to the success of a primer walking sequencing strategy.

Possibly the greatest advantage of mass-spectro-metric sequence readout is the potential to reduce significantly the overall error rate of the process. Errors arise because the enzymatic reactions used to grow the copies of the template DNA are not uniformly efficient. Certain fragments are made less efficiently than others; in particular, termi-nation with the same base at two adjacent sites may often result in reduced abundance of the second (heavier) fragment. If the low-abundance fragment is not detected, its absence may not be noted in electrophoresis given the lack of an abso-lute size vs. distance scale, resulting in an error. Even if the absence of a fragment is noted, it is often not possible to determine in which of the four base sets (A, G, T or C) the missing peak would have placed. When such errors arise, multi-ple runs are needed to resolve ambiguities, and if errors are systematic rather than random they may remain undiscovered or unresolved. In contrast,

P. Williams!Int. J. Mass Spectrom. Ion Processes 131 (1994) 335-344 343

the time-to-mass scale of a TOF mass spectrometer is absolute. There can be (and often are) deviations from ideality in the time to mass calibration equation due to initial kinetic energies acquired by the ions from the supersonic matrix vapor expan-sion, but the calibration curve as a whole, and particularly in the vicinity of any possible error, is considerably overdetermined due to the other peaks in the mass spectrum. Thus, missing peaks can be flagged unambiguously so that no undetected errors are possible. Even better, the mass determination capabilities of the mass spectrometer in many cases will allow a determi-nation of which base corresponds to the missing peak. The four DNA bases have unique masses, with a minimum mass difference of 9 Da between A and T. The mass difference corresponding to the difference of the centroids of two mass peaks can be determined with a precision limited ultimately by the statistical noise in the peaks. It can be shown [12] that the error (standard deviation) in centroid position for a peak of width W containing N ions is given by

For a resolving power « 1000, the identity of the base terminating a missing peak can be deter-mined within 99% confidence limits given no more than about 600 ions in each of the surrounding peaks.

A final potential advantage of mass spectro-metry is that the technique may allow other changes within the overall DNA sequencing scheme which could offer additional simplification and/or acceleration. For example, the mass differ-ence measurement capability discussed above may ultimately allow "one-pot" rather than "four-pot" growth of sequence mixtures: the terminating base on each strand may be determinable from the mass difference between it and its neighbours. This would improve the throughput of the robot systems designed to automate such growth by a factor of four. Further developments might result from the lack of a need to label the different strands with radiolabels or fluorescent dyes as is current

practice. Then other means of generating sequence fragment sets might be considered: for example an exonuclease digestion from one end of a target strand of DNA, with the other end blocked, could be interrupted at different times to produce the required sequence sets.

6. Conclusion

Clearly, much remains to be done before mass spectrometry can begin to impact DNA sequen-cing, and particularly the Human Genome Pro-gram. TOF-MS will replace only one of the steps in a rather complex process. Neverthe-less, it is important to pursue such technology because the full impact of the Human Genome Project will not be realized unless a capability exists to very rapidly and inexpensively resequence selected regions of DNA in humans and other organisms, to search for genes for hereditary dis-eases and other aspects of human variation, to compare human DNA with that of other organ-isms to study evolutionary divergence, and to investigate the effects of toxins and radiation on DNA.

Acknowledgements

The work described here has benefitted from an enjoyable collaboration with Bo Sundqvist, and from the outstanding efforts of Randy Nelson, David Scheiltz, Bob Thomas, Chau-Wen Chou, and Cong-Wen Luo. I am grateful for the gener-ous support of the Department of Energy Human Genome Project Grant DE-FG02-91ER61127.

References

1 P. Williams and B.U.R. Sundqvist, in R.H. Stuhnel and M.L. Knotek (Eds.), Desorption Induced by Electronic Transitions, DIET HI, Springer, Berlin, 1988, p. 230.

2 P. Williams and B. Sundqvist, Phys. Rev. Lett., 58 (1987) 1031.

3 R.W. Nelson, M.J. Rainbow, D.E. Lohr and P. Williams, Science, 246, (1989) 1585.

4 K. Tanaka, H. Waki, Y. Ido, S. Akita and T. Yoshida, Rapid Commun. Mass Spectrom., 2 (1988) 151.

344 P. Williams/Int. J. Mass Spectrom. Ion Processes 131 (1994) 335-344

5 M. Karas and F. Hillenkamp, Anal. Chem., 60 (1988) 2299. 6 R.W. Nelson, R.M. Thomas and P. Williams, Rapid

Comm. Mass Spectrom., 4 (1990) 349. 7 D.M. Schieltz, C-W. Chou, C-W. Luo, R.M. Thomas and

P. Williams, Rapid Commun. Mass Spectrom., 6 (1992) 631.

8 P. Williams and R.W. Nelson, in K.G. Standing and W. Ens (Eds.), Methods and Mechanisms for Producing Ions from Large Molecules, Plenum, New York, 1991, p. 265.

9 E.F. Cromwell, K. Reihs, M.S. de Vries, S. Ghaderi, H.R. Wendt and H.E. Hunziker, J. Phys. Chem., 97 (1993) 4720.

10 J. Kieleczawa, J.F. Dunn and F.W. Studier, Science, 258 (1992) 1787.

11 L.E. Kotier, D. Zevin-Sonkin, I.A. Sobolev, A.D. Beskin and L.E. Ulanovsky, Proc. Natl. Acad. Sei. U.S.A., 90 (1993)4241.

12 J.O. Meredith, F.C.G. Southon, R.C. Barber, P. Williams and H.E. Duckworth, Int. J. Mass Spectrom. Ion Phys., 10 (1972/73) 359.


Factors affecting the resolution in matrix-assisted laser desorption-ionization mass spectrometry

Arnd Ingendoha, Michael Karasa, Franz Hillenkampa, Ulrich Giessmannb

3Institute for Medical Physics and Biophysics, University of Münster, Robert-Koch-Strasse 31, 4400 Münster, Germany bFinnigan MAT, PO Box 144062, 2800 Bremen 14, Germany


Abstract

The influence of the physical parameters laser spot size, laser irradiance and angle of incidence on the mass resolution of matrix-assisted UV-laser desorption-ionization spectra has been investigated for a reflector-type time-of-flight (TOF) mass spectrometer. Minimal peak widths are in the range of the laser pulse duration, i.e. a few nanoseconds. Such narrow ion signals can be obtained only for spot sizes of more than about 70 ìôç in diameter, at the lowest laser irradiance for ion generation, and in the mass range below about 2000 Da. Mass resolutions of up to m/6m = 6000 (FWHM) have been obtained for peptides up to about 3000 Da. The reduction of mass resolution observable under non-optimized experi-mental conditions can be rationalized by the increasing amount of material ablated, enhanced number of collisions in the desorbed plume, and by an increase of the effective ionization time. At larger masses the non-resolved isotopic distribu-tions also contributes to the decline in mass resolution.

Key words: Matrix assisted laser desorption-ionization; Mass resolution; Time of flight mass spectrometer; Biomolecules

Introduction

Matrix-assisted laser desorption-ionization (MALDI) has become an established method for mass analysis of macromolecular compounds [1-3]. Besides a documented broad applicability to the analysis of proteins up to a molecular weight of about 300 kDa, several other classes of biomole-cules such as carbohydrates [4], glycolipids [5] and oligonucleotides [6,7] and also technical poly-mers [8,9] etc. became accessible by this method. The wide acceptance of MALDI-MS for the ana-lysis of large (bio)molecules is also reflected by the increasing number of commercially available, dedi-cated instruments. So far MALDI instruments


have been mostly used for the determination of the mass (molecular weight) of macromolepules and the identification of different molecular com-ponents in a mixture. In both cases the peak widths of the ion signals and the resulting "appar-ent" mass resolution are very important perfor-mance characteristics, because they determine the accessible mass accuracy as well as the limitations of the separation of components of similar mass in the spectrum; they may originate from the sample mixture as given for the analysis, or from adduct or fragment formation in the desorption process. For clarity of discussion this paper distinguishes between the mass resolution of the instrument, determined e.g. from the peak widths of a mono-isotopic mass peak in the low or intermediate mass range under optimal operational conditions, and

SSDI0168-1176(93)03873-K

346 A. Ingendoh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 345-354

the apparent mass resolution as derived from the peak width of signals in mass spectra of real analytes. The latter includes all the effects listed below and is the quantity usually directly accessi-ble to the user (and the one of prime interest to users). Whereas the instrument resolution usually varies only slightly with varying mass, the apparent resolution typically exhibits a quite dramatic deterioration with increasing mass.

The MALDI process is initiated by a short pulse of intense laser light generating a current pulse with a large number of (high molecular weight) ions. Therefore time-of-flight (TOF) mass spectro-meters fit ideally as mass analyzers, firstly because of their unlimited mass range and secondly because of their ability to record the entire spectrum for every single laser shot. Linear as well as reflector TOF instruments are extensively used for MALDI-NIS. Both types of instrument are basically simple in design and construction, even if the more recent developments such as gridless ion reflectors are implemented. Special attention should, however, be paid to the design of the ion optics in order to achieve optimum ion collection and mass resolu-tion because of the complex nature of the MALDI ion formation process. The increasing interest in TOF mass spectrometry is documented by a series of review papers which have appeared in recent years, e.g. ref [10].

A major drawback of TOF analyzers in conjunction with MALDI ion sources so far has been the relatively low mass resolution. Instru-ment resolutions in the order of a thousand have been achieved in the low to intermediate mass range [11-13] declining to values of only a few hundred in apparent resolution for ions with masses above about lOkDa. These values fall short by almost one order of magnitude of the mass resolution of up to about 20 000 for organic molecular ions, reported for other ion source-TOF combinations [14-17].

Broad ion peaks, limiting the apparent mass resolution in MALDI analyses, can result from a number of quite diverse factors.

(i) Broad initial energy distributions (and

energy deficits). These can in principle be compen-sated for by ion reflectors or the effects can be reduced by high acceleration potentials in linear instruments. It should be observed though that the width of this distribution increases quite dramati-cally with increasing acceleration fields due to collisional processes in the expanding plume.

(ii) A spread in apparent ion generation time, originating from either the finite laser pulse width (minor influence) or the ion formation process. This cannot be compensated for by an ion reflec-tor. Again higher acceleration fields aggravate this problem.

(iii) Ion acceleration regions following the field-free ion separation flight tube, as frequently used in conjunction with ion detectors. These effects can add another time (i.e. peak width) spread if metastable decay has occurred in the drift region or collision induced dissociations have taken place owing to residual gas molecules or field-defining meshes in the ion path.

(iv) Time spreads due to the formation of sec-ondary ions in addition to secondary electrons at the conversion dynode of the secondary electron multiplier (SEM) [18-20].

(v) The natural (particularly 12C/13C) isotope distribution of hydrocarbon molecules. This limits the peak width to the envelope function, unless well resolved, which causes particular problems in the intermediate mass range (see below).

(vi) Cationization of the formation of adducts between analyte and matrix molecules or fragments thereof. This factor can be controlled to a consid-erable extent by a careful desalting of the sample (where needed), and in particular, by the right choice of matrix [21].

(vii) Fragmentation of analyte ions by cleavage of small neutral components such as H 2 0 , NH3 or small ligands, as frequently observed in MALDI [22]. This fragmentation also increases with increasing strength of the ion extraction field. Frag-mentation due to metastable decay in the field free drift region is usually not a problem for the limited mass resolution of a few hundred in linear TOF instruments, unless there is further ion accéléra-

A. Ingendoh et al. /Int. J. Mass Spectrom. Ion Processes 131 (1994) 345-354 347

tion following the drift region (see above). It can, however, seriously degrade the peak width in reflectron instruments, if not well resolved.

This paper focuses on the experimental condi-tions in the ion source assuming that the boundary conditions for the obtainable mass resolution are set by the ion formation process and its important physical parameters such as laser irradiance-fluence, focal diameter and angle of incidence.

Ion initial energy and ion generation time distributions in M ALDI ion sources

Two strategies have been pursued to compensate for the energy distributions of MALDI-generated ions: either the implementation of ion reflectors, usually in conjunction with only moderate ion acceleration potentials of a few thousand volts, or acceleration potentials of up to 50 000 V in linear instruments. Reflectron TOF instruments capable of compensating the flight times of ions due to energy spreads of up to ±15% are state of the art today. The use of ion mirrors has been described and discussed in several review papers with regard to problems and requirements for various ion sources [15]. For ions generated by MALDI a strongly forwarded ion emission and a mass inde-pendent initial velocity distribution has, on the one hand, been determined [23]. An additional energy spread (and energy deficit) has, on the other hand, been reported for ions accelerated promptly by a static acceleration field [24]. Two reasons are believed to cause the additional energy spread and ion energy deficit: "delayed" ion formation resulting from (photo)chemical reactions [25] and acceleration impediment due to multiple collisions of ions [24], both taking place in the expanding high density ablation plume above the sample surface. Indeed, it has been observed that the ion yield can be enhanced to some extent by a delayed ion extraction [26]. Both processes will not only lead to a loss and spread in ion energy, they will more importantly for the mass resolution, also result in a delay and spread of the apparent ion formation time, which cannot, in principle, be

compensated for by either a large acceleration potential or an ion reflector and will, therefore, necessarily degrade the mass resolution, even though energy spread alone is easily compensated by the reflector.

The influence of isotope distributions

The problems arising in mass spectrometry with increasing mass and a given series of isotopic peaks was discussed in the early days of high-mass analysis by Yergey et al. [27,28]. Within the limited mass resolution, attainable even under optimal conditions in MALDI-TOF analysis, only the peak distributions arising from the 12C/13C ratio in typical biomolecules need to be considered; the fine structure due to the isotopic distribution of other atomic constituents will not be resolved under any realistic circumstances. Surprisingly, mass accuracy is only indirectly affected by a non-resolved isotopic peak distribution in the high mass range. For proteins with a mass above about 4500 Da the binomial peak distribution becomes symmetric within the usual measurement precision (mathematically this limit is defined as the mass at which the most abundant and the average (chemical) mass differ by no more than the desired accuracy of mass determination, say one in 104). Centroiding even of only the sym-metric top part of a mass peak will result in an accurate value for the chemical mass (molecular weight). This is important in practical work because peaks are often symmetric at the top but become more or less asymmetric in the bottom part owing, e.g. to minor contributions of adducts or fragments. Discrimination of nearly equal mass components in a mixture will, of course, still suffer from the limited apparent mass resolution in this mass range.

The real problem arises in the middle mass range between a few hundred and the above given upper limit of about 4500 Da if the distribution of isoto-pic patterns is not resolved. Though the accurate mass could in principle still be obtained by inte-grating the total peak area (yielding the chemical


mass), this is very difficult, if not impossible, in most real cases. As described above, the bottom parts of the peaks are often severely distorted by contributions from other than the isotopic distribu-tions, and the base line is also rarely well-enough defined, particularly if the spectrum contains very strong low-mass (matrix) signals as typically observed in linear instruments, frequently requir-ing blanking of this mass region. Even minor non-isotope contributions to such peaks will severely limit the mass accuracy and errors of one mass unit for peptides in the 1000-1500 Da range are not uncommon. An apparent mass resolution, suf-ficient for a reasonable solution of the isotopic distribution is therefore highly desirable in this mass range. Such a resolution can only be achieved by reflector instruments.

A word of caution concerning mass resolution is also in place here. In TOF mass analysis, mass resolution is usually given using the FWHM defi-nition. It must be kept in mind that this definition is somewhat misleading if a unit mass (isotope) resolution is required. A mass resolution of m jam = 2000 at a mass of 2000 Da will result in a modulation of only 8% for two adjacent (Gaus-sian) peaks one mass unit apart, certainly not good enough to resolve the isotopic pattern. Simi-lar considerations apply to the resolution of the true envelope function of the isotope distribution in the high mass range, though this symmetric peak broadening will not directly affect the accuracy of the mass determination as discussed above. Still, as high a mass resolution as possible, certainly one in the range of several thousand, is highly desirable.

Experimental

The mass spectrometer used for the investiga-tions was a VISION 2000 instrument (Finnigan MAT, Bremen, Germany). Ions are formed using a Q-switched (5 ns) frequency-tripled Nd:YAG laser (Speser 600, Spektrum, Berlin, Germany) which operates at 355 nm. Fine attenuation of the laser irradiance to values typically in the range of 10 6 to l0 7 W cm 2 is achieved by a dielectric mirror

using its angle-dependent reflection-transmission rate. The focal spot size of the laser beam on the target is 70 ìéç, unless mentioned otherwise; the angle of incidence is about 15°. Ions are acceler-ated in two stages to 5keV and focused by an Einzel lens. The complete mass spectrum is recorded in all cases. Ion beam blanking, as typi-cally used to suppress the high-intensity low-mass background signals in high-ion energy instruments, is not required. The system is equipped with a grid-less reflectron. The equivalent drift length of the instrument amounts to about 1.7 m. In conjunc-tion with the low source potential this drift length provides for relatively long flight times of ions (e.g. flight time for Angiotensin, mol.wt. 1296.5 Da, is about 60 ^s).

For most of these investigations a microchannel plate (MCP) was used. As high molecular weight proteins need high kinetic energies for an effective detection, an additional post-accelaration stage with a potential of about 20 kV in front of the MCP can be used for enhanced generation of elec-trons and/or small secondary ions by impact of high mass primary ions. Data acquisition of the analog signal was mostly performed with a LeCroy 9450 digital oscilloscope (4 x 108 samples s"1). Data are then transferred to a PC for further evaluation. In special cases a Hewlett-Packard 54510A 1 x 109 samples s"1 digital oscilloscope was used allowing for the acquisition of faster signals. All spectra shown represent unprocessed single-shot data or are accumulated from typically 10 single-shot laser spectra.

2,5-Dihydroxy benzoic acid (10gl~l in a 10% ethanol, 0.1% trifluoroacetic acid (TFA) aqueous solution), sinapic acid and a-cyano-4-hydroxycin-namic acid (both 10 g 1_1 in 1 : 2 acetonitrile/0.1% TFA) purchased from Aldrich were used as matrices. Peptides were purchased from Sigma and dissolved at a concentration of 5 x 10~5 M in 0.1 % TFA. The analyte and the matrix solutions were mixed in a ratio of 1 : 5. One microliter of the mixture was deposited on a flat metal target, dried in a stream of air and transferred into the mass spectrometer by a vacuum lock.

A. Ingendoh et al. j Int. J. Mass Spectrom. Ion Processes 131 (1994) 345-354 349

100 (W+Na)+

5t=4.6ns (M+K)

»A^JU^vJ UUA J 1 — I 1 1 1 1 U

100

50

Calculated isotopic distribution

65.72 65.97

(a) tfiight / M *

I l I I | I I I I | l I I I l

2845 2850

(b) Mr/z

Fig. 1. MALDI mass spectra of (a) cyclodextrin (permethylated, mol. wt. 1429.6 Da) showing a mass resolution of R = 7200 and (b) mellitin (mol. wt. 2847.5 Da) with R = 6000.

The VISION 2000 allows the observation of samples by a CCD camera coupled to a microscope with a lateral resolution of about 20 ìéç. The laser spot is adjusted to the optical axis of the instrument. By moving the sample in the x-j>-plane with respect to the laser focus, selected sample areas can be irradiated. Thus relevant information about the state and optimization of sample preparation is obtainable.


Obtainable resolution

The maximum resolution obtained at an optimized adjustment of the instrument including the laser focal spot size (see below), the detector system (MCP) and the data acquisition (1 x 109

samples s"1 digital oscilloscope) is in the range of several thousand.

Figure 1 shows a MALDI spectra of per-methylated cyclodextrin with a resolution R = 7200 and of mellitin with R = 6000. A better mass resolution has only been reported recently for MALDI performed in a FT-ICR mass spectrometer [29]. The minimal peak width of 4.6 ns detected for MALDI conditions is even shorter than the half width of the applied laser pulse; this value can be regarded to set the final limitation of the obtainable time reso-lution. Whether a picosecond-laser offers the pos-sibility for an improvement needs to be explored, but it seems questionable that the time scale involved in the desorption-ionization event can be considerably reduced as the phase transition even of a shallow, but nevertheless finitely-thick, surface layer will require a finite time. It is furthermore noteworthy that this high resolution is not obtained at the expense of signal intensity; we did not observe any significant loss of the signal intensity (integrated peak area) compared to a (low

100

5650 5750 Mr/z

(a)

00-

60-

20-

/ \ R=950

/ rv\ \ R=1900

rvp^ iv^ypr r r r r^Fr r ' 5728 5733 5738 (b)

Mr/Z

Fig. 2. MALDI mass spectrum of bovine insulin (mol. wt. 5733.5 Da) showing an apparent resolution of Äa = 950 (a) and computer simulated deconvolution of the signal (b) leading to a calculated value of R = 1900.


100

I I I I l I I l I I I I l l I I I | M

12300 12400 M,z

Fig. 3. MALDI mass spectrum of cytochrome c (mol. wt. 12 359.2 Da) with an apparent mass resolution of about 1150.

resolution) linear TOF mass spectrum as detected by a straight-tube SEM.

