A Dynamic Reaction Cell for Inductively

A Dynamic Reaction Cell for InductivelyCoupled Plasma Mass Spectrometry (ICP-DRC-MS). II. Reduction of InterferencesProduced Within the Cell

Scott D. Tanner and Vladimir I. BaranovPE-SCIEX, Concord, Ontario, Canada

An rf-driven reaction cell offers the potential for enormous efficiency in the removal ofinterfering ions prior to mass analysis in inductively coupled plasma mass spectrometry.Concomitant with this efficiency is the potential for interference production within the reactioncell. Operation of a low rf-amplitude quadrupole reaction cell in a bandpass mode, eitherrf-only or with additional dc voltage between pole pairs, is shown to suppress these in-cellinterferences. Examples of various types of interferences (spectral and continuum) are shown,and their elimination without sacrificing sensitivity is demonstrated. The advantage ofdynamically sweeping the bandpass in concert with the mass analyzer is shown. The necessityof efficiently evacuating the reaction cell (in this case by venting it to the high vacuumchamber) when reaction gas is not added is demonstrated. (J Am Soc Mass Spectrom 1999,10, 1083–1094) © 1999 American Society for Mass Spectrometry

The dynamic reaction cell is evolved from a ge-neric rf-driven multipole reaction cell. The latterhas gained much interest recently as an “on-line

test tube” for removal of isobaric interferences, notably(but not exclusively) for inductively coupled plasmamass spectrometry (ICPMS). The reaction cell itself isoften considered a variation of the collision cell used forion fragmentation in tandem MS/MS. Although thesedevices share the characteristics of employing rf toconfine the ion beam and, often, are enclosed in order tooperate at pressures above ambient, the dynamics andfunction of the cells are quite distinct. The collision cellis characteristically operated at sufficiently low pres-sure (;0.1 mtorr) that the ion kinetic energy remainsrelatively high (;20–100 eV). With a non-reactive col-lision gas, the collision cell is intended to promotefragmentation of polyatomic ions by conversion ofkinetic energy to internal excitation of vibration. Doug-las [1] reported an attempt to reduce the relative inten-sity of CeO1, derived from an ICP source, by collisionalfragmentation. However, he showed that scatteringlosses were greater than the fragmentation yield. None-theless, he also showed that the addition of a reactivegas (O2) to the cell promoted ion–molecule reactionwhich permitted discrimination of the rare earth ele-ments and their oxides through the specificity of oxida-tion (e.g., Ce1 and Tb1 readily react to form their oxideions, but further oxidation of CeO1 occurs much more

slowly). For the ICPMS application, for which a pre-ferred application is the reduction of isobaric interfer-ences derived from plasma ions, ion–molecule chemis-try is expected to be more efficient than collisionalfragmentation. For example, the elimination of the Ar1

interference on Ca1 cannot be achieved through frag-mentation (at least at normal laboratory energies!).Shortly thereafter, Rowan and Houk [2] showed thatargide ions react more rapidly with a number of gases(such as CH4 and Xe) than do some of the isobaricelemental ions. This should have ushered in the use ofthe reaction cell for ICPMS, but the method remainedfraught with difficulties, primarily those of distinguish-ing between residual atomic analyte ions and isobaricpolyatomic ions produced within the reaction cell.Much more recently, Koppenaal and co-workers [3–6]have shown that great efficiency in interference rejec-tion by reaction with H2 and/or H2O can be achieved inan ion trap. Their early reports indicate that similarprocesses may be effective in a linear multipole, butwith reduced efficiency [4, 5]. Turner et al. [7] havepublished preliminary work that claims efficient chem-ical resolution, but showed results only at relativelyhigh ion signal levels.

As noted above, Rowan and Houk [2] recognizedearly that the reaction of ions in the multipole cellproduced new ions in the cell, and that these mayinterfere with the measurement of other analyte ions.They demonstrated that, for a non-thermalized reactioncell, there is a difference between the axial kineticenergies of the analyte ions (which have suffered suffi-

Address reprint requests to Scott D. Tanner, PE-SCIEX, 71 Four ValleyDrive, Concord, Ontario L4K 4V8, Canada.

© 1999 American Society for Mass Spectrometry. Published by Elsevier Science Inc. Received January 29, 19991044-0305/99/$20.00 Revised May 27, 1999PII S1044-0305(99)00081-1 Accepted May 27, 1999

ciently few collisions that they retain some of theirinitial kinetic energy gained from the expansionthrough the interface) and ions produced within thecell. Application of a potential barrier downstream ofthe reaction cell allows some resolution of the analyteions from the newly formed interference ions, wherethe degree of resolution obtained is dependent on thedifferences in the ion energy distributions. Hence, themethod is most effective when the pressure is suffi-ciently low that few thermalizing collisions occur, un-der which conditions the efficiency of reaction is rela-tively low. A compromise is thus required between theefficiency of reactive removal of isobaric interferencesderived from the plasma and kinetic energy discrimi-nation against new interferences produced within thereaction cell.

Koppenaal and co-workers [3–6] have achieved ex-cellent analytical performance from an ICP coupledwith an ion trap mass analyzer by employing reactivechemistry within the trap. The ions stored in an ion trapare not thermalized, but long storage times are feasiblewhich allow for promotion of efficient reactive chemis-try. In addition, the ion trap may be operated in afiltered noise field (FNF) mode (or other auxiliaryexcitation modes) in order to selectively trap certainions or to “tickle” ions to higher energies at whichcollisional fragmentation may be achieved. Operationof the trap with H2 as a reaction gas leads to theformation of H2

1 and H31 ions. The provision of a low

mass cutoff above 5 amu rapidly removes these productions from the trap, and this was reported [3–6] to offeran opportunity to reduce space charge effects (whichadversely affect the dynamic range of the ion trap).Eiden et al. [5] noted that removal of Ar1 from the ionbeam before introduction to the ion trap/reaction cellresulted in lower levels of chemical ionization of thebackground gases in the ion trap (with a consequentimprovement in sensitivity). They also noted that ion-ization of background gases in the trap could produceinterferences for the determination of 129I, and thatoperation of the ion trap at higher rf potential (so thations with m/z , 40 amu were unstable in the trap)reduced this interference.

If the pressure of the collision cell is raised, ionenergies are damped [8, 9] and ion collection at the exitof a closed cell passes through a maximum (collisionalfocusing). The reduction of the ion energy facilitates theuse of thermal ion–molecule chemistry. This approachis particularly attractive because of the enormous spec-ificity and efficiency of ion–molecule chemistry (“chem-ical resolution”). The rate of an ion–molecule reactionprocess is directly proportional to the partial pressure ofthe reaction gas. Collisional focusing is most efficient atpressures in the range of 10 mtorr (and depends on themass ratio of the ion and neutral in addition to otherfactors, such as cell length). Because the efficiencyrequired of the reaction cell may be exceptionally large(e.g., 10 orders of magnitude to eliminate the Ar1

interference on Ca1), the optimum pressure of the

reaction cell may be greater than optimum for colli-sional focusing.

