TIME RESOLVED FLUORESCENCE SPECTROSCOPY WITH A FAST ANALOG

TECHNIQUE: APPLICATIONS TO MICROSCOPIC SPECIMENS

by

MATTHIAS W. PLEIL, B.S.

A THESIS

IN

PHYSICS

Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for

the Degree of

MASTER OF SCIENCE

Approved

Accepted

May, 1987

\ ^ ^ ^ ACKNOWLEDGEMENTS

I would like to thank the following individuals who have worked

closely wlth me in the lab and with whom I have had many fruitful

discussions: Greg Sullivan, Qiaries Landis, and Dr. Shubhra

Gangopadhyay.

I gratefuliy thank my mentor Dr. Waiter Borst for his guidance

ajid support; only through him was this work possible.

I also like to thank Drs. Roland Menzel ajid Thomas Gibson for

their advice and enlightening discussions.

I would like to thank my parents for their love and

encouragement.

Most importantly, I thank my wife, Oriajia, for her constant moral

support, love and encouragement. Without her I would not have

completed this work. I dedicate this thesis to her.

11

TABLE OF CDNTENTS

11 ACKNOWLEDGEMENTS

LIST OF TABLES iv

LIST OF FIGURES v

INTRODUCTION 1

aiAPTER

I. THEORY OF FLUORESCENCE 3

Li terature ci ted 18

II. APPARATUS 19

Li terature Ci ted 31

III. METHOD 32

Li terature Ci ted 55

IV. APPLICATION TO SCINTILLATORS 56

Li terature Ci ted 69

V. APPLICATION TO GEOLOGICAL SPECIMENS 71

Coal 71

Crude Oi i 75

Li terature Ci ted 86

VI. APPLICATION TO CRIMINALISTICS 87

Li terature Ci ted 93

CONCLUSION 94

COMPREIIENSIVE BIBLIOGRAPHY 97

111

LIST OF TABLES

2-1. Instrument Parameters and Fluorescence Signals 25

4-1. Scinti ilator Fluorescence Lifetimes 64

4-2. Summary of Effective Decay Times Determined by Others 65

6-1. Decay Times of Fingerprint Background Materials 89

IV

LIST OF FIGURES

1-1. Absorption of a photon by a molecule as governed by the Frajick—Condon principle 14

i-2. Possible paths of return to the ground state of a system in the exci ted s tate 15

1-3. The derivation of the single exponentiai fluorescence decay law applied to a collection of identical systems (molecules) 16

1-4. Fluorescence emission spectrum of Rhodamine 6G in ethanoi (0.15 g/L) by continuous wave mercury lamp excitation (365 nm line) 17

2-1. Schematic diagram of the apparatus 26

2-2. Internal side view of the Leitz MPV3 microscope 28

2-3. Digitized single photon response of the apparatus 29

2-4. Typical output from the photodiode trigger source 30

3-1. Graphical description of the convolution process 46

3-2. Total instrument response of the system 47

3-3. Totai system spectral response S(A) for a given optical configuration (20X Rolyn objective, K399 high pass filter, 0.5 mm exit slit. 0.3 mm measuring diaphragm) .. 48

3-4. Raw c.w. spectrum R (X) of a green-yellow resinite coal maceral 49

3-5. Spectrally corrected c.w. spectrum of the same green-yellow resinite maceral as in Figure 3-4 50

.3-6. Emitted fluorescence pulse of p-bis 2-(5-Phenyloxazolyl) benzene (or POPOP) in ethanol solvent fitted with a mono-exponential decay function 51

3-7. Self-deconvoluted instrument response assuming a mono-exponential decay 52

3-8. Time-resolved A-coefficient spectra and corresponding decay curves for seperate anthracene (T = 4.18 ns) and POPOP (T = 1.35 ns) solutions r>.3

V

3-9. Time-resoived fiuorescence spectra and decay times for an anthracene-POPOP mixture as a function of emission waveiength 55

4-1. Emitted fluorescence pulse from a Pilot-U scintillator fitted with a mono-exponential decay function 66

4-2. Dependence of fluorescence decay time on emission waveiength 67

4-3. Fluorescence spectra of the seven scintiilators obtained with continuous wave excitation 68

5-1. Fluorescence pulse emitted from an alginite maceral after laser exci tation 77

5-2. Example of time-resoived fiuorescence from an alginite coal maceral 78

5-3. Resolution of the composite fluorescence spectrum emitted from algini te macerais of New Albany shaie 79

5-4. A typical contiuous wave emission spectrum of an alginite coai maceral with the corresponding spectral parameters 80

5-5. Percent contribution to the total intensity averaged over the entire emission range as a function of sample maturi ty 81

5-6. Distribution of fluorescence decay times in coai macerals 82

5-7. Resolution of composite fluorescence spectra of two major maceral types found in the Kittaning coals 8.3

5-8. Fluorescence emission spectra and lifetimes from a Texas crude oi 1 84

5-9. Dependence of the three resolved fluorescence decay times (see also Figure .5-8) on API index for several crude oi 1 and condensate samples 85

6-1. Fluorescence spectra emitted from a cotton string fiber and from a f iber from a white sock 90

6-2. Fluorescence pulse emitted from a white sock fiber fitted with a bi-exponential fitting function 91

6-3. Fluorescence pulse emitted from a cotton string fiber fitted with a bi-exponential fitting function 92

VI

INTRODUCriON

A fast analog technique developed for the determination of

fluorescence emission parameters of microscopic geological particles

is applied to a variety of samples. Presented here are fluorescence

data from coal macerals. anaiysis of fluorescence emitted from crude

oils, applications to fiber analysis and piastic scintillators.

Characteristic decay times in the nanosecond and subnanosecond range

are given along with time-resolved emission spectra.

The new fast analog technique enabies us to resolve individual

fluorescing components from a mixture of one, two, or three

non-interacting fluorophores by differentiating between the individual

lifetimes. The microscopic samples are excited with a near

uitraviolet pulsed nitrogen-pumped tunable dye laser. The emitted

fluorescence is detected at specific wavelengths with a micro-channel

piate photomultiplier tube and fast waveform digitizer. A number of

pulses (64-2048) are signal averaged and passed on to a computer for

data analysis. Using iterative reconvolution techniques and least

squares fitting, the instrument response is removed and the

fluorescence emitted from a variety of sampies is modeled with a sum

of exponential terms, each term corresponding to an individual

fluorophore. Complex systcms are characterized by rcsoiving the

individual fluorescence spectra and lifetimes.

This technique has been under deveiopment since the beginning of

1983 by W. Borst ejt aT. The author joined the project as an

undergraduate in November of the same year. Much of the author's time

in research has been spent on learning the operation of the equipment,

developing software for data acquisition, anaiysis, and calibration

routines, checking and rechecking the quality of the procedures

developed and maintaining ail of the equipment in peak operating

condition. From the outset, the physics group has been working in

cooperation with the geology departments of Southern lilinios

University (1983-1986) and Texas Tech University (1986-present). As a

result of this cooperation, the system was designed for ease of use,

requiring as little as possible of the operator in terms of technical

skills, allowing him to concentrate on the samples under investigation

and enabling him to acquire large amounts of data, which is necessary

in geological analysis. The system is routinely used by Charles

Landis, the resident geologist, who has provided the group with a

weaith of geological data and who has directed the geological

research. The technique has been tested under diverse sample

conditions and its limitations determined. It has also been appiied

to other specimens more directly related to physics including plastic

and acrylic scintillators, and various fibers.

CHAPTER I

THEORY OF FLUORESCENCE

Luminescence properties are extensively used for characterization

of materials. The term 'luminescence' is broad and covers all

phenomena related to the emission of light from materiais which have

been excited by the absorption of radiation or by other means.

Luminescence is made up of two catagories: phosphorescence and

fluorescence. Fluorescence is a fast phenomenon which stops

-9 practicaily at once when the excitation source is removed (10

seconds) and it is the result of a transition between two electronic

states having the same spin multiplicity. Fluorescence is commonly

observed in gases, solutions, and solids. Phosphorescence is a slow

phenomenon and is the result of a transition between two states of

different spin multipiicity; it is primariiy observed in solids where

the re-emission of radiation may continue for hours [1].

Energy must be absorbed before luminescence can occur. Whether

or not light of a given frequency can be absorbed by a moiecular

system is governed by the Franck-Condon principie. Figure 1-1 is a

plot of the potentiai energy as a function of internuclear separation

of a diatomic system. The lower curve is for the ground state

configuration (S^). The upper curve is for the first excited singlet

state (S.). The closely spaced horizontal lines within both potential

curves represent the various vibrational modes the system may be in

for a given electronic state. The Franck-Condon principle basically

states, that, since the absorption process takes very little time (on

-15 the order of 10 seconds), the absorption from the ground state to

the first excited state must be represented by a vertical line. In

other words, the system does not change its internuclear configuration

during the absorption process (the internuciear seperation, R, is

constant). At room temperature, the occupation of vibrationai states

is governed by the Boitzmann distribution:

^ ^ e"^^^'^ (1-1)

or equivaiently:

^= -(E2-E,)/kT (J.2)

where ^ (=N /N ) is the ratio of the population of level 2 cind level 1

and AE (=E„-E.) is the energy difference of the two vibrational states

in question. T and k in the above equations represent the temperature

and Boitzmann's constant, respectively. If one lets T=300 K, and

AE=i500 cm~ (approximately the spacing of two successive vibrational

energy levels), then ã = O.Oi. Hence, the molecule will be in its

lowest vibrational state at room temperature [2]. Since the molecule

is most probabiy in its lowest vibrational levei in the ground state

prior to absorption (v=0, S ), the internuclear separation will have

its equilibrium value R . Note, for higher vibrational levels (v>l).

the probability is that the internuclear separation will be close to

the ailowed extremes of the potential curve (this is analogous to the

harmonic osciilator where it is found from the wavefunction that the

probability distribution for the particle is highest close to the

cxtremes of its motion for v>l). Quantum mechanical iy, the lare;csL

.3

overlap between the two wavefunctions involved in the absorption

occurs for the strongest transition (in Figure 1-1, the most probable

transition is S„, v=0 —* S^ , v=4).

Once the system has gone through the absorption process and is in

the excited state, it may go through vibrational relaxation or it may

emit a photon from the vibrational level to which it was initially

excited (through spontaneous or stimulated emission). Wliich of these

processes is dominant depends on the environment of the system. In

the gas phase, a given molecule is essentiaily isolated, hence, the

only way for it to iose vibrational energy is to either emit an

infrared photon (ending up in the lowest vibrational level of the

first excited state) or it can make the transition directly to the

ground state. The direct transition to the ground state is the most

probable, hence, in rarified gas phase studies, the vibrationai levels

of the excited state are easily discernible in the emission spectra

obtained by broadband excitation. The work presented here consists of

liquid and solid phase studies. A collection of fluorophores

(fluorescing molecules) in liquid or solid phase lose all excess

vibrational energy in the first excited singlet state prior to the

emission of a photon. This energy transfer is accomplished by thermai

relaxation through interactions wilh other molecules and is very

-13 -11 rapid. 10 ' to 10 seconds [3]. Therefore. absorption. inciuding

vibrational relaxation, can be considered instantaneous with respect

-9 -3 lo the emission process (~10 - 10 scconds).

Shortly after absorption the systcm is in the lowest vibrational

mode of the first excited singlet slate. Tlie systcm can return to thc

ground state via radiative or nonradiative paths (see Figure 1-2). If

the transition is direct from the first excited singlet state to the

singiet ground state and involves emission of a photon, the phenomenon

is termed fluorescence. If there is inter-system crossing. where the

first exciled singlet state makes a transition to a triplet state of

iower energy (through spin-orbit interactions via vibrational coupling

between the two states), and then an emission of a photon occurs as a

result of the T. -» S„ transition, the phenomenon is termed

phosphorescence. Of course the system may also return to the ground

state (in general, any of a number of lower states) through numerous

non-radiative paths which invlove energy dissipation through phonon

interaction with the surrounding media (consequently, the system

environment heats up); this is called inLernal conversion in the case

of no change in spin muitiplicity (As = 0) and inter-system crossing

in the case of a change in spin muitiplicity (As = 1). Energy may

aiso be transferred to other molecular systems by dipoie-dipole

interaction and they in turn may or may not radiate.

The radiative transition from thc first excited singlet state to

the ground state in an atom is governed by the quantum mechanics of

spontaneous and stimulated emission. Spontaneous emission occurs

without any externai perturbation (hence the term spontaneous).

Stimulated emission is the rcsult of an external electric field

perturbing the system thereby inducing the transition to a lower

electronic state resulting in the emission of a photon. A proper

quantum mechanical treatment of the absorption and emission proccesses

would involve the use of quantum elcctrodynamics (the quantization of

the electric field). However, one may use Einstein's detailed

balancing approach coupled with the electric dipoie approximation to

get an accurate description of the emission processes involved [4. 5].

The developement of detailed baiancing is applied below to the

simpiified case of a two state electronic system; there is no

degeneracy nor is there any interaction by collisions with other

systems.

Detailed balancing makes use of the fact that the induced

absorption rate is

R(^.^^^ =B...p A(o..) (1-3) ij ij rad'' ij^

where B. . is the Einstein "B" coefficient and p ,(w. .) is the energy ij rad^ ij'

density of the radiation:

"radt^ij^ = (l^l^/^TT) . (1-4)

ã is the magnitude of the electric fieid. The excitation light has to

be of energy hv = AE where AE is the energy difference between the two

electronic states i and j for absorption to be allowed. The frequency

of excitation must therefore be w. .(= 2TTV) . The Einstein B

coefficient can be approximated by

^^ 3 \i •-'

using the first order electric dipole approximation where

e = the charge of an eiectron

h = Planck's constant/2F

új = the wavefunction of the i and j states i.J

and

r = the average position vcctor of thc electron and is in the

direction of the electric dipole moment (p = e*£. where p is the

dipoie moment) [4] <//. |i:p .> is the matrix element and is a measurc

of the degree of coupling between the two states.

The spontaneous rate for electronic transitions. Ni/.> to !>/'.> J ' 1

states, can be written as

j^(spont) ^ ^ >j,

Ji Ji ^ J i' (1-6)

where A.. is the Einstein "A" coefficient and E., E. are the energies Ji J 1

of the j and i states, respectively.

The number of molecuies making the transition from the lower to

the higher energy state is

dN. . ^-J.^N.-R?^.^^) =N.B .-. . p ,(w. .)

1 1j rad 1j (1-7)

dt 1 ij

and the number of transition being made from the higher to the lower

state is:

dN. . - r ^ = N . dt j

j (stim) ^ j (spont)

ji Ji (1-8)

where R Ji

(stim) . , ^ r . 1 . 1 • • n(stl" ) ^ '^ is the rate for stimulated emission. K. .

is Ji

related to the energy density p i w ) by'-

„(stim) „ , .. R.. ^ = B.. p j(w ..) ji ji ' rad^ ji'

where

"^ 3 h^ '

and is cailed the Einstein B coefficient for stimulated emission

(1-9)

(1-10)

Hence,

dN. . J-^i

d t = N.