As mentioned before, the isotopic cluster can lead to a deterioration of the apparent resolution. For bovine insulin (mol. wt. 5733.4 Da) this apparent resolution determined from the FWHM value is about 950 only; the instrumental resolution can be calculated to be about 1900 by a computer simulated deconvolution (Fig. 2). Comparable values have been obtained with MALDI performed in a magnetic sector instrument for ubiquitin (8565 Da). Also for cytochrome c (12 359.2 Da) (see Fig. 3) the value determined by deconvolution of the isotopic distribution amounts to about 1600. It has to be noted, however, that already minor changes in the actual peak width considerably change the calculated mass resolution. The possibility of compound identification by "high resolution" TOF analysis is exemplified in Fig. 4 for a Zn-porphine-complex. It shows good agreement between the theoretical isotopic pattern of the

compound and the spectrum obtained by MALDI-TOF MS.

The fact that high ion kinetic energy is also not required for systematic reasons in a reflectron instrument for a MALDI ion source is substan-tiated by another experiment. The electrical poten-tials of both ion source and reflectron were reduced to a quarter, corresponding to an ion energy of 1250eV. MALDI spectra of small peptides can nevertheless be registered with high resolution. Figure 5 shows the respective mass spectra obtained for angiotensin. Owing to the low ion energy and restrictions of the experimental set-up to low post-acceleration potentials at the channel plate and thereby low ion detection efficiency, these investigations could not be extended to higher mass compounds.

The influence of the laser irradiance

Irradiance (or fluence) is known to be the most critical experimental parameter for matrix-assisted laser desorption. Generally, the best results with respect to the signal quality, both in terms of reso-lution and intensity, are obtained in a small irradi-ance range above threshold irradiance. As shown in Fig. 6 for matrix-assisted desorption of a peptide and an oligosaccharide, the width 6t of the mole-cular ion peak increases considerably with increas-ing irradiance. At the same time the fluctuation of the measured values becomes larger, indicated by the increasing length of the error bars. It is assumed that this degradation of resolution is mainly caused by the steeply increasing amount of material

680 685 Mr/z

Fig. 4. MALDI mass spectrum of a Zn-porphine complex (5, 10, 15, 20-tetraphenyl-21//, 23//-porphine-Zn, mol. wt. 678.1 Da) and its calculated isotopic pattern.

A. Ingendoh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 345-354 351

100 If)

I 7 5

< 50

"c - 25

1 1 1 I | I I I ' I | I I I

1295 1300

(a)

I | I I

100

75

50

25

z

I º ß )/ À/ÀËíÊËËË 1 | 1 | I Ã I | I I I

1296 1300 ML

(b)

Fig. 5. Comparison of two MALDI spectra of angiotensin (mol. wt. 1296.5 Da) obtained at full (5kV) (a) and reduced (1250 V) (b) acceleration potential.

ablated. This will lead both to an extension of the effective ionization zone and to increasing prob-ability of impeding collisions in the acceleration phase; the energy spread and deficit connected to this will be compensated for by the reflector. Furthermore, space charge effects due to the increasing number of ions can also contribute to the observed degradation of mass resolution.

The influence of laser spot size and angle of incidence

It has been observed in direct UV laser desorption-ionization experiments that the thresh-old irradiance is a function of the laser spot size at the sample surface [30]. To investigate this for MALDI conditions, the focal spot size was varied and threshold irradiances as well as the obtained peak widths was determined. Whereas a single lens

1,0 1,2 1,4 1,6 1,8 2,0 Laser Irradiance (Rel. Units of Threshold Irradiance)

Fig. 6. Measured peak widths of gramicidin s (mol. wt. 1141.5 Da) and permethylated cyclodextrin (mol. wt. 1429.6) as a function of the applied laser irradiance.

of 140 mm focal length resulting in a focus diameter of approximately 70 ìéç is used as a standard device, the focus diameter can be varied between 10 and 280 ìðé by using a variable telescope. This is

L

É_Ú Telescope L2 ö , L3

xC ^x. y y

"X

Fig. 7. Schematics of the optical set-up for laser beam focusing using a single lens (top) and a telescope (bottom) for increase in the laser spot size; exchanges of lenses LI and L2 will result in a decrease of the laser spot size; on the right side, the laser spot size is shown as monitored by a linear photodiode array.

10°i

E . 0

§

o § 107--ó (0

¼

o 0)

106-

■ ^ ■\

► Gram. S (DHB) o Perm. Cyclo. (DHB) Ä Bradykinin (Sin. Acid) ▼ Gram. S (Sin. Acid) o Mellitin (DHB) ■ Brilliant Green (direct)

■ — — 8 ▼ ■

1 1 r 1 1 1 1 1 1 1 1 1

50 100 150 200 Focal Diameter (ìóé)

250 300

Fig. 8. Plot of threshold irradiances vs. focal diameters for MALDI of different test compounds and matrices.


done in practice by adding two more pre-adjusted lenses of adequate focal length. The principal optical set-up is shown graphically in Fig. 7. The obtained focal diameter was controlled with a linear photodiode array of 25 ìéôé interdiode distance and agreed well with the calculated value; two examples are given on the right-hand-side of Fig. 7. The threshold irradiance was determined separately for each given spot size by a variation of the incident laser energy per pulse with a continously variable attenuator. Figure 8 summarizes the results for some selected com-pounds. 2,5-Dihydroxybenzoic acid (DHB) and sinapic acid were used as matrices. For both matrices and all compounds investigated only a minor influence of the focal diameter on the thresh-old irradiance is observed for focal diameters in the range between about 70 and 280 /xm. The absolute value of (2-3)xl06Wcm~2 agrees well with the values as described in the literature [31]. Below 70 μπι a steep increase in the threshold irradiance is found. It is interesting to note that the highest value of (2-3)xlO7Wem- 2 corresponds to the value given in the first MALDI paper [32] by the authors; indeed MALDI was investigated origin-ally with a laser microprobe instrument with a typical focus diameter of ^ÉÏìçé. As a compar-ison, an organic dye (brilliant green) directly applied to the metallic substrate was examined: it exhibits essentially the same behavior. In agreement with the results reported in the influence of the laser

11

10

9

^ 8

S 7

Perm. Cyclo. (DHB) Gram. S (DHB Bradykinin (Sin. Acid) Gram. S (Sin. Acid) Brilliant Green (direct)

5jo o

50 100 150 200 Focal Diameter (ìéç)

250 300

Fig. 9. Plot of peak width as a function of the focal diameter for MALDI of different test compounds and matrices.

irradiance section, the peak width of MALDI ions increases considerably with decreasing focal dia-meter i.e. with the correspondingly increasing irradiance. Some representative results are given in Fig. 9. There is, furthermore, a small but significant difference between DHB and sinapic acid. All values determined for sinapic acid are slightly larger leading to a lower mass resolution obtainable with a sinapic acid matrix. For the organic dye, however, even though it shows the same increase in the thresh-old irradiance with decreasing spot size, an influence on peak widths is not observed.

The steep rise of the threshold irradiance for ion detection with decreasing spot size can be rationa-lized by a geometric consideration taking into account the gaussian beam profile on the target. It is assumed that a minimum absolute number of ions must be generated for a signal to be detected; this number of ions will depend on the size of the irradiated area as well as the distribution of the irradiance over this area. For a gaussian profile of given diameter (defined as the l/e2 width of the distribution) and average irradiance, about 40% of the area experiences an irradiance above the average and 7.2% one of more than twice the average irradiance. For small spot sizes the smaller available area for ion generation must be compen-sated by a correspondingly higher average irradi-ance. A similar argument was used by Beavis [33] to rationalize the increase of the ion signal with increasing irradiance near threshold. The higher required irradiance and the thereby increased den-sity of the ion/neutral cloud, increased probability for collisions, and expansion of the effective ioniza-tion zone is regarded as being responsible for the observed reduction in minimal peak width. In con-trast to this, ions stemming from direct laser desorption of the organic dye are preformed and thus generated within the laser pulse width at the equipotential surface of the metallic substrate; no influence of the focal diameter is therefore to be expected. The results indicate that focal diameters in the range of 70 to 100 ìéç are best suited for UV-MALDI as they do not impose limitations with regard to the obtainable mass resolution and still

A. Ingendoh et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 345-354 353

-J

J .Ä

J ■

J ■

1 °

Ä

Q

, Ä

■ O

■ Sinapic acid I o a-cyano-4-hydroxycinnamic acid I Ä DHB |

Ä

4 "

■

0

1 ■ ' I ' ■ ' ' '—' ■ ' U i . . . . . .— I 1000 10000 100000

Molecular Mass (Da)

Fig. 10. Relative threshold irradiances for two angles of inci-dence (80° and 15° to the surface normal) for different matrices and test compounds; full and broken lines are given as guides for the eye.

allow for optimization of the desorption by select-ing individual, homogeneous sample areas.

In a further experiment the influence of the angle of incidence was investigated. In the experimental set-ups used for MALDI by different groups, many different angles between 70° and 15° to the surface normal were used. Threshold irradiances as well as the obtainable peak widths have been inspected for two largely different angles of incidence, (15° and 80°) using three different matrices as well as test peptides and proteins as large as bovine serum albumin (66kDa). Whereas no significant differ-ences were found with regard to the obtainable peak widths, considerable differences were found for threshold irradiance. Figure 10 summarizes the results. Whereas a DHB matrix shows only a minor influence of the angle of incidence, these values decrease to about 50% for a glancing incidence as compared to the nearly normal one in the case of the two cinnamic acid derivatives.

Both DHB and sinapic acid crystallize into highly anisotropic structures, but sinapic acid forms very flat sheets [34] lying preferentially flat on the metal substrate. This may cause the observed reduction in the threshold irradiance when changing from normal to glancing incidence. For DHB forming considerably thicker needles located near the rim of the droplet [35] and

exposing different crystal surfaces in all directions, this effect is not expected. One interesting implica-tion of these observations is that it may at least partially explain the different value for threshold irradiances given by different groups.

Acknowledgments

Part of this work was done in partial fulfillment of the requirement for the Ph.D. of A.I. at the University of Münster. Financial support of this work by the Bundesministerium für Forschung und Technologie (BMFT, grant No.: 13N5640) is greatly appreciated. A.I. was supported by a grant of the Graduiertenförderung des Landes Nordrhein-Westfalen (GrFG NW).

References

1 M. Karas, D. Bachmann, U. Bahr and F. Hillenkamp, Int. J. Mass Spectrom. Ion Processes, 78, (1987) 53.

2 M. Karas and F. Hillenkamp, Anal. Chem., 60 (1988) 2299. 3 F. Hillenkamp, M. Karas, R.C. Beavis and B.T. Chait,

Anal. Chem. A, 63 (1991) 1193. 4 B. Stahl, M. Steup, M. Karas and F. Hillenkamp, Anal.

Chem., 63(1991)463. 5 P. Juhasz and CE. Costello, J. Am. Soc. Mass Spectrom., 3

(1992) 785. 6 K. Tang, S.L. Allman and C.H. Chen, Rapid. Commun.

Mass Spectrom., 6 (1992) 369. 7 E. Nordhoff, A. Ingendoh, R. Cramer, A. Overberg, B.

Stahl, M. Karas, F. Hillenkamp and P.F. Crain, Rapid Commun. Mass Spectrom., 6 (1992) 771.

8 P.O. Danis, D.E. Karr, A. Holle, F. Mayer and C.H. Wat-son, Org. Mass Spectrom., 27 (1992) 843.

9 U. Bahr, A. Deppe, M. Karas and F. Hillenkamp, Anal. Chem., 64 (1992) 2866.

10 R.J. Cotter, Anal. Chem. A, 64 (1992) 1027. 11 A. Brunelle, P. Chaurand, S. Della-Negra, J. Depauw, Y.

LeBeyec and G.B. Baptista, Proc. 39th ASMS Conference on Mass Spectrometry and Allied Topics, Nashville, TN, May 19-24, 1991, p. 1703.

12 F.J. Mayer, A. Holle, R. Schäfer and R. Frey, Proc. 40th ASMS Conference on Mass Spectrometry and Allied Topics, Washington, DC, May 31-June 5, 1992, p. 378.

13 M.L. Vestal and R.W. Nelson, Proc. 40th ASMS Confer-ence on Mass Spectrometry and Allied Topics, Washing-ton, DC, May 31-June 5, 1992, p. 350.

14 K. Walter, U. Boesl and E.W. Schlag, Int. J. Mass Spec-trom. Ion Processes, 71 (1986) 309.




17 X. Tang, R. Beavis, W. Ens, F. Lafortune, B. Schueler and K.G. Standing, Int. J. Mass Spectrom. Ion Processes, 85 (1988)43.

18 B. Spengler, D. Kirsch, R. Kaufmann, M. Karas, F. Hil-lenkamp and U. Giessmann, Rapid Commun. Mass Spec-trom., 4 (1990) 301.

19 R. Kaufmann, D. Kirsch, H.A. Rood and B. Spengler, Rapid Commun. Mass Spectrom., 6 (1992) 98.

20 J. Martens, W. Ens, K.G. Standing and A. Verentchikov, Rapid Commun. Mass Spectrom., 6 (1992) 147.

21 R.C. Beavis and B.T. Chait, Anal. Chem., 62 (1990) 1836. 22 R.S. Annan, HJ . Köchling, J.A. Hill and K. Biemann,

Rapid Commun. Mass Spectrom., 6 (1992) 298. 23 R.C. Beavis and B.T. Chait, Chem. Phys. Lett., 181 (1991)

479. 24 J. Zhou. W. Ens, K.G. Standing and A. Verentchikov,

Rapid Commun. Mass Spectrom., 6 (1992) 671. 25 H. Ehring, M. Karas and F. Hillenkamp, Org. Mass Spec-

trom., 27 (1992) 472.

26 B.H. Wang, K. Dreisewerd, U. Bahr and F. Hillenkamp, J. Am. Soc. Mass Spectrom., 4 (1993) 393.

27 C. Fenseleau, J. Yergey and D. Heller, Int. J. Mass Spec-trom. Ion Phys., 53 (1983) 5.

28 J. Yergey, D. Heller, G. Hansen, R.J. Cotter and C. Fense-leau, Anal. Chem. A, 55 (1983) 353.

29 J.A. Castoro, C. Köster and C.H. Wilkins, Anal. Chem., 65 (1993) 784.

30 U. Bahr, M. Karas and F. Hillenkamp, in Proc. Third International Laser Microprobe Mass Spectrometry Workshop, 27-28 August, 1986, Antwerp, (published by Department of Chemistry, University of Antwerp, Uni-versiteitsplein, B-2610 Wilrijk, Belgium).

31 R.C. Beavis and B.T. Chait, Rapid Commun. Mass Spec-trom., 3 (1989) 233.

32 M. Karas, O. Bachmann and F. Hillenkamp, Anal. Chem., 57 (1985) 2935.

33 R.C. Beavis, Org. Mass Spectrom., 27 (1992) 864. 34 J.N. Bridson and R.C. Beavis, J. Phys. D, 26 (1993) 442. 35 K. Strupat, M. Karas and F. Hillenkamp, Int. J. Mass

Spectrom. Ion Processes, 111 (1991) 89.


Sequencing of peptides in a time-of-flight mass spectrometer: evaluation of postsource decay following matrix-assisted laser desorption ionisation (MALDI)

R. Kaufmann*, D. Kirsch, B. Spengler Institute of Laser Medicine, Heinrich-Heine-University, Moorenstrasse 5, D-4000 Duesseldorf Germany

(Received 21 March 1993; accepted 2 July 1993)

Abstract

In matrix assisted laser desorption ionisation (MALDI) a large fraction of analyte ions undergo postsource decay (PSD) during flight in the field free drift path. By means of a modified two-stage reflectron, daughter ion time-of-flight spectra of medium sized linear peptides (up to 2800 u) were recorded containing full sequence information. Precision, accuracy and mass resolution of fragment ions were almost as good as obtained in high energy CAD studies performed in four-sector instruments. Instrumental sensitivity was better by at least one order of magnitude. In reflectron time-of-flight mass spectrometry (RETOF-MS) the cleavage pattern of PSD products is different from that obtained by high energy and low energy CAD. In our instrument, conditions which were energetically comparable to high energy and low energy CAD could easily and comparatively be studied in the same experiment by varying instrumental parameters. Activation mechanisms of PSD were found to be largely determined by collisional events (ion/neutral) induced by the acceleration field during early plume expansion. Future potentials of PSD analysis after MALDI are discussed.

Key words: Matrix-assisted laser desorption-ionization; Postsource decay; Collisionally-activated dissociation; Reflectron; Peptides

Introduction

Molecular structure elucidation by mass spectro-metry has become a promising and rapidly expand-ing tool. The exploitation of collisionally-activated dissociation (CAD) of precursor molecules [1-3] has found its way into practical mass spectrometry and has considerably extended analytical applic-ability to molecular structure determination of e.g. oligopeptides. The most successful approach is based on tandem mass spectrometry, (MS-MS) with collisional activation (CAD) occurring in between a precursor-ion-selecting and a daughter-ion-analyzing mass spectrometer in sequence.


So-called high performance instruments, based on double focussing (four-sector) setups, allow effi-cient high energy activation and optimal mass reso-lution for both precursor selection and fragment analysis, but, due to their high cost, they are diffi-cult to afford for routine analysis. More popular approaches use quadrupole filters either as triple-quad arrangements or as hybrids. The latter employ a double focussing mass spectrometer for improved precursur ion selection. Both hybrids as well as triple-quads share the problem of accepting precursor ions at rather low kinetic energies only (£lab = 10-50 eV).

Several comparative studies [4-6] indicate that for peptide sequencing low energy CAD is restricted to precursor ion masses below about

356 R. Kaufmann et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 355-385

800 u and, thus, is of limited usefulness. Some improvement was recently attained by increasing the kinetic ion energy in such hybrids to 500 eV (£iab) [7]. There is, however, a rather general con-sensus that, even under the conditions of high energy CAD (up to lOkeV £lab), which is typical for four-sector tandem instruments, fragmentation efficiency rapidly degrades with precursor masses above 2000 u. Some recent attempts to circumvent the efficiency problem by employing surface induced dissociation (SID) instead of CAD [8,9] have not yet found broad application.

It appears then, that all practical means to considerably improve efficiency of collisionally activated dissociations in larger peptides by dump-ing more collisional energy into the analyte mol-ecule have been exhausted.

There is, however, a second but widely neglected parameter in CAD (and possibly SID) experi-ments, namely time. Dissociations induced by col-lisional energy transfer are anything but prompt events, especially in large polyatomic ions. Calcu-lations of metastable decay rates in model oligo/ polypeptides [10-12] based on the RRKM theory

Fig. 1. Comparative schemes of setups for mass spectrometric analysis of collisionally activated product ions, (a), Four-sector instrument (EBEB); (b), hybrid instrument (EBqQ); (c), triple Quad; (d), RETOF-mass spectrometry ( ), stable precursor ions: ( ), transmitted ions: ( · · · ) > non-transmitted daughter ions. Areas indicating possible sites of fragmentation are shaded.

or derivatives thereof indicate that, for ions larger than say 1500u, the chance to decompose within typical mass spectrometric time scales (10~5 to 10~4s) steeply diminishes.

From this point of view four-sector instruments are rather unfavourable devices. On the one hand, precursor ions must survive the long passage through MS 1 (« 50 /xs), thus only "cool" species will reach the collision chamber. On the other hand, once they have collided they are expected to decay almost immediately, i.e. in a short time interval of e.g. 1 /xs before they enter MS 2 (see Fig. 1). Hence, with increasing precursor masses (and thus lower decay rate) this type of instrument must run into conflicts even at the highest possible levels of collisional activation (see also ref. 13).

Hybrids or triple quads, however, provide for a much longer time window (40-100 ^s) but suffer from an inherently low level of collisional excita-tion unless, perhaps, ions are multiply charged as in electrospray ionisation (ESI) and/or are pre-excited [14-17].

In this situation we propose to evaluate time-of-flight (TOF) mass spectrometry as a possible alter-native to, or, for some applications, even as a sub-stitute for conventional MS-MS. Although this suggestion might appear rather ambiguous at first glance, we will demonstrate, by means of only a few minor modifications, that a simple reflectron TOF mass spectrometer (RETOF-MS) can be operated so as to reach performances in product ion analysis comparable to those of the most advanced tandem instruments.

Metastable ion analysis is a well known tech-nique in RETOF mass spectrometry [18-22]. The first successful attempts to sequence peptides by metastable-ion analysis have been demonstrated by Tang et al. [19,20] employing particle induced desorption for precursor ion formation. With the advent of matrix-assisted laser desorption (MALDI), extremely efficient pulsed ion sources for larger peptides and proteins emerged [23, 24] and are about to become (along with ESI) the standard approach to practical mass spectrometry for larger peptides and proteins.

R. Kaufmann ei al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 355-385 357

Recent investigations in our laboratory [25,26] have demonstrated that in MALDI a large frac-tion of the desorbed analyte ions undergo fragmen-tation-neutralisation reactions during flight. The activation energy for this kind of "postsource-decay" (PSD) apparently stems from: (1) multiple early collisions between analyte ions and (neutral) matrix molecules during plume expansion and ion acceleration; (2) from collisional events with resi-dual or admitted gas molecules in the field free drift path. Thus, MALDI-RETOF mass spectrometry is rather unique in providing for: (1) highly pre-excited precursor ions which (2) move at high kinetic energy over a relatively long distance (time window) where they can undergo (3) either unimolecular decomposition and/or (4) can be further activated by collisions with residual or other gas molecules (see Fig. 1). The main charac-teristics of the different instruments used for CAD Table 1

Main characteristics of instrumentation used for CAD product ion

product ion mass spectrometry are listed in Table 1 for comparison.