In this perspective, the reaction cell may be moreproperly viewed as an embodiment of a multipole ionguide reactor [10, 11]. This device is advantageouslyused to measure the rate constants of ion–moleculereactions and to follow sequential reactions. Operatedat high frequency and low rf amplitude (low q) and atrelatively high pressure, the ion kinetic energies arenear thermal and therefore near-thermal ion processescan be observed. It is advantageous to use a high-ordermultipole (such as a hexapole or octapole) for the ionreactor because the effective potential is wide, and thisfurther ensures near-thermal conditions. Because it is anobjective of such studies to study both the rate ofreactant ion loss and the appearance of product ions,the device is operated in the rf-only mode. Again, low qoperation combined with a higher order multipoleimproves the mass range of confined ions, and istherefore clearly preferred. With such a device, sequen-tial processes involving ions of widely varying massmay be studied.

In an earlier article [12], we reported that operationof an rf-only quadrupole at high q results in an increasein the ion kinetic energy, and consequently enablescertain otherwise thermodynamically prohibited reac-tions. It is to be expected that an ion gains more kineticenergy from the rf-field in a quadrupole configurationthan in a higher order multipole. In addition, the massrange that is stable within a quadrupole field is morerestricted than in a higher order multipole. Accord-ingly, it is conventional wisdom that the preferredconfiguration for the study of ion–molecule reactions isa high-order multipole operated at low q (high fre-quency and low rf amplitude) [11].

We describe in this article an improved multipolereaction cell which takes advantage of the uniquecharacteristics of the quadrupole. To be an effectiveanalytical tool for the rejection of isobaric interferencesin ICPMS, the reaction cell must be characterized byenormous reaction efficiency. Inherent in the need forefficiency of reaction of interference ions is the concom-itant efficiency of interference production within thereaction cell [12]. Certainly, multiple collision condi-tions are required; a simple Beer’s-law type calculationshows that the ions must average approximately 20reactive collisions to achieve reaction efficiencies on theorder of 1010. Under these conditions, sequential reac-tion chemistry is to be expected, with concomitantproduction of intermediate product ions of variousmasses. Each product ion has the potential to interfereat the mass of another analyte ion. Further, despiteheroic attempts to purify the reaction gas, trace levels oforganic impurities will always be present in the cell.These may derive from impurities in the reaction gasitself, from backstreaming pump oil, or from outgassingof ion optical components. The high efficiency of thereaction cell ensures that at least a portion of theseimpurities will be ionized within the reaction cell. It

1084 TANNER AND BARANOV J Am Soc Mass Spectrom 1999, 10, 1083–1094

should also be noted that an additional source of“impurity gas” in the reaction cell may result frominclusion of the plasma gas within the cell. The reactioncell is typically placed within the ion optics chamber (ofa differentially pumped ICPMS), or communicates be-tween the high vacuum chambers. In the absence ofadded reaction gas, the pressure in the cell is at orbelow the pressure of the ion optics chamber (depend-ing on the conductances of the entrance and exit aper-tures), and is typically of the order of 0.1–1 mtorr. Thegas that fills the cell under these conditions is predom-inantly plasma gas that has been sampled through theskimmer. This gas contains, in addition to Ar, signifi-cant amounts of H and O and perhaps N and mixedpolyatomics of these. When sufficient reaction gas isadded to the cell to raise the pressure above ambient,gas flows out of the cell through both the entrance andexit apertures and shields the cell from entrainment ofplasma gas (and ions must be focused through thecounterflow of reactant gas effusing from the entranceaperture).

We recall that a quadrupole is characterized by awell-defined stability region that is normally expressedin terms of the Mathieu parameters, a and q:

a 5 ax 5 2ay 54eVdc

mv2r2 (1)

q 5 qx 5 2qy 52eVrf

mv2r2 (2)

where e is the electronic charge, Vdc is the dc voltagepole-to-pole, m is the ion mass, v is the rf angularfrequency, r is the field radius of the quadrupole array,and Vrf is the peak-to-peak rf amplitude. The quadru-pole is most often operated within the first stabilityregion (0 , q , 0.908; 0 , a , 0.24), which is shown inFigure 1. Under collision-less conditions, ions that havemasses such that a and q fall within the stability regionhave stable trajectories regardless of their initial condi-tions once they have entered an ideal, infinite quadru-pole field. This is a result of the independence of motionin the X and Y directions, and is uniquely characteristicof the quadrupole; higher order multipoles may beexpressed in a similar manner, but the region of stabil-ity is a function of the initial position of the ions withinthe field. Therefore, a quadrupole finds use as a massanalyzer, whereas higher order multipoles find greateruse as rf-only confinement guides (they have wider, butmore diffuse, stability regions). Operation of the quad-rupole under multiple collision conditions is expectedto result in a modified stability region. We expect thatthe fundamental characteristics of quadrupole stabilityremain, but this has not yet been thoroughly investi-gated (and will be the subject of a subsequent article).

In the rf-only mode of operation, the quadrupole actsas a high pass filter. Ions having masses such that q .0.908 do not have stable trajectories and are ejected

from the device. The low mass cutoff may be scannedby increasing the rf amplitude or by decreasing the rffrequency. If conditions are provided that promotespecific reaction of an isobaric interference ion, theproduct ion will most often appear at a different mass.If this ion has a mass that falls within the stability regionand it satisfies the acceptance conditions of the cell, itwill be retained within the cell. Under these conditions,the product ion may react further with the reaction gasor with impurities in the cell and produce new ionswhich may interfere with the analyte ion (i.e., a se-quence of reactions leading to a new interference at thesame mass). Provided that one of the intermediate ionshas a mass which falls below the stability cutoff, it willbe ejected from the cell, and further reaction (potentiallyproducing new interference ions) will be suppressed.Clearly, the quadrupole offers substantial control overthis sequential chemistry by allowing the selection ofconditions (rf amplitude and frequency) that provide anadjustable and well-defined low mass cutoff.

Further, the quadrupole reaction cell may be oper-ated with an additional dc voltage between pole pairs,providing a non-zero a. While a impacts somewhat onthe low mass cutoff, its primary impact is through theintroduction of a high mass cutoff. This mode of oper-ation allows rejection of higher mass intermediate prod-uct ions which may subsequently react to produce anew interference ion at the mass of interest. Therefore,the quadrupole reaction cell offers the potential todefine a mass bandpass window which may be selectedin a manner to reject the intermediate product ions ofsequential reactions which may eventually result in anew isobaric interference. Notably, it is not necessarynor desirable to operate the quadrupole reaction cell asa mass filter; for this application it is operated as adynamically adjustable bandpass device, where thebandpass is selected according to the specific chemistrywhich is to be promoted or suppressed.