J B..P ,((.)..) + A.. ji' rad' ji' ji

(1-11)

In equilibrium. the time rate of change of the number of molecules in

the upper and lower states must be equal. Therefore.

dN. . dN . . (1-12)

dt

or

N. J B. .p ,(w.. ) + A.. j r rad^ ji'' ji

= N.B. .p ,((j. .) 1 1j rad 11'

(1-13)

Taking the ratio of N . to N. J 1

N. _i _ N. " 1

B. .p ,(o). .) ij^rad^ ij^

B . .p ,(GJ . . ) + A .. ji'^rad^ ji'' ji

(1-14)

and equating it to Equation (1-2) (the molecular system is in thermai

equilibrium with its environment), one establishes the result:

B. .p ,(w. .) ij'^rad^ ij^

B. .p ,(w.. ) + A.. ji' rad^ ji^ ji

We have w. . = w.., and ij Ji

p .(w. .) = p ,(w ..) rad 1 j rad j i'

Solving for p ,(w..), rad j 1 •'

= e -(E.-E.)/kT ^"^jiXkT

j 1' = e ^ (1-15)

(1-16)

P j(í^--) = rad j1' JLl

B. . e^^'j^'^^'^ - B

(1-17)

ij Ji

and noticing that the energy density follows Planck's distribution

law

~ 3 h(i).

Prad<-'ji) = Vî •

the relation:

A . . âl

h(0 . . Xrr

B. . e J ' / ' '^ - B

TT C hcJ . . / , .T-

e J ^ ^ ^ ^ - 1

(1 -18)

1 h o j . . J ^ .

2 3 --ir c h ( i ) . . / , .T-

e J^-^^^ - 1 ij Ji

is obtained. Here, c is the speed of light in vacuum.

(1-19)

Thc only way

that Equation 1-19 can be true, is if Lhc following relations arc met:

B.. = B. . Ji iJ

(1-20)

and

10

A .. h(ú ..

B. . 2 3 (1-21)

IJ TT C

or equivalentiy,

'"' 3 h(i)..

ji ij 23 vi - - ; TT C

Note. the detailed baiancing results given above apply to the gross

behavior of a molecular system. Originally, this approach was applied

to electronic transitions as would occur in rarefied gas studies of

atomic systems. If one denotes the degeneracy of the electronic

states |v//> and \yp .> by g. and g., respectively, then Equation (1-20) becomes:

giB.. =g.B.. (1-23)

while Equation (1-22) remains the same with B. . replaced by g.B. . [.51. ij J ij ^ -

In time resoived studies, one looks at the fluorescence of the

specimen after the excitation source has been turned off. In most

cases p .((j. .) is negligible and essentially nonexistent (except

wi thin a laser cavity where p ,((J. .) can be appreciabie) . This

results in the depopulation of the excited state being governed only

by the spontaneous emission rate A... The radiative lifetime of the

system is just the inverse of the spontaneous emission ratc:

T " , = A.. . (1-24) rad j 1

The fluorescence lifetime observed from a specimen by measuring the

photons emitted in time includes contributions from nonradiative as

well as radiative rates.

The average amount of time a molecular system is in the excited

state can reveal much about its environment. By exciting N sysLems

into Ihe excited state and observing the emission of photons with

11

time, one may determine parameters such as Lhe decay time of the

system in the excited state, which includes non-radiative as well as

radiative decay rates. For most systems the probabiiity of emission

is directly proportionai to the number in the excited state. hence,

the emission follows a single exponential decay law. The number of

photons emitted in the time interval from t to t+dt is dN where dN is

proportionai to the number of systems (molecules) in the excited state

N:

dN = -R-N-dt . (1-25)

R is the rate constant governing the efficiency of transition from the

first excited singlet to the ground state and the right side of the

equation is negative indicating a reduction in the number of systems

in the excited state with time. R takes into account radiative and

nonradiative energy transfer rates including intersystem crossing.

Figure 1-3 shows the derivation of the single exponential decay law.

Since the intensity emitted from the collection of systems is directly

proportional to the number of photons emitted at any specific time,

one may write for a single fluorophore:

I(t,A) = I^(A)-exp(-t/T) , (1-26)

where I(t,Á) is the intensity of the flourophore at time t and

wavelength A, I (A) is the intensity at t=0 for the given A and T is

again the fiuorescence or observed lifetime. In the absence of

competing processes, where all non-radiative rates are suppressed. the

observed decay time T is just the intrinsic or radiative lifetime of

the excited state ~^^..^^-

12

The extension of the above derivation to a number of non-

interacting fluorophores is straight forward. For N. non-interacting

fluorophores, the intensity as a function of time at a given

wave1eng th i s

N

I(t.A) = ^ A.(A) • exp(-t/T.) . (1-27)

i = l

Here A.(A) is the pre-exponential coefficient and it is equivalent to

I (A) in Equation (1-26) for the individual fluorophores.

Most organic specimens are made up of a conglomerate of N

fluorophores which emit a total signal I(t,A). If one can determine

the individuai A.'s and T.'S which make up the total I(t,A) over a

wavelength range, then one may separate the conglomerate spectrum into

its components, hence, the individual constituents may be identified.

The absorption and emission spectra for most organic molecules in

the liquid or soiid phases are in generai quite broad at room

temperature. These spectra show very little structure. The lack of

weil defined peaks, which would correspond to transitions between

vibrational levels, is a result of the large number of vibrationai

levels and rotational sublevels. The rotational levels are also

extremely numerous at room temperature. hence. the resulting spectra

are effectiveiy continuous.

Figure 1-4 shows a typical emission spectrum of a pure substance

(Rhodamine 6G at 0.15 g/L concentration in ethanol). The peal< at 566

nm represcnts the emission from the lowest vibrationai level of the

first excited singlet state (this experiment was done at room

temperature) to an upper vibrational level of the ground state. This

13

most probable transition of emission corresponds to an energy

difference of .576 nm or AE = 17361 cm . Thc emission spectrum is

red-shifted relative to the absorption spectrum due to Stokes' shift.

The primary reasons for Stokes' shift is 1) there is some cnergy loss

due to vibrationai relaxation when the system is in the first excited

singlet state, 2) the first excited state of most molecules decay into

an upper vibrational level of the ground state due to the

Franck-Condon principle (see Figures 1-1 and 1-2) and 3) the excited

molecule becomes more polar resulting in a reorientation of the

surrounding solvent molecules which in turn lowers the energy of the

system (the energy gap between the two states decreases) [2]. The

vibrational levels of S„ and S.. are usually very similiar and the

reciprocal transitions have proportional probabilities (i.e., if the

S, , v=0 -* S v=4 absorption transition has the highest probability.

so does the S^. v=0 -» S„, v=4 emission transition, see Figure 1-1).

As a result, the absorption and emission spectra are oftentimes mirror

images of each other. This is indeed the case of Rhodamine 60 in

dilute ethanol soiution [6].

The work presented here depicts a fast analog technique used in

determining the pre-exponential coefficients and corresponding decay

times of fluorescence emitted by complex organic systems made up of

one, two or three non-interacting fluorophores. From these

parameters. the component spectra may be resolved allowing one Lo

"fingerprint" Liie organics cind in some cases even determine Lhe

chemical composilion of these unknowns.

14

P o t e n t i a 1

E n e r g y

Thc Franck-Condon l ' r i n c i p i e

F i r s t Exci tcd

S i n g l e t S t a t e

Ground S t a t e

= G

= 6 5

1 0

i

0

Internuclear ScparaLion

Figure 1-1. Absorption of a photon by a moleculc as govcrned by the Franck-Condon principle. The graph reprcsents the potcntial energy curves (not to scaie) as a function of internuclear separation for the first excitcd singlet and ground sLatcs of the bond rcsponsible for fiuorescence. The horizontal lines represent the vibrational modcs (rotational modes are not given for clarity). The absorj^tion and emission takes piace over negligible timc, hencc, the sysLem docs not changc its internuclear scparation. As a resuit, vertical lines are drciwn to represent the change of eiectronic state. In this dcpiction, the molecule is at room temperature.

15

Possiblc Excitcd —* Ground SLate Energy Transfer Path;

«1 -

Vibrationai Rclaxation

Fluorescence Emission

or Internai

Conversion

o

Inter-system Crossing

- T,

Phosphorescence

or Internai

Conversion

Process

Exci tation

Rela.xation (Internal conversion)

Inter-system crossing

Phospiiorescencc H

l'luorescencc

Time (seconds)

=;io-'-^ ;.io-^2

:.10-'2 -3 4 -10 - 10

-9 - 10

H Timcs include contribution from non-radiative transitions

i- igurc 1-2. i'ossiblc patlis of rcturn to thc ground sLate of a systcm in tlie excited staLe. Oncc cxciLed, tiie systcm gocs tiuougli vibrational relaxation jîrior to inter-sysLem crussing or dirccL rcLurn Lo tiie ground state. Return to ' iie ground state may or may not include emission of radiation. 'l'imes associatcd witií tlíc various processcs are also givcn.

16

Ratc of Decay of an Initially Excitcd StaLe

dN(t) _ dt

l = -(k + k )'N(t)

i^adiaLivc Non-I^adiaLive Dccay RaLcs

NumÍDer of Molecules i n the c x c i t e d s t a t c

i^ewr i t ing :

M>,( = - ( k + k ) d t N ( t ) ^ r n'

In l . cg ra t i ng :

N(L) = N -e o

-L/T

1 r = (I< + k ) and N is Llie numbcr of cxcitcd molccuics at t=0, ^ r n^ o

I'igure 1-3. 'I lie derivation of tiie single exponential fluorescence dccay law applied to a collcction of idcntical systcms (moiccules). T is the decay time or lifetime of the molecule in Liie first excited singlet sLate in Lhe presence of competing. non-radiaLive processes. T is Lhe Lime iL Lakes for Lhe number of excitcd molccules Lo decrease Lo 1/c of Liic iniLial numbcr of cxciLcd sysLcms, and r is commonly referred Lo as Liie observed lifetime.

17

>-»—

> h-<

I

LU CC

1.0

0.8 .

0.6 .

0.4

0.2 •

0.0 400 500 600 700

WAVELENGTH (nm)

800

I-'igure 1-4. F luo re scencc cmiss ion specLrum of i^hodamine GC in e t i ianol ( 0 . 1 5 g/L) ijy con t inuous wave mercury lamp c x c i t a L i o n (.3G.5 nm l i n c ) . Tlie specl rum iias Ijcen smooLlicd and normalizcd Lo uni t I ie ight ; i ts pcak o c c u r s at .566 nm.

18

Literature Cited

1. Hirschlaff, E. Fluorescence and Phosphorcsccnce, Chemical Publishing Co. Inc., New York, 1939.

2. Lakowicz, Joeseph R. Principles of Fluorescence Spectroscopy. Plenum Press. New York, 1983.

3. Hercules, David M. Fluorescence and Phosphorescence Analysis. Interscience Publishers, New York. 1966.

4. Weider, Sol The Foundations of Quantum Theory, Academic Press, New York. 1973.

.5. Bransden, B.H. and Joachain, C.J. Physics of Atoms and Moiecules, Longman, London, 1983.

6. Berlman, Isadore. B. Handbook of Fluorescence Spectra of Aromatic Molecules, Academic Press, New York, 1971.

aiAPTER II

APPARATUS

The laser fluorescence microscopy system is represented

schematicaliy in Figure 2-1. Further specifications are given in

Table 1-1. A pulsed laser and three sources of continuous wave (c.w.)

illumination are interfaced with a Leitz MPV3 microscope system. The

nitrogen pumped dye laser provides intense near-ultraviolet light

pulses. The laser is coupled to the microscope via a liquid light

guide. The modular construction of the microscope system allows

various optical components to be readily interchanged. The emitted

fluorescence is detected by a fast two-stage microcliannei plate (MCP)

piiotomul tiplier tube. The output of tlie MCP is directed either to a

picoammeter when acquiring conventionai c.w. spectra or to a fast

waveform digitizer when acquiring time-resolved spectra. The

digitizer sums a number of MCP output pulses and passes the average to

a desktop computer where data reduction and anaiysis take place.

Four light sources are linked to the microscope. The three c.w.

sources are part of the Leitz MPV3 system. The tungsten lamp

(Philiips 100 W #7023), mounted directly onto the right side of the

microscope body, is used primarily for spectral calibration. However.

it is also used as a white liglit source for viewing the samples

because i ts output is primarily in tlie visible part of tlie spectrum.

Thc 75 W high pressure xcnon lamp is a broadband source cmittini:!:

continuously from the ultraviolet tlirougii the visible parts of tlie

19

20

spectrum. The intensity of the xenon lamp at any particuiar

wavelength is much lower than the 365 nm line of the high pressure.

100 W mercury arc lamp (IIBO-lOOW/2). The ,365 nm line is selected by

the input monochromator and it is the primary c.w. fluorescence

excitation source. This line was chosen for its relatively high

intensity and its wavelength is in the near-ultravioiet (u.v.) making

it compatible with the glass optics on the excitation side. (Glass

absorbs in the u.v.) The iaser excitation line was chosen close to

this wavelength for comparison purposes. The output of the input

monochromator is directed through a collimator and enters the left

side of the microscope. The collimator contains a collimating lens

and a turret, which has three settings, two of wliich transfer filters

into the light path (380-800 nm or 200-390 nm band pass), and the

remaining setting is just an open diaphragm. The monochromator and

collimator are bypassed when a higher intensity is desired by mounting

the lamp directly to the microscope housing.

The EG&G 2100 nitrogen-pumped tunable dye laser may utilize any

of a number of dyes ailowing a wide range of pulsed excitation

wavelengths to be used. The nitrogen laser pumps the dye laser with

1.2 ns FWHM, 337 nm, 0.6 mj pulses at a repetition rate of 1-100 pps.

The dye is in a standard fluorescence quartz cuvette situated within

the laser cavity. This cavity uses a grating set at grazing incidencc

and a totaiiy reflecting mirror. which is coupled to a stepping moLor.

This combination resulLs in a tuning range of .360-800 nm. Utilizing

BPBD dye. the laser provides ncar u.v. pulses at 373 nm and is tiie

source of fluorescence excitation for timc-resolved spectral analysis.

21

The full width at half maximum (FWHM) of an individual dye laser pulse

was determined to be about 500 ps by a standard time-correlated single

photon counting technique presently under developement. The energy

per pulse was found to be greater than 10 pj using standard

calorimetric measurements and a Scientech 362 calorimeter. The iaser

generailly operates at a repetition rate of 10 Hz.