In a recent report [26] we have already demon-strated that a RETOF setup can be exploited to obtain full sequence information from PSD frag-ment ions of medium sized peptide precursors (1600 u) produced by MALDI. In the present paper we explore this approach more extensively with respect to the requirements of practical mass spectrometry and to its possible merits vis-a-vis conventional MS-MS.

Experimental

General description of the setup and sample preparation

The RETOF-MS used was basically a modified LAMM A 1000 type setup (Fig. 2). For laser

mass spectrometric analysis

Ion source Collisional

activation

Time window available for ion decay

Transmission of precursor ions

Transmission of products (fraction of total product ions formed)

Interference with matrix product ions

Precursor ion selection

Four-sector tandem MS

FAB/CF-FAB High energy

(up to lOkeV

£lab)

Short (2-5/is)

15%

0.03% 3% with array detector

Yes

High resolution required and possible

Hybrids (BEqQ)

FAB/CF-FAB Low energy

(20-50 eV)

Long (20-30/xs)

6-8% (Mismatch of energy acceptance)

0.1%

Yes

High resolution required and possible

Triple Quad

FAB, ESI Low energy

with pre-excitation (ESI)

Long (20-30 /xs)

2%

0.1%

Yes

High resolution required, not possible

RETOF (PSD-MALDI)

MALDI High energy

(in flight), adjustable high pre-excitat. (in source), adjustable

Very long (30-300/xs) adjustable in limits

« 50%

>50% 8-10% at acceptable mass resolution

No (no isobaric ions formed)

High resolution not required, low resolution possible

358 R. Kaufmann et al.jlnt. J. Mass Spectrom, Ion Processes 131 (1994) 355-385

desorption, pulsed laser light of either a N2 laser (A = 337nm) or a fourth-harmonic Nd-YAG laser (ë = 266 nm) could be transmitted and focussed onto the sample by means of 90° incident focussing optics, described below. Vacuum conditions were typically (5-6)xl0~7 Torr through a 16 m3 h_1 rotary pump followed by two turbomolecular pumps (150 lmin"1 and 330 1min-1 respectively). Samples were prepared in the usual way by blow drying about 1-10 ìÀ of sample solution (^ 10~2M dihydrobenzoic acid (DHB) and « 10"5M analyte in aqueous solu-tion) on a polished aluminium substrate. The sample could be imaged through the focussing optics by means of a CCD camera; the x-y-z posi-tion was controlled manually by micrometer trans-lation stages.

Ion formation and extraction, and precursor selection

Laser focussing, sample imaging and ion acceleration-extraction were all performed coaxi-ally and perpendicularly to the sample surface by means of a combined optical-ion-optical device. The optical part consists of a quartz triplet lens (numerical aperture « 0.3) bearing a central bore (6 mm diameter) for ion transmission. Ion accelera-tion and collimation is performed by an assembly of diaphragms located within the free working dis-tance (« 32 mm) between the front lens and the sample plane. The first part of the field free drift path is formed by a rather narrow tube (diameter 6 mm, length 180 mm), tapering the central bore of the triplet lens assembly. At the exit of this tube a

0D trigger- f—[ diode

imaging

Hg light source

N 2 laser

Nd:YAG laser, 266nm

3 =*B beam blanking \\w\

objective with central hole

ion optics MCP-detector

HV-power-supply

¾ delay-gate

► ,i i 0 0 · • oo 0 · 0 .eCrey9450

i 1 ■ 7 ·I HV-power-supply j· -

A DA-controller

J HV-power-supply (-.. -

Fig. 2. Schematic setup of MALDI-RETOF-MS.

R. Kaufmann et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 355-385 359

beam blanking and precursor ion selecting device is located. Precursor mass selectivity with this sim-ple device is about 1 : 50 which is good enough to separate e.g. (M + H) + from (M + Na)+ precursor ions for 1000 u parent molecules. It should be kept in mind that, in M ALDI, the yield of prompt frag-ments is extremely low. Thus, precursor selection with parent molecules > lOOOu is needed only in heterogeneous samples (mixture analysis) with multiple precursors within about the same mass range.

SIMION (D.A. Dahl and J.E. Delmore, EG&G Inc, Idaho) trajectory modelling indicate that the complete arrangement transmits ions within a ±60° take off angle under the conditions that: (1) the lateral kinetic energies of the ions do not exceed 2 x 10~3 of the forward component (e.g. 20 eV at lOkV acceleration voltage); (2) the aspect ratio (length to diameter) of the central extraction tube through the focussing optics is about 30; (3) off-axis misalignments are carefully avoided.

It was found that the arrangement not only helps to bring mass resolution of the instrument (for parent molecules up to « 3000 u) close to its theo-retical limit (Μ/ΔΜ « 5000), but, acting as a kind of space filter element, the whole device has the further advantages of feeding the TOF spectro-meter with a fairly well collimated ion beam (divergence limit about ±1.5°) and of preventing most of the unwanted particles (early neutrals, decay products formed during acceleration, stray-or secondary particles) from reaching the spectro-meter.

Standard conditions for ion extraction were an acceleration voltage « lOkV and an initial extrac-tion field strength « 3 x 104 Vcm"1, unless other-wise indicated.

TOF-spectrometer and reflectron

The geometrical scheme of the TOF reflectron-MS is illustrated in Fig. 3. Here, s designates the acceleration distance over which ions are acquiring their drift energy Εά. The split extraction field allows the adjustment of the initial field strength

(£i) independently of the total acceleration voltage applied. Lx is the length (2034 mm) of the field free drift path up to the entrance of the reflectron, dx

(20 mm) and d2 (239 mm) the length of the first and the second retardation field respectively of the reflectron, and L2 (1228 mm) the second field free drift path between reflectron and the large-area MCP detector (72 mm diameter).

In Fig. 3 ion flight paths of either parent or fragment ions illustrate the way in which a reflec-tron distinguishes between masses of isovelocic fragment ions by flight time dispersion. Fragment ions produced in the first field free drift path have the same velocity as their parents, and, hence, enter the reflectron all at the same time but with different kinetic energies. The kinetic energy (£f) of a frag-ment ion is given by

£f = Epfm (/m = Wf//Wp) ( 1 )

where Ep is the kinetic energy of the precursor ion and mp and raf are the masses of the precursor and fragment ion respectively. Since fragment ions do not penetrate into the retarding field of the reflec-tron as deeply as their parents, their turn-around time is shorter. Hence, they leave the reflectron earlier and arrive sooner at the detector than their unfragmented precursors.

For a single-stage reflectron with a homogeneous deceleration field a simple linear relationship exists between the flight times (tp and tf) and the masses (mp and mf) of the precursor and fragment ions respectively:

2(ff/'p) = («f/wp) + 1 (2)

Thus, in principle, a complete mass spectrum of fragment ions can be simultaneously recorded by employing a single-stage reflectron in its usual mode of operation [19,20]. The problem here is mass resolution. In all kinds of reflectrons optimal time focussing of ions with an initial energy spread requires that ions spend a certain time in the reflec-tron. If this focussing condition is mismatched because the passage through the reflectron is either too long or too short, mass resolution is more degraded, the more the actual ion flight path (or

100

75

3 a ►g

50

I 25

]

•S

[M+

HJ+

I

smal

l P

SD

fr

agm

ents

| I

I I

I |

I I

I I

| 80

.0

85.0

90

.0

"Yh

95

.0

I I

l l

l l

I l

100.

0 10

5.0

T

H I

|l

I 1

Ã|

I I

110.

0 11

5.0

HV

*

Fig

. 3.

Sch

emat

ic i

llust

rati

on o

f fr

agm

ent

ion

mas

s sp

ectr

omet

ry i

n a

two-

stag

e re

flec

tron

TO

F-M

S (s

, =

4 m

m,

s 2 =

4 m

m,

L,

= 20

34 m

m,

d x =

20

mm

, d 2

= 2

39 m

m,

L 2 =

122

8 m

m).


flight time) in the reflectron deviates from optimal conditions. Therefore, the recording of a complete fragment ion mass spectrum with acceptable mass resolution requires many consecutive "tuning" steps of the reflectron voltages in order to match optimal time focussing conditions for any (small) segment of the full daughter ion mass range recorded.

As illustrated in Fig. 4 a two-stage reflectron is less feasible against deviations from the opti-mal penetration depth or optimal residence time than its single-stage counterpart [21]. In an opti-mally designed device, conditions can be found where the window of daughter ion energy (or mass), over which flight time focussing (i.e. AT/T of fragment ions) would not exceed say 2 x 10~4

can be made as wide as one fifth of the full fragment ion energy (or mass) range, whereas a single-stage reflectron, for the same resolution, would allow the accommodation of only a mass segment of 0.09 full scale. Consequently, in a two-stage reflectron operated e.g. at voltages set for optimal mass reso-lution of 12keV precursor ions (U\ — 7162V, U2 = 12.500 V in our case), the first segment of fragment ion energies or masses over which ions will be recorded with an acceptable resolution of say 4 x 10"4, is 8000 eV < Ef(m{) < 11 000 eV, i.e.

115.4 T

115.2

r 5

115.0-h

114.8 4-

9000 10000 11000 12000

kinetic energy [eV]

Fig. 4. Comparison of ion flight time as dependent on kinetic energy in a RETOF MS equipped with either a single-stage or a two-stage reflectron. Calculation is based on the geometrical and electrical parameters applied in the actual RETOF configura-tion assuming a parent ion of 1350 u. Parameters for the single-stage reflectron were adjusted to the same virtual flight path.

one quarter of the full fragment ion mass scale. Also, fragment ions with 5800eV < Et(mf) < 7000 eV or 11000eV< E{ < 12.000 eV will still be distinguishable but with increasingly poorer mass resolution. All other fragment ions with Ef(m{) < 7162eV will not penetrate into the sec-ond stage of the reflectron but will be mirrored by the short first stage with virtually no mass dispersion at all (see inset to Fig. 3).

In order to obtain full mass spectra of PSD frag-ment ions with a two-stage reflectron, typically 10 to 14 consecutive mass scale segments must be recorded with the reflectron adjusted to the respec-tive fragment ion energies. This approach gives suffi-cient segment overlap to compile complete fragment ion mass spectra at optimal mass resolution.

Mass calibration of fragment ions

While, in a single-stage reflectron TOF-MS, a simple relationship exists between the flight time and the mass of a fragment ion (see above), mass calibration of fragment ions dispersed in a two-stage reflectron is anything but straightforward, since the full motion equation describing the flight time in the reflectron cannot be solved in terms of fragment ion mass. Therefore, if one wants to avoid tedious calibrating procedures via model meta-stables of known precursor and daughter masses, one needs to know, to the highest possible accu-racy, the geometrical parameters of the spectro-meter as well as the actual (variable) voltage settings applied. Based on this set of data and the actual flight times of the fragment ions recorded, a computer can calculate by iteration the fragment ion masses with the desired accuracy.

The motion equation of the RETOF-MS in its general form can be written as

T{ = [faten Uv mp)> + ' d ( £ l i L2, <*, â , Wp)

+ tT(dx,d2,OL, £/a, Uu U2jmpJm)} (3)

where: 7} = total flight time of the fragment ion (w); /a = flight time in the acceleration region(v); td = flight time in the field free drift region (v);


tT = flight time in the reflectron (v); mp = mass of the precursor ion (v); sa = length of the accelera-tion region (f); t/a = acceleration voltage (v); Lx

and L2 = first and second field-free drift length (f); a = half angle of deflection of the ion flight path versus the sample surface normal (f); dx and d2 = length of the first and the second stage of the reflectron (f); U\ and U2 — applied voltages at grid 2 and 3 of the reflectron relative to the flight tube potential (v); fm = fragment/precursor mass ratio (c). In the preceding list (f) indicates a fixed (instrumental) parameter, (v) an experimental vari-able, (m) a measured variable and (c) a parameter to be determined.

For practical work, an algorithm based on the above set of parameters has been integrated into the PC-based "Ulisses" TOF data acquisition pro-gram (Chips at work, Bonn, Germany) to perform this task.

As will be shown in the Results section the aver-age mass resolution of fragment ions is mainly deter-mined by the width of the precursor ion signal and reaches typically Μ/ΔΜπ 800-1000 (FWHM) if precursors are in the 1000-2000 u range (see exam-ple given in Fig. 7).

Detector and signal recording

One of the inherent problems of using a reflec-tron for PSD fragment ion mass discrimination relates to the fact that fragment ions of different masses are also laterally dispersed by the reflec-tron at least if — as is usually the case — the reflectron axis deviates from a 180° reflection angle by a few degrees (2-3° vertically in our case). Thus, if one wants to avoid mass dependent transmission losses, a large area detector is manda-tory. Under our actual geometrical conditions we found a 75 mm diameter MCP detector to repre-sent just the lowest theoretical area limit which would accommodate the lateral (geometrical) spread of fragment ions recorded over a typical mass scale segment. We are, however, fairly sure that, owing to collisional momentum transfer and kinetic energy release, the angular spread of frag-

ment ions is by far larger than the spread based simply on geometrical calculations.

To avoid postacceleration by fringe fields extending from the front side of the detector (usually at -1.8 kV), a grounded grid was placed about 15 mm in front of the first channel plate, or the MCP detector was offset such as to shift its front-side potential to zero. While the latter proce-dure is certainly preferable with respect to correct mass assignment of, in particular, low mass frag-ment ions, the former prevents sensitivity loss (very low kinetic energies) for these ions.

Usually 50-100 spectra were summed for each voltage setting (mass window) at a cycle frequency of 1 Hz. No smoothing or filtering routine was applied to the spectra presented in this paper.

This means that recording of a full sequence spectrum with typically 10-12 overlapping spec-tral segments requires « 30 min of recording time in our actual setup.

Results

State of the art performances: a model case {Substance P)

This paragraph aims to demonstrate the instru-mental performances reached so far. Since PSD-RETOF mass spectrometry for peptide sequen-cing after MALDI is an innovative approach not yet common to, or established amongst, mass spec-trometrists we start with a model case to illustrate conditions and achievements in detail.

As a model peptide we have chosen Substance P, a neuropeptide (monoisotopic mass 1346.7 u) con-sisting of 11 amino acid residues which has been frequently and extensively investigated in high energy as well as in low energy CAD MS-MS studies [5,6,27-29]. Substance P was also included in the pioneering work of Tang and co-workers [19,20] on the application of reflectron (single-stage) TOF mass spectrometry to analyse either metastable or prompt product ions after particle induced desorption and was the subject of a recent attempt to explore multiphoton photoionization-dissociation (MUPI) for peptide sequencing [30].


Fig. 5. Consecutive segments of PSD fragment ion spectra of Substance P. Labels in each spectrum indicate the voltage settings of U2 in the two-stage reflectron with U\ = 0.57377 U2. See text for a further explanation.


PSD ion spectra recording

Figure 5 contains a series of consecutive PSD fragment ion segment spectra obtained from Sub-stance P in a DHB matrix after laser desorption (ë = 337 nm). On the left hand side each segment spectrum starts with a broad peak containing all those fragment ions which, at the indicated retard-ing field, have not penetrated the second stage of the reflectron and have been mirrored by the first stage without mass dispersion. This "junk-hill" is a fair indicator of the total fragment ion yield (about 50% of unfragmented parent molecules in the case of Substance P) and helps to roughly scale frag-ment ion peak amplitudes within a series of conse-cutive segment spectra. On the right hand side of the "junk hill" each segment spectrum displays a number of mass resolved fragment ion signals. Each spectrum might be visualised as part of a string which, moving left to right through a win-dow, is being continuously pulled out of the "junk hill" when the reflectron voltages are reduced from standard settings (upper left hand panel) down to the lowest values (lower right hand panel). Since useful sequence information is not expected from ions smaller than 70 u, recording was usually ter-minated with the last spectrum tuned to the appro-priate mass scale segment.

The spectral window over which mass separated fragment ion signals can, in principle, be identified is indicated by two broken vertical lines (w\) in Fig. 5. They delineate borders given by the foot of the "junk hill" on the left side and, on the right side, by a point where fragment ions start to disappear (because they are leaving the reflectron on the rear side). The mass range of each segment spec-trum extends over a range of 0.56 to 1 of the largest fragment ion (mfl) accommodated. This means that, basically, five consecutive segment spectra recordings could cover the whole product ion mass range extending from the precursor ion mass (mp) down to say 0.07 mp. Since, however, acceptable mass resolution is obtained only in a smaller spectral window which extends from say 0.7 Wfl to 0.9 rafl (indicated by the borders labelled

u>2), 10-14 segment spectra are usually needed for full mass range analysis.

Complete PSD fragment ion mass spectra, as shown in Fig. 6, are composed of a series of con-secutive overlapping cut outs from segment spectra as displayed in Fig. 5. Such composed spectra are infected with the problem of fitting a mass scale on the abscissa since, in a two-stage RETOF, the rela-tionship between time of flight and fragment ion mass neither follows a simple steady function nor does the same expansion factor apply for each segment. Thus, in the absence of an available rescaling algorithm, the fragment ion signals of the spectra displayed in this work are individually mass assigned.

Accuracy and precision, and lower limit of analysis

In the model case of Substance P, fragment ion mass assignments and cleavage interpretations according to the scheme of Roepsdorf and Fohl-man, and Johnson et al. [31] are given in Table 2 and Fig. 6. Fragment ion mass resolution M/AM (FWHM) is in the range of 800 to 1200 (see also Fig. 7) which is typical for precursors the size of Substance P. Accuracy of mass assignments is 5 x 10~4 (averaged over the whole fragment ion mass range) with slightly better figures (3 x 10~4) applying for fragment ion masses above 700 u (relative errors in the digital control of reflectron voltages are larger at lower absolute voltage settings). Precision within one experiment done on the same sample is about 2 x 10~4 and worsens to 3-4 x 10"4 on a day to day basis.

With 60 single spectra summed for each mass segment, the signal-to-noise ratio is excellent in cases like Substance P (no filtering or smoothing was applied to the spectra shown). The total yield of fragment ions in this particular case was about 0.5 with respect to the precursor ions reaching the detector unfragmented. (For a discussion on effi-ciency of fragmentation dependent on various experimental parameters, see the mechanisms of activation and energy transfer below.

The amount of analyte consumed is never known

R. Kaufmann et al.jlnt. J. Mass Spectrom. Ion Processes 131 (1994) 355-385 365

Fig. 6. Complete PSD fragment ion spectrum of Substance P (1347.7 u). This is a composed spectrum consisting of 14 segments taken from recordings as shown in Fig. 5. For each segment 50 single shot spectra were summed up. Mass labels are those which the computer program calculated on the basis of geometrical data and actual voltage settings. Total acceleration voltage = 10 kV; initial acceleration

field strength « 3 x 104 Vcm"1.


Table 2 Calculated and observed masses of PSD fragment ion of Substance P

Peak No.

Fragment mass

Calc. Obs.

AM(u) Accuracy (KP3)

Assignment

1302.7 1171.7

1302.1 1171.9

-0.6 -0.2

0.46 0.17

M-44

10

11 12

13 14

15 16

17 18

19 20 21

22

23 24

25 26

27 28

29 30 31

32 33 34

1154.2

1029.4 1041.6 1001.6

984.6

876.5 854.5

837.5

718.5 707.4

690.4 607.4

590.4 579.4

562.4 473.4

451.3 382.3 365.3

354.3

337.3 254.2

237.2 226.2

209.2 140.1

129.1 120.1 112.1

101.1 87.1 84.1

1154.2 -0.5

;.6 :.i .4

0.8 -0.5 -0.2

984.7

875.8 854.4

837.7

-0.1

-0.7 -0.1

+0.2

718.3 707.0

689.8 607.3

590.9 579.1

562.2 472.9

451.3 382.2 364.8

354.6

337.4 254.2

237.1 226.1

209.1 140.1

129.1 119.9 112.1

100.9 87.0 84.0

-0.2 -0.4

-0.6 +0.1

+0.5 -0.3

-0.2 -0 .5

-0.1 -0.5

+0.3

+0.1

-0.1 -0 .1

-0.1

-0.2

-0.2 -0.1 -0.1

0.43

0.80 0.50 0.20

0.10

0.80 0.12

0.24

0.27 0.57

0.86 0.16

0.84 0.51

0.36 1.06

0.26 1.37

0.85

0.30

0.42 0.44

0.48

1.60

2.00 1.10 1.20

dio a 9 - 1 7 a8

a8-17

a7 + Na - 1 a7

a7-17

b 6 - 17

a6

a6-17 b5

b5 - 17

a5

a5-17 a4+Na - 1

a4

b3

b3 - 17

»3

a 3 - 1 7 b2

b 2 - 1 7 a2

a 2 - 17 b, - 17

a, (Phe) a, - 1 7

(Lys/Glu) (Leu) C3H6N3

35 70.1 70.1 (Pro)

R. Kaufmann et al./Int. J. Mass Spectrom. Ion Processes 131 (1994) 355-385

1 0 0 1 Al

367

75 H

eu

50 H

25H

M=226.2u =17% of

parent mass Ì/ÄÌ = 1130 AM = 0.2u

1 1 2 . 7 1 1 2 . 8 1 1 2 . 9 1 1 3 . 0 1 1 3 . 2 1 1 3 . 4 ps

Fig. 7. PSD fragment ion mass resolution demonstrated for the a2

Matrix was DHB, ë =

exactly in M ALDI but can be approximated (from data such as the diameter of the laser focus, the amount of sample deposited per unit area, and the number of laser shots needed to exhaust one sample position) to a value in the range 20-40 amol per laser shot under the present conditions. Hence, with 1000 laser shots for complete fragment ion analysis, analyte consumption amounts to 20-40 fmol This figure is certainly not the lower limit for analysis. As the example in Fig. 8 demonstrates, signal-to-noise ratio would have permitted a reduction in the number of spectra summed in Fig. 6 by at least one order of magnitude without loss of essential information. Thus, the question of detection limits lends itself to the problem of how sample preparation protocols can be devised to match such ultra-low detection capabilities. It is obvious that the way samples are usually prepared for MALDI-TOF mass spectrometry are unsatis-factory. However, with the foreseeable adaptations of MALDI-TOF mass spectrometry to ultra-small

fragment of Substance P (full width at half maximum definition). 337 nm, £/acc = 10 kV.

volume separation techniques such as capillary zone electrophoresis, instrumental sensitivity will certainly be greatly improved.

