Spectrometry

Resonance-enhanced multiphoton ionization of argon: reactivity asa probe for the

M. Schweizer, S. Mark, D. Gerlich*

fur Physik, Freiburg, 79104 Freiburg, Germany

(Received 27 October 1993; accepted 22 November 1993)

Abstract

The guided ion beam (GIB) technique is not only well suited for determining precise integral cross-sections but is alsouseful for analyzing state populations in simple systems based on their chemical reactivity. In this paper, the method has been applied to characterize some multiphoton ionization schemes via different resonant intermediates of Ar utilizing theendothermic charge-transfer reaction + + Ar as a probing reaction. Cross-sections measured in the threshold region indicate that ionization via the and states leads to andproduction with a purity of (97 3)% and (95 respectively, whereas other transitions lead to mixtures. Addi- tional studies of collisions betweenstate-selectedArf ions with and and comparison with published values notonly corroborate these conclusions but also provide new state-specific information for these systems, especially at lowcollision energies.

Key words: Resonance-enhanced multiphoton ionization; State-selected reactions; Charge transfer; Guided ion beam, ions

1. Introduction

Many attempts have been made in the last two decades to measure very detailedsections for collision processes. One of thegoals of such studies is, for example, to findspecific forms of energy which influence pre-dominantly the outcome of a chemical reac-tion. A more fundamental aim is to understandthe dynamicsof the interaction between the

*Corresponding author. New address: Fakultatwissenschaften, Technische UniversitatPF 964, Chemnitz, Germany.

Dedicated to Professor Christoph Ottinger on the occasion ofhis 60th birthday.

tants from first principles and to develop theore-tical methods which are able to predict theexperimental results. For critical comparisons between theory and measurement very accuratestate-to-state cross-sections are needed.Although the expression state-to-state has beenin use many years, there are few experimentswhere the states of the reactants are preparedas well as those of the products are analyzed.

For preparing ionic reactants suitable prop-erties of the ionization process itself can beutilized. Among the various methods such aselectron bombardment, chemi-ionization orphotoionization, the last one has the largestpotential. Single-photon ionization with

1994 Science B.V. All rights reserved0168-1

2 M . Schweizer J . Mass Spectrom. Processes 135 (1994) 1-17

monochromatic VUV photons from a contin-uous discharge lamp was pioneered byChupka and Russell today in most instances synchrotron radiation or VUVlasers are applied. Alternatively one can com-bine the energy of several photons for ejectinga single electron. This has the advantage thatthe required wavelength is in the visible or nearUV region, where intense laser sources areavailable. If one uses, in addition, excitationvia a resonant intermediate state, ions can beproduced with high efficiency and also in stateswhich otherwise cannot be reached.

Photoionization methods have the advan-tage that the manifold of formed states canbe limited by energy constraints or restrictedby propensity or selection rules, in particular ifresonant intermediates or autoionizing levelsare involved. Nonetheless in most situations,it is advisable to analyze the state distribution of the formed ion ensemble, by recording aphotoelectron spectrum (PES). If ionizationwith monochromatized photons leads to ionsin several states, one can pick out individualones by coincidence techniques, correlating one ion with one energy-selected photoelec-tron. Especially well suited for this purposeare threshold electrons and first promising developments have also been reported withzero-kinetic-energy photoelectrons Thehigh-energy resolution of this technique allows one to prepare molecular ions inselected rotational states

If one uses for ionization other methodsthan photoionization the state of the createdion cannot be determined by electron spectro-scopy. There are also other approaches toinfluence the state population of reactantssuch as chemical quenching, thermalization of an ensemble at various temperatures (inparticular at low ones), optical excitation andlaser pumping. In all these instances more gen-eral methods are needed to analyze the result-ing populations; based on lasers orchemical The last techniaue

has been used for many years to characterizeautoionizing transitions in and it isapplied routinely in drift tube experiments for analyzing reaction products Morerecently this approach has also been com-bined successfully [6] with the guided ionbeam (GIB) technique which is known to bea powerful method for determining preciseintegral cross-sections in particular, incombination with VUV and multiphotonionization sources

In this paper we discuss the potential butalso the limitations and problems of the GIBtechnique for determining state-specificsections and for probing ionic states by chemi-cal reactions. As a test case we use a simpletwo-state mixture of ions prepared byresonance-enhanced multiphoton ionization(REMPI) via several intermediates. Reactions of these ions with several molecules are examinedfor the specific purposeof seeing whether they aresuited for determining the : fine-structure ratio of the reactants.