The laser pulses are directed via a iiquid iight guide to the

rear of the microscope. Figure 2-2 gives a cut-away side view of the

microscope interior. The laser pulse enters the microscope (left side

of Figure 2-2) and bypasses mirror M . which is removed from the path

for laser excitation. (M. is in place when exciting the sample with

c.w. sources. M^ allows one to chose either the right or left ports of

the microscope housing. Proper adjustment of M^ and M^ also allows

one to chose between transmitted or reflected illumination of the

sampie). After passing through the diaphragm D^, which is situated

between the two excitation lenses L. and L^, the laser pulses are

directed through the illumination diaphragm ID, and fieid diaphragm

FD. The laser beam is out of focus at the illumination diaphragm,

hence, by varying its diameter, one effectively controls the intensity

of excitation. The excitation beam is focused at the field diaphragm

ailowing one to continuously adjust the size of the illumination spot

on the sample (when the objective is focused). After passing through

tiie field diapliragm, tlie excitation liglit is reflectcd to the sample

by eitlier a diciiroic mirror or a neutral density beam splilter. The

dichroic mirror reflccts all (>90%) light below 400 nm and ailows

light above 400 nm to pass through. Tlie nuetral density beam spliLter

22

reflects 50% of tlie incident light regardless of its waveiength

(.300-800 nm). The dichroic mirror is used when coilecting

fiuorescence data while the l eam splitter is used when determining the

instrument response, which includes the laser pulse characteristics.

These mirrors are each part of a "cube" and there are four cubes

making up the "Ploemopak," which is a turret of cubes allowing for

easy exchange of optical elements. The Ploemopak is unique in the

MPV3 construction, for it allows the user to interchange optics

readily by rotating a knurled knob, which effectiveiy places one of

four opticai cubes into position. Tlie laser pulses are reflected down

through the objective and then onto the sample. The light excites the

sample fluorescence, which passes through the objective and dichroic

mirror, (the emitted fluorescence is generally above 400 nm) or

through the beam splitter. A K.399 high pass filter (Oriel) is used

when acquiring fluorescence above 400 nm (F. in Figure 2-2) to cut out

any leakage from the Hg lamp (or laser), which shows up as a second

order contribution in the range of 725-745 nm and is appreciable for

low intensity samples. The light is either directed to the oculars,

top port (used in taking photographs of the sample) or to the exit

monochromator. Before reaching the exit monochromator, tiie

fluorescence must pass tlirougli the measuring diapiiragm whicli allows

the user to select the area on the specimen he wishes to measure.

Tiiis diaphragm can be chosen to be either circular with incrcmental

diameters from .0'/" to 5 mm or in may be rectangular with continuously

variable widLÍis and lengths in the same rangc. Tlie image of the

measuring diaphragm can be seen ly the user through the oculars if

23

desired. The measuring diaphragm may also be rotated to fit

irregularly shaped specimens; this is done by actually rotating the

image under investigation using the prism arrangement situated in

front of the measuring diaphragm. Hence, one can be very specific

about the measured sample area; samples down to a few microns and up

to hundreds of microns in size have been easily selected and their

fiuorescence measured. The wavelength to be detected is selected by

the exit monochromator. This grating monochromator is in Littrow

arrangement and has 600 lines/mm resulting in a dispersion of 6.6

nm/mm exit slit width [1]. Interchangeble slit widths of 1.0, 0.5,

0.3, and 0.15 mm results in a resolution range of approximately 7 to 1

nm, respectively. The principle wavelength is scanned lineariy and

continuously either manually or by two computer controlled motors.

Tlie wavelength at which the monochromator is set is recorded by the

computer with an analog to digital converter from a potentiometric

output.

The Hamamatsu R1564U-01 two stage, proximity focused microchannel

plate (MCP) photomultipiier tube detects the wavelength-selected

fluorescence. This state-of-the-art MCP has 12 pm in diameter pores

resulting in a transit time jitter of approximately 90 ps [2]. Figure

2-3 represents a typical singie photon (or delta function) response of

the MCP convoluted with tlie digitizer response when a potential

difference of 3.0 kV is applied.

Tlie output from the MCP is directed either to a picoammeter when

acquiring continuous wave spectra. or to a Tektronix 7912AD fast

waveform digitizer when acquiring fluorescencc decay data. Fhe

24

picoammeter converts the MCP output current (-- nA) to a 0-10 V signal

appropriate for the analog to digital (A/D) converter of the computer.

Tlie fast waveform digitizer consists of tiie main control uni t (the

7912AD) and two plug-in units, the 7129 amplifier and the 7890P

programmable time base. The frequency response of the input amplifier

is about 1 GHz, and the maximum digitization rate is 100 GHz (10

ps/channei). We generally use the 40 ps/channel rate. The vertical

scale is divided into 512 channels (and so is the time scale). A

photodiode monitors the laser pulses and triggers the digitizer. A

typicai photodiode output pulse (convoluted with the digitizer

response) is shown in Figure 2-4. The individual phomultiplier pulses

are summed by the digitizer. The number of pulses ranges from 1 to

64. For acquisitions over 64 pulses, the digitizer sums in blocks of

64, sending the average of each block to the Tektronix 40.52A computer.

The computer system includes a Gemini-10 printer (Epson), Hewlet

Packard 7475A 6-pen plotter, Tektronix 4631 hard copy unit, three

Tektronix 8 inch floppy disk drives. a 40.50E01 ROM expEuider and a

built in memory expansion unit. which is used for data storage and

programming.

The entire system has been designed to allow easy data

acquisition. storage and analysis. With the above mentioned

components we can routinely acquire c.w. fluorescence and

time-resolved specLra of microscopic specimens. The relative ease of

use faciIiLates interaction with other. lcss technical ii,roups and

allows greater precision in repeatabi1ily of experiments.

25

Table 2-1. Instrument Parameters and Fluorescence S]

Size of sample Diameter of analyzed area Tuning range of emission monochromator Range used Monochromator bandwidth

Continuous Fluorescence Excitation

Tuning range of excitation monochromator Typical excitation wavelength Monochromator bandwidth

Fiuorescence Excitation by Pulsed Laser

Dye laser pulse duration (FWHM) Puise energy (BPBD dye, .37.3nm, lOHz) Peak power Laser bandwidth Dye iaser tuning range

Excitation wavelength used Pulse repetition rate Photons per pulse (BPBD dye) Photons reaching sampie Photons onto measured region C^ 10 pm)

Typical fluorescence yield Photons reaching MCP Typicai number of photoelectrons per pulse Instrument response risetime FWHM of instrument response Single pulse digitization rate Number of pulses signal averaged

10 5

220 380

220

.370

1

Lgnals

- 750 pm - 2.50 pm - 800 nm - 750 nm 1 - 7 nm

- 800 nm 365 nm

1 - 7 nm

< 600 ps > 10 nJ > 10 kW 0.04 nm - 780 nm

.373 nm - 100 Hz > 10''' > 10'^ - io'°

0.001 - 0.3 lO '• - l O ^

10 - lO''

<

10 -

< 700 ps . 1.25 ns - 100 GHz 64 - 2048

26

HARU

COPY UNir COMPUTER

IN/OUT

DEVICCS

XENON ARC LAMP

MERCURY ARC LAMP

PLOTTER

PRINTER

WAVEFORM DIGITIZER

MONOCHROMAÎOR

TRIGGER

MCP

MONOCHROMATOR

LEITZ MPV 3

MICROSCOPE

TUNGSTEN

LAMP

TUNEAULE

DYE LASER

NITROGEN

PUMP LASER

Figure 2-1. Schematic diagram of the apparatus. Fluorescence puises from the sample surface are excited by pulsed laser. spectrally analyzed, and acquired with the fast waveform digitizer for subsequent time-domain analysis by computer. For continuous wave excitation, mercury and xenon lamps are used and a picoammeter (not shown) is connected to the MCP photomuitiplier and computer.

27

(D

m 3; X <T> M. P n- W

C W '1 »—' M .

!-• 3 f^ Oq

CL M -

P "a D' n P oq 3

n: • - < •

ap D '

• 0 P U) U)

• - • í M .

• —

r t (\> -1

< • — ^ 50% at 399

nm. ~96%

above 401

nm

0 cr C_i.

a> 0 r-r (-!• < ft>

' o ' M .

0 Í T 1 0 (-•• 0

3 >-•• 1 T 0 - 1

0 -J

neutral density beam

splitter

*r) M -

n> M '

a a M -

t3 ir - 1 P

09 3

MH ^ ^ ^ .

c 3 M ^

3 fi3 r-t M .

0 D

a M .

&3 'a =r-ragm

r a> D W fD W

S M .

•1 '1 0 1 (/)

(A fD (—' a> o r t n> a

a

w (fl a) a

o fa

a :r a 0) M.

o 0) ar X n M. O rt M.

o 3 O 3 3 M. O n O -î p f O -1 -1

•-n -1 O 3

a> o - j

a

(/)

o - 1 o o nr P

a>

•d P <-r 0)

X}

o (-r O 3 C

X)

a) - 1

tr a fD M. -1 -1 0) a>

o r-r <-r D ' a> a> a í a> (a M. < <-r a> rr H - a> fD -1

09 <-r <-r O :r

<-r rt :r O 0) cr o fl> o

c a M-0) P3 <-T T a> (fl o rr » - • 1

a> o a -j

c cr a>

(/) a> M.

oq

•a 3 O M

f-r -J O

fO - r t

(fl 3 : a> O to

X P P a> rr P 3

a o 3: p • r r p

rr a c/) a> P 3 3 • - 0»

*0 ••'• -J •—09 O 0) :r c

r r -J '-'1 »<

C r r M-O C P -! D 3 (D 09 t3 (/)(/)(/! O <-r fD fD fD 3 3 P O <-r fD •— fD

P -! r r 3 (fl - 'O P w r-r < nr (D fD -J •— P O (/) <-r C

0» 09 C -I r

Xí (fl <-r

r r r r pr'

:r zr (D -J -J o o •— C C 0)

09 09 -•^ = r D- r r <-r <-r " 0 =r D- o (D fD -J

<-r O -1 cr M. p

CM. 09 r r 0) n r O <-r

rr p <-r :r D-(D (/) -1

r r O cr -J C 0) P 09 p M. izr 3 09

:r r r M. r r l ^ (fl fD

•o a p 1 M. <-r 0)

< ir p fD "1 T '—» r r 3 '--s fD M. K— a 1 0) O O <-r 3 -J rr (/) 0 3 : —

a =r M^ fD U)

0)

(/) -J p 0) a 3 3 •o o — < • a 3

^ 0 9

P

M. M.

09 09 :r c <-r -J

(\) Xi p ro <-r I :=r to

(/) >—i

M. ID 3 r t

a a> M. -1 o 3 P P <-r ^-a> a (fl

M cr a '^ a> r r < 3 - — 0) a> a o o

fD r t a 3-

O r r M. r -O O O 3 fD 3 'i-) fD M.

M. i r 3 0) c o a C M. (n O

3"

r t N 3" a> H 3 :

3 - -T3

3 a> <

p < M. (D O

^- 3

C 3 M' 3 P

O 3

-J -O - J

« 3" 0) - j (D

(fl fD O T O

TI 0)

< a>

P 3 (/) >-••

o -J o

M- U) M. n

09 O P 3"a 3 <-r a> a ••n 3 O <-r H- fD •—' -J

O (A

(/)

H 3-0)

28

oc ^ '^ O uo a: <

29

=3 Q .

•— ' - '

Z) > O E Q_ ^ \J

-

4.99

3.90

2.81

1.71

0.62 -

- 0 . 4 7 '

SINGLE PHOTON INSTRUMENT RESPONSE

RISE TIME 0.52 ns

FALL TIME 1.00 ns

1/2 WIDTH 0.86 ns

y V - ^ jirui ^r^^^

12 r

16 20

NANOSECONDS

Figure 2-3. Digitized single photon response of the apparatus . This pulse represents the convolution of a delta function signal (photon) with the response of the detector. which is made up of the 2-stage microchannei plate photomultiplier tube. RG-223, 50 Q coaxial cable and fast waveform digitizer.

30

124

99

> E

Z) CL.

74

O 49 UJ (21 O 5 24

TRIGGER SIGNAL vs. TIME

RISE TIME: 1.55ns

FALL TIME: 6.23ns

TIME (ns)

Figure 2-4. Typical output from the photodiode trigger source. The diode is situated close to the dye laser head and detects the back reflected laser signal. The photodiode output can be changed ly varying its position in the reflected beam.

31

Literature Cited

1. E. Leitz Inc. Product Literature Detailed List of Microscope Equipment and Accessories, Part No. 610-126 p. 5/5 1983.

2. Hamamatsu Photonics K.K. Proximity MCP PMT Data Sheet.

CHAPTER III

METHOD

The acquisition and reduction of the pulsed fluorescence data

requires much care. The fluorescence decay received by the computer.

M(t), is not the actuai fluorescence signal emitted by the specimen

F(t). M(t) is the convoiution of F(t) with the instrument response

function, I(t).

The individual fluorescence pulses emitted from the specimen.

F'(t). are actually a convolution of the laser pulse profile, L(t),

with the actual fluorescence decay of the specimen, F(t). The

convolution of the laser pulse with the fluorescence decay is

represented in Figure .3-1. The mathematical analog to the

laser/fluorescence folding ( or "faltung") can be represented by the

convolution integral:

t F'(t) 3. / L(T).F(t-T) dT

— 00

= L(t)^F(t) (3-i)

The MCP itseif will foid its temporal response, Mcp(t), into the

signai it detects, its final output being S(t):

t S(t) = / Mcp(T)-F'(t-T) dT . (-3-2)

—00

The MCP signai is transfered iy a RG-223. .50 Q coaxiai cable. which

distorts the signal negligibly by folding its time rcsponse. C(l) into

S(t) resuiting in a new signal S'(t)

t S'(t) = / C(T).S(t) . (3-3)

—00

32

33

In the finai leg of the signai path, the overall measured signal,

M(t). is a convolution of S'(t) with the digitizer response, Dig(t):

t M(t) = / S'(T).Dig(t-T) dT . (3-4)

—00

Since we assume all of the above functions have a linear response

in time, the foilowing general statements hold.

If

A(t) = B(t))C:(t)

then

A(t) = C(t)HB(t) (3-5)

and if

A(t) = B(t)^ C(t) D(t)

then

A(t) B(t) :(t) > D(t) (3-6)

From the above relations for the generai functions A.B.C, and D, we

may write the measured signal, M(t) as:

M(t) = L(t))<F(t)HMcp(t) C(t)MDig(t) (3-7)

From reiations (3-5) and (3-6) we see that, mathematically, the order

of the convolution does not matter. Hence M(t) may also be written

as

M(t) = L(t) Mcp(t)HC(t)xDig(t)HF(t)

Le 11 i ng

I(t) = L(t)HMcp(t)>C(t)*<Dig(t)

allows one to write:

M(t) = I(t)»«F(t)

or equivalently,

(3-8)

(3-9)

34 t

M(t) = / I(T).F(t-T) dT . .3_10) -00

The task is to obtain the fluorescence decay function F(t) and,

from this, the various dccay times and percentage contributions of the

individuai fluorescing components. This is done with an iterative

reconvolution technique and least squares fitting (see below). First,

the instrument function has to be determined. If F(t-T) in the above

integrai is a deita-function ô(t-T), then the measured singnal M(t) is

just the instrument response I(t). Experimentaily, either a mirror or

a glass piate is used in piace of the sample, which corresponds to

repiacing F(t-T) with ô(t-T). Thus, the instrument response is

obtained and contains the contributions from the iaser pulse, MCP

temporal response, cabie contributions and preamplifier bajidwidth

(approximateiy 1 GIIz) of the fast waveform digitizer. Figure .3-2

represents a typical instrument response curve acquired experimentally

using a glass plate as the sample with the neutral density beam

splitter cube in place (see Chapter II). It is the convolution of

L(t), Mcp(t), C(t) and Dig(t) which is equivaient to the convolution

of Figure 2-3 witli the laser pulse shape.