Product ion pattern and type of cleavages

As to the pattern of favoured backbone clea-vages, there are some similarities but also some remarkable differences between PSD-RETOF and CAD-MS-MS for which the example of Substance P appears to be already fairly representative.

First of all there is the near complete absence of side chain specific cleavages (dn in the case of Sub-stance P) in contrast to high energy CAD spectra which regularly contain almost all possible dn frag-ments at yields comparable to, or even exceeding, those of the corresponding aM fragments. However, high energy CAD spectra of Substance P are usually free of b„ fragments [28,29] in contrast to low energy CAD spectra obtained in triple quad or hybrid setups which are largely dominated by bn

b 1-1

7

UWIJu

J Fi

g. 8

. Ten

seg

men

t sin

gle

shot

PSD

fra

gmen

t io

n sp

ectru

m o

f Su

bsta

nce

P. T

otal

am

ount

of

sam

ple

cons

umed

« 4

00am

ol.


fragments [5,6]. MALDI-PSD-RETOF mass spec-tra apparently represent some intermediate situa-tion featuring both a complete a„ and a nearly complete (n = 1-9) bn series together. (These find-ings will be reconsidered and further discussed below in the Mechanisms of activation and energy transfer section).

Another obvious peculiarity of our MALDI-PSD fragment ion spectra is the predominance of the (a„ - 17) and (b„ - 17) signals which usually occur at much larger abundances than the corre-sponding a„ and bn counterparts. The formation of these products is usually thought to be due to ammonia loss from an N-terminal arginine resi-due. Although such satellites are not uncommon

in CAD spectra they never reach yields as observed under the present conditions. We cannot exclude the possibility that a small fraction of our prompt precursor ions are already desaminated, although such ions have never been found in the prompt MALDI spectra recorded by linear TOF instruments. Since precursor ion selection in our actual instrument does not yet allow discrimination between MH+ and (M-17)H+ species at 1348u (Substance P) we must leave this question open for now. However, the only other experimental condition under which (a„/b„ - 17) satellites of comparable intensity have been observed is reported in the work of Tang et al. [20] on PSD-RETOF spectra after particle induced desorption

Table 3 Comparative performance data obtained in MS-MS studies of Substance P

Mass spectrometer

Type of device

Four-sector BEEB

Hybrid BEqQ

Four-sector Four-sector

EBEB

Four-sector EBEB

Four-sector EBEB

RETOF one stage (ion count, m.)

RETOF one stage (ion count, m.)

RETOF two stage (analogue m.)

RETOF two stage (analogue m.)

Ion source

FAB 35keV cesium

FAB 8keV xenon

CF-FAB FAB

8keV xenon

FAB 8keV xenon

FAB 6keV xenon

SI lOkeV cesium

SI lOkeV cesium

MALDI (DHB/ 337 nm)

LD/MUPI

Activation mechanism

High energy CAD(lOkeV) helium

Low energy CAD (14-27 eV) argon

High energy CAD High energy

CAD(6keV) helium

High energy CAD(4keV) helium/argon

High energy CAD(5keV) helium

PSD(8keV)

Prompt decay

PSD lOkeV

Prompt Photo-dissociation

Fragment mass accuracy (u)

±0.3

±0.5a

±0.3 ±0.3a

±0.3

±0.3

±0.3

±0.3

±0.3

±0.3

Peak half width (u)

0.3

3.0

0.3 0.3a

0.3a

0.2 at 225 u

0.2

0.2

0.2

0.3

Predominant backbone cleavages

a„, d„

a„, b„

a„, (d„) a», d„

a„, d„ (He) a„, d„ (Ar)

a„,d„

a„, a„ - 17 b„, b„ - 17

a„

a„ - 17, a„ (PSD) b„ - 17, b„ a„, d„ (CAD)

ÆË» y«> ÷Ë

(a„, c„)

Estimated limits of analysis

15pmol

370pmol

200fmol lOOpmol

Not determined

0.5pmol

0.1-1 pmol

0.1-1 pmol

0.4-4 fmol

Not determined

Ref.

[5]

[5]

[27] [6]

[28]

[29]

[19]

[20]

This work

[30]

Estimated from published data.


of Substance P. It is interesting to note that prompt (SIMS) fragment ions obtained under essentially the same conditions did not contain such satellites at all.

Thus, for the time being, we conclude that loss of ammonia is primarily a secondary (PSD) process which, due to a relatively low rate constant, requires a rather long time window to reach that yield which is obtained under our conditions. This type of cleavage seems to run in parallel with the backbone fragmentation processes and, to a lesser extent, also occurs in pep tides containing: (1) either arginine in non-terminal positions or (2) aminated side chains other than arginine.

Comparison with MS-M S

From the viewpoint of practical mass spectro-metry it is necessary to compare these results with those obtained by MS-MS. In a recent study, Bean et al. [5] have investigated the performance of four-sector instruments (BEEB) versus hybrid designs (BEqQ) in the analysis of collisionally activated peptide fragment ions including Substance P. Their data appear to be representative of the many other CAD investigations carried out on

Table 4 List of peptides investigated

this frequently used model peptide (refs. 6 and 27-29). Table 3 summarises most of the relevant data collected from the literature. It demonstrates that, at this point, MALDI-TOF mass spectrome-try of PSD product ions can already easily compete with standard MS-MS. Accuracy and precision of mass assignment, mass resolution, and (although not explored in Table 3) time of analysis, all reach the performance of four-sector tandem instruments equipped with an array detector [29]. Instrumental sensitivity in PSD-RETOF analysis is better by at least two orders of magnitude; the upper precursor mass limit (see next paragraph), up to which full sequence information can be obtained, is about 2500 to 3000 u.

General observations

Our actual experience with PSD fragment ion analysis of peptides extends over about 60 compounds (conjugates included) ranging from Leu-enkephalin (five residues, 556 u) to ACTH (39 residues, 4567 u). Table 4 lists some of the investi-gated peptides larger than lOOOu which gave full sequence spectra, with the exception of apamine internally crosslinked by two disulphide bonds.

Substance Average mass (u)

Amino acid sequence

Synthetic peptides

Renin-inhibitor #421 Substance P Tyr8-Substance P #435 Gamma-MSH Bombesin #520 Tyr4-Bombesin # 493 (Lipid) Apamin

Melittin

I II III IV V

1030.3 1085.3 1085.3 1043.3 1071.3 1156.4 1184.6 1347.7 1363.7 1545.8 1570.8 1619.9 1668.9 1669.9 1864.1 2021.5

2846.6

H-GAKAVGEAKAAG-OH . -L G- . . . -G L- . . . -G A- . . . -G V- . . H-PHPFHLFVY-OH H-IGEGTYGVVYK-OH H-RPKPQQFFGLM-NH2

H-RPKPQQFYGLM-NH2

AC-RKSATTKKVASSGSP-NH2

H-YVMGHFRWDRFG-OH H-XQRLGNQWAVGHLA-NH2

H-QRKEAADPLASKLNK-OH H-XQRYGNQWAVGHLA-NH2

Fatty acid-GLTISSLFSRLFGKK-OH H-CNCKAPETALCARRCQQH-NH2

H-GIGAVLKVLTTGLPALISWIKRKRQQ-NH2


The largest peptide which gave a full set of iden-tifiable sequence ions was melittin (26 residues, 2849 u). Figure 9 shows a typical spectrum.

There are a few recent reports on attempts to fragment melittin by ESI in combination with either a four-sector instrument [32], a triple quad [16], or an ion-trap MS-MS device [15]. While the instrumental conditions of these approaches differ in many respects they share multiple low energy collisions as a common denominator for precursor ion (pre)excitation. These attempts have been partly successful insofar as they could produce comparable product ion spectra of melittin (4+ or 2+) precursors which, in the absence of product ion charge state information and with the rather limited mass resolution attained, could only tenta-tively be assigned to at least some partial series of y- and b-type sequence ions.

Taking those melittin product ion spectra as refer-ences we notice that we have certainly got much more complete and unequivocal sequence informa-tion with the C-terminal (y, and y-17-type) frag-ment ion signals dominating the spectrum (two arginine residues near the C-terminus at the 22-and 24-positions). There is, however, also a series of rather abundant b- and a-type fragments, parti-cularly in the low mass region (n = 2-8), which, together with the y-type fragments, provide full sequence information on this 26-residue peptide. Although the number of spectra accumulated for each segment in this case had to be increased to n = 250 in order to attain an acceptable signal-to-noise ratio, no more than 150fmol of analyte (out of a sample prepared containing 20 pmol) had to be consumed for the analysis.

The case of melittin, although remarkably suc-cessful in principle, can also serve to outline the accuracy limitations inherent so far in MALDI-PSD mass spectrometric analysis. For reasons of optimal yield and "in-source activation" (see below), laser desorption of larger peptides requires irradiances at least 30-50% above the threshold of ion formation. As a consequence, mass resolution drops significantly below the nominal instrumental performance. Thus, isotopic discrimination is

usually lost with peptides above say 1800u and product-ion as well as precursor-ion signals become envelopes of isotopic distribution pattern folded with a T0 (start time) spread typical for the M ALDI ion formation process. Although smooth-ing and center-of-mass routines can be run on such signals, the degradation of mass assignment cannot be fully offset. In consequence, the factors affecting precision and accuracy of fragment ion mass assign-ment add up to a typical relative error of (0.8-1.2) x 10"3 for fragment ion masses above 2000 u. This means that, in the present case of the melittin spectrum, mass determination of the larger frag-ments is within absolute error limits of 1.5-3 u.

Attempts to fragment peptides larger than melit-tin, e.g. ACTH, gave only limited partial sequence information, usually from the small-sized (n = 2-5) and also from a few large-sized products close to the precursor ion mass.

Remarkably enough, the overall efficiency of fragmentation did not diminish with increasing precursor size. From the height of the "junk-hill" relative to the signal amplitude of the unfragmen-ted parent molecule, it even appears that total frag-ment ion yield increases rather than decreases with larger precursor masses (see Fig. 10). With respect to the increasing number of possible cleavages in such molecules, one might be tempted to see the degradation of the signal-to-noise ratio simply as the consequence of statistics which might be counteracted by accumulating a larger number of spectra. In practice, however, this turned out to be of only limited success. Occasionally, "spectral islands" of fairly-well-discernable signals appeared on top of a "wavy" but otherwise unresolved back-ground which, more or less, extended over all spec-tral windows. Our actual, but admittedly rather vague, interpretation is based on a conjunction of two phenomena. (1) As already discussed for the example of melittin (see above), mass resolution in M ALDI spectra of larger peptides is not as good as one would expect from the instrumental perfor-mances. Thus, with the spectral number density of possible cleavages increasing and the isotopic distri-bution pattern broadening, more and more of the


y22-17

Fig. 9. PSD fragment ion spectrum of melittin (2847 u). Acceleration voltage 12kV, initial electric field strength « 5 x 104 Vcm i (250 spectra summed in each of 14 segments).


3.U

100 148 ps

a.u

146 168 190 212 ps

Fig. 10. Increase of total fragment ion yield (compare "junk hill" area versus parent molecule signal amplitude) with increasing precursor ion mass. Spectra were recorded from the same mixed peptide sample. Acceleration voltage was 14kV, initial field strength

= 4 x 104VcnT'.


smaller signals will smear out in a noisy back-ground. (2) Since, in addition, the rate constants for decomposition are decreasing with larger pre-cursor ion size, the fraction of "late" fragmenta-tions which occur during the residence of an ion in the reflectron or thereafter augments, thereby con-tributing to an abundant but unresolved fragment ion background.

As to the analogies of PSD with CAD fragment ion spectra, we also observe charge-directed frag-mentations, i.e. a predominance of either C- or N-terminal fragment ion series depending on where the charge bearing residues are located. Further congruent findings are: (1) fragment ions ending in a glycine residue appear usually at low abun-dance, those ending in a proline at high abun-dance; (2) hydrophobic residues (Leu/Ile and Val) are rather unfavourable while hydrophilic residues (Thr and Ser) are generally more favourable with respect to site directed cleavages.

The presence or absence of very abundant residue-specific immonium ions, in particular

those of proline (70 u), threonine (74 u), lysine/glu-tamine (101 u), histidine (HOu), phenylalanine (120 u), and tyrosine (136 u) is a rather unequivocal indicator for the occurrence or the lack of the cor-responding residue, whereas arginine (129u) and tryptophane (159u) coincide with b-type fragments of Gly-Ala or Ser-Ala combinations respectively.

We have concentrated our attention primarily on protonated precursors by avoiding alkali contami-nated samples or, in cases of minor contamina-tions, by trying to select the protonated from the cationised (usually Na) species. In a few favourable instances (about equal yields of cationised and pro-tonated precursor species not larger than about 1000 u), where our gated beam blanker permitted a good enough precursor selection, we have tried to compare PSD product ion spectra of both kinds of precursors. The data in Table 5 demonstrate the result of such an attempt in a synthetic peptide (peptide V in Table 4, 12 residues, 1071 u) which had been Na contaminated to such an extent as to yield more than 70% of the precursor ions as catio-

Table 5 PSD fragment ion observed with either (M + H)+ or (M + Na)+ precursor of peptide (V)a

(M + Na)' (M + H)H

Product mass (m/z)

Assignment Product mass (m/z)

Assignment

995 877

722

635 607 478

449

328 311 257 129

b u + N a + 1 8 a,0 + Na - 1

c8 + Na - 1

b7 + Na - 1 a7 + Na - 1 a6 + Na - 1

b5 + Na - 1

b4

b4

b3

b2

17

954 883 866 855 812 684 656 627 613 585 484 457 439 427 399 328

257 129

bu bio b , o - 17 aio b9

b8

ag c7

b7

a7

b6

a6

8 6 - 1 7 b5

a5

b4

b3

b2

See Table 4.


nised species. While the product ion spectrum of the protonated precursor is dominated by a series of N-terminal b„ and a„ fragments, the product ions of the cationised precursor are mainly of the type (a„/b„+Na-l). With cationised precursors, the ladder of sequence ions is terminated by an abundant signal attributed to a product of the type (bn+Na+18) . These preliminary results are in agreement with the recent analysis of Teesch et al. [33] employing metastable and CAD frag-ment ions of gas-phase cationised peptide precur-sors to investigate the location of the alkali metal in such complexes.

"Internally stabilized" peptides

The presence of internal disulphide, amide, or ester bonds in the investigated peptides is clearly reflected in the pattern of PSD product ions. With the still-preliminary results at hand we can state that such conditions tend to reduce the overall yield of fragment ions in general. In all cases inves-tigated so far, the bond forming residue(s) could easily be located since the series of sequence ions (either N-terminal or C-terminal) usually ended with the fragment next to the stabilised part of the backbone.

b?-18

781

# 52G

H-QRKEAAD Ü J »

- 1 8

— b 7 - i 8 — ;

(781 u)

PLASLLNK-OH

(MH+-18)

Ç^^Ë^^Ë

«267.35

Fig. 11. PSD fragment ion spectrum of a synthetic peptide (# 520, 1668 u) internally crosslinked as indicated by the loss of 18 u (H20) with respect to the calculated parent mass. Possible crosslinks (supposed to involve a carboxyl and an NH2 group) are proposed by the inset. Note that more than 90% of all fragment ions formed show up in an extremely abundant b7 - 17 fragment signal which is

supposed to represent the cleavage located in front of the internally stabilised part of the backbone.


Occasionally, we even saw an extraordinary pre-dominance of that "last possible cleavage", espe-cially if the residue forming the internal crosslink was an aspartic acid. Figure 11 shows an example of a synthetic peptide (# 520 in Table 4, 15 residues, 1668u) for which an ESI triple quad was unsuc-cessfully used to try to establish its amino acid sequence. MALDI-RETOF mass spectrometric evidence suggests first of all that 90% of the parent molecules had a mass deficit of 18 u (see inset Fig. 11), indicating the loss of H 2 0 most probably due to internal cyclisation. The vast majority (more than 90%) of PSD fragment ions in this case appeared in the form of one single, very abundant signal at 781 u which has been identified as the b7

fragment of the (M-18)H+ precursor ending in a C-terminal aspartic acid residue. Since N-terminal fragments smaller than n = 1 were completely absent, except for a rather prominent signal at 129.1 u (bO, we suppose that an internal amide was chemically formed between the asp7 and the arg2 residues prior to mass analysis. Similar predo-minances of asp-directed b- or y-type fragmenta-tions have been observed in gamma-MSH and ACTH, both containing asp at the n = 9 and 29 positions respectively.

Mechanisms of activation and energy transfer

We are not yet in a position to draw a clear-cut picture of the activation, energy transfer and relaxation mechanisms ruling PSD fragmentation under our experimental conditions. Nevertheless, some preliminary evidence allows us to suggest at least a rough scenario.

Basically, in line with current concepts, our pre-sent understanding of MALDI-PSD mechanisms sees fragmentation of a polyatomic ion as the result of two distinct processes separated in time, namely: (1) activation by energy transfer; (2) dis-sociation after intramolecular energy redistribution. Most contemporary concepts distinguish between collisionally activated and so-called unimolecular dissociations, the latter denoting those fragmenta-tions which occur in the absence of a collision gas

and as the consequence of some "in-source" exci-tation. Under the present experimental conditions this dichotomy is not as clear-cut as in classical MS-MS technologies, since: (1) "in-source" activa-tion is essentially (multi)collisional; (2) the prob-ability for "post-source" collisions in the field free drift path certainly cannot be neglected under the actual conditions [23] but, depending on residual gas pressure and ion size, can be large or small. Thus, making no ab initio distinction between the different sources of internal energy one may, in the time domain, simply distinguish between post source and prompt dissociations, although in MALDI-TOF mass spectrometry the observation of prompt fragments is a rare exception. It will, however, be shown below that in MALDI-RETOF mass spectrometry, experimental condi-tions can be varied such that either in source activa-tion or high energy post source collisions become the predominant sources for energy transfer and subsequent fragmentation.

"In source" activation

The principles of desorption and ion formation in MALDI are still poorly understood (for discus-sion of current concepts see ref. 34). It is clear by now that, with the actual irradiances applied, photoionization and/or photodissociation cannot play any major role. Instead, energy deposition in a superficial layer of the solid state matrix, via thermal and/or pressure pulse "sputtering" [35] leads to a jet-like ejection of matrix particles with intact analyte molecules entrained. While the plume of matrix-derived particles expands with a velocity of typically 1000-1500 m s - 1 over a cos2 0-cos3 Θ angular distribution [36], analyte molecules are being dragged by this expanding cloud and reach about half this velocity (« 700 to 800 ms"1). Results of Beavis and Chait [37], Huth-Fehre and Becker [38], and from this laboratory [36] consistently demonstrate that analyte ions emerging unaccelerated from the plume are roughly isovelocic irrespective of their mass and have a quasi-gaussian velocity distribution extending over about one fifth


of their mean velocity (full width at half maxi-mum). Although there is certainly some analogy between plume expansion after MALDI and the classical pulsed gas jet used to entrain and cool labile compounds, it is not clear whether the hydro-dynamic model proposed by Vertes et al. [39] is correctly describing the energy transfer mechan-isms especially between neutral matrix and

entrained guest molecules. This applies especially to the degree of expansion cooling which, if their predictions were correct, would certainly be diffi-cult to reconcile with the extent of PSD observed during flight.

It is, however, not unreasonable to assume that the major proportion of the activation energy acquired in source stems from ion/neutral

4

k T

P(M

, " f r f t o t a l ) '\

Y x f r ( l - 9 )

! 3 se 1 0 4 Vcm"

1

JL

1

1 ■

♦ H ) 1

L 1

108 126 142 159 ìå

5 -x. 0.O V e r a

L

4 >t 1 O V c m — 3.

\

Fig. 12. Three PSD fragment ion spectra of Substance P (5 kV) with the reflectron set for parent ion recording (most of the fragment ions formed are contained in the "junk hill"). Voltage labels indicate initial field strength applied in the first distance of the split acceleration

field. Note that fragment ion yield largely increases with higher extraction field strengths.

378 R. Kaufmann et al./Int, J. Mass Spectrom. Ion Processes 131 (1994) 355-385

collisions under the condition that the ions rapidly gain kinetic energy from the acceleration field. Thus, one might predict that the initial electric field strength acting on analyte ions during early plume expansions, when they are still surrounded by a high number density cloud of potential collision partners (mainly neutral matrix molecules), should be a main deter-

minant for the efficiency of PSD product forma-tion.