It should be noted that in the present experi-ment the ion ensemble is probed by collisionsin the same scattering cell which is otherwiseused to determine state-specific cross-sections.This has the advantage that collisional relaxa-tion is accounted for, as is radiative decay orfragmentation of molecular ions en route tothe interaction region. In the specific case of REMPI the in situ analysis is also superior toelectron spectroscopy in certain respects. Here the electron energy distribution may be per-turbed and the ion state population candepend critically on parameters such as laserpower density. It is therefore problematic totransfer results from one experiment to another.

2. Experimental

2.1 Instrumental details

The was in a well

M . Schweizer et J. Mass Spectrom. Processes 135 (1994) 1-17

Fig. 1. photoionization source. Part (a) shows the most important features in the plane which is formed by the laser beam, the gas beam and the quadrupole axis: (1) piezoelectricvalve, (2) skimmer, (3) ion repeller, (4) exit ion lens. Part (b) illustrates that thehyperbolic boundary conditions of the field are approximated by four groups of 15 rods each. For diagnostic purposes, ionization is also possible with an electron beam which is perpendicular to the laser beam.

characterized GIB apparatus which has beendescribed in detail elsewhere One of theoutstanding features of this machine is that itcan be used to measure absolute values forintegral cross-sections with high accuracyfrom thermal energies to several electronvolts.

The main modification for the present study concerns the ion production. Figure 1 shows asketch of the photoionization source. The ions are created inside a linear quadrupole. The hyperbolic boundary conditions of this deviceare approximated by 4 x 15 rods with dia-meter l mm which are arranged as depictedin Fig. 1. The inscribed circle of this structure has a diameter of This rather largedimension has been chosen to reduce the influ-ence of surface potential distortions in thevicinity of the center line of the quadrupole where the ions move. Another advantage ofthe open construction is that the neutral pre-cursor gas can be pumped away efficiently.

The use of an guiding field in aionization source has the advantage of high

collection and transmission efficiency. In addi-tion, the and energy-dependentfocusingproperties of the harmonic effective potential[7] allow us to prepare a primary beam with avery narrow translational energy distribution.We succeeded in producing an ion beam with an axial energy of only and with anenergy half width of In the presentstudy the quadrupole has been operated in aless selective mode using higher guiding fieldstrength. The resulting energy half width,

= 25 is still smaller than the over-all energy resolution which is determined byother effects such as the thermal target gasmotion. For preparation of an ion beamwe have used a frequency of =

2.4MHz, an amplitude of = 245 V and anaxial kinetic energy of 1

The ionizing laser (30 Hz Quantel pumped dye laser with doubling and mixing)passes through the quadrupole perpendicu-larly and intersects the molecular beam onthe axis of the ion guide (see Fig. 1). The

4 M. Schweizer et J. Mass Spectrom. Ion Processes135 (1994) 1-17

TableREMPI of Ar using several intermediate states. Columns give the required wavelength, the number of photons, the cross-sectionmeasured for the + charge-transfer system at and the fraction of ground-state ions determined for the indicatedtransitions