Figure 1. The stability diagram (first region) for a linear quad-rupole under collision-less conditions. The region bounded by theby 5 0 and bx 5 1 curves provides stable ion motion in both xand y.

1085J Am Soc Mass Spectrom 1999, 10, 1083–1094 A DYNAMIC CELL FOR ICP-MS

Recall that ions gain kinetic energy from the rf fieldin proportion to the magnitude of the rf amplitude.Further, for a given value of q, the rf frequency in-creases with the rf amplitude. Therefore, for a givenpressure and q, the number of collisions per rf cycledecreases as the rf amplitude is increased. Because wehave earlier shown that the rf field energy contributionto the collision energy increases inversely as the ratio ofthe collision frequency to the rf frequency [12], anincrease in the rf amplitude increases the collisionenergy. This gain in collision energy, which is bothdirectly and indirectly proportional to the rf amplitude,may result in the promotion of otherwise endothermicprocesses. In turn, this introduces the potential for newinterference ions and also contributes to reactive loss ofthe analyte ion. It may, then, be preferable to scan the rffrequency of the reaction cell rather than the rf ampli-tude. In this manner, it remains feasible to adjust theamount of energy gained from the rf field in a moreuniform manner by adjustment of the rf amplitude as aglobal parameter. This is orthogonal to the excitationachieved by operation at high q, near the bx 5 1boundary, which also provides control of the collisionenergy. An additional benefit of scanning the rf fre-quency rather than the rf amplitude is that the dcamplitude at a given a is independent of the mass of theion.

It is convenient to reference the rf frequency to themass of the ion being analyzed in the downstream massfilter. This provides a dynamic bandpass which isadjusted with the mass of analysis. Because a and q areadjusted by variation of a low amplitude dc voltage andrf frequency, respectively, they may be adjusted on aper mass basis. Thus, we have a quadrupole reactioncell which is characterized by a dynamic mass band-pass, which is dynamically moved in concert with theanalyte mass and whose bandpass width may be ad-justed dynamically. It is these dynamic characteristicswhich distinguish the dynamic reaction cell (DRC) fromgeneric rf-only multipole reaction cells.

It is not the intent of this article to discuss the meritsof a particular reaction gas. In fact, a wide range ofreactive gases may be employed with advantage forspecial applications, particularly if the reaction cellprovides for control of the secondary reaction pro-cesses. We have developed successful analytical meth-ods using hydrogen, ammonia, methane, ethane, ethyl-ene, nitrous oxide, and oxygen. Nonetheless, it will beapparent that we favor ammonia as a reaction gas ofrelatively general application. In part, this is becauseammonia has a high proton affinity and thus acts as aconvenient terminal ion under thermalized conditions,and ionization of hydrocarbon impurities by protontransfer is suppressed. Further, ammonia has an ioniza-tion potential of 10.2 eV [13], which is intermediatebetween those of many analyte elements and argon(and most of its polyatomics), which accordingly offersthe potential for effective chemical resolution of thesespecies. Certainly, ammonia reacts with some impor-

tant analyte elements, notably Ti, Ni, and As, and isaccordingly not prescribed for their trace determina-tion. Further, NH3 acts as a good ligating agent, andforms cluster ions readily. These clusters would inter-fere with the determination of other elements unless thereaction cell provides for selective suppression of thesereactions (which, it might be recognized, is an impor-tant advantage of sweeping the bandpass in concertwith the analytical mass). The reader is referred to theprinted and electronic database maintained by Anicich[14] for information on the reaction rates and productsof thermal ion–molecule reactions.

In this article we present data which demonstrate theexceptional efficiency of the reaction cell for eliminationof interference ions derived from the plasma. Examplesare given of interference ions which are producedwithin the cell, and of the use of the dynamic bandpassto control these without sacrificing sensitivity to theanalyte ions (provided, of course, that the analyte iondoes not react rapidly with the reaction gas, and that thebandpass of the reaction cell is properly selected andnot too narrow). It is shown that sweeping the bandpassof the reaction cell in concert with the mass analyzerallows optimum transmission of the analyte ions acrossthe mass spectrum while simultaneously providing forthe reactive elimination of isobaric interferences andsuppression of formation of new interferences withinthe cell. Conventional ICPMS results are obtained byventing the reaction cell to the high vacuum chamberwhen reaction gas is not added, which eliminates theinterferences which otherwise would form throughreactions with plasma gas entrained into the cell underthis condition. Analytical figures of merit will be pre-sented in a subsequent publication, as will data relatingto quadrupole stability under conditions of multiplecollisions.

Experimental

Two different research-level prototype instrumentswere used to obtain the data reported here, and the datawere recorded on these instruments a year apart. Con-sequently, the data reported for the two instrumentsshould not be compared in detail, but the qualitativecharacteristics are comparable. The earlier instrument(used for the data presented in Figures 4–6, 8, and 9)operated with a fixed-bandpass for the reaction cell,which was not synchronized with the mass analyzer.Accordingly, for this instrument the mass analyzerscanned through the bandpass of the reaction cell, andsensitivity as a function of mass was convoluted by thetransmission characteristics of the reaction cell band-pass. The later instrument (used for the remainder ofthe presented data) provided a reaction cell bandpassthat was scanned in concert with the mass analyzer. Forthis instrument, each ion that passed through the massanalyzer had passed through the DRC at the sameoperating point (a, q) during the time that it was beingmeasured. It might be noted that the commercial ver-


sion of this instrument (the ELAN 6100 DRC) differsfrom both of these prototypes in several aspects, butnotably in that the DRC power supply now providesdigital frequency synthesis (as opposed to the analog rfsupplies used in this work).

The basic configuration of these instruments hasbeen described previously [12]. The drive circuitry forthe dynamic reaction cell was modified by the additionof a dc voltage between pole pairs (the a parameter).The fixed-bandpass configuration (used for the datapresented in Figures 4–6, 8, and 9) used a frequencygenerator (Hewlett Packard, model 33120A) to producea low voltage sinusoidal waveform of fixed frequencyand amplitude. The output from the frequency genera-tor was amplified (in-house manufactured amplifier) tothe desired rf amplitude (typically 200 Vrf peak-to-peak). This DRC power supply was known to exhibit animbalance that generated a dc potential between polepairs, and this imbalance was a function of the rffrequency. Hence, a small non-zero value of a wasalways present, and varied with the frequency. Theintentional addition of a dc component in this configu-ration was obtained from an external dc power supply(Xantrex, model LXQ 30-2, Burnaby, British Columbia,Canada) and was symmetrically applied about the DRCrod offset potential. In this manner, the rf amplitudeand frequency (and dc amplitude) were fixed so that themass bandpass window of the DRC was fixed. Thisconfiguration allowed scanning of the mass analyzeracross the mass bandpass of the DRC.