Various deconvolution techniques have been examined by other

investigators [1,2,3]. Software has been developed by Greg Sullivan

to mathematicaily modei the fluorescence pulses for testing three of

these deconvolution techniques [4]. Tiie testing routine synthesizes

an instrument response, inciuding random Gaussian noise, and

convoiutes it with an assumed multi-exponential fluorescence decay.

The synthesized signal M(t) is then deconvoluted using the different

techniqucs to see how the calculated decay times and pre-exponential

35

coefficients compare with the known values set at the beginning. It

was found that the Fourier transform method requires intensive

operator interaction, and its ability to discern a sum of two or more

exponentials was found to be questionable. The method of moments

couid resolve single and double exponential fluorescence decays.

However. extension to three exponential decays was difficult. The

technique which was found to give the best results is the iterative

reconvoiution method, which was also found to be most satisfactory by

O'Conner e^ al. [5]. The ability to successfully deconvolute data

containing up to three different fluoresence decays, provided that

their lifetimes differ by a factor of two ajid that they each have

contributions greater than 1.5%, has l een established [6]. Results

were also tested experimentally for two and three component

non-interacting dye mixtures [6,7]. The iterative reconvolution

method assumes that the fluorescence decay F(t) from the sample is a

sum of exponential terms corresponding to emission from N individuai

fluorophores:

N

F(t) = ^ A..exp(-t/T.) . (3-11)

i = l

In general, A. is a function of wavelength. Combining this with

Equation (3-10) gives the fitting function

N

l''(t) = Y ^i'í I(T)-exp M"(t) = > A.'j I(T).exp -(t-T.)/T. —00 '-

i = l

dT (.3-12)

where A and T. are tiie parameters to be adjusted until a best fit to 1 1

the measured signal M(t) is found. This best fit is obtained wiien tiie

error sum \ is minimized, wliere

V h s j = i

) - M"(t.)

36

(3-13)

1

and n is the totai number of channels. All the 512 channels stored by

the digitizer can be used in the deconvolution routine, however, to

enhance the speed of computation, only 128 channels are generally

used. It was found that by lining up the fitting function, M°(t) with

the measured signal M(t) at 10% of the peak along the rising edge, a

large improvement in the fit results [4]. This lineup is

experimentally valid since there is a difference in the optical path

length between the acquisition of the instrument response and the

acquisition of the measured signal due to the insertion of the K399

filter and the changing of the cube from the beam splitter to the

dichroic mirror

2 The weight a. is the square of the uncertainty in the

th measurement of the i channel and is given by

a.2 = a ^ + C.M(t.) + 1 o 1

clM(t) dt " . ' ^

a^M(t) . At^ 2

dt 2 (3-14)

Here a is the baseline noise (typically 0.01 mV ). C.M(t ) the o ^

statistical counting error (^ 1 mV ), a^ the sample variance of the

time axis jitter caused by small fluctuations in tlie trigger pulse and

signal pulse rise time (a^ ~ 100 ps, depending on the number of pulses

averaged), and At {X 0.25 channels) thc fractionai channel shift used

in the time axis lineup routine. For a typical 100 mV pulse. tlie 2

maximum value for tiic third term in Equation (3-14) is 10 mV . 2

Average values for the last two terms in this equation are 2 mV and

37 2

0.25 mV . Extensive work was done in determining the uncertainties 2 ,

C7. . and more work is necessary to better understand how changes in

experimental conditions affect the various terms in Equation (3-14).

It is the author's opinion that the time shift which occurs as the

result of path length difference between the acquisition of the

instrument response and the fluorescence signal shouid be taken care

of experimentaily and not in the fitting routine. Information on very

fast kinetics may be iost in the alignment of the fitting function

with the measured signal. To the best of our knowledge, we are the

first to inciude the third and fourth terms in the uncertainty which

deal with the jitter and time axis lineup.

2 Minimizing the \ error sum is an iterative process, which

invoives incrementing the parameters A. and T. and thus successively

2 reducing \ . The method empioyed [8,9] is a combination of a gradient

2 search along the \ hypersurface and linearization of the fitting

function. Wlien far from the surface minimum, the gradient plays the

dominant roie in determining the increment chcmges. As the search

cioses on the minimum, the linearization of the fitting function

dominates. The parameter increments are given by

2n

k=l

where B. = (A.,T.), and /3, and a., are given by the relations: j ^ i i ' k jk

. , a. k 1 = 1 1

^ , c?M''(t.) aM^'^t.) r T Y 1 e" i' e'' i^ , ^ - f~ ,,-.

"jk = 2 ~ • —ôã: ôB— • h ^ jk • '^ '^

38

M^(t.) is the best fit from the preceeding iterations, A is a

weighting factor and ô ^ tlie Kronecker delta. The weighting factor

determines the relative contribution to the increments from tiie

gradient search on the hypersurface to the linearization of the

fitting function. This basic iterative reconvolution method was

developed by Marquardt [9] and has been foilowed by the author in a

new IBM/AT computer to be used with the single photon counting system

currently under developement.

The operator begins by assuming a given number of exponential

decays (1,2 or 3) and giving the computer some starting values for the

A.'s and T.'S. These input parameters are used to construct the first

fitting function M (t) by convoluting F(t) with the measured

instrument response. The program determines the change in the

parajTieters based on the above equations and forms a new set of

2 parameters. A new fitting function is then determined. The \ is

checked and if there is an improvement, tlie new values for the

parameters are kept and the proceedure again determines new ÔB . xJ

2 vaiues. If the vaiue of \ is not improved, X is incremented by an

2 order of magnitude and ôB . values are again found. Once \ improves

from the last best fit, A is reset (to 0.001). This interative

reconvolution process continues until the improvement in \ is iess

than 3% of the last best fit, at which point tiie program terminates

and displays the best fit curve with thc measured signal along with

tlie final parameters and residuals.

Time-resolved fluorescence emission spectra are obtained in 10 nm

steps by scanning the emission monochromator of the MI'V3 from 100 to

39

7.50 nm and acquiring the fluorescence decay at each wavelength. This

takes about 10 seconds (for 64 pulses, including data storage) at each

wavelength or 5 minutes total. The data reduction yields the muitipie

decay times, component spectra, and relative intensities and takes

about 4 minutes at one wavelength when fitting with a two exponential

fitting function. The total data reduction time for ail wavelengths

is 1-2 hours. If a three exponential fit is done, the total data

reduction time may increase to over 6 hours time. Programs have been

written to reduce an entire data set over an evenings time thus

increasing the efficiency of the system.

Conventionai spectra are obtained with c.w. illumination and

continuously scanning the emission monochromator. This takes about 60

seconds (depending on the scan speed selected by the operator, in this

case it is 10 nm/s), including correction for optical system response

and smoothing. Equivalent spectra are obtained by integrating tlie

time-resolved results.

For routine work, a tungsten lamp is used to determine the

spectral response of the system. The lamp is cross calibrated with a

standard lamp of a known spectral output (Eppley seriai no. ES-8319,

traceable to National Bureau of Standards reference standards, serial

numbers: F-94 and EV-8). Tliis is done by measuring the raw spectrum,

which includes the system spectral response, of both tlie tungsten and

standard lamps. The ratio of the raw tunsten spectrum T (A) to the

raw standard lamp spectrum E (A) is independent of the systcm

response. and when multiplied witli the known standard lamp spcctrum

E(A). results in the true tungsten spectrum T(A):

40

^^^^ = rJX) • ^(^) • (3-18)

The determination of the relative spectral response S(A) of a chosen

optical configuration (objective, dichroic mirror. filter. apertures.

etc.) is done routinely by measuring the raw tungsten spectrum T (A)

and dividing by the true tungsten spectrum T(A):

T^(^)

^ ^ = TJX) • (3-19)

The corrected fluorescence emission spectrum R(A) from a specimen

is then determined by taking the ratio

R (A)

^ ^ = STÃT ' ^ " ^

where R (A) is the raw fluorescence spectrum emitted from the

specimen. The above calibration procedures are built into the

software. In this way. normaiized fluorescence spectra are obtained

independently of the particular optical components chosen. The system

spectrai response corresponding to the use of a given objective,

measuring diaphragm. exit slit, and filter combination is given in

Figure 3-3. and a typicai normalized raw c.w. spectrum is given in

Figure 3-4. The upper curve of Figure 3-5 is the result of dividing

the raw spectrum by the system spectral response giving the spectrally

corrected fiuorescence emission spectrum of the specimen. The lower

curve is the smoothed version of the corrected curve; the smoothing is

done by a running average routine where the averaging window is

defined by the user along with tiie region over which smootinng is to

occur. The unsmoothed. spectrally corrected version of the curve is

stored on disk.

41

Fo test ail of the above mentioned procedures experimentally. we

measured a number of solutions containing one or more non-interacting

components with known spectra and decay times. Figure .3-6 is the

emitted fluorescence of the fluorophore POPOP in ethanol. The

measured decay time is within 10 ps of the accepted value of 1..35 ns

[10]. Many other samples were tested with similar results [6].

In order to determine the temporai sensitivity of the system, we

acquired tlie instrument response I(t) twice, once as the instrument

response and once as the assumed fluorescence (see Figure 3-7). In

theory, the decay time determined from such an acquisition should be

zero. We found that eui assumed, monoexponentiai fitting function gave

a good fit with a decay time of 7.3 ps. This result is nearly the same

as the uncertainty determined with POPOP and other solutions

(anthracene, NADH, Rhodamine etc). After determining the accuracy

and precision of our system with monoexponential decays, we went on to

biexponentiai systems. The fluorescence emitted from these binary,

non-interacting systems can be treated as a two-component exponential

function as given in Equation (3-11). Figure 3-8 represents the

result of a complete time-resolved data acquisition and reduction of a

mixture of POPOP and anthracene in ethanol. The top two graphs are

the A-coefficient spectra and a fitted fluorescence decay curve

derived from the separate solutions. The bottom graph represents the

resolved spectra and one of the corresponding decay curves acquired

from tlie mixture. It is important to note that we are able to resolve

with confidence a two-component non-interacting mixture into its

component spectra ly fitting witii a two-exponential decay function.

42

Figure 3-9 gives the A^T product spectra (equivalent to the puise

integrated spectra) of seperate anthracene and POPOP solutions. which

is aiso equivalent to the c.w. emission spectra. The bottom curve

shows the time-resolved spectra of the components along with the

mixture. Comparison of Figures 3-8 and 3-9 show that the A'T spectra

are a much more reliable means of determining the component spectra of

a mixture of non-interacting fluorophores than just the A-coefficient

spectra. This is obvious from the spurrious spike at 400 nm of the

resolved A-coefficient spectrum for POPOP in Figure 3-8. Notice also

that the structure in the anthracene spectra is better resolved in the

bottom of Figure 3-9.

When trying to fit a two-exponential system with a three

exponential fitting function, one of two things occur: either the fit

resuits in two dominant terms of high percentage and one of the

contributions is negligible, or the fit results in a redundancy of one

of the components by showing equal decay times. In other words, when

fitting with a function containing more terms than necessary, a

redundancy or negligible contribution of one or more of the components

results. For reliable results. the actual decay times have to differ

by a factor of two and the percent contribution from any component has

to be 10-20% or above.

The percent contributions of the individual components are given

by

A..T. P. = ' ' (3-21) ^ 2 A..T.

1 1 1

43

and represent the fractional fluorescence intensity of the i^^

component. The total fluorescence spectrum emitted from the specimen

is given by the sum of the components:

00

F(A) = / F(t,A) dt t=0

00

= 2 i(' ) / exp(-t/T^) dt (3-22) t=0

1

wlîich may he w r i t t e n as

F(^) = 2 ^i(^)*'^i • (3-23) i

Hence, the fluorescence intensity from the i component at A is just

F.(A) = A.(A).T. . (3-24)

This is the basis for plotting A.'T. as a function of A and 1 1

designating this the i component spectrum.

If these components are truely pure, their lifetimes will be

independent of emission wavelength as alluded to above and the

A-coefficients will vary correspondingly with emission wavelength. It

should aiso be noted that the A-coefficients are corrected for the

system spectral response. A.^T. spectra often times produce better

resuits than just A-coefficient spectra alone, especially if the

components are weak. The fitting routine compensates a large

estimation of one parameter by reducing the value of the other (A's

and T'S) for a given term in the fitting function (Equation 3-12)

resulting in an accurate result for the product, A..T, . For some

samples, sucii as coal or crude oil, tiiere are probably more than three

components making up tlie fluorescencc emitted. Hence, our thrce

component model oftcntimes gives decay times wiiich are not independent

44

of emission wavelength. We iiave found that in these circumstances,

the A.(A).T.(A) product, which corresponds to the c.w. spectrum of

this "component" (now not really well defined), produces consistent

results and sometimes the spectrum can be matched to a known spectrum

(see crude oii section).

The accuracy of the fit can be assessed from the residuals and

2 2 the reduced \^ [4], where \ ~ 1 indicates a good fit:

\ 2 ^

2

V n-j

(3-25)

where j is the number of fitting parameters to fit n channels. (n-j)

is the total number of degrees of freedom. The reduced chi squared

value of one for a good fit can be understood if one considers that

the difference between the fitted and real data should be on the order

of the uncertainty of the data, hence, the ratio of the difference and

the uncertainty shouid be one. Some of the fits presented here have

2 X values which are much less than one. This dilemma magnifies the

V

difficuity in accurately assessing tiie experimental uncertainty. It

2 should be noted that the value of \ is not important in any specific

fit as its reiative change between one, two and three exponential

2 fits. If the value of \ does not change by more than a factor of

2 three (i.e., a change in \ from 6 to 1.8) when going from a singie

to a double or a double to a triple exponential fit, then the lower

number of exponentials can be used to determine the number of

non-interacting fluorophores making up tlie sample fluorescence. This

conclusion comes from three years of experience and our work on

non-interacting mixtures.

45

The residuals, which are sho\m in the plotted fits, are defined

as the difference between the best fit M"(t) and the measured signai

M(t). Ideally, the residuals should be random and this has in fact

been observed in testing of the fitting routine with synthetic

functions when random Gaussian noise was added. However, real data

and the corresponding fits give a slight rolling in the residuals.