By employing a split acceleration field we could experimentally confirm this prediction. As illu-strated in Fig. 12, we have determined the frag-ment ion yield as depending on the initial field strength (E{) in the split acceleration distance. From the data shown in Fig. 13 it is apparent

0.9

0.8

0.7

0.6

0,5

0.4

0.3

0.2

0.1

5000 10000

Ei(Vcm-1)

15000 20000

Fig. 13. Total PSD fragment ion yield (Substance P) as dependent on (1) the strength of the initial electric field and (2) the kinetic energy to which the ions are finally accelerated. /fr is the intensity (peak area) of the fragment ions (the "junk hill"), and Ip is the intensity (peak

area) of the precursor ion. D, Uacc = 14 kV; Ä, Um = 10 kV; O, Um = 3kV; o , C/aCc = 1.5 kV.


that, for a given acceleration voltage (C/acc) of say 10kV, the fragment ion yield of e.g. Substance P increases from 0.15 to 0.51 if E-x rises from 1.66 x 102to 1.66 x lO'VcnT1.

It is particularly noteworthy that, for a given initial field strength (E{)9 the dissociation curve is shifted upwards to higher yields if the total accelera-tion voltage (£/acc) is lowered. We suspect that this is due to the prolongation of the ion flight time increasing the probability that low rate decays will occur in time. Assuming that first order kinetics rule this type of process we expect a mono-exponential decay characteristic which was found to be true after transformation of the data into the time domain (see Fig. 14).

In a given experimental situation, in source activation increases with laser irradiance in about the same way as described in one of our previous papers [25] investigating the stability of peptide

ions produced by MALDI in a linear TOF arrange-ment.

"Inflight" activation

As demonstrated in the preceeding paragraph, "in source" activation can be nearly fully prevented by choosing a very low initial field strength and irradiances close to the ion formation threshold such that the majority (more than 90%) of all par-ent ions formed survive the passage through the first field free drift path unfragmented. This opens the opportunity to perform "classical" CAD experiments by admitting gaseous atoms or mole-cules into the TOF spectrometer for high energy "in flight" collisions. To this end we simply turned off one of the two turbo pumps of our spectro-meter, thereby increasing the residual gas pressure from 5 x 10"7 to « 5 x 10~6 Torr. At this pressure

Time of flight

0.1

20 120

Fig. 14. Flight time versus decay rate ((/p//fr + /p) plots) of Substance P for two different initial field strengths: · , E{ = 3.3 kVcm- 1; ►, E{ = lOkVcm-1. Data were taken from the data set of Fig. 13 by calculating the flight times in the first field free drift path for the

different i/acc values applied.


the intensity of Substance P parent ions decreased to about 30-40% of its initial value. Thus, the criterion of "optimal collision pressure" as defined by e.g. Neumann et al. [40] is fulfilled, meaning that

the probability of a collisional encounter during flight is unity.

Figure 15 illustrates the experimental protocol: the experiment performed on Substance P started

Ô0Ã 192.9 194.9 196.9 198.9 119.9 112.9 114.9 US.9 118.9 129.9 ps

E. = 1 0 3 Vcm"1

1

5 x 1 0 T o r r

I ' I ' I « I ► I ► I ' I » I » I ► É0Ã 192.9 104.9 106.9 108.9 119.9 112.9 114.9 116.9 118.9 129.9 ìå

E. = 1 0 3 Vcm 1

1 — 6

5x 10 Torr

10

M^W - I | l | I | I I i | i — , i | i — j i i i | -

Ô0Ã 192.0 104.0 106.0 108.0 110.0 112.0 114.0 116.0 118.0 120.0 ps

Fig. 15. Fragment ion spectra of Substance P (first segment with reflector potentials set for parent molecule recording) under different experimental conditions. See text for a further explanation.


311

H i

co \

V 1 7 8?"17

t I i J I I

ag-17

\ \ î° l ,

/Vv

b2-17

X4

l|k IWn

b3-17

WWWWi C4H8N4 aj-17

C3H6N3

d/i* VwJw L,AJ ivUP

A

\JUV\ Fig. 16. CAD (residual gas) fragment ion spectrum of Substance P obtained at low extraction field strength (5 x 102 Vcm l) but

increased residual gas pressure (5 x 10~6 Torr) of the RE-TOF vacuum. See text for a further explanation.

382 R. Kaufmann et al.lint. J. Mass Spectrom. Ion Processes 131 (1994) 355-385

with recording a usual PSD fragment ion spectrum at high initial field strength (3.3 x 104 Vcm-1) and low residual gas pressure (5 x 10~7 Torr). The seg-ment spectrum shows the typical cleavage pattern characteristic for fragmentation under unimolecular PSD conditions (see above). Reducing the initial field strength to 103Vcm~l at the same back-ground pressure led to a 90% suppression of the

fragment ion formation. In a third step background pressure was raised to 5 x 1CT6 Torr keeping the initial field strength low. Under these conditions fragmentation occurs again but with a different fragmentation pattern. While the a„ - 17 satellites are largely reduced, the fragment ion spectrum is now merely dominated by a„ and some dn and b„ cleavages (see, for comparison, the full fragment

a5-17

.*5S2.21 (a 5 -17)+ l

*%3.<7

5

573,77

a 5 +l

«568.J3

I I I 1 1 1 1 1 FOF 113.9 114.9 115.9 116.9 117.9 118.9 113.9 129.9 [)S

® m.u

Fig. 17. Section of the fragment ion spectrum of Substance P at instrumental conditions forcing collisions with residual gas molecules ((A) postsource CAD at £/acc = 10 kV) and forcing in-source activation ((B) unimolecular PSD, high energy collisions with residual gas

molecules suppressed).


ion spectra shown in Figs. 16 and 6 respectively). These fragment ion spectra very much resemble classical high energy CAD spectra of Substance P as obtained in four-sector instruments [5,29].

Particularly noteworthy is the formation of άη

fragments which, from the viewpoint of its ener-getic requirements, is a rather demanding process. It involves a two-step reaction of first a homolytic backbone (C-C) cleavage (free radical) followed by a side chain a-cleavage [41] and, thus, is an impor-tant indicator for a high degree of internal activa-tion. In line with the experimental evidence given by Johnson et al. [41], in our CAD-TOF mass spectra of Substance P we nearly always saw a dis-tonic a„ + 1 product ion occurring in conjunction with the appearance of dn fragments (see Fig. 17).

Another peculiar difference between "in source" activated PSD and "post source" CAD spectra concerns small fragments (i.e. below 70 u). Not only is the relative yield of these products much larger in the latter than in the former case, but also the fragmentation patterns are distinctly different.

As shown in Fig. 18, the two fragment ion spec-tra share the immonium ion signals for proline (70 u), dissociated Na+ (23 u), and the common C3H6N3 (84 u) fragment moiety, but are otherwise rather incongruent. Particularly noteworthy in the CAD spectrum are the grouped signals at 39, 41, 43, 44 u and at 55, 56, 58, 59 u which, together with the prominent signal at 18 u (NH^), would be rather indicative of partial side chain cleavages.

These findings support a model of collisional acti-vation of macromolecules proposed recently by Uggerud and Derrick [42]. In their theory of impul-sive collision the efficiency by which collisional energy (in the center of mass frame) is transferred into internal energy is determined by the ratio of the mass of the collision partner and the mass of that atom (or atom group) within the macromolecule which is actually "hit". Efficiency of energy transfer becomes maximum if these two masses are about equal. This would indeed be the case under our conditions if we assume H20 or N2 being the most probable collision partner and say C, O or N (with their H satellites) being the entities hit during a

®

\jr www» v** 18

j#ft4^% SW V tyitorfW W r W*V

Fig. 18. Comparison of small {m/z < 100 u) fragment ions obtained from Substance P after either PSD (A) or CAD (B). Experimental conditions for PSD are as given in Fig. 6, for CAD as given in Fig. 16. See text for a further explanation.

384 R. Kaufmann et aL/Int. J. Mass Spectrom. Ion Processes 131 (1994) 355-385

collisional encounter. One of the consequences of this theory is the prediction that, with collisional energy increasing, such highly "localised" momen-tum transfer reactions will eventually induce immediate fragmentation at that particular loca-tion. In the case of a linear peptide this would mean that numerous small fragments should be formed by collisional "dissection'' of protruding side chain groups being at much larger risk of being hit than the core of the backbone chain.

If this picture is basically true, increasing the collisional energy in the classical (high energy) CAD experiment would not be a good strategy, since it would merely augment the yield of "unspecific" fragmentations. However, the case of multiple low energy collision to sum up internal energy by relatively small increments might be much more favourable. Here, by "soft" activa-tion, more energy can eventually be dumped into the internal energy pool without the risk of a hit-and-break event occurring. We assume that under the conditions of matrix-assisted laser desorption, and possibly also "field" excited ESI, these require-ments can be met. Besides an increase of the unim-olecular decay rate constants the efficiency of subsequent high energy collisional activation is probably increased by this mechanism.

What makes MALDI-RETOF mass spectrome-try rather unique in the frame of approaches to fragmentation of large polyatomic ions is the con-junction of three circumstances. These are: (1) an efficient "in-source" activation; (2) the ease of pro-viding for additional alternative high energy colli-sional activation during flight; (3) a large time window to accommodate even those subsequent fragmentations which occur at comparatively low rate constants.

Acknowledgements

Financial support by the Bennigsen-Foerder-Program of the Ministry of Science and Research (NRW, Germany) is gratefully acknowledged. Some of the peptides tested were provided by E. Jaeger (Max-Planck-Institute of Biochemistry,

Martinsried, Germany) and M. Mann (European Molecular Biology Laboratory, Heidelberg, Germany).

References

1 F.W. McLafferty (Ed.), Tandem Mass Spectrometry, Wiley, New York, 1983.

2 G.L. Gisle and R.G. Cooks, Chem. Eng. News, 59 (1983) 40.

3 R.A. Yost and CG. Enke, Anal. Chem. A, 51 (1979) 1251. 4 A.J. Alexander, P. Thibault, R.K. Boyd, J.M. Curtis and

K.L. Rinehart, Int. J. Mass Spectrom. Ion Processes, 98 (1990) 107.

5 M.F. Bean, S.A. Carr, G.C. Thorne, M.H. Reilly and S.J. Gaskell, Anal. Chem., 63 (1991) 1473.

6 L. Poulter and L.C.E. Taylor, Int. J. Mass Spectrom. Ion Processes, 91 (1989) 183.

7 C D . Bradley, J.M. Curtis, P.J. Derrick and B. Wright, Anal. Chem., 64(1992)2628.

8 R.B. Cole, S. LeMeillour and J.C Tabet, Anal. Chem., 64 (1992) 365.

9 R.G. Cooks, T. Ast and A. Mabud, Int. J. Mass Spectrom. Ion Processes, 100 (1990) 209.

10 D.L. Bunker and F.M. Wang, J. Am. Chem. Soc, 99 (1977) 7457.

11 E.W. Schlag and R.D. Levine, Chem. Phys. Lett., 163 (1989) 523.

12 L.L. Griffin and DJ. McAdoo, Proc. 40th ASMS Con-ference on Mass Spectrometry and Allied Topics, May 31-June 5, 1992, Washington, DC, p. 1831.

13 J. Adams, Mass Spectrom. Rev., 9 (1990) 141. 14 CJ. Barinaga, CG. Edmonds, H.R. Udseth and R.D.

Smith, Rapid Commun. Mass Spectrom., 3 (1989) 160. 15 G.J. van Berkel, G.L. Glish and S.A. McLuckey, Anal.

Chem., 62 (1990) 1284. 16 R.D. Smith, J.A. Loo, CJ. Barinaga, CG. Edmonds and

H.R. Udseth, J. Am. Soc. Mass Spectrom., 1 (1990) 53. 17 C.K. Meng, C.N. McEwen and B.S. Larsen, Rapid Com-

mun. Mass Spectrom., 4 (1990) 151. 18 S. Della-Negra and Y. Le Beyec, Anal. Chem., 57 (1985)

2035. 19 X. Tang, W. Ens, K.G. Standing and J.B. Westmore, Anal.

Chem., 60 (1988) 1791. 20 X. Tang, W. Ens, F. Mayer, K.G. Standing and J.B. West-

more, Rapid Commun. Mass Spectrom., 3 (1989) 443. 21 A. Brunelle, S. Della-Negra, J. Depauw, H. Joret and Y. Le

Beyec, Rapid Commun. Mass Spectrom., 5 (1991) 40. 22 U. Boesl, H.J. Neusser and R. Weinkauf, J. Phys. Chem.,

86(1992)4857. U. Boesl, R. Weinkauf and E.W. Schlag, Int. J. Mass Spec-trom. Ion Processes, 112 (1992) 121.

23 M. Karas and F. Hillenkamp, Anal. Chem., 60 (1988) 2299. 24 R.C. Beavis and B.T. Chait, Rapid Commun. Mass Spec-

trom., 3 (1989) 432. 25 B. Spengler, D. Kirsch and R. Kaufmann, J. Phys. Chem.,

96 (1992) 9678.


26 B. Spengler, D. Kirsch, R. Kaufmann and E. Jaeger, Rapid Commun. Mass Spectrom., 6 (1992) 105. B. Spengler, D. Kirsch and R. Kaufmann, Rapid Commun. Mass Spectrom., 5 (1991) 198.

27 R. Orlando, 40th ASMS Conference on Mass Spectrometry and Allied Topics, May 31-June 5,1992, Washington, DC, p. 1935.

28 J. Bordas-Nagy, D. Despeyroux and K.R. Jennings, J. Am. Soc. Mass Spectrom., 3 (1992) 502.

29 J.A. Hill, S.A. Martin, J.E. Biller and K. Biemann, Biomed. Environ. Mass Spectrom., 17 (1988) 147.

30 G.R. Kinsel, J. Lindner, J. Grotemeyer and E.W. Schlag, Proc. 39th ASMS Conference on Mass Spectrometry and Allied Topics, May 19-24, 1991, Nashville, TN, p. 350.

31 P. Roepstorff and J. Fohlman, Biomed. Mass Spectrom., 11 (1984)601. R.S. Johnson, S.A. Martin and K. Biemann, Int. J. Mass Spectrom. Ion Processes, 86 (1988) 137.

32 R. Orlando and R.K. Boyd, Org. Mass Spectrom., 27 (1992) 151.

33 L.M. Teesch, R.C. Orlando and J. Adams, J. Am. Chem. Soc, 113(1991)3668.

34 M. Karas, U. Bahr and U. Giessmann, Mass Spectrom. Rev., 10 (1991) 335.

35 R.E. Johnson and B.U.R. Sundqvist, Rapid Commun. Mass Spectrom., 5 (1991) 574.

36 B. Spengler and V. Bökelmann, Nucl. Instrum. Methods, (1993) in press.

37 R.C. Beavis and B.T. Chait, Chem. Phys. Lett., 181 (1991) 479.

38 T. Huth-Fehre and C.H. Becker, Rapid Commun. Mass Spectrom., 5 (1991) 378.

39 A. Vertes, G. Irinyi, L. Balazs and R. Gijbels, Proc. 39th ASMS Conference on Mass Spectrometry and Allied Topics, May 19-24, 1991, Nashville, TN, p. 927.

40 G.M. Neumann, M.M. Sheil and P.J. Derrick, Z. Natur-forsch., Teil A, 39 (1984) 584.

41 R.S. Johnson, S.A. Martin and K. Biemann, Int. J. Mass Spectrom. Ion Processes, 86 (1988) 137.

42 E. Uggerud and P.J. Derrick, J. Phys. Chem., 95 (1991) 1430.


Energy-isochronous time-of-flight mass analyzers

H. Wollnik 2. Physikalisches Institut, University Giessen, 35392 dessen, Heinrich-Buff-Ring 16, Germany

(Received 14 July 1993; accepted 4 October 1993)

Abstract

Techniques of time-of-flight mass analyzers are discussed for systems of reflector-type geometry as well as for systems that employ sector fields. Ion-optical calculations of particle trajectories are shown as well as some technical details of the recording of ion bunches.

Key words: Energy-isochronous time of flight; Ion optical calculations; Reflector-type geometries

1. Introduction

Classical magnetic sector-field mass analyzers separate ions of different momentum-to-charge ratio p/q oc \/Km/q laterally in their focal planes. Thus, ions of different mass-to-charge ratios (m/q) are separated from each other if they had all been accelerated by the same potential, i.e. that their energy-to-charge ratios K/q are all equal. For ions that were not all formed at the same poten-tial, the energy-to-charge ratios K/q vary over some usually small range which results in a beam broadening and thus in a reduction of mass resolving power. This can be avoided, however, by adding to the system an electrostatic sector field [1,2] that deflects ions of different energy-to-charge ratios differently, where this deflection is independent of the ion masses (see Fig. 1).

Such an energy-achromatic mass analyzer can be used as a mass spectrograph [1,3] in which the ions of different mass-to-charge ratios are recorded simultaneously by a photographic plate or by some electronic equivalent [4]

The same system can be used also as a mass spectrometer in which, at any one moment, only

those ions are recorded whose mass-to-charge ratio allows them to pass through a specific exit slit, while all ions that arrive laterally displaced at the focal plane are lost. In this spectrometer mode the fields in the system are varied with time such that ions of different mass-to-charge ratios are recorded at different times and that the intensity of the beam changing with time finally reveals the mass spectrum. Though the achieved mass resolving powers R = m/Am in the two modes of operation are quite similar, one records close to 100% and perhaps 20/R% of all ions in the two modes, respec-tively. For R = 2000, mass spectrometers scanning in this manner are about 10000 times less sensitive than mass spectrographs. The same reduced sensi-tivity is found in other scanning mass spectrometers such as quadrupole mass filters [5,6] or in some operating modes of quadrupole ion traps [7,8].

In time-of-flight (TOF) mass analyzers all initially formed ions are finally recorded as in a mass spectrograph. However, here the ions of different masses arrive after each other as in a classical mass spectrometer. The difference is that they arrive quite rapidly after each other so that the whole mass spectrum can and must be recorded in

SSDI0168-1176(93)03888-S

388 H. Wollnikjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407

(K/qyi+äê)

(m/q^l-tfj

(m/q)b

Fig. 1. A double focusing, i.e. angle- and energy-focusing sector field mass analyzer is shown. In this system ions of two energy-to-charge ratios (K/q) = (K/q)0(\ ±δê) and three mass-to-charge ratios (m/q)0 and (m/q) = (m/q)0(l ±6m) are shown with all (m/q)0 ions passing through the same exit slit, while ions of different mass-to-charge ratios are eliminated. Note also that the width of the beam of the one "ion species" should be larger than the slit opening in order to finally obtain the intensity distribution in this beam when the beam is scanned across the exit slit. Usually, one chooses this slit width to be a factor of 2 or 3 smaller than the FWHM of the ion beam.

a very short time, perhaps in 100 ßs if all ions were started at the same time1. This allows the moni-toring of relatively rapid variations in the amount of sample material as would be the case in a GC-TOF combination in which a TOF system continuously monitors the mass distribution in the effluent of a gas Chromatograph or of a capil-lary zone electrophoresis instrument (see Fig. 10). For all such investigations it is also quite useful that a TOF mass analyzer has no limit as to how heavy an ion can be to be analyzed. Apart from problems in forming a heavy ion and in recording it, the performance of a TOF mass analyzer usually improves with increased ion masses and thus with increased flight times.

2. Principles of energy-isochronous TOF mass analyzers

TOF mass spectrometers separate ions of differ-ent masses by their different flight times. To deter-mine these flight times it is necessary to know the ion flight distance as well as the ion velocity. The

velocity v = \v\ of relativistically slow ions of the ion mass m in daltons (u) and the ion energy K in electronvolts is

V m y m

Using the correct conversions one finds the ion velocity v in millimeters per microsecond [8]:

v « 13.891332 J-= 13.891332WK-^ \ m V m (1)

1 Alternatively, all ions of mass m and charge q must enter in times proportional to yjmfq (see section 4.2.1.).

if V is the potential by which ^-times charged ions are accelerated. Thus, in a field-free region of length L, the flight time T = L/v depends on the ion's energy K = qV as well as on the ion's mass m. To make the ion's flight time independent of the ion's energy one must send the more energetic, and thus faster, ions on a properly dimensioned detour as com-pared to reference ions of mass m0 and energy K0.

One way to achieve such a custom-tailored detour is to let the ions under investigation move in a homogeneous magnetic field. In such a field of flux density B =\B\ all particles of mass-to-charge ratio {m/q) and energy-to-charge (K/q) move in circles of radii along which the centrifugal force mv2/p is balanced by the centripetal force qv x B.

H. Wollnikllnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407 389

E=0 ion source

J— Lo ^H

E=E0

ion detector

Fig. 2. An energy-isochronous TOF mass analyzer is shown that includes an electrostatic repeller field E = E0 into which more energetic and thus faster ions penetrate deeper. If properly dimensioned, this flight path elongation compensates for an increased ion velocity such that the overall flight time is inde-pendent of energy deviations [12] to first order. Note the posi-tions of two groups of ions of equal masses but different energy-to-charge ratios (K/q)x and (K/q)2 shown for different times. Note also that the ions must pass through the meshes of a grid that defines the extent of the repeller field.