The remainder of the data were obtained using adynamically scanned analog DRC power supply. Thedc voltage signals corresponding to the desired q and a,the rf amplitude, and the analyzer mass (taken from thedigital-to-analog converter of the mass analyzer quad-rupole power supply) were fed to an external analog rfpower supply. An rf signal was generated dynamicallywith a frequency defined by these inputs and wasamplified to the requested Vrf. A dc component be-tween pairs was generated according to the resultantfrequency and Vrf and requested a, and was addedsymmetrically around the DRC rod offset potential. Asmall imbalance in the circuitry resulted in a minimumvalue of a of about 0.005 (i.e., nearly, but not exactly,rf-only when a zero value for a was requested). For theexperiments reported here, the rf amplitude was fixedat 200 Vrf peak-to-peak. In this configuration, the massbandpass window of the DRC was dynamically ad-justed with the reference q and a defined for the massbeing transmitted through the mass filter. The bandpassis adjusted on a per element basis; that is, a value of qand of a is defined for each analyte in the peak hoppingmode and the bandpass is adjusted prior to measure-ment of each signal. For spectral scanning, a constantvalue of q and of a is applied throughout a scansegment; for this work, all scans were performed in asingle scan segment (so q and a are constant throughoutthe scan, this being achieved by frequency scanning theDRC).

For the analysis of elements that are not normallyinterfered, it may be desirable to emulate “convention-al” ICPMS operation with the ICP-DRC-MS. If reactiongas is not added to the DRC, plasma gas from the ionoptics chamber is entrained into the reaction cell. If thereaction cell remains enclosed, the plasma gas pressur-izes the cell and facilitates ion–molecule reactionswhich increase the spectral background. Suppression ofthis chemistry can be achieved by increasing the meanfree path within the reaction cell, and this is mostreadily achieved by increasing its conductance into thehigh vacuum (mass analyzer) chamber. In the presentconfiguration, the DRC acts as the interface between theion optics vacuum chamber and the high vacuumchamber. Although an obvious way to increase theconductance into the high vacuum chamber is to in-crease the diameter of the exit aperture of the DRC, orto remove the aperture plate competely, this has adeleterious effect on the continuum background signalobtained. Rather, the DRC employed in this work isequipped with an externally activated venting mecha-nism which opens the reaction cell around its circum-ference. Operation in this mode, with the radial ventopen and reaction gas flow stopped, is referred to as“vented operation.”

The DRC entrance tube lens, exit aperture lens, andac prefilter rod offset were electrically connected andmaintained at 230 V. The DRC rod offset potential wasvaried according to the pressure in the DRC; for dataobtained with a reaction gas, it was set to 21 V, andwhen vented to the high vacuum chamber (withoutreaction gas) it was set to 210 V. With the DRC vented(i.e., operating in the low 1025-torr range, and thereforecollision-less conditions), the ion kinetic energies areunaffected by the DRC rod offset and the mass analyzerrod offset was held at 0 V (i.e., ions were mass analyzedat energies corresponding to their gas kinetic valuesobtained from the isentropic expansion through thevacuum interface). With the DRC pressurized (multiplecollision conditions), the ion kinetic energies were cen-tered near the DRC rod offset potential with a narrowdistribution (collisional energy damping), and the massanalyzer rod offset was set to 25.5 V (i.e., ;5-eV ions).All other parameters of the ICPMS system were set asusual for a PE-SCIEX ELAN 6000 ICPMS operatedunder normal plasma conditions [plasma power was1000 W, injector flow was ;0.9 L/min, sampling depth9 mm, lens voltage (E1) ramped linearly with ion mass].

Most of the experiments were conducted in a rela-tively clean laboratory (Class 1000) with sample prep-aration and introduction in Class 100 enclosures. Dou-ble deionized water (DDIW) was prepared freshlyusing an ELIX/Gradient purifier (Millipore Corpora-tion, Bedford, Massachusetts). Analytical samples wereprepared by serial dilution from standards (Spex Certi-prep, Metuchen, New Jersey) using DDIW and 1%Seastar Baseline Nitric Acid. An analyte mixture con-taining B, Na, Mg, Al, K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn,As, Sn, Sb, Au, and Pb at 1 ng/mL (1 ppb) was used.


The DRC could be remotely vented to the highvacuum chamber or closed for pressurized operation.When vented, the flow of reaction gas was stopped. Thepressure in the DRC under vented conditions is calcu-lated to be ;0.025 mtorr. For most of the pressurizedDRC work, the reaction gas was semiconductor gradeammonia that was passed through a getter purifier(Nupure Corporation, Manotick, Ontario, Canada) toremove oxides, including water. The getter was oper-ated at room temperature under which conditions itdoes not efficiently remove hydrocarbon impurities. Forsome of the early work the reaction gas was a mixtureof 50% NH3 in N2, and one experiment employed areaction gas mixture of 40% H2 in He. The reaction gaspressure was controlled by regulating the flow ofreaction gas into the cell, using an all-metal 1-sccm (or3-sccm) mass flow controller (MKS Instruments, An-dover, MA, model 1479 MFC); the minimum detectionlimit for Fe was obtained at a flow of ;1.2 mL/min andthat for Ca at ;1.5 mL/min, corresponding to calcu-lated DRC pressures of 10 and 12 mtorr, respectively.

Experimental Results and Discussion

A number of spectra were recorded in order to assessthe importance of venting the reaction cell when reac-tion gas is not introduced (Figures 2 and 3), the impactof the bandpass on the intensities of background andanalyte signals (Figures 4–6), and to demonstrate thesynchronization of the bandpass with the mass analyzer(Figure 7). As noted above, these data were obtained ondifferent instruments a year apart. Because the experi-ments were intended to provide comparative datawithin a series, no particular attempt was made tooptimize analytical performance (this will be the subjectof a separate communication). For convenience, the ionsignals at the m/z of several of the analyte elements havebeen extracted from these spectra, and are provided inTable 1. These data are identified by the correspondingfigure number and by the sample identification, beingeither a blank (DDIW) or a 1-ppb mixed analyte sample.While some information relating to the DRC conditionsis provided, full reference is to be made to the figurecaptions and to the text associated with the figures. Insome instances the ion signal saturated the pulse count-ing channel, and the entry is left blank in this case. Theearliest data (obtained on instrument A) were recordedwith a measurement time of 0.1 s per data point, and sothese signals (cps) appear in increments of 10 cps; thelater data (obtained on instrument B) were recorded at2 s per data point. The net signals (signal for standard 2signal for blank) are provided in Table 2.