This rolling is most probably due to radio frequency pickup by the

eiectronics. The source of this r.f. signal is the spark gap of the

nitrogen iaser; this is verified because the use of double sheilded

coaxial cables and sheilding the MCP with aluminum foil have reduced

the effect. Increasing the number of signal averaged pulses aiso

reduces the rolling in the residuals. The quality of the fit can be

ascertained by looking at the residuais themselves. Experience shows

that the maximum residual should be less than 2% of the emitted

fluorescence peak pulse value.

The mean lifetime presented in the fitted plots is defined as:

1 P..T,

" ~ " 100

1 L (3-26)

and has proven to be useful on occasion for comparison.

46

(a) CONVOLUTION

to

TiME

(b) CONVOLUTION

>-

H-

CO

Z

TIME

Figure 3-1. Graphical description of the convolution process. (a) F(t-T) is the resuit of convôiuting a function F(t) with a delta function occurring at t=T. (b) If one has a continuous distribution of delta functions given by L(t) then convoluting F(t) with L(t) results in F'(t). F'(t) has been normalized to the amplitude of L(t) for comparison purposes.

47

TOTAL INSTRUMENT RESPONSE

37 -

30

Z) Q.

u

22

14

- 1

RISE TIME 0.64 ns

FALL TIME 1.05 ns

1/2 WIDTH 0.99 ns

1

16 20

NANOSECONDS

Figure .3-2. Total instrument response of the system. This curve is obtained by substituting a reflector (a ô-function equivalent) for the sample. The response is a convolution of the single photon response of the apparatus with the laser puise form.

48

>-t—

Z

Q LU

M I

<

ûí o z

0.8

0 .6

0 .4

0 . 0

SYSTEM SPECTRAL RESPONSE

1.0 F

0.2 -

400 500 600 700 800

WAVELENGTH (nm)

Figure 3-3. Total system spectral response S(A) for a given optical configuration (20X Rolyn objective, K399 high pass filter. 0.5 mm exit siit. 0.3 mm measuring diaphragm). This curve includes the MCP spectral response.

49

R F I E L N L A T I V

U T 0 E R N E S S I

E C T E Y N C E

o.a

0.6 -

0 .4

0 .2

400 500 500 700

WAVELENGTH /nm

800

Figure 3-4. Raw c.w. spectru R (A) of a green-yeilow resinite coai maceral.

50

24-MAR-87 09:32:16

Vp (raw) -2774 mv

PEAK AT: 519nm Q : 0.48 Qm: 1.04

AHEA PAHAMETERS: V: 21 8: 29 G: 25 Y: 17 0: 7 H: 0

SAMPLE INFORMATION

GHEEN-YELLOW HESINITE

R E L A T I V E

U T

R E L U A T I V E

F I L N T E 0

R N E S

0.8 -

0.5

0.4

0.2

400

0.8 -

0.5 -

0.4 -

0.2 -

40Q

500 600 700 WAVELENGTH /nm

800

500 500 700

WAVELENGTH /nm 800

Figure 3-5. Spectrally corrected c.w. spectrum of the same green-yellow resinite maceral as in Figure 3-4. The top curve represents the corrected spectrum and the bottom curve is the smoothed version. The spectral parameters are defined as:

V = Peak Voltage of picoammeter P

Q = Intensity at 650 nm / Intensity at 500 nm Q = Intensity at the spectral peak / Intensity at 500 nm m

The area parameters represent the area under the curve for the following wavelength intervals:

V = 420 - 480 nm B = 480 - 540 nm G = 540 - 600 nm Y = 600 - 660 nm 0 = 660 - 720 nm R = 720 - 780 nm

51

> E

33

30

23

15

a 12 16

NANOSECONDS

20

1 i -ocT-es I 1 : ; 8 : M 7

^ e í î i = '527 nm

l ^ - 0 .527 [0.527]

Q^- 1.421 mV a = 0.249 mV P o

64 PULSES

A l =

P l =

72.5

100

± 0.5

± 0

m V

%

T1 - 1.345 ± 0.011 n s

T^ = 1.346 ± 0.011 n s

1.1

- 1 . 1 RESIDUALS

POPOP IN ETHANOL

Figure 3-6. Emitted fluorescence pulse of p-bis 2-(5-Phenyloxazolyl) benzene (or POPOP) in ethanol solvent fittcd with a mono-exponential decay function. The resulting lifetime of 1.346 ns is very close to the 1..3.5 ns determined by Lakowicz [10]. whose value is judged to be within ± 0.2 ns. The accuracy of our system is better than ± 100 ps.

52

166

- 93 > E

3 CL y—

Z) O a. o 5

69 -

46

23

-1 8 12

NANOSECONDS 16 20

A^ .i - 3 7 3 nm

X^ = 0 . 7 1 3

64 PULSES

Ai - 4782 mV

Pl = 100

T^ = 0.073 ns

0.8

^ 0.0

-0.8 RESIDUALS

Figure 3-7. Self-deconvoluted instrument response assuming a mono-exponential decay. The instrument function was acquired twice, one being deconvoiuted with the o ther . The r e s u l t i n g "decay time" of 73 ps r ep resen t s the s e n s i t i v i t y of the system.

53

AN HRACENE »-COEFriCIeHT SPECTRUH

â:.m AHn 47B

UAVELENOIH Cnmi

POPOP »-COEFriCIENT SPECTRUM

iM 4nn 47B UAVELENOTH tr,i.>

mV

ANTHRACENE/POPOP HIX A-COEE. SPECTRUH

mV

_--_-;

\ MFASUREO \ Fir \ : FITTrn \ : REGION

?0 30 40

NANOSECONOS

RESIDUALS

nr.ASUREO

12 16 NANOSECONOS

RESIDUALS

MFASUHED

FiTTFD REOIOM

0 17. 10 NANOSECON S

mV

4pn *nn 470

UAVELEHOTH <r.«)

1.5

0

-1.5 RESIDUALS

Figure 3-8. Time-resolved A-coefficient spectra and corresponding decay curves for separate anthracene (T = 4.18 ns) and POPOP (r = 1.35 ns) solutions. The lower graph shows the precision of our technique in resolving components by their decay times (TI = 1.22 ns and r^ = 4.20 ns) and A-coefficient spectra. (No energy transfer takes place in this case.)

54

í iMl.SMON SPLCíRA

1 —

n _

l U

•—

_

111

A

1

l U

OC

] .0

O.U

0.6

0.'\

0.2

0.0

1.0

o.u

0.6

0.4

0.2

0.0

1.0

-.

/ /

/ >

1

1 1

A t

1 1 1

•

'• •.

"••

«

i 1 1

I 'OPOI ' •N

" • N.

N • » .

N •^ \ N \

V

1 1 1 1 1

ANTI iRACLNL

— •

• * • • • • • . .

•-•.. 1 1 1 1 7

o.u

POPOI ' •)• ANTI-IKACLNG

0.0

'.. ANTI-IRACtNE •• •. ••• ••••..

4 00 420 4 40 4 60

1

•

1

•

<>

1

DLCAY T1ML5

"C, = 1.29 nz

• • •

•

1 1 1 1

T. i - 4 .22 n.-.

• •

1 1 1 1

•

1

•

1

1 1

1

L

L

t

1»

_

-

-

T(

"C 1 - 1.27 iiG

. j * • _

11.1= 4.11 ns

I 1 i 1 \-

1.4

1.3

H 1.2

\A

\:i

).o

1/1

o > _ O \^j U l

\rt

O - T

-" i.4

1.3

1.2

4.4

4.2

4.0

3.0

H 3.6

3.4 400 4 20 440 4 60

<

WAVtLLNGTH (nm)

Figurc 3-9. Timc-resolved fluoresccnce si)cctra and dccay timcs for an anthraccnc-POPOP mixturc as a function of cmission wavclcngth. Thc top iialf of the figure shows thc rcsults for thc purc subsLanccs. the bottom half for the resolution of the mixturc. Thesc specLra (cxccpL for thc POPOP + anthracenc mixturc) arc dcLermincd by pIoLLing Lhc A -T producLs at 10 nm intcrvals. Thc spcctrum of thc mixLurc was i i

delermined by inlcgrating cach cmiHcd f Ivioicsccncc pulsc ovcr Limc and corrccting for thc systcm .speclral rosponse.

55

Literature Ci ted

1. Grinvald. A., and Steinberg. I.Z.. Biochemistry. Vol. 13. p. 5170. 1974.

2. Grinvald. A.. and Steinberg. I.Z.. "On the Analysis of Fluorescence Decay Kinetics by the Method of Least-Squares." Anal. Biochem.. Yol. 54. pp. 583-598. 1974.

3. Cundall. R.B.. and Dale. R.E., Time Resolved Fluorescence Spectroscopy in Biochemistry and Biology. Plenum Press. New York, 1983.

4. Sullivan, G.W.. "Time Resolved Fluorescence Microscopy by Pulsed Laser." M.S. Thesis, Southern Illinois University, 1983.

5. O'Conner, D.V.. and Andre. J.C.. "Deconvolution of Fluorescence Decay Curves: A Criticai Comparison of Techniques," J. Phys. Chem.. Vol. 83. pp. 1333-1343, 1979.

6. Gangopadhyay, S., Pleil. M.W.. and Borst. W.L.. "Resolution of Interacting and Non-Interacting Fluorophore Mixtures by Laser Induced Fluorescence Spectroscopy." submitted to J. of Luminescence, Feb., 1987.

7. Borst. W.L.. Gangopadhyay. S.. Pleil, M.W.. "Fast Analog Technique for Determining Fluorescence Lifetimes of Multicomponent Materials by Pulsed Laser." Presented the in Sixth SPIE Conference on Fluorescence Detection. Vol. 743. January 15-16, 1987.

8. Bevington, P.R., Data Reduction and Error Analysis for the Physicai Sciences, McGraw-Hill, New York, 1969.

9. Marquardt, D.W., "An Algorithm for Least-Squares Estimation of Non-Linear Parameters," J. Soc. Ind. Appl. Math.. Vol. 11. No. 2, pp. 431-441, June. 1963.

10. Lakowicz. J.R., Principles of Fluorescence Spectroscopy. Plenum Press, New York. 1983.

aiAPTER IV

APPLICATION TO SCINTILLATORS

Plastic and acrylic scintillators are extensively used in

particie detection because of their short fluorescence decay times.

high quantum yields and the ease in which they Ccui be shaped aná

fabricated. The transfer of kinetic energy of the detected particle

from the soivent to the primary scintillant foilows one of many

possibie paths. Since the emitted fluorescence pulse shape is

dependent not only on the scintillator but also on the source of

excitation [1-6], a detailed understanding of the energy transfer and

the contribution to the overall pulse shape from each stage of the

process is important. The overall pulse shape has been catalogued for

a number of particle, scintillator, and detector combinations, and

estimates of the contributions from the individuai energy transfer

stages have been made by various fitting techniques [1,3,7-12]. We

have measured the contribution from the last energy transfer stage of

seven common scintiliators, namely NE102A, NE104, Pilot-U, PSIO,

PS15A, BBQ90 and BBOT150, by using puised near ultraviolet laser

excitation. Our results conclusively give the fluorescence lifetime

of the final stage soiute together with its corresponding fluorescence

spectra. The measured fluorescence is caused by the transition from

the first excited singlet state to the ground state of the primary

scintiilant molecuie situated within the plastic or acrylic

scintillator enviroment. Accurate determination of these fluorescence

56

57

lifetimes effectively eliminates one fitting parameter in the overail

puise shape function. Previous work has not provided the fluorescence

lifetimes of these moiecules within the scintillators themselves.

There has been much discussion concerning fitting functions that

may describe the fluorescence emitted (or prompt response) from

unitary, binary, ternary, thick or thin scintillators following

nuciear particie excitation. The simplest system consists of only the

primary scintillant and is described by most investigators with a

unitary function (single exponentiai) [1-3,8,13]

i(t) = A exp[-t/T] (4_1)

Here it is assumed that the excitation has a negligible duration and

that the formation time of the first excited state, X as defined by

Birks [8,13], which includes ionisation, excitation, ion recombination

and vibrationai releixation, is very short compared to the fluorescence

decay time r of the primary scintillant. Aiso. the scintillator is

assumed to be transparent to the emitted fluorescence. Wlien an energy

transfer process occurs between a solvent and primary solute (or

scintillant). as is the case with binary scintillator systems. the

function which best describes the fluorescence output becomes more

complex and may be dependent on sample thickness [1]. If the system

is a liquid, it can be described well with Stern-Voimer kinetics

[1,8.10], and the emitted scintillation puise can be expressed by the

relation

i(t) a r

^ - • ' i

exp(-t/T) - exp(-t/T.) (4-2)

where r again is the fluorescence decay time of the primary solute and

T is the decay time of the solvent in the prescence of the

58

fluorescing soiute [1.2,8,10]. The energy transfer for this binary

system can be described according to Birks [8,13] as

ly- . Ix ^ ly . X'* . (4-3)

where Y represents the solvent, X the solute, and the superscripts 1

£md X represent the singlet and excited states, repectively. The

environment for binary plastic and acrylic soiid soiutions is

different from liquid solutions. Another relation is needed to

describe the emitted fluorescence. (The plastic and acrylic

scintillators are hardened liguid solutions, hence the term "solid

solution.") The binary soiid soiutions foliow Forster kinetics, which

according to Birks [5,8] results in the relation

1/2" i(t) a exp[-t/T] - exp -t/T. - 2T7(t/T ) (4-4)

where T. and r are the decay times of the solvent and soiute.

respectively, and T] is dependent on the soivent concentration. Both

Equations (4-2) and (4-4) have been used to describe plastic

scintiliator solutions with varied success [1,8,10]. In order to get

the fluorescence into a range suitable for certain photodetectors,

waveiength shifters are added to the scintillator, thus forming a

ternary soiution. These ternary systems are the least understood.

Because of the complexity of the emitted fluorescence pulse, many

different fitting functions have been tried. Two functions

successfuily employed to fit fluorescence pulses from binary and

ternary piastic solutions involve a Gaussian function muitiplied with

either a unitary fuction such as Equation (4-1) or a function similar

to that in Equation (4-2). resulting in the overall fitting functions

[1.2.14-16]:

59

i(t) = f^(t) exp[-t/T] (4_5)

and

i(t) = f^^t^^e"'/"" - e"'/''^] . (4-6)

The function f^^t) of the binary soiution in Equation (4-5) is a

Gaussian characterized by a standard deviation a, which represents the

energy transfer rate from the detected nuclear particie to the opticai

levels (or first excited state) of the solute molecule. and r is the

fluorescence decay time of the primary solute. For the ternary

piastics, fp(t) represents a Gaussian describing the rate of energy

transfer from the detected radiation to the primary solute. The

exponentiai terms in Equation (4-6) describe the energy transfer

process from the primary soiute to the wavelength shifter and the

final emission of the light from the wavelength shifter. T. is the

decay time of the solute in the presence of the wavelength shifter,

and T is the fiuorescence decay time of the wavelength shifter.