Configuration Designation REMPI 0.4 Fraction

impact

light beam is focused by a lens. For3+ 1 and 3+ 2 REMPI of Ar the dye laserhas been operated with DCM. By doublingthe output a wavelength between 314 and

has been obtained with an energy of2 per pulse and a resolution of 0.08 .For 4 + 1 REMPI a mixture of Rh B and Rh6G has been used. Mixing the output of thedye laser with the fundamental of the YAG leads to 373 nm with an energy of 8 perpulse. Since there was no interactivity etalon in the YAG laser the resolution of the mixedlaser beam was only about 1 Dependingon the imaging and focusing conditions the power density of the laser light in the ioniza-tion volume could be varied between and

wThe neutral precursor gas is injected coaxi-

ally with the quadrupole axis. In order to pro-duce 20-30 ions per laser shot an argondensity of about is required in the laser focus. To achieve this, there are two alter-native beam sources, a pulsed supersonicbeamas depicted in Fig. or a 1 mm diameter tube ending very close 3mm) to the laser focus. The former has the advantage of cooling theneutral gas during the adiabatic expansion, leading for example to low rotational tempera-tures. The effusive gas inlet is superior if adia-batic cooling is not necessary, as for atoms, ornot desired, if one needs neutral precursors in high rotational states.

2.2 REMPI of Ar

The ionization potential of Ar lies atBy spin-orbit interaction the

ground state is split into two levelswhich are separated by AE = Inorder to ionize Ar with four or five photonsthe wavelength must be shorter than

or respectively. For aresonant transition via an intermediate withthe main quantum number 4 one can selectour of a total of seven accessible states(Table 1) resulting in 3+ 1, 3+ 2 or 4 + 1REMPI schemes In the present work wehave examined only those four transitionswhich are shown schematically in Fig. 2.With three photons of orthe 4s or states are reached,while with four photons of 373.6 nm or372.8 nm we reach the or 4p1

states, respectively. Note that the illustration uses different scales in different regions, butthe relative position of the intermediate states can be compared and the range in the vicinityof the ionization limit (grey area) is at the samescale, too. A typical spectrum for 4 + 1REMPI measured at a laser power of5 x is shown in Fig. 3. The asym-metric line shape is due to stark broaden-ing and varies strongly with the laser power. For example, the half width of the 3+ 1 tran-sition at has been observed to

M . Schweizer et Mass Spectrom. Ion Processes 135 (1994) 1-17

Fig. 2. Resonance-enhanced multiphoton ionization of argon via some intermediate states with the main quantumnumber 4. Note thatthe energy scale in the region of the intermediate states is different from that in the vicinity of the ionization limit.

increase from 0.7 at 2.6 x w toat 2.5 x

The most important question for our appli-cation is whether there are any obvious criteria for finding a correlation between the inter-mediate state and the state of the formed ion. In general, propensity rules are based on the assumption that the core of the intermediate state has already or character andthat this configuration is conserved during the absorption of the subsequent photon. Thisoften holds for Rydberg states but for the main quantum number 4 as in the presentexample one must expect significant devia-tions due to electron-electron correlation.Other criteria are based on accidental resonances with autoionizing states as in the

case of the 4s' transition. For this transition the energy of four photons lies justabove the ionization limit of the state,

in the re ion of Rydberg states whichverge to the state (see Fig. 2). Transitionsvia this path can compete with the absorption of a fifth photon, which means thatture changing (near-) autoionization can com-pete with core-preserving ionization.

Photoelectrons from these two processeshave very different energies, and3.78 and, disregarding discrimination effects, the detection of the two groups of elec-trons should be no problem with a simplephotoelectron spectrometer. Unfortunately, almost no PES work on Ar multiphoton ionization has been reported, and there is

M . Schweizer J. Mass Ion Processes 135 (1994) 1-17

Fig. 3. 4 1 REMPI spectrum of argon via and recorded with a laser power density of 5 x w in the focus. A curve hasbeen superimposed for help with visualization.

especially no information on excitation via theor the intermediate. Some notes on

electron spectra from transitions via the 4s'and the 4s intermediate can be found in the

thesis of Orlando et al.have ionized Ar, mainly for calibration pur-poses, via the 4s' intermediate. For this transi-tion both studies conclude the electron energy distribution that there is a strong preference for forming in the groundstate, which is energetically possible with a3 + 1 process. We have corroborated the factthat in this instance only four photons with314.5 nm are utilized for ejecting the electronby examining the dependence on the laserpower P. For this transition ionization isvery efficient and varies proportionally with

whereas all other transitions produce fewer ions and the yield depends much morestrongly on the laser intensity since at least fivephotons are necessary.