Spectra obtained for a blank (DDIW) solution and fora 1-ppb mixture of 18 elements are shown in Figure 2.They were obtained with the dynamic reaction cellvented to the high vacuum chamber and without addi-tion of reaction gas (cell pressure ;2 3 1025 torr) and atlow q (q 5 0.11, where the rf amplitude was fixed at

200 Vrf and the frequency is scanned with mass). Underthese conditions, the dynamic reaction cell behaves as aconventional rf-only ion guide. The rod offset potentialof the cell in this mode is 210 V so that the (mass-dependent) ion kinetic energies are in the range of13–20 eV, and the ions experience ;60 cycles of the rfduring transit through the cell. These spectra are similarto those obtained using a conventional ICPMS instru-ment, with the exception of a relatively low continuumbackground (see, for example, Figure 3a of [15]). It issignificant that nearly identical spectra are obtained athigher q (e.g., q 5 0.25). The latter result indicates thatthe ionic composition is not affected by the low masscutoff of the bandpass. This is to be expected undercollision-less conditions. Illumination of the off-axisdetector by photons and Ar* metastables, being themajor contribution to the continuum background inICPMS, is attenuated by the shadow stop and bycollimation through the entrance and exit apertures ofthe reaction cell.

Spectra obtained with the DRC “closed” (that is, withthe vent to the high vacuum chamber sealed, so that gasenters and exits only through the entrance and exitapertures) are shown in Figure 3. Under these condi-tions, the pressure in the cell is ;1.0 mtorr and ion flowthrough the cell is dominated by collisions. The gas thatfills the cell is entrained plasma gas sampled from theion optics chamber through the entrance aperture. Forconventional pneumatic nebulization of aqueous sam-ples without desolvation, this gas contains up to 17% Hand O (derived from the water sample) in the bulk Ar[16], in addition to other contaminants derived eitherfrom the sample (i.e., acid) or the vacuum system (i.e.,vacuum pump oil). As a result, the conditions providefor a rich and complex ion–molecule chemistry. For themost part, the sensitivity to the analyte ions is largelyunaffected; that is, there is not significant reaction orscattering of the analyte ions. (The exceptions to thisappear to be Na and Mg, which are the lowest massanalytes and are most likely to suffer scattering in thepredominantly Ar gas, although the result may alsoreflect the imbalance of the DRC power supply, whichwill be most apparent under collisional conditions.)However, the DDIW blank spectrum is significantlymore populated. Relative to Figure 2, the blank spec-trum is elevated at nearly every mass (with the excep-tion of argide ions, which are suppressed by less than afactor of 3). The most notably increased backgroundions include m/z 5 27, 28, 29, 30, 55, 64, 66, 68, and 69.These ions are probably oxides, nitrides, and hydratesof plasma-based ions, and perhaps ionized organiccontaminants (particularly the latter two ions). Unlikethe vented cell results, the background DDIW spectrumis affected by the operating point, q. Increasing q to 0.25yields a reduction in the spectral background, mostnotably at higher m/z. Clearly, operation of the cellunder collision conditions in the presence of gas havingeven trace level contaminants (e.g., the organic compo-nent in the present case) can produce new interference


Tab

le1.

Ion

sign

als

(cps

)fo

ran

alyt

eel

emen

tsex

trac

ted

from

Figu

res

2–7

Fig

ure

nu

mb

er2

23

34

45

56

67

7

Inst

rum

enta

BB

BB

AA

AA

AA

BB

Sam

ple

DD

IW1

pp

bD

DIW

1p

pb

DD

IW1

pp

bD

DIW

1p

pb

DD

IW1

pp

bD

DIW

1p

pb

Rea

ctio

ng

asn

on

en

on

ep

lasm

ag

asp

lasm

ag

as1:

1N

H3/N

21:

1N

H3/N

21:

1N

H3/N

21:

1N

H3/N

21:

1N

H3/N

21:

1N

H3/N

2N

H3

NH

3

Gas

pre

ssu

re10

25

torr

102

5to

rr1

mto

rr1

mto

rr20

mto

rr20

mto

rr20

mto

rr20

mto

rr20

mto

rr20

mto

rr9

mto

rr9

mto

rrq

or

freq

uen

cyq

50

.11

q5

0.1

1q

50

.11

q5

0.1

13.

58M

Hz

3.58

MH

z1.

53M

Hz

1.53

MH

z1.

53M

Hz

1.53

MH

zq

50

.45

q5

0.4

5a

or

Vd

ca

50

a5

0a

50

a5

00

Vd

c0

Vd

c0

Vd

c0

Vd

c14

.25

Vd

c14

.25

Vd

ca

50

a5

0

Ele

men

t23

Na

156

1586

512

380

2647

084

140

00

012

6064

9924

Mg

2110

801

1649

8050

4071

00

00

526

6327

Al

196

1512

210

440

2305

460

7493

010

00

860

6139

K44

3148

347

1950

4061

222

074

0310

4741

243

40C

a87

054

267

650

5269

880

1007

540

722

193

48T

i17

222

675

1200

1980

220

079

010

610

022

07

571

52C

r35

037

6352

815

219

3875

631

800

5862

140

2434

00

3391

3820

394

55M

n43

2040

422

1115

650

256

120

4254

010

3591

1135

469

56Fe

5068

879

878

420

1486

110

2220

1217

658

58N

i19

315

3501

110

746

2933

783

638

8917

644

052

8120

350

219

6936

60N

i6

6197

183

7919

8757

284

526

190

2020

012

076

2809

63C

u11

1309

461

1680

992

8420

301

190

5101

026

023

674

4965

Cu

3362

9221

282

3127

055

2230

2120

040

9635

5364

Zn

5867

2440

3312

299

250

7623

2047

610

8011

5913

66Z

n69

3892

1108

6087

7044

1130

2820

020

734

8475

As

1234

5577

5436

5812

5612

1190

1250

00

763

aIn

stru

men

tA

isth

eea

rly

vers

ion

(fix

edb

and

pas

sD

RC

po

wer

sup

ply

),d

ata

reco

rded

07M

ay19

97;i

nst

rum

ent

Bis

the

late

rve

rsio

n(d

ynam

ical

lysc

ann

edD

RC

po

wer

sup

ply

),d

ata

reco

rded

10A

pr

1998

.


ions within the cell itself. Further, the intensity of theinterference is a function of the operating point, q. It isconcluded that the reaction cell provides an environ-ment which promotes sequential reactions which mayeventually produce a ubiquitous ionic background.Increasing the value of q, by reducing the rf frequencyin this case, increases the low mass cutoff. If one of the

intermediate product ions of the sequence of reactionsfalls below this cutoff, the remainder of the sequence isterminated and the background interference is reduced.This is of particular importance at higher masses, be-cause these interference ions almost certainly arise froma complex series of reactions, and operation at higher qat higher mass allows destabilization (ejection) of awider range of lower mass (intermediate) ions. It willlater be shown that this broad-band ejection can beapplied for the analysis of low mass analyte ions as wellby application of a dc component (a) which provides ahigh mass cutoff in addition to the low mass cutoff.