For relativistically slow ions the flight path per turn thus has a length of 2ðñ with p in millimeters

ñ * 0 . 1 4 3 9 7 1 1 ^ = 0 . 1 4 3 9 7 1 1 ^ â ( 2) qB B

where K is given in electronvolts, V in volts, B in teslas and m in atomic mass units. Since both p and v in Eqs. (1) and (2) are proportional to y/K = y/qV the overall flight time T of a particle of given m/q is calculated for one turn in microseconds as

T = 2πρ_ 0.0651195 m

v B a (3)

Thus, for ions that move in a homogeneous mag-netic field the flight times for a full turn or for a specified fraction thereof are proportional to the mass-to-charge ratios m/q and inversely propor-tional to the magnetic flux density B, but they are independent of the ions' energies K. This fact was recognized early on and has led to the development

of the Cyclotron accelerator [9] but also to the ion cyclotron resonance mass spectrometer [10].

For slightly different mass-to-charge ratios m/q = {mo/qo) + A{m/q) = (m0/q0){\ + 6m) the ion flight times are T+AT=T{l+6t) so that the time resolving power T/AT = \/6t and the mass-to-charge resolving power (mQ/qQ)/A(m/q) = \/bm

are equal:

1 = T = 1 = m0/q0

6t~ AT 6m~A(m/q) (4)

Note here that ion mass differences of N per cent result in flight time differences of N/q per cent. Highly charged ions thus become increasingly difficult to separate.

The postulated existence of a homogeneous magnetic field requires high costs and limits the choice of ion sources and ion detectors severely. Thus, alternative systems are very desirable. Since, in most cases, the energies K of the ions vary only slightly around the energy K0 of a reference ion, we may define

(5)

with 6K < 1 and postulate only that the ion flight path is designed such that the overall flight time T is independent of 6K, but that it can still depend on KQ9 a fact which allows a larger variety of systems than the homogeneous magnetic field. This includes systems that are comprised of homo-geneous and inhomogeneous magnetic and electro-static sector fields separated by field-free regions but amended also by magnetic and electrostatic quadrupoles and rotationally symmetric lenses.

2.7. TOF mass analyzers that use an electrostatic repeller field

In Fig. 2 an idealized TOF mass analyzer is shown in which an ion beam is sent into a homo-geneous electrostatic repeller field [11] in which the detour of more energetic ions is achieved by their deeper penetration into the field. A q-times charged ion of mass m, and velocity v = ivx +jvz penetrates


into the repeller field E = jEz to a depth z = K/(qE) = V/E, where V=K/q is the potential by which the ion under investigation had been accelerated in the z-direction in the ion source and acceleration region.

Denoting the distance between the ion source and the entrance to the repeller field (see Fig. 2) by L0 one finds the total ion flight time to be twice the flight time Tt from the ion source to the apex of the individual ion trajectory in the repeller field. This total flight time equals 2Tt with

v7

J__U 2V_ vz/2 vz Evz

(6)

where the first term describes the flight time in the field-free region in which the ion moves in the z-direction with the velocity vz, and the second term describes the flight time in the repeller field in which the velocity component in the z-direction is on average vz/2. For K— qV this vz is found from Eqs. (1) and (5) as

v7 « 13.891332

with vZo = 13.891332v/K0/(m/?). Thus, Eq. (6)

can be written as

Tt = T0 + (T\6)6K + (T\66)6i +

with

(7)

T0 =

(T\S) =

(Τ\δδ) =

IL· m/q 0 E J2v2o ' 13.891332 "V Vt

2V° æË 1

3 L n -2ΫΛ 1

8u zo

Here, the right-hand terms are obtained for L0 = 2V0/Em which case (T\6) vanishes [11,12]. Thus, the ion flight times T0 reveal the ion mass distri-bution quite accurately even if the ion energies and velocities are only approximately equal, i.e. δê ^ 0.

Note here that the condition (Τ\δ) = 0 prevails also for ions of different charges q.

For slightly different mass-to-charge ratios (»V?o) + A(/w/tf) = (m0/?o)(l + àm), the ion flight times are T0 + AT= T0{\ + δ,) oc [(m0/q0) (1 +<5W)]1//2 so that in such TOF mass analyzers the mass-to-charge resolving power is half as large as the time resolving power

1 T = 1 m0/qQ

2δ( 2 Ä Ú 6m A{m/q) (8)

Note also that ion mass differences of N per cent result in flight time differences of N/(2q) per cent. Highly charged ions thus become increasingly difficult to separate.

2.2. Energy-isochronous TOF mass analyzers that use sector fields

In deflecting fields the flight path of more ener-getic ions can also be increased relative to the flight path of ions of lower energies. In the case of homo-geneous magnetic fields it was shown above that the ion flight times per turn are completely inde-pendent of the ion velocities. However, besides technological problems one should note that a homogeneous magnetic field has focusing proper-ties only in the plane of deflection so that the ion path is a helix. With each turn the ions thus move more and more away from the magnet mid-plane.

If one limits oneself to ions that have at least approximately equal energy-to-charge ratios K/q — (K0/q0)(l + δê) with ( 5 ^ < 1 , one can build high performance TOF mass analyzers [13-15] in which the ion flight time depends only on 62

K, 63K...

but is independent of the linear term in 6K. In such systems, however, one can achieve lateral focusing not only in the plane of deflection but also in the perpendicular plane [2] by using one or more magnetic or electrostatic sector fields. However, though such TOF mass analyzers have been built with a single sector field, the real value of such systems becomes apparent only if several such sector fields are combined.

H. Wollnikllnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407 391

3. TOF mass analyzers for energetic ions

Ions of different energies can be mass analyzed by laterally dispersive energy- and angle-focusing systems [1,3], so called double-focusing mass spectro-graphs, or by systems that achieve energy and mass dispersions in different directions [16-18]. For ener-getic ions, in both cases strong electrostatic fields are required that can only be reached by sophisticated technology. A fully equivalent mass analysis for ions of different energies can be achieved, however, longitudinally by energy-isochronous TOF mass analyzers built from magnetic fields only [2,13-15,19,20]. Since the flight times of individual ener-getic ions are optimally measured by determining the times when such ions pass through start and stop detectors, one needs fast electronic circuitry and small detectors which in turn postulate that the ions must laterally stay close together.

The first TOF mass analyzer of this type [19,20] has been built for the mass determination of nuclear fragmentation products. This TOF system (see Fig. 3) consists of four identical sector magnets

Start Stop-Detector

Fig. 3. An energy-isochronous TOF mass analyzer for energetic ions is shown consisting of four identical sector magnets and an overall flight path of 14 m length [19,20]. Trajectories are indi-cated for initially divergent ions of equal mass-to-charge ratios and two energy-to-charge ratios. Note the particle positions after different flight times.

of deflection angles of φΒ = 8Ã each that deflect the reference ions along radii of p0 — 1 · 1 m a nd ensure that the ion flight time is independent of the relative energy deviation 6K to first order. Because of its four entrance and four exit field boundaries, which are all inclined by 23.3°, the fringing fields help to stigmatically focus all ions that leave a point of the start detector divergingly to a corresponding point on the stop detector inde-pendent of 6K to first order, i.e. achromatically. Thus, the stop detector can be as small as the start detector, i.e. about ±10 mm. In the TOF mass analyzer as described in Refs. 19 and 20 the ions move through the 14 m long system in about 500 ns with calculated flight time deviations of < 0.01 ns full width at half maximum (FWHM). Since the particle detectors, however, have timing errors of 0.1ns, only mass resolving powers of about (m0/q0)/A(m/q) « 3000 can be reached. The accuracy of the determination of the m0/q0

value, however, is much better than 1/3000 since this accuracy is the precision with which the center of mass of the mass peak in question is determined. This accuracy is proportional to (m/Am)y/N where N is the number of ions that contribute to the mass peak in question.

It should be noted that ions formed in fragmen-tation reactions are always multiply charged so that there are always several flight times for ions of a given mass (see Fig. 4). Higher mass resolving powers can be reached only if the ion flight path is elongated. This can be achieved by a larger size of the TOF mass analyzer or by building the system as a ring [21] through which the ions must move repeatedly (see section 5).

Though it seems to be the method of choice to build a TOF mass analyzer that is energy-isochronous, one can also measure the ion flight time Toe y/K/m as well as the ion energy K of each individual ion. In this fashion one can deter-mine exact ion masses even if the ions have different energies. This has been done for instance in Ref. 22 for an approximately 100 m long flight path and a large magnetic sector field analyzer that deter-mined the ion energy to about ±0.01%.

392 H. WollnikjInt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407

103 I

10z l·

g 101

c

694 744 794 844 894

Time of Flight

Fig. 4. A mass-to-charge spectrum obtained by the TOF mass analyzer of Fig. 3 is shown, in which the obtained ion intensity is plotted as a function of the ion flight time for the investigated proton-induced Th-fragments which have energies of « 1 MeVu-1 and usually spread over about five charge states. Note that there are many ions that have the same m/q for instance m/q = 24/8 = 27/9 = 30/10 = 33/11 = . . . but that there are gaps in the neighbourhood of m/q values 2.5, 3.0 etc since q can take up integer values only. To distinguish ions of the same m/q value one can determine additionally, though mostly only approximately, the flight time of each ion in a field-free region and thus the true ion velocity. The ion's charge q is then found from a comparison of this flight time and the one in the TOF mass

analyzer.

4. TOF mass analyzers for low-energy ions

In the case of low-energy ions, one can build energy-isochronous systems for which in Eq. (6) the term (T\6) vanishes [2,11,12] by systems of magnetic or electrostatic sector fields or by electro-static repeller fields. If the repeller field region of a system such as the one shown in Fig. 2 is split by an intermediate grid into two regions of different field strengths [23-25] one reduces the ion transmission slightly, but by proper choice of the two fields one can even achieve (T\6) = (Τ\δδ) = 0, so that the first non-vanishing term is proportional to δ\. Unfortunately, the ion losses by grids are not only geometrical but, since the equipotential surfaces in the neighbourhood of a grid wire are not flat, some of the ions are bent away from their initial trajectory quite considerably. This effect is especially large if the grid separates

[26] regions of differing field strengths (see Fig. 5).

To avoid this deterioration in the optical beam properties one can use grid-free ion repellers or mirrors [26]. To predict the ion-optical properties of such ion mirrors is mathematically difficult; however, to build the systems is rather simple. A feasible TOF system that incorporates one grid-free ion mirror is shown in Fig. 6. Such grid-free ion mirrors feature a more or less homogeneous field in the region in which the ions turn around, and exhibit strong lens actions in the region in which the ions enter and leave the repeller field. Since an axis of rotational symmetry exists here, the beam must be adjusted more precisely than in gridded mirror systems, which requires beam deflection plates and at least one rotationally-symmetric lens as shown in Fig. 6.

The mass resolving power of any such TOF mass

H. Wollnikjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407 393

II*

TF

| f»

|"P~

| (·

I I

1 Ã f»

1 1 1 1

100V 8C >V 61 )V 4 1 ov \

/ · \

'""Ã·

~T ~ ' — il ( · J '~ — ) ~~

1 (* .,

~~V~ ~~

Λ 1 T \ W ov v 2 0 V \

90V 70V 50V 30V 10V

Fig. 5. An initially parallel ion beam is shown passing through an idealized grid of infinitely thin wires that separate regions of different field strengths. As one can see, the ion trajectories stay widely undisturbed if they pass close to the middle of a grid mesh. Ion trajectories that pass close to a grid wire, however, can be bent noticeably. Thus, the optical properties of an ion bundle can deteriorate considerably,

i.e. the phase-space area of the ion beam is increased as is the beam diameter downstream from a wire grid.

analyzer is limited by the ratio of the total ion flight time and the length of the ion pulse or the jitter of the start and stop pulses. The smallest principal jitter can be expected if the ion to be investigated

creates its own start and stop signals [27] by producing secondary electrons when passing through a very thin foil and then impinging on a stop detector (see Fig. 7). The start pulse also can

ion source einzel lens

grid-free ion reflector

Fig. 6. A TOF mass analyzer is shown including a pulsed ion source and particle detector as well as a grid-free ion mirror and the necessary adjustment devices, i.e. a beam deflector and an adjustment lens. Note that the shown ion bunches are short at both the front end and the back end as well as close to their apices in the middle of the repeller field, but that they are long in between these positions.


symmetry-axis

electric sector 2

Fig. 7. An energy-isochronous TOF mass spectrometer for solar wind investigations. The ion under consideration passes through a thin foil on which it creates secondary electrons that serve as the start signal and then flies energy-isochronously to a stop detector, where the energy isochronicity is achieved using electrostatic sector fields. Note that in this special system, [27] poloidal electrostatic sector fields are used that are rotationally symmetric to the axis of rotation shown. Thus, all ions that are initially perpendicular to the axis of

rotation shown are mass analyzed, where the azimuthal arrival position characterizes the initial azimuthal angle of incidence.

be obtained from the formation process of an ion to be investigated [28-30]. Similarly good results can be expected if the ion formation process is intimately linked to the ion start pulse [31].

In almost all cases the performance of a TOF mass analyzer improves with an increased overall flight time. A long total flight time, however, requires a large system or one that uses the existing path length several times (see also section 5) but it still requires that either the ions are produced in a very short time or that they are bunched together by static or by dynamic field arrangements [12,26,29,32,33].

4.1. Pulsed ion production

In a pulsed ion production the pulse length ATis mass independent but the flight time T is propor-tional to y/m/q (see Eqs. (1) and (7)) if all ions are accelerated by the same potential difference. Thus, the mass resolving power improves with increasing mass-to-charge ratio. A TOF mass analyzer that achieves (m0/q0)/A(m/q) « 1000 for ions around

m0/q0 = 100 should thus obtain (mQ/q0)/A(m/q) « 30000 for ions around m0/q0 = 100000. Pulsed ion production is thus most useful and desirable for the investigation of large molecular ions [34].

One way to form ions in a short time is to shoot a short pulse of laser light into a relatively dense puff of gas or into a gas jet [35], thus forming ions by multi-photon excitation. This allows the investi-gation of the ions of interest with good mass resolv-ing power even though relatively large quantities of material are required.

Bombarding a solid or liquid surface by individual energetic fission products, i.e. plasma desorption mass spectrometry [28], short bunches of ions, i.e. secondary ion mass spectrometry [29], or short bursts of laser light [30] can form one or several ions of heavy organic molecules in each shot. Although the ionization process in all three cases is not well understood, one can say that energy is coupled into the solid or liquid matrix in which the molecules of interest are individually embedded and that these pre-ionized molecules are simply

H. Wollnik/Int. J. Mass Spectrom. Ion Processes 131 (1994) 387-407 395

lifted from the matrix and pulled away by an accelerating field.

Of the three mentioned methods the pulsed laser technique is today the most effective and easiest to handle. This method has become efficient, however, only after the matrix was modi-fied to effectively absorb the laser light [36,37]. At this point one should also note that all three desorption methods [28-30] make extremely effi-cient use of the sample material. In special cases the unused material can even be recouped for additional analyses.

The described methods and, here especially, matrix assisted laser desorption ionization (MALDI), see Refs. 37 and 38, are capable of ionizing and accelerating ions larger than 300 000 u. These molecular ions have also been mass analyzed in a TOF mass spectrometer, however, so far the mass-to-charge resolving powers achieved have been limited. The reason is that large molecular ions move slowly, so that the impact of such ions on a metal surface does not produce secondary electrons that could quickly be accelerated. What is recorded are probably only secondary electrons formed by post-acceler-ated fragment ions that, because of their smaller velocities, arrive over a longer time.

Since the peak widths due to the impaired ion recording are large, there was no need to use very sophisticated TOF systems. Thus, up to now, mostly straight non-isochronous TOF mass analy-zers have been used for the MALDI procedure [37,38] or the method of Ref. 28, that do not employ repeller fields.

4.2. Quasi-continuous ion production

To produce ions efficiently it is often advan-tageous to produce them continuously, store the

2 A very flexible way to build a corresponding system is to store ions in an ion trap, possibly tuned to be non-mass-selective, extract them in a pulsed mode and mass analyze them in a TOF mass analyzer [39]. If this ion trap stores only ions of a limited mass range the system has also MS-M S capabilities.

grid 2

L;

grid 1

gridO

detector

f C ^ R ^ 1 % s 1 ^\. x

A \ z \

1 \ \

\

I

Vz! f __ \

vx \

\ Ë

v=v2 X r

\ v=v,

\ v=o

v=o Fig. 8. Sketch of an ion buncher in which the x-component of the ion velocity vx is undisturbed while the z-component vz is changed to v2 by a pulsed field ( V2 - V\ )/L2. The magnitude of this field must be chosen such that the ion flight times to the ion detector — a distance L0 downstream — become independent of the initial z-position of the ions under investigation. If an ion that starts from point A is not at rest initially but moves towards point B, it arrives at the final ion detector delayed. This delay time is equal to the time it needed to reach point B where the z-component of the ion's velocity vanishes. If an ion that starts from point A seems to come from point C, it arrives early at the final ion detector with the difference in arrival times being the time the ion would have needed to move from point C to point A.

formed ions for some time and extract them during short intervals only2. Assume that after some time of continuous ion production an ion cloud has been formed between the grids of Fig. 8 and that all these ions are at rest if V\ equals V2. A sudden change of V2 then establishes an electric field E2 = (V2 — V\)/L2 over the ion cloud that expels these ions towards the detector plane in Fig. 8. Ions that are initially close to grid 2 in this set up are accelerated to higher velocities than ions that are initially close to grid 1. However, the latter are already closer to the detector plane. Thus, both groups of ions can arrive at the detector plane simultaneously if all potentials and distances have been chosen properly [26].


The overall ion motion in the system of Fig. 8 is very similar to the ion motion in the second part of the system of Fig. 2, i.e. from the apex of the parabolas to the final ion detector. Thus, the resulting flight time T can be calculated from Eq.(6)if^ = VxjLx is equal to E2 = (V2- Vx)/L2

with

F„ = i ^ p } fc = ( r 2 - K , ) £ (9)

For the more general case Ε2φ Ex a similar rela-tion can be derived

Τ=Τ0 + (Τ\δ)δê+(Τ\δδ)δ2ê

+ {Τ\δδδ)δί+--·

T0vZo=L0+ —

1 +

2(r|*K = -L0 + ^

ß-ÏÏ-ÁÃ

•*'f,-')('-v. ÊË-é/2

(10)

S(T\6S)vIO = 3L0 + ^

,+º-» Ê F ] \ -3 /2 -

\6{T\66S)vZo = -5L0 + 2V±

, + , l - · 0-3 ÊË-5/2"

with vZo « 13.891332y/V0/(m/q). For the most interesting case {Τ\δ) — 0 the other three coefficients

simplify to

7>z0 = Lo[2

{Τ\δδ)νÆϋ=ΐψ

(Τ\δδδ)νÆο = .8

- » -

\VQ-Vj 4 V V^Vy

16 V0(2V-Vl) L0

Equation (10) was derived under the assumption that in the initial cloud the ions were at rest as far as the motion in the z-direction is concerned, though vX9 the x-component of the ion velocity ϋ9

can have arbitrary values. If the ions have an initial z-component of the

ion velocity, there is an additional term to the flight time of Eq. (10). An ion that starts at point A, i.e. at zA in Fig. 8 and moves towards B, arrives at zB, after a time AT = (zB - zA)/(vz/2) and starts to move from B towards the detector plane as would any other ion that had been at rest at point B initially. The potential difference AV between A and B is Ä V = (zB - zA)E2 found from the initial vz « \3.S9\332\/AV(q/m) in milli-meters per microsecond and the initial energy Kz = ^ÄÊÀç electronvolts. Thus, one finds ÄÃÀç micro-seconds as:

Ä Ã = 2ÄÊ 2VAV(m/q) 2^fWz~m v7E>) 13.89^2

Vz{™/q) 96.48£7

13.89^2

(H)

Analogously one can consider an ion that starts from point A with a velocity vz towards the detector. This ion behaves like one that has started from point C in Fig. 8 a time AT earlier.

For an ion of initial velocity vz or an equivalent energy qA V the overall flight time to the detector in Fig. 8 is thus T± AT as determined from Eqs. (10) and (11). For a typical initial energy of K=qAVz*±0.25eV and E2™25VmnTl one thus finds from Eq. (11) that Af&±y/rn/ (13.89133 x 25#)//s,i.e. « ±28.8/#nsfora#-times ion of lOOu. Note here that for constant

H. Wollnik/Int. /. Mass Spectrom. Ion Processes 131 (1994) 387-407 397

qAVz one finds Ä Ã á y/m/(qE2) which together with fcx^/m/q (see Eq. (10)) yields f / Ä Ã á E2y/q. Note further that, different from the case of a pulsed ion production of section 4.1, here the TOF resolving power f/ATand thus (see Eq. (7)) the mass resolving power m/Am = T/(2AT) is the same for ions of different masses but improves with increasing field strength E2. Since there is always some electronic jitter and pulse broadening, there is a lower limit for ÄÃ and consequently an upper limit for the mass resolving power. This is mainly important for the low mass ions for which ÄÃ is smaller than for the high mass ions.