The parameter common to all of the above relations is T, the

fluorescence decay time of the primary scintillant in the presence of

the other scintillator components. The decay time r has in the past

been determined by muitpie parameter fittng of the puise emitted from

the scintillator after T-ray or nuclear particle excitation, using the

above relations convoluted with the instrument response of the

particuiar experiment. By fitting the emitted pulses with a large

number of parameters (which is necessary for high energy excitation)

previous investigators did not determine the real fluorescence decay

time T precisely nor the actual contribuLion of Lhis last stage of the

energy transfer process. We have measured the fluorescence of the

60

primary scintillant directly using near ultraviolet excitation by

puised laser, thereby eliminating the energy transfer mechanisms

present with high energy excitation but maintaining the environment of

the primary scintillant. Hence, we were able to determine r much more

accurately, and the values presented here can be used in future

studies for eliminating one important fitting parameter.

The NEÍ02A, NE104, and Pilot-U plastic scintillators were

supplied by Thorn EMI Gencom Inc., whereas the PSIO. PS15A, BBOT150

and BBQ90 acrylic scintillators were supplied by Polycast Technology

Corporation. The fiuorescence was monitored from a very small area on

the surface of the specimen. Hence, effects due to self absorption

and sample thickness did not have to be considered.

Figure 4-1 shows a typical decay piot for Pilot-U. Because we

are exciting the first excited singlet state of the primary

scintillant direcLiy, we are justified in fitting the data with a

monoexponential decay function:

A(t) = A^exp(-t/T) , (4-7)

where A(t) and A are the fluorescence intesities at time t and t=0, ^ ' o

respectively, and T is the characteristic fluorescence decay time of

the primary scintillant in its scintillator environment. This

function is the same as Equation (4-1) for unitary scinti1lators if

one also assumes, as is the case here, no self-absorption in the

samples. The accuracy of the fit was assessed from the residual and

2 2 reduced \ , where \ < l and peak residuals of lcss than 2% indicates

a good fit. The results for the other samples are given in Table 4-1.

all of which could be fitted accuratcly with a monoexponential dccay.

61

We feei that the decay times are correct within 100 ps. as was

verified by measuring the lifetimes of several substances. namely rose

bengal. PBO, anthracene, fluorescein and others, and comparing our

resuits with the known iifetimes. The reproducibility of our

experiment is weil within 100 ps. The measurements of the dependence

of the decay time on emission wavelength showed that ali decay times

(except for BBQ90) are independent of wavelength. As an example.

Figure 4-2 shows the decay time for NEÍ04 and BBQ90 as a function of

emission wavelength. NE104 is typical of the other samples showing a

tight, random distribution of lifetimes about the mean for all

waveiengths.

Figure 4-3 give the fluorescence spectra obtained with continuous

wave excitation at 365 nm. The scintillators PSiO and PS1.5A have the

same fluoresence spectra, which closely resemble those from NE102A and

NEÍ04. According to Polycast Technology Corporation, the fluorescence

we see in the two PS sampies is caused by POPOP. This is apparent

from the emission spectra. Thorn EMI Gencom Inc. was not able to

provide us with information on the primary scintillant (fluorescent

component) for "proprietory" reasons. It seems. however, that the

fluorescence detected in the NE102A and NE104 samples also is from

POPOP because of the strong resemblance of the emission spectra with

those of POPOP in solution [19]. The emission spectrum for BBQ90

matches that for BBQ obtained by Auronet [9]. and BB0T1.50 matches that

of BBOT in ethanoi [19].

The optical levels investigated in the present work are those

used in nuclear particle detection. while ihe excitaLion process is

62

different. The emission spectra of the plastic scintiilators NE102A.

NE104 and Pilot-U matched the manufacturers' spectra well. whiie the

emission spectra for the acrylic scintillators PSIO. PS15A, BBQ90. and

BB0T150 matched the spectra of POPOP in the case of the PS samples,

BBOT in the case of the BB0T150 sample and BBQ in the case of the

BBQ90 sample. As a resuit of the good match of these emission

spectra, we know that the decay times determined at those wavelengths

(corresponding to the first singlet to ground state transitions) are

those of the last step in the total energy transfer process when the

scintiliator is subjected to high energy excitation.

Our measured fluorescence decay times r are different from those

given by others (Table 4-2). The reason for these differences is in

the definition of the vaiues iisted: The decay times quoted are for

the entire pulse shape from a particular experiment, including the

energy transfer mechanism. These are not to be confused with r in the

above reiations. The uncertainty resulting from the use of a iarge

number of fitting parameters may be the reason why the investigators

who used gamma particles or high energy electrons do not give the r

values for the components of interest. Only Waynant [3] gives the

value of T. Waynant used a nitrogen laser in his determination of the

decay time of NEÍ02 and found a value of 2 ns by fitting a photograph

of a sampling osciiloscope output with a monoexponentiai function.

This value is twice as large as ours. This discrepancy is attributed

to the relative inaccuracy of Waynant's method. (NE102A and NE102 are

essentially the same type of scinti1lator; they could not be

differentiated by spectra nor by decay times.)

63

Our experience with dye soiutions teils us that the environment

of the fiuorophore can strongly influence its decay time because of a

variety of concentration and solvent effects. Birks also found

variations in the decay time of plastic scintillators as a function of

the primary scintillant concentration and composition of the solvent

(polyvinyltoluene or poiysterene; see reference [3]). These

environmental differences probably cause the variations in the decay

times measured for the various POPOP scintiilators (PSIO, PS1.5A,

NE102A, and NE104). The differences in the BB0T150 and BBQ90 decay

times as compared with the corresponding liquid soiutions also can be

attributed to environmental differences. Hence, it is important that

the fluorescence decay time of the emitting soiute be known within the

scintiliator environment. This has been accomplished in the present

work by excitation with near ultraviolet laser pulses, thus

eliminating the energy transfer contributions.

We are confident that the lifetimes determined here are those

designated as r in ail of the above relations and hence are important

in previous and future data on emitted pulses from particle detectors.

These lifetimes can be used to better understand the complex

mechanisms involved in the energy transfer of kinetic energy from the

nuclear particle to the excited singlet state of the primary

scintiilant. This may facilitate the development of better detectors

and improve the future analysis of data.

A paper on the scintillator work presented here has been

submitted to, and acccpted by Nuclear Instruments and Methods [21].

64

Table 4-1.

Scintillator

NE102A NE104 Pilot-U PSIO PS15A BBQ 90 BB0T150

Scintillator Fluorescence

Wavelength Range

.370 - 510 385 - 505 370 - 480 400 - 500 400 - 560 420 - 600 400 - 560

Lifetimes

T ± 2'0 m 1.10 ± 0.15

0.81 ± 0.03 0..58 ± 0.02 1.32 ± 0.17 1.09 ± 0.05 8.81 ± 0.81 ^ 1.23 ± 0.13

^BBQ90 exhibited a dependence of the lifetime on emission wavelength.

65

Table 4-2. Summary of

Samp1e

NE102 NEÍ04 Piiot-U PSIO PS15A BBQ 90 BB0TÍ50 BBOT in cyclohexane

(l.Og/1 concentration) POPOP in benzene

(O.ig/1 concentration) POPOP in cyciohexane

(O.lg/1 concentration)

Effective Decay Times Determined by Others

Effective decay times in ns

2.4[2],2.0[3],2.5[8],2.2[i4],2.4[17] 2.0[8],1.74[14],1.9[17] 1.36[2],1.36[10],1.38[14],1.3[17] 3.9[18] 3.9[18] 20.0[18] 28.0[18] 1.1[19]

1.26[19]

1.50[19]

Except for references [3,19], all times were determined using high energy excitation and various fitting techniques. The technique used by Waynant [3] is described in our conclusion. Berlman's fluorescence decay times were obtained by hydrogen flashlamp excitation and direct measurement of the photomultiplier output current pulse with a sampling oscilloscope [19]. The acquired signal was fitted in a manner analogous to ours, resuiting in the given decay times. Beriman's vaiues are for pure solutions at the specified concentrations.

66

1 -3 O. I -Z)

o CL

> Lli)

u

140

12

8 4

56

28

- I 0

F IT

P ILOT-U

MEASURED

2 4 6

NANOSECONDS

8 0

0 .6

^ o

- 0 . 6

- - • I

j-s^ : t^'^^m\^^ :::

• •

RESIDUALS

WAVELENGTH: 4 3 0 nm

1024 PULSES

A = 411 mV

r = 0 .583 + 0.003 ns

X^= 1.02

Figure 4-1. Emitted fluorescence pulse from a Pilot-U scintillator fitted with a monoexponentiai decay function. The resulting lifetime of 0.583 ns has a statistical error of 0.003 ns. The absolute accuracy is 100 ps as determined by direct comparison of standards.

67

co c:

LU

>-< O LLI Q

10 •

9 -

8 •

7 J

6 -

BBQ 9 0 . . • • • • • , - ^ - - . - - - ^ , . - - . . -

•

•

Tm «= 8.81 ns

1 1 1 i 1 1 1 1 1 í 1 1

i .o 4

0.9

0.8

0.7

J L » t 1

NE 104

Tm = 0.82 ns

j _ j I 1 1 1

360 4 0 0 4 4 0 480 520 560 600

WAVELENGTH (nm)

Figure 4-2. Dependence of fluourescence decay time on emission wavelength. BBQ90 is the only sample which showed a lifetime dependence on emission wavelength. The NE104 scintillator is typical of all other samples showing a tight, random distribution of lifetimes about the mean.

68

7,

O LU

M < <

o

J60 600 640

Figure 4-3. Fluorescence spectra of the seven scintiliators obtained with continuous wave excitation. The excitation source is the 365 nm line of a high pressure mercury arc lamp. The similarities between the two NE and two PS samples exist because POPOP is the common fluorophore.

69

Literature Cited

i. M. Moszynski and B. Bengston, Nucl. Instr. and Meth. vol. 158 p.Í. 1979.

2. G.F Knoll, Radiation Detection and Measurement. John Wiley, New York, 19797^

3. R.W. Waynant and R.C. Eiton, in: Organic Scintillators and Liquid Scintiilation Counting. eds.. D.L. Horrocks and C.-Tz. Peng Academic Press. New York. 1971.

4. G. D'Agostini. G. Marini, B. Martellotti. F. Mass. A. Rambaldi and A. Sclubba. Nucl. Instr. and Meth. vol. 185. p. 49, 1981

5. J.B. Birks and A.J.W. Cameron. Proc. Phys. Soc. B. vol. 69. p. .593, 1956.

6. S. Lipsky and M. Burton, J. Qiem. Phys. 31 (5), p. 1221, 1959.

7. L.M. Bollinger and G.E. Thomas, Rev. Sci. Instr., vol. .32, p. 1044. 1961.

8. J.B. Birks and R.W. Pringle, Proc. R. Soc. Eden., vol. 70, p. 233, 1971/72.

9. C. Auronet, J. Blumenfeld, B. Boso, M. Bourdinaud, P. Evard, C. Jeanney and C. Laford, Nucl. Instr. and Meth., vol. 169. p. 57. 1980.

10. F.J. Lynch. lEEE Trans. Nucl. Sci. NS-22 p. 58. 1075.

11. I.B. Berlman. J. Qiem. Phys. 33 (4). p. 1124. 1960.

12. P.B. Lyons and J. Stevens. Nucl. Instr. and Meth. vol. 114. p. 313. 1972.

13. J.B. Birks. The Theory and Practice of Scintillation Counting. Pergamon. Oxford. 1964.

14. M. Moszynski and B. Bengston. Nucl. Instr. and Meth. vol. 142. p. 417. 1972.

15. B. Bengston and M. Moszynski. Nucl. Instr. and Meth. vol. 155. p. 221. 1978.

16. B. Bengston and M. Moszynski. Nucl. Instr. and Meth. vol. 117. p. 227. 1974.

70

17. Thorn EMI Gencom Inc. product literature (Gregory Kapp).

18. Polycast Technology Corporation product literature (John Lee).

19. A.B. Berlman. Handbook of Fluorescence Spectra of Aromatic Moiecules. Academic Press. New York. 1971.

20. W.L. Borst, M.W. Pleil, G.W. Sullivan, J.C. Crelling and C.R. Landis, in: New Applications of Analytical Techniques to Fossil Fueis. American Chemical Society. Division of Fuel Chemistry. New York. April 13-18. (1986) vol. 31 (1) pp. 7-16,

CIIAPTER V

APPLICATION TO GEOLOGICAL SPECIMENS

Coal

Fluorescence microscopy has been a usefui tooi for the

characterization of organic materiais in coals and rocks [1].

Carbon-rich coal is composed of microscopic organic equivalents of

minerals called macerals. We studied macerais under near ultraviolet

excitation and are the first to investigate the fluorescence emitted

from these organic compounds in the time domain. By applying the

present technique. we have shomi that the maceral fluorescence can be

modeled adequately with a sum of exponential decays. Hence. more than

one fluorophore is present.

In order to determine the usefulness of time-resolved

fiuorescence analysis as an extension of conventionai fluorescence

analysis of organic materials found in coals and rocks, two sets of

samples were studied. One set was a series of shale samples at

different levels of maturation from the New Albany Formation from the

Illinois Basin. While maturation trends of this material in the

Illinoes Basin have been studied using vitrinite reflectance (a

standard measure of maturity) and qualitative and quantitative

fluorescence techniques [2.3,4], nothing previously was kno\Yn of the

time-resoived fluorescence properties of these (or any) sampies. The

other set was a series of samples from the lower Kittanning seam from

Pennsylvania and Oliio, which has a more extended range in rank, from

71

72

.47 to 1.01 vitrinite reflectance (from high-volatile bituminous C to

medium-voiatile bituminous).

Two primary fluorophores were found in alginite macerais from the

New Aibany Shale Group of the Illinois Basin. Alginite macerais are

generally yeiiow to orange in coior, microscopic in size (10-100 im)

and are believed to originate from prehistoric aigea. The data

acquisition consisted of 21 decay curves, one tai<cen every 10 nm over

the fiuorescence emission range of 450-650 nm for each alginite

macerai investigated. Each decay curve acquired was individually

deconvoluted with an assumed two exponential fitting function

resulting in the determination of the component A-coefficients and

corresponding decay times. Figure 5-1 represents one such acquired

curve and its corresponding fit. Figure 5-2 represents the results of

fitting the 21 decay curves acquired from one sample at 10 nm

intervais; A., T. and P. are plotted as a function of emission

wavelength. The coefficients A.(A) make up the component

A-coefficient spectra. Since the decay times remained relatively

constant over the emission wavelength range, which is expected for two

non-interacting fluorophores, our two component model is justified for

this sample set, and the spectra represent the individual component

fluorescence, which make up the total fluorescence. The A-coefficient

spectra were averaged over 45 alginite macerals covering a limited

range of maturity and the resulting spectra are given in Figure 5-3.