The spectra of Figures 4–6 were obtained by holdingthe frequency and amplitude of the rf constant through-

Table 2. Net sensitivity (cps) for 1-ppb mixed element solution

Figure number 2 3 4 5 6 7

Instrumenta B B A A A BReaction gas none plasma gas 1:1 NH3/N2 1:1 NH3/N2 1:1 NH3/N2 NH3

Gas pressure 1025 torr 1 mtorr 20 mtorr 20 mtorr 20 mtorr 9 mtorrq or frequency q 5 0.11 q 5 0.11 3.58 MHz 1.53 MHz 1.53 MHz q 5 0.45a or Vdc a 5 0 a 5 0 0 Vdc 0 Vdc 14.25 Vdc a50

Element23Na 15709 7903 7944 0 0 523924Mg 10780 4964 4021 0 0 265827Al 14926 12614 7433 10 0 605339K 43916 38662 7183 4019640Ca 53397 52048 9995 2178548Ti 22503 18602 590 600 220 56452Cr 28491 23537 26821 24300 3391 2035655Mn 36103 39100 42420 3581 3545856Fe 29190 14441 2210 1764558Ni 15695 18591 5538 4841 330 671760Ni 6191 7736 23046 1830 120 273363Cu 13082 16748 11016 4911 260 721365Cu 6259 8018 5252 2090 40 345764Zn 6666 8265 7373 4741 80 590266Zn 3823 4980 4341 2790 20 347775As 3443 5359 2200 60 0 56

aInstrument A is the early version (fixed bandpass DRC power supply), data recorded 07 May 1997; instrument B is the later version (dynamicallyscanned DRC power supply), data recorded 10 Apr 1998.

Figure 2. Spectra recorded without reaction gas and with thereaction cell vented to the high vacuum chamber. The electricalconditions within the reaction cell included the following: the rfamplitude was 200 V (peak-to-peak), the dc voltage between polepairs was near zero (a 5 0.005), and the rf frequency was scannedin concert with the mass analyzer to yield q 5 0.11 (calculated forthe mass currently being filtered in the downstream mass analyz-er). The darker shaded spectrum was obtained for a mixture of 18elements (B, Na, Mg, Al, K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, As, Sn,Sb, Au, Pb) at 1 ppb (1 ng/ml) each. The lighter shaded spectrumwas obtained for a blank DDIW sample.

Figure 3. Spectra obtained under the same conditions as Figure2, but with the reaction cell closed (not vented to the high vacuumchamber).


out the mass scan. Hence, q and a (when Vdc isnon-zero) decrease linearly with mass. That is, thedynamic reaction cell was held at a constant bandpasscondition without reference to the mass analyzer. Forthese data, the reaction cell was pressurized with amixture of 1:1 NH3:N2. Neither the pressure within thecell nor the flow of reaction gas into the cell weremeasured at that time; the cell pressure (;20 mtorr) wasinferred from the pressure in the high vacuum chamberby assuming effusive flow through the exit aperture.Throughout this article, we reference mass bandpasswindows which are appropriate for collision-less con-ditions (i.e., that are given by the conventional quadru-pole stability limits). It is recognized that stability isaffected by collisions, and this will be presented in afuture article. Our experience is that collisions make theboundary conditions somewhat diffuse, and particu-larly the bx 5 1 boundary (the high-q boundary) isaffected by promotion of endothermic reaction loss dueto the nonthermal contribution of the rf-field energy[12].

The spectra of Figure 4 were obtained with operationof the reaction cell at 3.58 MHz at 200 Vrf and withVdc 5 0 (a 5 0). In this mode, the reaction cell acts as ahigh pass filter with a low mass cutoff near m/z 5 5amu (corresponding to q 5 0.908). Under these condi-tions, most ions are stable within the reaction cell, andthis provides for a rich ion–molecule chemistry. Com-parison with Figure 2 shows the significant increase inchemical background, at nearly every m/z, resultingfrom this chemistry. A few of the analyte ions (notablyTi, Ni, and As) are reactive with NH3, and the sensitiv-ity to these is accordingly suppressed. Whereas thesensitivity for nonreactive analyte ions remains roughlysimilar to that observed in the vented mode (Figure 2),the detection limits for normally uninterfered ions isdegraded by the elevated chemical background.Clearly, the chemistry provided by the reaction cell iseffective at reducing certain plasma-based interferences

(note the reduction of the argide ion signals at m/z 536–41, 56, and 80), but the conditions promote theformation of new interferences within the reaction cellthat limit the effectiveness of the cell for analyticalpurposes.

Decreasing the rf frequency to 1.53 MHz provides adramatic improvement, as seen in Figure 5. Still oper-ating as a high pass filter (rf-only), the reduced fre-quency increases the low mass cutoff to ;m/z 5 27amu. In this mode, the dominant primary and second-ary product ions in the cell, NHx

1, are unstable. Becausethe instability is achieved quickly, these ions have littletime to react with other neutral species in the cell, andthe chemical background is substantially reduced. Thisis a clear case of reduction of interferences by ejection ofthe intermediate products of sequential reactions withinthe dynamic reaction cell. Whereas there remains aremnant of the interferences produced within the cell,particularly at higher m/z, the detection limits aregreatly improved (notably for the normally interferedions 40Ca1, K1, and 56Fe1). As will be shown, there isadvantage in yet reducing the bandpass of the reactioncell, and this may be enacted even for rf-only operationby scanning the frequency (affecting the low masscutoff) with the analyte mass being passed through thedownstream mass filter. It will be recognized thatsignals for low mass ions (e.g., Na1, Mg1, and Al1) aremissing from the spectra of Figure 5, and that this is aresult of their instability below the low mass cutoff.Analyte ions that are reactive with NH3 (Ti, Ni, and As)remain suppressed. The sensitivities to the higher massanalyte ions (above m/z 5 60) are suppressed relative toFigure 4, although it is not clear why this should be.This may relate to the frequency dependence of the dcimbalance of this early DRC power supply.

A yet further improvement in signal to noise isachieved with the application of a dc voltage betweenpole pairs (i.e., the introduction of an a component). Thespectra of Figure 6 were obtained under the sameconditions as those of Figure 5, except with the additionof 14.25 Vdc between pole pairs. Operation in this mode

Figure 4. Spectra obtained for the samples of Figure 2, with thereaction cell closed and pressurized to ;20 mtorr with 1:1 NH3:N2.The rf frequency was held constant at 3.58 MHz (Vrf 5 200, Vdc 50).