Due to the lack of PES data and for further characterization of the various REMPI pro-cesses we have used our GIB apparatus, andstudied the reactivity of the produced ionsin collisions with molecules such as

and In these studies, the charge-transfersystem of + plays a central role sinceit is, among others, one of the best and mostrecently examined systems

3. Results and discussion

3.1 Calibration of statepopulation

It is known from measurements in a drift tube that the rate coefficient of thecharge-transfer system + has a pro-nounced energy dependence. At room temperature it has only about 10% of thegevin limit. With increasing energy the ratecoefficient drops steeply and reaches a mini-mum at a collision energy of aboutUp to this energy, only the exothermic reac- tion channel + + Ar cancontribute to formation. The significantincrease at collision energies above 0.3 iscaused by the opening of the endothermic channel (a)+ Ar. Experiments with selected ions have shown that in this

M . Schweizer et Mass Spectrom. Ion Processes 135 (1994) 1-17

Fig. 4. Integral cross-sections for the charge-transfer reaction + Ar measured at = 0.1 and for the four ionization pathways depicted in Fig. 2. The observations are in accordance with the fact that at low energies the integralsection is independent of the fine-structure state whereasat energiesin the threshold region for forming ions, excited ionsare more reactive.

energy range the integral cross-sectiondepends on the fine-structure state ofwhereas it is independent of this at low energies.

These observations have also been repro-duced for ions created by REMPI, firstin a more qualitative way. Figure 4 shows afew typical results for two collision energies,

= and = Thetions are plotted for the four different ioniza-tion paths depicted in Fig. 2. At = 0.1they all have the same value within the errorbars, whereas, at = a significantdependence on the transition is obvious. Thisbehavior has been used to calibrate the popu- lation of the fine-structure states and

in the prepared ion beam. The indi-vidual steps of the calibration procedure aredescribed below.

First the measured cross-section has beencorrected for contributions from pro-duction. For this purpose has beenrecorded from thermal energies to a fewtronvolts. The results are depicted in Fig. 5. Inorder to separate the and chan-nel and to account for the thermal broadening,

the data points have been fit by the weightedsum of two empirical analytical functions(broken lines) representing the cross-sectionfor individual channels. The solid lines repre-sent effective cross-sections calculated with weighted sums of these three functions andtaking into account the thermal 300 K targetgas motion

Figure presents results for 3 + 1REMPI via 4s' and 4 + 1 REMPI via

These are the two transitionswhere the reactivity of the resulting ionsis the least and the most, respectively.Theenergy part of the cross-section, where only ground-state ions are energeticallyaccessible, has been described with thealready reported and Gerlich

however with slightly different para-meters (broken line in Fig. 5):

In accordance with other experimentsour results also show that this exothermic charge-transfer channel does not depend onfine-structure energy, which can be taken as

M. Schweizer et J.Mass Spectrom. Ion Processes 135 (1994) 1-17

Fig. 5. Integral cross-sections for the charge-transfer reaction + + as a function of collision energy for four

different ways to prepare ions by REMPI. Upper panel: 3+ 1 REMPI via 4s' and 4 + 1 REMPI via [3/

Lower panel: 3 + 2 REMPI via 4s (x) and 4 + 1REMPI via (0).The solid lines are effective cross-sections calculated using weighted sums of the analyticalcross-sections shown as broken lines. The weights for the boldlines are given in Table 1, the thin solid lines are predictions for pure states.