The dotted lines reprcsent twice the sample variancc; hence, there are

definitely two separate componenLs conLributing to the fluorescence.

Each individual A-coefficient spectrum was normalizcd to unit area

73

prior to averaging (we were interested only in the general shape of

the spectrum and not the absolute intensity which is dependent on

sampie size as well as aperture settings). The average spectra were

then normalized to unit height for comparison. The blue component of

Figure 5-3 has a lifetime in the subnanosecond range (0.2 ns). while

the red component iifetime is in the nanosecond range (3.0 ns).

Figure 5-4 represents a typical c.w. emission spectra of an alginite

maceral. Figure 5-5 shows that the percent contribution to the total

intensity (P.) averaged over the entire emission range of each

fluorescing component. The percent contribution has a noticeable

correlation with sample maturation (or rank). The maturity or rank of

the sampie was determined by standard vitrinite reflectance

measurements. which ranged from 0.4 to 0.62 in vitrinite reflectance.

With increasing rank. the two primary fluorophores of alginite

macerals change only in their percent contributions (relative

intensities). and the individual component spectra remain the same. as

also indicated by the small sample variance in Figure 5-3. It seems

that as maturation occurs. one fluorophore may be transforming into

another type or both fluorophores are changing into non-fluorescing

products at different rates.

The Kittanning coais have a wider rank range (0.47 to 1.01

vitrinite reflectance) and an abundance of different maceral types.

All of the time-resolved data was again fit with a two exponential

fitting function. This Lwo-componenL analysis was performed on Lhree

of the niain maceral types. Figure 5-6 shows the unique distribution

of decay times as a function of imiccral type. The short component

74

iifetime T^ is shortest for sporinite. followed by alginite (New

Aibany) and fluorinite. The longer component r^ is shortest for

alginite, foilowed by fluorinite and sporinite. The averaged

A-coefficient spectra for two of the three macerai types are given in

Figure 5-7 (aiginite component spectra are given in Figure 5-3). Only

the fluorinites showed a spectral seperation of the quality in Figure

5-3, hence the two-component analysis is probabiy valid here. The

other maceral type shows larger deviations in the A-coefficient

spectra, which is a direct result of both the complexity of the system

(this maceral is most probably comprised of many fluorophores) and the

quality of the data (only 64 pulses were averaged for each decay

curve). Some of the two component lifetimes showed a correlation with

rank suggesting that our two-component model may be inadequate.

however, the fluorophore environment is also changing resuiting in a

different non-radiative decay rate which changes the observed

component lifetimes. (This is analagous to the scintillators where

the lifetimes depended on the type of solvent and its concentration.)

There may also be some energy transfer mechanisms invoived or there

may be a distribution of iifetimes present. (The fluorescence emitted

from a single protein fluorophore, for example, can be represented

with a distribution of lifetimes since there is probably a

distribution of configurations the fiuorophore may be in [5,6].) Even

with these complications, the technique has shown that the resolved

component spectra from this two component model can be used to

identify the macerals. Fluorinite and sporinite spectra differ

significantly. For fluorinitc, the short lifctime spectrum peal<s in

75

the biue. while the long lifetime spectrum peaks broadly in the

yeilow. Conversely. sporinite spectra are more broad; the short

lifetime component peaks near 5.50 nm (yeliow) and the longer lifetime

component peaks in the green. This relationship between the decay

times and spectra of sporinite is opposite to that of alginite.

thereby distinguishing these important macerals and providing a basis

of identification of those petrographicallly troublesome constituents

of sedimentary rocks. Resinite fluorophores also peak in the

yeiiow-orange region of the visible spectrum but the shorter iifetime

has a larger contribution from the blue.

Our resuits for coal groups indicate that temporal fluorescence

analysis is an extension of conventional coal characterization

techniques by measuring the fluorescence lifetimes for various

macerals. The observed change in percent contribution of the

fiuorescing components with increasing rank represents the first

measurement of this kind in coal petrology. The resolution of the

totai fluorescence emitted from a geologicai specimen into its

component fluorophore spectra and the tracking of the component

spectra through a range of maturity levels has never been done before.

Cude Oil

We also have used our technique to characterize crude oil samples

from petroieum deposits in Alabama. California. Texas. and Venezuela.

The technique is capable of analyzing three dominant fluorophores in

the samples. Figure 5-8 shows Lhe Lime resolved specLra and

fluorescence decay times of a Texas crude oii. Also shown are the

76

continuous wave excitation spectrum and pulse integrated spectrum for

the sample. Characteristic decay times and spectra of the short lived

component (T^ = 1.68 ns) and a longer lived componcnt (T„ = 18.9 ns)

resemble standard reference spectra and lifetimes of anthracene and

pyrene. respectively [7]. These component spectra were found by

plotting the product A.-T. as a function of wavelength in 10 nm steps.

The spectra are the actual component spectra. obtained by integrating

the fitting function over time. Large deviations in r resulted for

the short lived component beyond 500 nm; these deviations are most

probably caused by the component's low fluorescence intensity. Figure

.5-9 shows three plots of API index (proportional to the inverse of

specific gravity and a standard geological measure of maturity) versus

the component lifetimes of all the oil sampies. The lifetimes of all

three components decrease with decreasing API index. This decrease in

the lifetimes is probably due to the increasing percentage of

asphaltenes. which act as fluorescence quenchers. The quenchers ailow

the excited fluorophores to return to the ground state via more

numerous nonradiative paths. The decrease in API from condensates to

heavy crude oils is related to the decrease in maturation levei.

Hence the lifetime of the components can give direct information about

the maturity of these oil samples. The peak of the c.w. spectra is

also shifted from red to blue with incrcased maturation levels for

these samples.

77

241

193

m V 144

96

48

m V

- 4

18-MAR-85 15: 15:10

A MEASURED / \ FIT 1 l

\ : ; FITTED \ : ; REGION

1 \

1 \

/ ^-~~~^

0 2 4 5 8 1

NANOSECONDS

RESIDUALS

0

1

'^em

' " p ^

54

Al

A2

P1

P2

Tl T2

^ .

=

—

500 nm

0 . 5 5

9 .97

PULSES

=

=

=

=

=

=

1188 .8

51

7 4 . 8 7

2 5 . 12

0 . 2 4 2

1.897

0 . 5 5 8

[ 0 .

mV

+

+

+

+ +

+

55]

O o = 0

18 .2

1.5

0 .B4

0 . 8 4

0 . 0 0 5

0 . 0 3 2

0 . 0 1 8

243

\ 1

mV

% o/ /o

ns

ns

ns

•p.V

iOILOlLl ALGINITE

Figure 5-1. Fiuorescence pulse emitted from an alginite macerai after puised laser excitation. The pulse is the convolution of the instrument temporal response and the actual fluorescence of the sample. The exit monochromator was set at 500 nm. The low \ value and a peak residuai of less than 2% indicate a good fit.

78

2 140aA ALG.2

10000 r

.000

A^(mv) ^oo

10 .

10 r

T^ (ns)

100

Pi(X) 50 •

0

FLUDRE5CEi\CE DECAY PARAMETERS

450 550 550

WAVELENGTH/nm

750

A2

Tí T2

iii

Pl

P2

Figure 5-2. Example of time-resolvcd fluorescence from an alginite coal maceral. The wavelength dependence of the A-coefficients. fluorescence decay times. and percent contributions of the fluorescing constituents are given.

79

NEW ALBANY SHALE RANK SERIES

C 0 E F F I C I E N T

400 500 600 700

WAVELENGTH (nm)

LEGEND

800

2 X SamplB Varlanca

Average Spectrum

TOTAL NUMBER F SAMPLES: 45

WAVELENGTH RANGE: 450 - 550 nm

Figure 5-3. Resolution of the composite fluorescence spectrum emittcd from alginite macerals of New Albany shale. Two time-resolved component spectra are obtained (normaiized to unit height). The •'blue" component has a lifeLime in the sub-nanosecond range (0.2 ns). and the "red" component in the nanosecond range (3.0 ns). The individual component spectra remained the same for 45 macerals. but Ihe percent contribution of cach component depended on sample maturi ty.

80

19-FEB-85 t4108I30

Vp (raw) -5000 mv

PEAK AT: 536nm Q : 0.56 Qm: 1.07

AREA PARAMETERS: V: 16 B: 32 G: 31 Y: 20 O: 1 R: 0

SAMPLE INFORMATION

A3 lOILOlLl ALGINITE

R E L A T I V E

R E L A T I V E

400 500 600 700 WAVELENGTH /nm

8 0 0

0 . 8

0 . 6

0 . 4

0 . 2

4 0 0 5 0 0 6 0 0 7 0 0

WAVELENGTH /nm

8 0 0

Figure 5-4. A typical continuous wave emission spectrum of an alginite coai macerai with the corresponding spectral parameters

81

r fíW ALOANY O I L SMALE RANK S E n i E ES

r = 0.5743

PERCENTAGE 1

91

81

70

59

43

38

Slope = ^3.858

0.4 0.4-1 0.43 0.53

nEFLECTArsCE IX)

0 . 5 8 0 . 6 2

r = - 0 . 6 6 5 9 S l o p e = - 5 9 . 4 3 5

PEnCEriTAGE 2

61

50

40

10

0.4 0 . 4 4 0 . 4 9

f lEFLECTANCE C

0 . 5 3 0 . 5 0 0 . 6 2

Figure 5-5. Perccnt contribution to thc total intcnsity avcraged over the cntirc cmission range as a function of sample maturity. The percent contribution of each component has a noticeable correlation with sample maturation (vitrinite reflectance).

82

QJ ^ W

P^ : 3 o CJ

o o

o IS w O' w o^ \M

w >

H

H-I

w

20 -1

10

0

0

0

SPORINITE

T^ = 0 .18 ns

T r

T^ = 4 .13 ns

0 1 2 3 4

10 -

FLUORINITE

T-j = 0.32 ns ^2=3.17 ns

íli 0 1 2 3 A

20 -

10

mT-j_ = 0.21 ns ALGINITE

^2=2.15 ns

=] P=^

0 2 A 5

=i .

7

-1 1 1 1

6 7

7

FLUORESCENCE DECAY TIME / ns

Figure 5-6. Distribution of fluorescence decay times in coal macerals. Two characteristic groups of decay times are discernible for each maceral type.

83

c

F. F F I C I A H T

400

C L) F; F F I c [ A tl T

500 GOO 700

WAVELENGTH (nm)

800

KITTANNING HANK GEfUES

LEGENU

i P O n i N I T E :

2 X Sampl t í V a n a n c e

A v e r a g c S p e c t r u m

TOTAL NUMUER OF SAMPLES: 3G

WAVELENGTfl flANGE: 450 - G50 nm

400 500 GOÛ / 0 0

WAVELLMGll-l (nm)

000

KITTANr i IMG IIA'i;-; GL f l I LS --- ILUU11INITL5

TOFAL NUMULÍI Or GAMPLES: 34

Figure 5-7. Resolution of composite fluorescence spectra of the two major maceral types found in the Kittaiining coals. The fluorinite macerals were the only type which could be modeled well with two components. The other maceral shows larger standard deviation in the component spectra indicating the complexity of the system.

84

&-• M LO 2; w [-1

M

CU

u w u íf)

D.

o T

1.0

0. 0

0. 6

0. 4

0. 2

0. 0

12

10

PULSE INTEGRATED SPECTRuM

c.w. EMISSION SPECTRUM

0

/'''' '~*\ TIME - RE S 0 L VE E - T =18. 9ns/ \

// T.-,-^ T3ns ^\(

1 1 1 1 1

) SPECTRA

DECAY TIMES

0 n n n o o ' u " u «•

0 o o

^2

W

0

u

X

n " « A "

• . • • •

• .

1

300 420 460 500 540 500 300 420 460 500

WAVELENGTII (nm)

540 500

Figure 5-8. Fluorescence emission spectra and lifetimes from a Texas crude oil. The top graph shows the c.w. emission spectrum and pulse-integrated data. The lower graph contains three time-resolved component spectra and the corresponding decay times found by fitting the fluorescence pulses with a thrce exponential decay functon.

85

a ^

<:

u Q

20

16

12

8

1

2

0

-10 0 10 20 30 AO 50 60

API-INDEX

Figure 5-9. Dependence of the three resolved fluorescence decay times (see also Figure 5-8) on API index for several crude oii and condensate samples. (A higher API index means greater sample maturity). A direct relationship is indicated between the component lifetimes and maturity.

86

Literature Cited

1. Borst, W.L.. Pleil. M.W.. Sullivan. G.W.. Landis. C.R.. and Creiling. J.C, "Laser-Induced Fluorescence Microscopy of Coal Macerals and Dispersed Organic Material." in New Applications of Analytical Technigues to Fossil Fuels. American aiemical Society. Division of Fuel Chemistry Vol. 31. No. 1. pp. 7-16, New York. 1986.

2. Barrows. M.H.. Cluff. R.M.. AAPG Memoir 35: Petroleum Geochemistry and Basin Evaluation. p.lll. (1984).

3. Barrows. M.H., Cluff. R.M., Harvey, R.D., Proc. 3rd Eastern Gas Shale Symp., p. 85, (1979).

4. Cox. G.E.. M.S. Thesis. Southern Illinois University at Carbondale. (1985).

5. Dowben. G.. and Lin. G. "Use of Nanosecond Pulse Fluorometry for Lhe Study of Protein Structure," Sixth SPIE Conference on Fluorescence Detection, Vol. 743, Jan. 15-16, 1987.

6. James, D.R.. Liu. Y.S.. Siemiarczuk. A., Wagner. B.. and Ware, W.R. "Recovery of Underlying Distributions of Lifetimes from Fluorescence Decay Data" Sixth SPIE Conference on Fluorescence Detection, Vol. 743. Jan. 15-16. 1987.

7. Berlman, A.B.. Handbook of Fluorescence Spectra of Aromatic Moiecuies. Academic Press, New York, 1971.

aiAPTER VI

APPLICATION TO CRIMINALISTICS

Application of our fast analog technique lends itseif very weli

to the study of forensic materials. We have applied our technique to

hair, fibers and also common surfaces encountered in fingerprint

detection. The microscope allows us to select the specific areas

where we can then determine the c.w. fluoresescence and time-resoived

spectra.

Fibers are a common material in criminal investigations and must

often be distinguished from one another. We have applied our

technique to fibers, which look alike under the microscope and have

the same fluorescence spectra (Figure 6-1). The fluorescence pulses

emitted from two such similar fibers are given in Figures 6-2 and 6-3.