Figure 5. Spectra as recorded for Figure 4, but with the rffrequency reduced to 1.53 MHz.


provides a bandpass window in the range between 30and 86 amu. (These limiting masses correspond to theboundary limits of a collision-less quadrupole; in real-ity, the bandpass width is significantly less than thisbecause of destabilizing or reactive collisions.) It isnoted that the values of a and q decrease with ion masswhen operating with a fixed bandpass (fixed rf and dcamplitudes and frequency). Accordingly, transmissionefficiency in this instance is a complex function of ionmass. In this case, the transmission efficiency maxi-mizes near m/z 5 40 amu, for which q ; 0.6 and a ;0.09. Although the addition of the a component reducesthe sensitivity, it also intercepts chemistry producinginterference signals which involve higher mass interme-diate products. The net result can be yet improvedlimits of detection for analyte ions that are within theoptimum window of the bandpass, with the virtualelimination of interferences which result both fromplasma-based ions and sequential chemistry within thereaction cell.

An improved version of the dynamic reaction celladjusts the bandpass of the cell in concert with theanalyte mass being passed through the downstreamfilter according to the user-selected values of Vrf, q, anda. These three parameters define the amplitude of thebandpass dc voltage, Vdc, and with the analyzer massdefines the rf frequency. Hence, the dynamic reactioncell is frequency scanned in concert with the massanalyzer. Accordingly, a bandpass window is not di-rectly observed in the resulting mass spectrum, but thebenefit of interfering chemical background reduction isobtained. Spectra obtained at Vrf 5 200, q 5 0.45, anda 5 0.005 with a 100% NH3 reaction gas pressure of ;9mtorr are given in Figure 7. Other than the plasma-based ions NO1, NO2

1 and Ar21 (for which NH3 is an

ineffective reaction gas) and an unknown interferenceat m/z 5 68, the residual signals in the blank solutionappear to result from elemental contamination (i.e., Cr,Ni, and Cu) as these ions appear at their expectednormal isotope distributions. Na is clearly present in theblank, and the residual signal at m/z 5 28 may be due to

Si. We believe that the residual signals at m/z 5 39 and40 correspond to 39K1 and 40Ca1 with some residual40Ar1 (optimization for Ca occurs at slightly higherNH3 pressure). The m/z 5 41 peak is too large to be theisotope of K; one of the primary reaction products ofAr1 with NH3 is ArH1, and the NH3 pressure used forthese spectra was insufficient to ensure the completesecondary reactive loss of this argide ion. Higher NH3

pressure reduces the m/z 5 41 peak to the correct Kisotope ratio, but also reduces the sensitivity to certainof the analyte elements (e.g., Ni) which react slowlywith NH3. Hence, there is an optimum reaction gas andreaction gas pressure, as well as q and a, for eachanalyte species. However, NH3 at 10-mtorr pressure atq ; 0.5 and a ; 0.005 is a good compromise for manyof the normally interfered analytes.

Although Al is not severely interfered under ventedoperation (Figure 2), the detection limit is yet improvedusing NH3 reaction gas (Figure 7). This is principallybecause N2

1 reacts rapidly with NH3, and this sup-presses the ion signal at m/z 5 28 and reduces itsabundance sensitivity overlap on m/z 5 27 to below thecontinuum background signal (as discussed byBollinger and Schleisman [17]). Operation with a reac-tion gas of 40% H2 in He at a cell pressure of ;5 mtorrresults in a dramatic increase in the background signalat m/z 5 27, most probably due to the ionization of ahydrocarbon contaminant and the formation of C2H3

1.This allows a convenient demonstration of the use of anon-zero value for the parameter a to improve signal/noise. The spectra shown in Figure 8a were obtainedwith 200 Vrf at 1.91 MHz and with Vdc 5 0. Under theseconditions, the cell operates as a high pass filter forwhich ions above m/z 5 17 amu are stable. The back-ground equivalent concentration of Al is approximately300 ppb. Addition of 14.25 Vdc (a 5 0.08 at m/z 5 27)dramatically reduces the spectral background withoutsignificantly affecting the analyte sensitivity (Figure8b). This changes the mass bandpass of the collision cell

Figure 6. Spectra as recorded for Figure 5, but with 14.25 Vdc

between pole pairs.Figure 7. Spectra obtained for the samples of Figure 2, but withthe reaction cell closed and pressurized to ;9 mtorr with NH3.The rf frequency was scanned in concert with the downstreamanalyzer mass to maintain q 5 0.45 at Vrf 5 200, with Vdc ; 0(a 5 0.005).


to include the range between 19 and 55 amu (calculatedfor collision-less conditions). The C2H3

1 interference ionmost probably arises from fragmentation of a highermass hydrocarbon ion. In this case, the introduction ofthe high mass cutoff apparently rejects the higher massprecursor ion from the cell, and hence the fragment ionat m/z 5 27 is concomitantly suppressed. It remains alesser possibility that the C2H3

1 ion is formed by dissocia-tive ionization of a hydrocarbon by a lower mass precur-sor ion, and that the change in the stability of low massions by introduction of non-zero a suppresses that precur-sor ion and its consequent product ion at m/z 5 27. Theuncertainty in the interpretation might have been resolvedby repeating the experiment under rf-only conditions atslightly lower frequency. In either event, the addition ofthe non-zero a improved the signal/noise for Al determi-nation by approximately an order of magnitude.

The use of non-zero a also has value when a reactiongas is not introduced. The effect is most pronouncedwhen the cell is not vented so that it is pressurized byplasma gas. (As noted above, this mode of operationhas little analytical value, but it does exaggerate theeffect to be discussed). In this case the major neutralspecies in the cell is Ar. Ar metastable neutrals, Ar*, are

produced in energetic collisions of Ar1 with Ar. Ourexperience is that this process is prominent when theAr1 ion kinetic energy exceeds about 15 eV. The pro-duction of Ar* is probable either within the cell or nearthe exit of the cell. Once produced, the Ar* may effusefrom the cell and are not affected by the electrical fieldswithin the mass analyzer. A signal may be produced atthe detector if the Ar* strikes the detector or emits aphoton which strikes the detector, and this signal con-tributes to the continuum background signal of theinstrument. Clearly, the contribution of this process tothe continuum background is reduced if Ar1 is unstablewithin the reaction cell. Two spectra are shown inFigure 9. Both spectra were taken for a sample contain-ing 10 ppb Li and with the reaction cell operated at 150Vrf at a fixed frequency of 2.92 MHz, without ventingand without added reaction gas. With Vdc 5 0, the cellpasses all ions above 5 amu, and the spectrum shows asubstantially elevated continuum background signal.With the addition of 12 Vdc, the bandpass of the cellincludes the mass range 6–16 amu, which excludes Ar1,and the continuum background signal is virtually elim-inated without substantial loss of signal for 7Li1 (the6Li1 signal is suppressed as this mass is at the edge ofthe stability boundary).