4.2.1. The storage ion source. One way to make use of the considerations of section 4.2 is to build a storage ion source [26,32,33] as shown in Fig. 8. In this case one chooses V2 Φ Vx only for a short

period while for most of the time V2 = V{. During this time an electron beam ionizes all gas atoms in this region and at the same time keeps all ions in the potential well formed by the space charge of the electron beam. During the period when V2 differs from V\9 the ions are extracted and bunched at the detector plane by proper choice of E2 in Fig. 8 (see Eq. (10)). Thus, ions may be formed for « 100/xs, extracted during « 1 /is and bunched to form an ion pulse of « 10-50 ns duration at the detector plane.

Of these stored ions often only the heavy mole-cule ions are of interest although the light ions are usually more abundant. Thus, one can try to elimi-nate the majority of the light ions by applying a high frequency field over the ion cloud that removes the quickly-accelerated light ions but leaves the heavy ions untouched or swings them a

DC ion source

Fig. 9. Orthogonal TOF spectrometer in which the ion beam initially moves perpendicularly to the electric field of the ion buncher. Note that the lens reduces the angles of inclination ±a0 of the initial ion trajectories to ±a\ and thus reduces the z-component vz of the initial ion velocity to vz(a0/a\) = vz. Note further that the final ion trajectories are not perpendicular to the incoming ion beam even if the z-component vz of the ion velocity has been enlarged considerably to vz by the electrostatic-field between grid 1 and grid 2. Note here also that for E = 0 the ions move obliquely to the grids 0,1 and 2 as in Ref. 42 while for a properly chosen E φ 0 as in Ref. 41 they can be bent electrostatically to be more or less perpendicular to the ion detector, in case it must be taken that the fringing fields of this electrostatic

deflector do not distort the flight times noticeably.

398 H. Wollnik\Int. J. Mass Spectrom. Ion Processes 131 (1994) 387-407

little back and forth [40]. Since, in the extracted ion beam the light ions that are formed during one half period of the applied high frequency are usually still predominant, one often deflects those to the side by a pulsed orthogonal field in order not to overload the ion detector.

The position where the ion bunching occurs can best be achieved at an arbitrary distance from the ion source by changing the extraction fields appro-priately (see Eq. (10)). However, this position can also be considered to be a new ion-source from which all ions start within a time of i A T a s calcu-lated from Eq. (11). The advantage of such an arrangement is that a TOF system as shown in Fig. 8 can be followed by a second TOF system as shown in Fig. 2 or Fig. 6.

In any case, the second TOF mass analyzer must handle an ion beam with a relatively large energy spread

AK=±qE2zmax^q(V2-Vl) (12)

with zmax < L2/2. This energy spread — that ideally should be small — increases with E2 while the flight time spread of Eq. (12), which also should be small, decreases with E2. Thus, compromises are unavoidable.

4.2.2 The "orthogonal" TOF mass analyzer. Another way to make use of the above considerations is to build the TOF mass analyzing system as shown in Fig. 9 and make use of the fact that vx can also be non-zero. Such an arrangement [41,42] allows the use of an external ion source that forms a continuous ion beam and sends it in the x-direction into the region between grids 1 and 2 shown in Fig. 9. This approach has the advantage that an otherwise tested, efficient and good d.c. ion source can be used that has been designed with no compromises with regard to a bunched operation.

In this case the angle of divergence a0 « ± vz/vx of the incoming beam determines the maximal ±vz for a given vx and thus the maximal Ä Ã accord-ing to Eq. (11). This angle a0 and consequently the maximal vz and AT can usually be changed by placing a lens of focal length F between the real

ion source and the bunching device of Fig. 9. Advantageously, this is done by forming a pupil or an image in the middle between grids 1 and 2 as indicated in Fig. 9. For an initial ion beam of diameter 2x0 and angle 2a0 one can thus (see for instance Ref. 2) form an image of size 2x{ = 2Mx0 with 2a! = 2a0/M or a pupil of size 2xx = 2Fa0 with 2ax = 2x0/F. Choosing M or F appropriately one can achieve a < a0. Obviously, this requires the image or pupil (of size 2xx) to be larger than the original beam diameter (2x0). If this size should exceed the space between grids 1 and 2, parts of the beam intensity must be sacrificed for a

Intensity Intensity

Fig. 10. TOF mass analysis of the GC peak of acetone [44]. Shown is the GC peak that has a base width of about 1 s. Note also that the mass spectra obtained at 100 ms and at 300 ms are almost identical. The only difference is that in the mass spectrum after 300 ms the signal-to-noise ratio is consid-erably better than in the mass spectrum after 100 ms. Summing all TOF mass spectra over the duration of the GC peak reveals, furthermore, few thousand times as sensitive a recording of the overall mass spectrum. This situation is drastically different from the single mass spectrum one would have obtained by a slow scanning quadrupole mass spectrometer in which the intensity of the mass spectrum is folded with the GC peak height.


reduced AT and thus an increased mass resolving power. As long as the beam diameter stays smaller than the space between grids 1 and 2, as indi-cated in Fig. 9, the full beam intensity is preserved. However, at least this is a way to trade intensity for mass resolving power.

Detrimental in the design of this orthogonal TOF mass analyzer is the fact that the flight time f « L0/vz of the ions measured between the time of the extraction, i.e. the time when the bunching pulse is applied to grid 2 and/or grid 1 in Fig. 9, and the ion arrival at the ion detector is usually larger than the time f « Lx/vx necessary to fill the ion source since, except for very heavy ions, L0/vz < Lx/vx. Thus, for some period ions will move in the x-direction beyond the grid arrangement shown in Fig. 9 and are consequently lost to any investigation.

4.23. TOF mass analyzers for the investigation of fast changing mass distributions. For some mass spectrometric investigations the sample material is only available for a limited time, a time that may be barely sufficient for the recording of a full mass spectrum in a scanning mass spectrometer. This situation exists, for instance, if the effluent of a gas Chromatograph must be analyzed [43,44]. If some portion of these molecules are ionized and mass analyzed in a scanning mass spectrometer, for instance a quadrupole mass filter [5,6], one decade of the obtained mass spectrum is usually recorded in a reasonable fraction of a second, while the GC peak has a width of about 1 s. Thus, it is obvious that at best the low-mass ions are recorded when the GC peak is at perhaps 20% of the maximum intensity and the high mass ions are recorded when the GC peak is at its maximum. Consequently, the intensity distribution of the mass spectrum becomes skewed.

In a TOF mass spectrometer, however, in which #-times ions of up to perhaps (q x 1200) mass units fly for 1 m accelerated by perhaps 1000 V, one finds from Eq. (1) an ion velocity of « 13mm/is~l and thus a flight time of « 80 //s. Over the short time of 80 /xs, however, the intensity in a typical 1 s GC peak (see Fig. 10) is practically constant. This

would even be the case if the GC peak was reduced to 0.1 s or less [43].

In addition to their short scanning time, TOF mass analyzers record all ions that are formed in the ion source if a storage ion source is used as described in sections 4.2.1 and 4.2.2. This should be compared to the performance of a classical scan-ning mass spectrometer that records only perhaps 1/3000 of these ions, as was outlined in section 1. Thus, a TOF mass analyzer should be very well suited for obtaining mass spectra of GC effluents. This is most important if the mass spectrum of the substance of a GC peak is unknown, since, in a slow scanning mass spectrometer, the timing could be such that the most intense mass peak is recorded when the GC peak intensity is low so that it could still go by undetected.

As an example we used a TOF mass analyzer as shown in Fig. 6 to mass analyze the effluent of a fast gas Chromatograph that used a 50 μτη capillary of 2 m length. Such a gas Chromatograph is known [43] to produce GC peaks of about or less than 100 ms width. Though a TOF mass analyzer can easily record many mass spectra in 100 ms it requires some effort [45] to add a large number of them and store the results with modest elec-tronic effort every 30 ms. From these data one then can reconstruct the fast GC chromatogram (see Fig. 11) and add the mass spectra for each GC peak separately in order to obtain the mass spec-trum for each GC peak.

Another case in which a fast scanning TOF mass analyzer equipped with a storage ion source is used advantageously is the study of fast surface reac-tions. In such investigations a puff of gas can be brought to a surface at which the molecules of this gas may undergo some surface reaction. To study this reaction one must obtain a full mass spectrum in a time short compared to the molecule reaction time. Additionally, such an investigation must be very sensitive since the number of molecules in the gas puff must be so small that only a small fraction of a molecule monolayer develops on the surface in question. Otherwise, the decomposition of the mol-ecules on the surface would not be analyzed, but

400 H. Wollnikllnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407

1 n-hexane 2 cyclohexane 3 n-heptane 4 methylcyclohexane 5 toluene 6 n-octane 7 chlorobenzene 8 ethylbenzene 9 o-xylene 10n-nonane

100 110 120 90 100 110 120

mass[u] mass[u]

Fig. 11. Fast GC recording of a mixture of 10 molecules. Note that in order to have about six intensity points per GC peak, the overall time for a GC chromatogram of about 5 s required 35 mass spectra per second to be sampled. Each of these mass spectra consisted of the sum of a large number of TO F mass spectra, each of which was the sum of about 20 mass spectra that were recorded within about 100^s after the ions of interest had been collected in the storage ion source. For each mass spectrum the signal strengths between the masses of 30 and 150 u were then added to obtain the shown GC chromatogram. The mass spectra for each GC peak were also added in parallel.

The resulting mass spectra are shown for two of the GC peaks.

rather the molecule decomposition on a molecule-covered surface. In Ref. 46 such a specialized TOF mass analyzer is described in which the decompo-sition of acetic acid molecules on a heated palladium surface was investigated (see Fig. 12) with a surface coverage of less than one thousandth of a mono-molecular layer.

5. Multipass TOF mass analyzers

In most TOF mass analyzers the mass resolving power is limited by the time needed to form or to record a short ion bunch. Thus, as stated in sections 3 and 4, increasing the length of the flight

path also increases the mass resolving power. This is done most efficiently by using the same physical path repeatedly.

5.1. Ring TOF mass analyzers

One way to make an ion fly along the same path several times is to form the flight path to be a ring (see Fig. 13). This can be done, for instance, by using an accelerator storage ring that is energy isochronous [14,15] by proper choice of the quad-rupole strengths. In this case one chooses opti-mally: (i) the beam diameters to be identical after each turn for a beam of multienergy ions; (ii) the


►►►►-. ►/►643KH ^7

0 20 40 60 80

time [msec]

Fig. 12. The time dependence of the intensity of C02 molecules recorded by a TOF mass analyzer after a puff of acetic acid molecules was applied to a palladium surface that was heated to different temperatures. The number of molecules of acetic acid here could be kept so small that at most one thousandth of a monolayer of such molecules existed at the palladium surface.

flight time per turn to become independent of the relative energy spread δê = AK/K0? One should note here, however, that for precise mass measure-ments a ring can also be used for which the second condition is not fulfilled, if the ions all have the same energy. At least for energetic ions this con-dition can be fulfilled if an electron cooler [47] is used in one of the straight sections of the storage ring. In such an electron cooler very monoenergetic electrons are forced to move along with the ions

As long as the ion energy is low in a ring accelerator, more energetic ions move around the ring faster, since the increase in path length for more energetic ions does not fully counter-balance the increased ion velocities. When the ions have been accelerated sufficiently, however, this increase in path length for more energetic ions overcompensates for the increase in ion velocities. For ions of a specific intermediate energy — the so-called transition energy — the ring is energy isochronous, i.e. the increase in path length for more energetic ions exactly balances the increase of their velocities.

at about equal speeds, thus slightly accelerating all slower and slightly decelerating all faster ions by Coulomb forces. Unfortunately, this cooling procedure requires relatively long times, perhaps several minutes or at least seconds, so that it cannot be easily applied for ions of short-lived nuclei easily.

If a small cloud of ions moves in a storage ring, the passing charges can produce signals in some pick-up electrodes with the recorded frequency being a measure for the ion mass very similar to a Fourier-transform mass analyzer [10,48]. If only individual ions move in the storage ring one can also record their passage through a thin foil by the release of secondary electrons [49], or one can eject the ions from the ring and determine their flight time after an additional flight distance plus n turns [50] in the ring, since the energy loss in such a foil can be kept below 1 keV, which is negligible if the ion energy is some 109 electronvolts. In the case of an energy-isochronous TOF system the corre-sponding signal is only « 0.1 ns wide [19,49,51] so that, for an ion flight time f per turn of 500 ns, about 200 turns or 100^s are necessary to deter-mine f and thus the ion mass to an accuracy of 1 or 2ppm. For two ions of the isobar 100, Qp values > 1 MeV could thus already be seen after 20 turns, where the error in the Qß value would perhaps be 50 keV if the center of gravity of an observed signal could be determined to one twentieth of the peak width at half maximum (FWHM).

5.2. Folded TOF mass analyzers

Another way to increase the ion flight path is to arrange several TOF systems, as shown in Fig. 6, in series [52,53] so that the beam is folded several times and the flight tube is used repeatedly. Here the optical quality of the ion mirrors is important since the ion beam passes through a relatively large number of such lenses thus achieving a quasi-continuous focusing.

Alternatively, one can use ion mirrors that are operated in a pulsed mode, for instance as indi-cated in Fig. 14. In such a system one could bounce ions back and forth repeatedly [53]

402 H. Wollnikjlnt. J. Mass Specirom. Ion Processes 131 (1994) 387-407

Septum Injection ^ » ^ < Ë | Ç Ç

e-Cooler

(m/qhj Detector 10m

Fig. 13. An energy-isochronous storage ring for fast ions is shown that consists of six sector magnets of 60° deflection angle each plus a number of properly powered quadrupoles. These quadrupoles can be tuned [21,49] to make each half of the ring energy-isochronous and at the same time laterally focusing. (#) , Ions of mass m0, (O), ions of mass m\. For both these ions, trajectories are shown for two ion energies K0 and K0 + AK where all ions were assumed to be singly charged. Note also the timing detector in which each passage of an

ion through a thin foil releases a few secondary electrons that can be recorded in a double channel plate detector.

between mirrors I and II. Mirror I here must be switched off when the ion bunch to be analyzed enters the system, but it must be switched on when the fastest ions of the bunch return after one passage through the system. However, mirror II must be switched off after a certain time in order to let the ion beam reach the final ion detector after many reflections. In such a system the time sepa-ration between neighbouring mass ions increases

with every passage through the system and so does the mass resolving power, provided the FWHM of the ion bunches stay about constant, as in Ref. 53.

53. Image aberrations in multiple pass TOF mass analyzers

It is not trivial that the FWHM of the ion bunches

ion source

microchannel plate detector

^ » ) | ( -

" ) ^ < : - . /

ΧΛ \Λ \Λ Χλ

il ion mirror I ion mirror II

collector plate

Fig. 14. A TOF mass analyzer is shown in which the ions are passed back and forth between two ion mirrors repeatedly.

H. Wollnikjlnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407 403

time-of-flight *-

Fig. 15. The calculated signal width after 0, 1, 2, 3 and 4 turns in a system as shown in Fig. 14.

stay constant with increasing number of passes through a multiple-pass TOF mass analyzer. There are image aberrations due to the geometric size of the initial phase space of the ion beam, i.e. due to those aberrations that are proportional to the second-order products of the initial beam diameters x, y and initial angles of inclination a « dx/dz, ß « dy/dz and there are image aberrations due to the energy spread 6K of the ions. Studying the image broadening due to those aberrations of second order one finds that after 4, 8, 12, 16 . . . passes through some systems, for instance the ones of Figs. 13 or 14, the beam broadening is always minimal (see Fig. 15) since in these cases the TOF aberrations of second order are all compensated for [54]. This finding is not limited to a repetitive special system, it is the result of the seemingly fourfold symmetry of the TOF mass analyzer for an ion that passes four times through the system. The reason for this is that, in a lateral achromatically focusing system of fourfold symmetry, the lateral geometric aberrations of second order vanish [55] which causes the TOF aberrations [56] to vanish also, a concept also used for the construction of the TOF mass analyzer of Fig. 3 described in Refs. 19 and 20.

6. Fast ion detection for TOF mass analyzers

Although one can easily appreciate the advan-tages of obtaining mass spectra quickly, one should also note the difficulties caused by the use of fast analog electronics and computers to record fast ion bunches. Usually, this is done by letting the ions impinge on channel plates or on fast open

dynode multipliers. In special cases converter elec-trodes are foreseen [57] on which the impinging ions form secondary electrons that are then guided to and amplified by double channel plates. Such converter foils are advantageously covered by, for instance, Csl to improve the yield of secondary electrons [51]. For the combination of a converter electrode with channel plates, furthermore, the ion detection efficiency is usually increased by about a factor of 2 compared to ions that impinge directly onto the first channel plate. This effect results from the fact that only ions that impinge on the channel plate surface create effectively secondary electrons that may be sucked into the amplifying channels, while ions that enter into the channels themselves release secondary electrons only in some depth in the channel, so that only a smaller fraction of the channel is available for amplification and the overall amplification factor is reduced for these electrons.

Thus, for each initial ion, at the end, several 106-107 electrons are obtained where the electron pulse rises in about 200 ps and falls off in about 800 ps. Thus, signals o f « 1 ns FWHM need to be processed further and if several ions are contained in one bunch, signals of several nanoseconds FWHM. All electrons that leave the last channel plate are usually accelerated to a collector electrode, i.e. a flat conductive surface of about the area of the channel plate, from where the signal must be

Such double channel plates can be built as chevron devices with no space between them. Usually, however, they are separated by fractions of a millimeter and the electrons are accelerated over this space by a few hundred volts as is the case also between the last channel plate and the electron collec-tor electrode (see Fig. 17).

404 H. Wollniki Int. J. Mass Spectrom. Ion Processes 131 (1994) 387-407

insulating foil

electron beam terminating resistor

ion beam

coaxial cable

collector channel plates 200 Ohm

200 Ohm

Fig. 16. An ion-registration system consisting of a double channel plate amplifier that usually has a gain of about 106 or 107. Note the capacitor, formed by a plastic foil between two relatively large conductive plates, that couples the fast signal to a coaxial cable. Note also

the short distance L between the final collector plate and the 50 Ù resistor of the coaxial cable.

conducted away through a good coaxial cable. Here, it is important to also use well-matched vacuum feed-throughs in which signal reflections caused by changes in their impedance are minimized.

To couple the signal from the collector electrode to a coaxial cable has been done by using conical coupling devices several centimeters in length [58]. Very good results have been obtained, however, also by simply starting the coaxial cable — usually with a 50 Ù resistance terminator — only a few millimeters away from the collector electrode (see the distance L in Fig. 16). In this case, the signals are reflected back and forth over this short distance but since the signal velocity is close to the velocity of light, the reflected signal overlaps with the primary signal after only about lOOps or less. Thus, the frequencies of these signal distortions are in the 10 GHz range and, thus, are not amplified in the downstream electronic circuitry that usually has frequency limits of some 100 MHz.

Rather important also is that capacitors are installed parallel to the channel plates and at least the last dynodes of an open multiplier. Quite important also, however, is a capacitor that

connects the last channel plate or dynode surface to ground (see Fig. 16) since this capacitor directly reduces high frequency ringing. In some cases it is also advantageous to have the collector electrode not at ground potential and to couple the fast signal through a capacitor into the coaxial cable — usually with no noticeable signal degradation. This can be done efficiently by placing a thin plastic foil between the collector electrode and another plate of equal size (see Fig. 16). This arrangement can even be improved [59] if this foil is replaced by an insulator that has a high dielectric constant.

For some applications in which the ion beam has a large cross section, it is of interest to not only record the arrival time of an ion but also its position. The most simple way to do this is to divide the electron collector into many electrodes, for instance arranged like a detector board with individual amplifiers. In order to obtain this posi-tion information quickly (for instance < 5 ms) an arrangement as shown in Fig. 17 can be useful, in which a grid system is installed behind a channel plate. The center grid records the time signal while the signals from the two meander-like grids yield


X - meander

Y - meander backplane

M

Fig. 17. A fast position sensitive read out system for fast ion recording. The time defining pulse here is taken from the middle grid, while the x and y positions are found from the delays of the pulses on the meander-like grids. Over a distance of 40 mm thus the position could be determined to ±0.5 mm.

delayed signals where the delay times describe the x and y position of the ion under investigation [60]. If, as usually is the case, the time signal is slightly delayed, depending on the x and y position this delay can be mathematically corrected for during data analysis if the position is known.

In cases in which ions are formed [30,36,37,38] by a single shot of a laser pulse at a low repetition rate, the mass spectrum can be recorded by a transient recorder which records the varying signal intensity in many channels of perhaps 1 ns widths. Here, usually one laser shot constitutes a full experiment, in which case there is plenty of time to transfer the recorded mass spectrum to a computer for display and manipulation.

In other cases, one usually adds several TOF mass spectra and then transfers only the sum spectrum to a computer for display and manipulation. If one can always record the start and stop of single ion as in Refs. 27 and 28 the electronic circuitry to obtain a mass spectrum is relatively simple. If each spectrum must be recorded by a transient recorder, the

summing is difficult and usually must be done using computer hardware. Such integrating transi-ent recorders has been built [61] coupled to the TOF mass analyzer of Refs. 26 and 33. Although there has been great improvements in fast electronic cir-cuitry and even more so in fast computers over the last few years, this part of the technology of TOF mass analyzers still has room for improvement.

7. Summary and outlook

TOF mass spectrometers were popular until 20 years ago. Then the quickly-improved performance of quadrupole mass filters pushed them off the main stage of mass spectrometry. In recent years, however, a renaissance of TOF mass analyzers has occurred. This renaissance is largely due to the improved technology that has been developed dur-ing the last decade and here especially the improved electronic circuitry and computer technology.