The results of fitting the pulses with a bi-exponential fit indicate

that for the white sock fiber of Figure 6-2. the fluorescence decay is

essentially mono-exponential (P. = 98.4%). Figure 6-3 gives the

resuits of using a bi-exponential fit to the fluorescence pulse

emitted from a string fiber. Clearly. there are two primary

components making up the total fluorescence. each having a

characteristic lifetime. We believe that the shorter component of 1.0

ns (T ) is the same for both samples. By looking at one emission

wavelength we have been abie to distinguish between the two fibers

based on fluorescence decay times. fibers which otherwise would be

considered identical.

87

88

We have also applied our technique to determine the fluorescence

lifetimes of materials commonly encountered in fingerprint detection.

Laser detection of latent fingerprints is oftentimes difficult on

surfaces such as wood and brown paper due to the large background

fluorescence from the surfaces. We measured the lifetimes of these

background materials so that Dr. Roland Menzel couid determine the

feasibility of using a gated imaging technique for fingerprint

contrast enhancement [6-1]. If the background has a short lifetime

while the desired image (fingerprint) has a long lifetime. then all

one needs to do is gate the detector enabling it to operate oniy after

the background fluorescence has decayed (ordinary photographic

techniques fai1 due to the short time scale). An image intensifier

tube with the proper gating electronics is ideal for this use and an

order of maginitude increase in contrast is expected [6-1].

The materials investigated and corresponding lifetimes are given

in Table 4. Both blue excitation (373 nm using BPBD dye) and green

(508 nm using Coumarin 500 dye) excitation were used. Green

excitation is preferred here since most fingerprint laboratories use

argon-ion or Nd-YAG (frequency doubled) lasers which produce green

light. It was found that fingerprints treated with

cyanoacylate/rhodamine 6G. which is often used for staining

fingerprints. have a lifetime of about 3.4 ns (see Figure 6-4). much

longer than most of the materials in Tabie 6-1. Hence. this proposcd

gated-imaging technique would work quite well.

89

rable 6-1. Fingerprint Background Materials

Material

Wood

Wood

Cardboard

Cardboard

Cardboard

Leather (rough)

Leather (smooth)

Brown Envelope

White Paper

Ye11ow Paper

Exci tation Wavelength

(nm)

373

508

508

508

373

373

373

373

508

373

Emission Wavelength

(nm)

587

587

599

559

585

540

439

434

588

539

Lifetimes (ns)

Ti = 0.568 T2 = 2.067

Ti = 0.573 T2 = 2.124

Ti = 0.999

Ti = 0.917 T2 = 3.234

Ti = 0.160 T2 = 1.502

Ti = 0.100 T2 = 1.841

Ti = 0.164 T2 .= 2.280

Ti = 0.462 T2 = 3.139

T, = 1.110

Ti = 0.451 T2 = 2.086

Percent

Pi P2

Pi P2

Pi

Pi P2

Pi P2

Pi P2

Pi P2

Pi P2

Pi

Pi P2

= 44.53 = 55.47

= 45.73 = 54.27

= 100.0

= 97.21 = 12.79

= 59.85 = 40.15

= 63.21 = 36.79

= 69.70 = 30.30

= 63.28 = 36.72

= 100.0

= 73.06 = 26.94

90

H

CO

w

W u w co w

o

1.0

0 . 8 -

0 . 6 -

0 . 4 -

0 . 2 '

0 . 0 400 500 600

WAVELENGTH/nm

700

Figure 6-1. Fluorescence spectra emitted from a cotton string fiber and a fiber from a white sock. Both fibers could not be distinguished under the microscope nor could they be separated on the basis of their c.w. fluorescence emission spectra.

91

591

472

354

m V

235

117

-1

m V

L

11.5

-11.6

MEASURED FIT

FITTED REGION

0 4 8 12 16 20 NANOSECONDS

A1 A2

Pl P2

Tl T2

=

=

=

=

=

=

1350.4

2 .2

98.42

1.57

0.G45

7.952

± 6.5

± 0.8

± 0.75

± 0.75

± 0.004

± 2.532

mV mV

/o

%

ns

ns

T„ = 0.957 ± 0.074 m

ns

RESIDUALS

Figure 6-2. Fluorescence pulse emitted from a white sock fiber fitted with a bi-exponential fitting function. The fluorescence decay is essentially mono-exponential (Pi = 98.-1%).

92

413 F

331

248

m V

165

B2

-1

m V 7.4

-7.4

MEASURED FIT FITTED REGION

4 8 12 16 20

NANOSECONDS

Al A2

Pl P2

Tl T2

=

=

=

=

=

—

753.5

127.1

75 .18

24.81

1.008

1.999

+

+

+

+

+

+

18.1

20 .2

3.15

3.16

0.019

0 . 1

mV mV

%

%

ns ns

T^ - 1.254 ± 0.076 m ns

RESIDUALS

Figure 6-3. Fluorcsccnce pulse emitted from a cotton string fiber fitted with a bi-exponential fitting function. The fluorescence decay is clearly made up of two components. It is believed that the shorter component of 1.0 ns (T^) is due to the same fluorophore for both the cotton and sock fibers.

93

Literature Cited

Menzel. E.R.. Pleil, M.W.. Gangopadhyay. S., "Enhancement of Fluorescent Fingerprints by Time-Resoived Imaging." Sixth SPIE Conference. Voi 743, Jan. 1.5-16. 1987.

CONCLUSION

The present teclmique has great utility for the characterization

of fluorescing microscopic substances. In criminalistic applications

the ability to differentiate between virtually alike fibers has been

shown. The fluorescing constituents of coal and crude oils can now be

resolved. The actuai fluorescence lifetimes of the primary

scintillant in seven commercially availible scintillators have also

been found and can be used to better understand the complex mechanisms

involved in the transfer of the kinetic energy from the nuclear

particle to the excited singlet state of the primary scintillant. thus

facilitating the development of better detectors and improving the

future analysis of data.

Since our preliminary studies on source rocks and crude oils have

sho>vn the existance of very short. picosecond constituents. it was

necessary to construct a new system employing a single photon counting

technique. A synchronously pumped, tunable dye laser system has been

chosen. purchased. installed and tested. and wiil be the new puised

excitation source. The output laser puises are at 368 nm and have a

full width at half maximum of less than 4 ps. This short pulse width

will excite the shorter components more efficiently than the 500-700

ps pulses of the nitrogcn pumped dye laser. Even though the

excitation efficiency for the longer lived components will decrease

wi th the use of the siiorter pulses. these components are in general

more intense. hence. they should still be resolvable (they can also be

94

95

resolved with the fast analog technique). The average power of the

new laser system at a repetition rate of 4 MHz is routinely between 4

and 5 mW (1-1.2 nj/pulse). At repetition rates of 400 kllz. a pulse

energy of over 3 nj has been measured. The computer hardware for the

new system consists of a PCA board which converts out IBM-AT personal

computer into a muitichannel analyzer. This hardware has been

installed. tested and linked to an EG&G Ortec time to emiplitude

convertor (TAC). The TAC is linked to a fast photodiode trigger

source and the MCP via two constant fraction discriminators and a

linear ampiifier. A total instrument response of 103 ps has been

measured. however. implementation of better triggering techniques and

better timing units should improve the instrument response to about 75

ps. Based on experience with the oid technique and the deconvolution

routine. it is hoped that a precision of less than 10 ps in the

lifetime determination can be achieved, hence allowing the resolution

of fast events other than fluorescence. A Fortran deconvolution

routine based on Marquardt's routine has been written ajid tested with

synthetic data. Further testing is planned to determine the

theoretical limitations of the routine, its precision as a function of

instrument response half width, channel number and total counts will

also be determined. After the theoretical limitations have been

determined, the system will be experimentally tested using the same

reference materials as with the old system (rose-bengal, POPOP.

rhodamine 6G, etc).

Current research is aimed at better understanding the origin of

oil. its maturation process and its basic chemical structure. It is

96

hoped that the chemistry of the individual fluorophores may be found

so that the fluorescence source can be determined and traced through

the evolution of the oil or source rocks. To better understand the

mechanisms involved in fluorescence emission, other, simpier systems

are being investigated. Pure polymers treated with dye rotors are

studied to determine the rotational properties of the rotors in the

poiymer environment. The observed fluorescence lifetime is a function

of the free volume availible to the rotor, hence, the more volume

available, the more the dye is allowed to rotate and the shorter the

observed lifetime. These rotor studies will help us understand the

fluorescence properties of dye stained insulators which have been

damaged through electric discharges. Possible time-resolved

luminescence studies on semiconductors have also been discussed. The

applications of both our new and old systems to all of these various

materials indicate the utility of the technique to materials research.

COMPREHENSIVE BIBLIOGRAPHY

Auronet, C. , Blumenfeld, J., Boso, B., Bourdinaud, M.. Evard. P.. Jeanney, C. and Laford, C. Nucl. Instr. and Meth.. voi. 169. p. 57. 1980.

Barrows. M.H.. Cluff. R.M.. AAPG Memoir 35: Petroleum Geochemistry and Basin Evaluation. p.lll. 1984.

Barrows. M.H., Cluff, R.M., Harvey, R.D., Proc. 3rd Eastern Gas Shaie Symp., p. 85, 1979.

Bengston B. and Moszynski, M. Nucl. Instr. and Meth. vol. 117, p. 227. 1974.

Bengston, B. and Moszynski, M. Nucl. Instr. and Meth. vol. 155. p. 221. 1978.

Berlman, Isadore, B. Handbook of Fluorescence Spcctra of Aiomatic Moiecules, Academic Press, New York. 1971.

Berlman. I.B. J. Chem. Phys. 33 (4). p. 1124. 1960.

Bevington. P.R. Data Reduction and Error Analysis for the Physical Sciences. McGra vTv-Hil 1. New-York. 1987.

Birks. J.B. and Cameron, A.J.W. Proc. Phys. Soc. B. vol. 69. p. 593. 1956.

Birks. J.B. The Theory and Practice of Scinti1lation Counting. Pergamon. Oxford. 1964.

Birks. J.B. and Pringle. R.W. Proc. R. Soc. Eden.. vol. 70. p. 233. 1971/72.

Bollinger, L.M. and Thomas, G.E. Rev. Sci. Instr., voi. 32, p. 1044, 1961.

Borst, W.L., Gangopadhyay, S., Pleil, M.W.. "Fast Analog Technique for Determining Fluorescence Lifetimes of Multicomponent Materials by Pulsed Laser," Sixth SPIE Conference, vol. 743, January 15-16, 1987.

Borst, W.L., Pleil, M.W.. Sullivan. G.W.. Crelling J.C. and Landis. C.R. in: "New Applications of Analytical Techniques to Fossil Fuels." American aiemical Society, Division of Fuel Chcmistry, New York, April 13-18, (1986) vol. 31 (1) pp. 7-16.

97

98

Bransden, B.H. and Joachain, C.J. Physics of Atoms and Moleculcs, Longman, London, 1983.

Cox. G.E.. M.S. Thesis. Southern Illinois University at Carbondale. (1985).

Cundall. R.B. and Dale. R.E., Time Resolved Fluorescence Spectroscopy in Biochcmistrv and Biolot>:v. Plenum Press. New York. 1983.

D'Agostini. Marini. G.. Martellotti. B.. Mass. F.. Rambaidi A. and Sclubba. A. Nucl. Instr. and Meth. vol. 185. p. 49. 1981

Dowben. G.. and Lin. G. "Use of Nanosecond Pulse Fluorometry for the Study of Protein Structure." Sixth SPIE conference. vol. 743. Jan. 15-16. 1987.

E. Leitz Inc. Product Literature Detailed List of Mlcroscope Equipment and Accessories. Part No 610-126 p. 5/5. 1983.

Gangopadhyay. S.. Pleil. M.W. and Borst, W.L., "Resoiution of Interacting and Non-Interacting Fiuorophore Mixtures by Laser Induced Fluorescence Spectroscopy," submitted to J. Luminescence, Feb.. 1987.

Grinvald, A. , and Steinberg, I.Z., Biochemistry, Vol. 13, p. 5170, 1974.

Grinvald, A., and Steinberg, I.Z., "On the Analysis of Fluorescence Decay Kinetics by the Method of Least-Squares," Anai. Biochem., vol. 54. pp. 583-598. 1974.

Hamamatsu Photonics K.K. Proximaty MCP PMT Data Sheet. 1983

Hercules. David M. Fluorescence and Phosphorescence Analysis. Interscience Publishers, New York, 1966.

Hirschiaff, E. Fluorescence and Phosphorescence. Chemical Publishing Co. Inc. New York. 1939.

James. D.R.. Liu. Y.S.. Siemiarczuk. A.. Wagner. B.. and Ware, W.R. "Recovery of Underlying Distributions of Lifetimes from Fluorescence Decay Data," Sixth SPIE conference, vol. 743. Jan. 15-16, 1987.

Knoll. G.F. Radiation Detection and Mcasurcmcnt. John Wilcy, New

York, 1979.

Lakowicz, Joeseph R. Principlcs of Fluorcsccncc Spcctroscopy. Plenum Press, New York, 1983.

Lipsky S. and Burton, M. J. aicm. Phys. 31 (5). p. 1221, 1959.

99

Lynch, F.J. lEEE Trans. Nuci. Sci. NS-22 p. .58, 1075.

Lyons, P.B. and Stevens, J. Nucl. Instr. and Meth. vol. 114, p. 313.

Marquardt. D.W.. "An Algorithm for Least-Squares Estimation of Non-Linear Parameters." J. Soc Ind. Appi. Math.. Vol. 11. No. 2. pp. 431-441. June. 1963.

Moszynski. M. and Bengston. B. Nucl. Instr. and Meth. vol. 142, p. 417, 1972.

Moszynski. M. and Bengston, B. Nucl. Instr. and Meth. vol. 158, p.l, 1979.

O'Conner, D.V. and Andre, J.C., "Deconvolution of Fluorescence Decay Curves: A Critical Comparison of Techniques," J. Phys. aiem.. Vol. 83, pp. 1333-1343, 1979.

Polycast Technology Corporation product literature (John Lee).

Sullivan, G.W., "Time Resolved Fluorescence Microscopy by Pulsed L^ser," M.S. Thesis, Southern Illinois University, 1983.

Thorn EMI Gencom Inc product literature (Gregory Kapp).

Waynant, R.W. and Elton, R.C. in: Organic Scinti 1 lators and Liquid Scintillation Counting, eds., D.L. Horrocks and C.-Tz. Peng, Academic Press, New York, 1971.

Weider, Sol The Foundations of Quantum Theory, Academic Press, New York, 1973.

PERMISSION TO COPY

In presenting thls thesis in partial fulfillment of the

requirements for a master's degree at Texas Tech University, I agree

that the Library and my laajor department shall make it freely avail-

able for research purposes. Permission to copy this thesis for

scholarly purposes may be granted by the Director of the Library or

my major professor. It is understood that any copying or publication

of this thesis for financial gain shall not be allowed without my

further written permission and that any user may be liable for copy-

right infringement.

Disagree (Permission not granted) Agree (Permission granted)

Student's signature í'^cT^i

•>'/•

Student 's signature

Date Date ^ ('

' • / - ' :