The stability of an ion in a quadrupole field withcollisions is affected by a number of parameters includ-ing scattering, energy loss, energy contribution from therf field, ion/neutral mass ratio, reactivity, and thermo-chemistry. Although it is important to understand theinteraction of these factors, the subject is beyond thescope of this work. However, it is necessary to point outthat the selection of q (and of a) is important for optimalperformance. Figure 10 shows the ion signal at m/z 5 56as a function of q obtained using NH3 reaction gas at ;9mtorr, with 200 Vrf and a 5 0.005. Data are presentedfor both a DDIW blank solution and a 100 ppt (0.1

Figure 8. Spectra obtained in the vicinity of Al (m/z 5 27) withthe reaction cell closed and pressurized to ;5 mtorr with 40% H2

in He. Spectra are shown for the blank DDIW sample (solid line)and for 100 ppb Na, Mg, and Al (dashed line). (a) The reaction cellwas operated at Vrf 5 200 and the frequency was held constant at1.91 MHz, with Vdc 5 0. (b) Spectra as per (a), but with theaddition of 14.25 Vdc between pole pairs of the reaction cell.

Figure 9. Spectra recorded for a sample containing 10 ppb Li.The reaction cell was closed and operated at Vrf 5 150, and the rffrequency was held constant at 2.92 MHz. The spectrum given bythe solid line was obtained with Vdc 5 0, while the dashed linewas obtained for Vdc 5 12. The spectra were recorded using a100-ms dwell time; the spikes to 10 cps correspond to single ioncounting during the dwell period.


ng/mL) Fe sample. It is evident that the transmission(stability) of elemental ions is suppressed at both lowand high q. This has been observed for every atomic ionthat we have studied, although the degree of suppres-sion, particularly at high q, is dependent on the ther-mochemistry of the reaction system (a slightly endo-thermic process may be enabled at high q due to theaddition of energy from the rf field [12]). Of significantimportance and interest is the response of the blanksignal at low q. Clearly, the increase of the blank signalnear q 5 0.15 is not due to Fe in the blank. Rather, thismust correspond to a m/z 5 56 interference ion which isproduced at low q, and its production must involve(either directly or indirectly) a lower mass ion which isstable in this region. Recall that q is inversely propor-tional to ion mass (eq 2). On the assumption that thelower mass precursor ion is NH4

1, which is the domi-nant ion within the reaction cell, we have provided aseparate q-scale in Figure 10 corresponding to m/z 518. It appears that the interference peak in the DDIWdata corresponds roughly to the stability for this ion (itcould equally well be ascribed to OHx

1 or NHx1, where

x 5 0 to 3). It is not necessary to identify the precursorion, but to recognize that it subsequently reacts eitherdirectly or indirectly to form a new interference atm/z 5 56 amu, and that this interference may be elimi-nated by operation at higher q. It will be understoodthat it is important to recognize the contribution of suchpotential interferences, particularly when complex (i.e.,indiscriminate) chemistry is employed. Of course, theproblem is most serious when the chemistry is directand the precursor ion is close in mass to the analyte ion;in this case, it may not be possible to efficiently sup-press the formation of the interference ion, and analternate chemistry is prescribed.

Conclusions

The dynamic reaction cell offers tremendous flexibilityin chemical approaches to isobaric interference suppres-sion. An enormous range of chemistries are known andmay be developed. The application of these chemistriesmay be limited, in some cases, by the appearance ofnew interferences produced in sequential reactionswithin the reaction cell. The use of a quadrupole con-finement device with an adjustable bandpass allowssuppression of these chemistries by rejection of inter-mediate product ions of these sequential reactions, withthe result that the interference is concomitantly re-duced. This additional “electronic control” of the chem-istry facilitates the use of yet more complex ion–mol-ecule chemistry. It is to be expected that substantialmethod development will occur, with the evolution of a“cookbook of recipes” of chemistries that are tailored tospecific applications. Nonetheless, we have found thatNH3 is an attractive reaction gas, as it offers ratherexceptional suppression of most argide interferenceions. It has also been shown that venting the cell to thehigh vacuum chamber, in order to reduce the pressureand thus the number of collisions within the cell, is abenefit which realizes “conventional” ICPMS perfor-mance in the reaction cell configuration.

We will report analytical figures of merit for thedynamic reaction cell in a separate paper. We intend tocontinue our investigation into the stability of ions inthe DRC.

References1. Douglas, D. J. Canad. J. Spectrosc. 1989, 34, 38.2. Rowan, J. T.; Houk, R. S. Appl. Spectrosc. 1989, 46, 976.3. Koppenaal, D. W.; Barinaga, C. J.; Smith, M. R. J. Anal. At.

Spectrom. 1994, 9, 1053.4. Eiden, G. C.; Barinaga, C. J.; Koppenaal, D. W. J. Anal. At.

Spectrom. 1996, 11, 317.5. Eiden, G. C.; Barinaga, C. J.; Koppenaal, D. W. Rapid Commun.

Mass Spectrom. 1997, 11, 37.6. Koppenaal, D. W.; Barinaga, C. J.; Smith, M. R. Rapid Commun.

Mass Spectrom. 1994, 9, 1053.7. Turner, P.; Merren, T.; Speakman, J.; Haines, C. In Plasma

Source Mass Spectrometry: Developments and Applications; Hol-land, G.; Tanner, S. D., Eds.; The Royal Society of Chemistry:Cambridge, 1997, p 28.

8. Douglas, D. J. J. Am. Soc. Mass. Spectrom. 1998, 9, 101.9. Douglas, D. J.; French, J. B. J. Am. Soc. Mass. Spectrom. 1992, 3,

398.10. Ervin, K. M.; Armentrout, P. B. J. Chem. Phys. 1985, 83, 166.11. Gerlich, D. Adv. Chem. Phys. 1992, 82, 1.12. Baranov, V. I.; Tanner, S. D. J. Anal. At. Spectrom. 1999, 14, 1133.13. Handbook of Chemistry and Physics, 53rd ed.; Weast, R. C., Ed.;

CRC: Cleveland, 1972.14. Anicich, V. G. Ap. J. Suppl. Ser. 1993, 84, 215. See also

http://astrochem.jpl.nasa.gov/asch/15. Tanner, S. D. J. Anal. At. Spectrom. 1995, 10, 905.16. Niu, H.; Houk, R. S. Spectrochim. Acta 1996, 51B, 779.17. Bollinger, D. S.; Schleisman, A. J. Plasma Source Mass Spectrom-

etry: New Developments and Applications; Holland, G., Tanner,S. D., Eds.; The Royal Society of Chemistry: Cambridge, 1999,p 80.

Figure 10. Ion signal profile for m/z 5 56 as the rf frequency isvaried to yield a scan of q (calculated for the m/z 5 56 ion) from0.06 to 0.9. For this data, Vrf 5 200 and Vdc ; 0 (a 5 0.005). Thereaction cell was closed and pressurized to ;9 mtorr with NH3.The solid line was obtained for a sample containing 100 ppt (0.1ng/mL) Fe. The dashed line was obtained for a blank DDIWsample (and the blank signals have been multiplied by a factor of10). The inset scale gives q calculated for an ion of m/z 5 18.


Documents

A Dynamic Reaction Cell for Inductively