The applications of TOF mass spectrometry so far utilize mainly:

(i) its unlimited mass range, which in the case of pulsed ion production, also features mass resolv-ing power that improves with increasing ion mass;

(ii) its fast scanning capability which allows the monitoring of fast changing substance concen-trations;

(iii) its efficient use of all formed ions. The application of the first property has especially come of age. The development of the PDMS [28] and the MALDI [36,37,38] techniques now allows molecules of 104 and 105u to be routinely determined. This technology has so far been estab-lished in many hundreds of installations worldwide and only awaits an improved ion-detection tech-nology that will increase the so-far limited mass resolving power.

Applications of the second property have so far only just started. Combined with the high detection sensitivity, the third good property of TOF mass spectrometers should especially improve the mass spectrometric investigations of surface reac-tants. Even more important, TOF mass analyzers could improve the investigation of the composition

406 H. Wollnik\lnt. J. Mass Spectrom. Ion Processes 131 (1994) 387-407

of the effluent of gas- and liquid-chromatographs as well as of capillary zone electrophoresis systems. Especially in view of the fast transients of fast gas chromatographs or new capillary zone electro-phoresis systems, TOF mass analyzers may become indispensible tools for chemical analysis.

8. Acknowledgement

For fruitful discussions I am indebted to A. Kraft and R. Becker. For financial support I am thankful to the 4'Bundesminister für Forschung und Tech-nologie" and to the "Deutsche Forschungsge-meinschaft".

References

1 F.W. Aston, Philos. Mag., 38 (1919) 709. 2 H. Wollnik, Optics of Charged Particles, Academic Press,

Orlando, FL, pp. 28, 220, 265. 3 J. Mattauch and R. Herzog, Z. Phys., 89 (1934) 786. 4 G. Boettger, G.E. Giffin and D.D. Norris, ACS Symp. Ser.,

102(1979)291. 5 W. Paul and H. Steinwedel, Z. Naturforschung Teil A, 8

(1953) 448. 6 P.H. Dawson, Quadrupole Mass Spectrometry and its

Applications, Elsevier, Amsterdam, 1976. 7 G.C. Stafford, P.E. Kelly, J.E.P. Syka, W.E. Reynolds and

J.F.J. Todd, Int. J. Mass Spectrom. and Ion Processes, 60 (1984) 85.

8 R.E. March and R.J. Hughes, Quadrupole Storage Mass Spectrometry, Wiley, New York, 1989.

9 E.O. Lawrence and M.S. Livingston, Phys. Rev., 37 (1931) 1707.

10 A.G. Marshall and L. Schweikhard, Int. J. Mass Spectrom. Ion Processes, 118/119 (1992) 37.

11 S.G. Alikanov, Soviet Phys. JETP, 4 (1957) 452. 12 W.C. Wiley and I.H. Mclaren, Rev. Sei. Instrum., 26 (1955)

150. 13 W.P. Poschenrieder, Int. J. Mass Spectrom. Ion Phys., 9

(1972) 35. 14 H. Wollnik, Nucl. Instrum. Methods, 186 (1981) 441. 15 H. Wollnik and T. Matsuo, Int. J. Mass Spectrom. Ion

Phys., 37 (1981) 209. 16 S. Neumann and H. Ewald, Z. Phys., 169 (1962) 224. 17 H. Wollnik, Thesis, Universität München, 1964. 18 E. Moll, H. Schrader, G. Siegert, M. Asghar, J.B. Bocquet,

G. Bailleul, J.P. Gautheron, J. Greif, G.I. Crawford, C. Chauvin, H. Ewald, H. Wollnik, P. Armbruster, G. Fiebig, H. Lawin and K. Sistemich, Nucl. Instrum. Methods, 123 (1975)615.

19 J. Wouters, DJ. Vieira, H. Wollnik, H.A. Enge, S.

Kowalski and K.L. Brown, Nucl. Instrum. Methods, 240 (1985) 77.

20 H. Wollnik, J.M. Wouters and DJ. Vieira, Nucl. Instrum. Methods A, 258(1987)331.

21 H. Wollnik, Nucl. Instrum. Methods A, 258 (1987) 289. 22 A. Gillibert, W. Mittig, L. Bianchi, A. Cunsolo, B.

Fernandez, A. Foti, J. Gastebois, C. Grégoire,* Y. Schutz and C. Stephan, Phys. Lett. B, 192 (1987) 39.

23 B.A. Mamyrin, V.l. Karataev, D.V. Shmikk and V.A. Zagulin, Soviet Phys. JETP, 37 (1973) 45.

24 W. Gohl, R. Kutscher, HJ. Laue and H. Wollnik, Int. J. Mass Spectrom. Ion Phys., 48 (1983) 411.

25 T. Bergmann, T.P. Martin and H. Schaber, Rev. Sei. Instrum., 60 (1989) 347, 792.

26 H. Wollnik, H. Grüner and G. Li, Ann. Phys., 48 (1991) 215. 27 D. Young and J.A. Marschall, Nucl. Instrum. Methods A,

298(1991)227. 28 D.F. Torgerson, R.P. Skowronski and R.D. Macfarlane,

Biochem. Biophys. Res. Commun., 60 (1974) 616. 29 A. Benninghoven, D. Jaspers and W. Sichtermann, Appl.

Phys., 11 (1976)35. 30 R. Kaufmann, F. Hillenkamp and R. Wechsung, Med.

Progr. Technol., 6 (1979) 109. 31 A. Duhr, R. Guckert and H. Wollnik, Proc. 41st ASMS

Conference on Mass Spectrometry and Allied Topics, San Francisco, CA, 30 May-4 June 1993, p. 929a.

32 H. Wollnik, R. Grix, G. Li, R. Kutscher and P. Feigl, in H. Matsuda and X. Liang (Eds.), Proc. Japan-China Sympo-sium on Mass Spectrometry, Bando Press, Osaka, 1987, p. 181.


34 R.J. Cotter, Anal. Chem. A, 64 (1992) 1027. 35 U. Boesl, J. Grotemeyer, K. Müller-Dethlefs, H.J. Neusser,

H.L. Selzle and E.W. Schlag, Int. J. Mass Spectrom. Ion Processes, 118/119(1992) 191.

36 Tanaka Y. Ido, S. Akita, Y. Yoshida and T. Yoshida, in H. Matsuda and X. Liang (Eds.), Proc. Japan-China Sympo-sium on Mass Spectrometry, Bando Press, Osaka, 1987, p. 185.

37 A. Overberg, A. Hassenbiirger and F. Hillenkamp, in M.L. Gross (Ed.), Mass Spectrometry in Biological Science: A Tutorial, Kluwer, Dordrecht, 1992, p. 181.

38 F Hillenkamp and M. Karas, in J.A. McCloskey (Ed.), Methods in Enzymology, Academic Press, New York, 1990, pp.280.

39 S.M. Michael and D.M. Lubman, Rev. Sei. Instrum., 63 (1992) 4277.

40 J. Tammoscheit, Thesis, Universität Giessen, (1992). 41 G.J. O'Halloran, R.A. Fluegge, J.F. Betts and W.L.

Everett, ASD TDR report 62-644, Bendix Co, (1964). 42 J.H.J. Dawson and M. Guilhaus, Rapid Commun. Mass

Spectrom., 3 (1989) 155. 43 P.A. Leclercq and L.A. Cramers, J. High Resolut.

Chromatogr., 11 (1988)845. 44 R. Grix, R.E. Tecklenburg and J.F. Holland, 2nd Interna-

tional Symposium on Applied Mass Spectrometry in Health Science, Barcelona, 1990 p. 259.


45 H. WoUnik, R. Becker, H. Götz, A. Kraft, H. Jung, C.C. Chen, P.G. Ysacker, H.G. Jansen, H.M.J. Snijders, P.A. Lecleucq and C.A. Cramers, Int. J. Mass Spectrom. Ion Processes, 130 (1993) L7.

46 H.J. Kelleter, G. Schüll, C D . Kohl, U. Grüner, A. Kraft and H. Wollnik, Surf. Interface Anal., 19 (1991) 581.

47 D. Liesen, GSI-Rep. 88-42, (1988). 48 M.A. May, A.G. Marshall and H. Wollnik, Anal. Chem.,

64(1992)1601. 49 J. Trotscher, K. Balog, H. Eikhoff, B. Franczak, B.

Franzke, Y. Fujita, H. Geissel, Ch. Klein, J. Knollmann, A. Kraft, K.E.G. Löbner, A. Magel, G. Münzenberg, A. Przewloka, D. Rosenauer, H. Schäfer, M. Sendor, D.J. Viera, B. Vogel, Th. Winkelmann and H. Wollnik, Nucl. Instrum. Methods B, 70 (1992) 455.

50 H. Wollnik, Nucl. Instrum. Methods B, 26 (1987) 267. 51 H.C. Seifert, D.J. Vieira, H. Wollnik and J.M. Wouters,

Nucl. Instrum. Methods A, 292 (1990) 533.

52 C. Su, Int. J. Mass Spectrom and Ion Processes, 88 (1989) 21. 53 H. Wollnik and M. Przewloka, Int. J. Mass. Spectrom. Ion

Processes, 96 (1990) 267. 54 H. Wollnik et al., in preparation. 55 K.L. Brown, IEEE Trans., Nucl. Sei., NS-26 (1979) 3490. 56 H. Wollnik and M. Berz, Nucl. Instrum. Methods, 238

(1985) 127. 57 N.R. Daly, Rev. Sei. Instrum., 31 (1960) 264. 58 G. Beck, Rev. Sei. Instrum., 47 (1976) 849. 59 A. Esper, Thesis, RWTH Aachen, (1991). 60 C. Klein, J. Trötscher and H. Wollnik, Nucl. Instrum.

Methods A, 335 (1993) 146. 61 J.F. Holland, B. Newcombe, R.E. Tecklenburg, Jr., M.

Davenport, J. Allison, J.T. Watson and CG. Enke, Rev. Sei. Instrum., 62 (1991) 69.

International journal of Mass Spectrometry and Ion Processes 131 (1994) 409 Elsevier Science B.V.

409

AUTHOR INDEX

Anex, D.S., 319

Baer, T., 295 Bargon, J., 319 Barofsky, D.F., 283 Bergmann, T., 21 Boesl, U., 87 Bondarenko, P.V., 181 Bönisch, M., 139 Brinkmalm, G., 283

Castleman, A.W., Jr., 233 Chan, T.-W.D., 67 Chien, B.M., 149 Colburn, A.W., Jr. 67 Colby, S.M., 125

Derrick, P.J., 67 Duncan, M.A., 307 de Vries, M.S., 319

El-Sayed, M.A., 265

Giannakopulos, A.E., 67 Giessmann, U., 345

Grant, P.G., 181 Griffiths, J., 265 Grotemeyer, J., 139 Grundwürmer, J.M., 139

Hâkansson, P., 283 Hillenkamp, F., 345 Hunziker, H.E., 319 Hwang, H.J., 265

Ingendoh, A., 345 Ioanoviciu, D., 43

Johnson, P.M., 193

Karas, M., 345 Kaufmann, R., 355 Kinsel, G.R., 139 Kirsch, D., 355 Knebelkamp, A., 319 Krause, H., 211

Lubman, D.M., 149

Macfarlane, R.D., 181

Mamyrin, B.A., 1 Martin, T.P., 21 Michael, S.M., 149

Neusser, H.J., 211

Reilly, J.P., 125 Reynolds, D.J., 67 Riley, J.S., 295 Robbins, D.L., 307

Schlag, E.W., 87, 139 Spengler, B., 355 Sundqvist, B.U.R., 283

Wei, S., 233 Weickhardt, C , 87 Weinkauf, R., 87 Wendt, H.R., 319 Willey, K.F., 307 Williams, P., 335 Wollnik, H., 387

Yeh, C.S., 307

Zhu, L., 193

International Journal of Mass Spectrometry and Ion Processes 131 (1994) 411-413 Else vier Science B.V.

411

SUBJECT INDEX

Atmospheric pressure ionization The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer 149

Background reduction Pulse amplitude analysis: a new dimension in single ion time-of-flight mass spectrometry 181

Biomolecules Design considerations in energy resolved time-of-flight mass spectrometry 67 Factors affecting the resolution in matrix-assisted laser desorption-ionization mass spectrometry 345

252Cf-plasma desorption mass spectrometry Pulse amplitude analysis: a new dimension in single ion time-of-flight mass spectrometry 181

Charge transfer resonance Decay energetics of molecular clusters studied by multi-photon mass spectrometry and pulsed field threshold ionization 211

Cluster ions Kinetic energy analysis in time of flight mass spectro-metry: application of time-of-flight methods to clusters and pyrolysis studies in supersonic expansions 295

Clusters Decay energetics of molecular clusters studied by multi-photon mass spectrometry and pulsed field threshold ionization 211 Using reflectron time-of-flight mass spectrometer tech-niques to investigate cluster dynamics and bond-ing 233

Collisionally-activated dissociation Sequencing of peptides in a time-of-flight mass spectrometer: evaluation of post-source decay following matrix-assisted laser desorption-ionisation (MALDI) 355

Direct laser desorption-ionization High-resolution mass spectrometry in a linear time-of-flight mass spectrometer 139

DNA Time of flight mass spectrometry of DNA laser-ablated from frozen aqueous solutions: applications to the Human Genome Project 335

Double focussing mass spectrometer Design considerations in energy resolved time-of-flight mass spectrometry 67

Electron impact ionization Photoemission electron impact ionization in time-of-flight mass spectrometry: an examination of experimen-tal consequences 125

Electrospray ionization The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer 149

Electrostatic lenses Design considerations in energy resolved time-of-flight mass spectrometry 67

Energy resolved time of flight Design considerations in energy resolved time-of-flight mass spectrometry 67

Energy-isochronous time of flight Energy-isochronous time-of-flight mass analyzers 387

Hydrogen bonding Using reflectron time-of-flight mass spectrometer techniques to investigate cluster dynamics and bond-ing 233

Intramolecular clocking The one dimensional photofragment translational spec-troscopic technique: intramolecular clocking of energy redistribution for molecules falling apart 265

Ion mirror The application of ion optics in time-of-flight mass spec-trometry 43

Ion/molecule complexes Photodissociation of magnesium ion/molecule com-plexes in a reflectron time-of-flight mass spectro-meter 307

Ion optical calculations Energy-isochronous time-of-flight mass analyzers 387

Ion optical system How to specify the ion optical system of a time-of-flight mass spectrometer 21

Ion optics The application of ion optics in time-of-flight mass spectrometry 43

Ion spectroscopy Photodissociation of magnesium ion/molecule com-

412

plexes in a reflectron time-of-flight mass spectro-meter 307

Ion trap storage The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer 149

Kinetic energy release Quantitative determination of kinetic energy releases from metastable decompositions of sputtered organic ions using a time-of-flight mass spectrometer with a single-stage ion mirror 283

Laser ablation Time of flight mass spectrometry of DNA laser-ablated from frozen aqueous solutions: applications to the Human Genome Project 335

Laser assisted reflectron Laser assisted reflectron time-of-flight mass spectro-metry 1

Laser desorption-ionization Laser ion sources for time-of-flight mass spectro-metry 87

Laser ionization Photoemission electron impact ionization in time-of-flight mass spectrometry: an examination of experi-mental consequences 125

Linear time of flight Laser ion sources for time-of-flight mass spectro-metry 87

Magnesium complexes Photodissociation of magnesium ion/molecule com-plexes in a reflectron time-of-flight mass spectro-meter 307

Mass analyzed threshold ionization spectroscopy Mass analyzed threshold ionization; structural infor-mation for a mass spectrum and mass information for ionic spectroscopy 193

Mass calibration High-resolution mass spectrometry in a linear time-of-flight mass spectrometer 139

Mass resolution Factors affecting the resolution in matrix-assisted laser desorption-ionization mass spectrometry 345

Mass spectra Mass analyzed threshold ionization: structural infor-mation for a mass spectrum and mass information for ionic spectroscopy 193

Matrix assisted laser desorption-ionization Design consideration in energy resolved time-of-flight mass spectrometry 67 Factors affecting the resolution in matrix-assisted laser desorption-ionization mass spectrometry 345

High-resolution mass spectrometry in a linear time-of-flight mass spectrometer 139 Sequencing of peptides in a time-of-flight mass spectrometer: evaluation of post-source decay following matrix-assisted laser desorption-ionisation (MALDI) 355

Microchannel plate Pulse amplitude analysis: a new dimension in single ion time-of-flight mass spectrometry 181

Multiphoton dissociation-ionization The one dimensional photofragment translational spectroscopic technique: intramolecular clocking of energy redistribution for molecules falling apart 265

Multiphoton ionization Decay energetics of molecular clusters studied by mul-tiphoton mass spectrometry and pulsed field threshold ionization 211

One dimensional photofragment translational spectro-scopy

The one dimensional photofragment translational spectroscopic technique: intramolecular clocking of energy redistribution for molecules falling apart 265

Optical spectra Mass analyzed threshold ionization: structural infor-mation for a mass spectrum and mass information for ionic spectroscopy 193

Penning ionization Using reflectron time-of-flight mass spectrometer techniques to investigate cluster dynamics and bond-ing 233

Peptides Sequencing of peptides in a time-of-flight mass spectrometer: evaluation of post-source decay following matrix-assisted laser desorption-ionisation (MALDI) 355

Perfluorinated polyethers Resonance-enhanced two-photon ionization time-of-flight spectroscopy of cold perfluorinated polyethers and their external and internal van der Waals dimers 319

Photochemistry Photodissociation of magnesium ion/molecule com-plexes in a reflectron time-of-flight mass spectro-meter 307

Photoemission electron impact ionization Photoemission electron impact ionization in time-of-flight mass spectrometry: an examination of experi-mental consequences 125

Photofragmentation Laser ion sources for time-of-flight mass spectro-metry 87

413

Postsource decay Sequencing of peptides in a time-of-flight mass spectrometer: evaluation of postsource decay following matrix-assisted laser desorption-ionisation (MALDI) 355

Pulse amplitude analysis Pulse amplitude analysis: a new dimension in single ion time-of-flight mass spectrometry 181

Pulse-laser ion sources Laser assisted reflectron time-of-flight mass spectro-metry 1

Pulsed field ionization Decay energetics of molecular clusters studied by multiphoton mass spectrometry and pulsed field threshold ionization 211

Reflector-type geometries Energy-isochronous time-of-flight mass analy-zers 387

Reflectron Decay energetics of molecular clusters studied by mul-tiphoton mass spectrometry and pulsed field threshold ionization 211 Laser ion sources for time-of-flight mass spectro-metry 87 Mass analyzed threshold ionization: structural infor-mation for a mass spectrum and mass information for ionic spectroscopy 193 Sequencing of peptides in a time-of-flight mass spectrometer: evaluation of post-source decay following matrix-assisted laser desorption-ionisation (MALDI) 355 The application of ion optics in time-of-flight mass spectrometry 43 The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer 149 Using reflectron time-of-flight mass spectrometer techniques to investigate cluster dynamics and bonding 233

Reflectron time of flight mass spectrometer Photodissociation of magnesium ion/molecule complexes in a reflectron time-of-flight mass spectrometer 307

Resolution High-resolution mass spectrometry in a linear time-of-flight mass spectrometer 139

Resolution enhanced time of flight High-resolution mass spectrometry in a linear time-of-flight mass spectrometer 139

Resonance enhanced laser ionization Laser ion sources for time-of-flight mass spectro-metry 87

Resonance enhanced multiphoton ionization The design and performance of an ion trap storage-reflectron time-of-flight mass spectrometer 149

Resonance enhanced two-photon ionization Resonance-enhanced two-photon ionization time-of-flight spectroscopy of cold perfluorinated polyethers and their external and internal van der Waals dimers 319

Rydberg states Mass analyzed threshold ionization: structural information for a mass spectrum and mass informa-tion for ionic spectroscopy 193

Sequencing Time of flight mass spectrometry of DNA laser-ablated from frozen aqueous solutions: applications to the Human Genome Project 335

Single-stage ion mirror Quantitative determination of kinetic energy releases from metastable decompositions of sputtered organic ions using a time-of-flight mass spectrometer with a single-stage ion mirror 283

Sputtered organic ions Quantitative determination of kinetic energy releases from metastable decompositions of sputtered organic ions using a time-of-flight mass spectrometer with a single-stage ion mirror 283

Supersonic expansions Kinetic energy analysis in time of flight mass spectro-metry: application of time of flight methods to clusters and pyrolysis studies in supersonic expansions 295

Thermally-labile molecules Laser assisted reflectron time-of-flight mass spectro-metry 1

Threshold photoelectron photoion coincidence Kinetic energy analysis in time of flight mass spectro-metry: application of time of flight methods to clusters and pyrolysis studies in supersonic expansions 295

Threshold photoelectron spectra Kinetic energy analysis in time of flight mass spectro-metry: application of time of flight methods to clusters and pyrolysis studies in supersonic expansions 295

Time of flight design How to specify the ion optical system of a time-of-flight mass spectrometer 21

Time of flight mass spectrometer Factors affecting the resolution in matrix-assisted laser desorption-ionization mass spectrometry 345

Van der Waals dimers Resonance-enhanced two-photon ionization time-of-flight spectroscopy of cold perfluorinated polyethers and their external and internal van der Waals dimers 319

Documents

Time of Flight Mass Spectrometry-Schlag