an experimental hint that fine-structure energy and translational energy are not equivalent. Bysubtracting the part described by Eq. (1) fromthe measured values, the specificcross-sectionsfor forming are obtained. These results show thresholds, which are consistent with thetheoretical values for the two ture states, indicated in Fig. 5 by the arrows. Indetail we calculate, at 0.4 in one instance alow value of 0.1 which shows thatphotoionization via 4s' is predestinated for production. For photoionization via

the created ion ensemble leads to1.9 0.2 indicating that this transition

predominantly produces the state of theSince only two ionic states are accessible

both measured cross-sections are linear combi- nations of and Therefore a first esti- mate of the maximum impurity can besimply achieved for that ion ensemble whichreacts with 0.1 With the assumption thatthis cross-section is exclusivelydue to ionsin the excited state and with the knowl-edge that the ions react with a cross-section larger than the measured 1.9 oneobtains for this impurity a value below 6%.

In order to estimate the fractional abun-dance of for the inter-mediate the cross-section is compared withthat for a statistical mixture of bothorbit states. Since a comparison of absolutevalues determined using different machines under different conditions is not trivial wehave remeasured a reliable absolute cross-section by using electron bombardment, butotherwise identical experimental conditions.For that purpose the ions have been createdusing an electron beam instead of the laser beam. At an electron energy of 60eV it issafe to assume that the two fine-structure states are populated in their statistical weight, : = 2:1. With this mixture we have measured a cross-section of1.0 0.1 at = 0.4 which is, withinthe error bars, consistent with the value pub-lished by Scherbarth and Gerlich Assum-ing again the extreme case thatstate ions do not react we find, as a firstcharacterization of the transition, thatmore than 86% of the produced ion ensembleis in the state.

For a more quantitative analysis, the thresh-old regions of the endothermic cross-section toform have been approximated with anansatz, using the line of center model:

As can be seen from Fig. a set of parametershas been found which lead to good overall

M . Schweizer et J . Mass Spectrom. Ion Processes 135 1-17

agreement with the experimental data. For the charge transfer as a calibration standard:ground state (threshold = 0.345

we have obtained a slow onset with = 1.90and = whereas for ions

= 0.167 the cross-section increases 2more steeply with = 0.95 and = .

These different threshold behaviors (see Fig. 5)which indicate different efficiencies forusing the fine-structure energy, can be under-stood qualitatively with the simple reactionmodel discussed by Scherbarth andGerlich Within that model, translationalenergy has to be transferred into vibrationalexcitation of the molecule by an intimatecollision, before the charge transfer to the

become energetically accessible duringthe dissociation of the collision complex. Starting with on the upper fine-structuresurface, less translational energy must beconverted, which may explain the steepercross-section onset for formation ofThis argument is based on the assumptionthat non-adiabatic coupling is similarfor both manifolds of surfaces converging asymptotically to the two fine-structurestates.

To illustrate the sensitivity and the limits ofour method in determining the fine-structurepopulation, Fig. shows the two effectivecross-sections for a beam of pure, 100%state-selected ions (thin solid lines) and forion beams with impurities of a few percent(bold solid lines). The best overall agreement with the data points has been obtained byassuming a : ratio of 97 : 3 for the4s' transition and of 5 95 for the transi-tion. Summarizing it can be concluded thatthese two transitions lead to almost pure pre-paration of in the andstructure state: the determined deviations have the same magnitude as the errors (seeTable 1).

On the basis of the preceding discussions wepropose to use the following functions for the state-to-state cross-sections of the +

Accounting for the individual experimentalinfluences, these functions allow one to deter-mine the fine-structure state distribution in anyother experiment where the primary ions are created in an unknown mixture of the twostates. As an example we use thesections shown in Fig. They have beenmeasured with ions produced by 3+ 2REMPI via 4s and by 4+ 1 REMPI viaThe data are in good agreement with theheavy lines calculated from the analytical func-tions (3)-(5) (broken lines) using the followingweighting factors. For ionization via 4s, 51of the resulting ions are assumed to be in the

state, for ionization via the com-parison leads to a slightly higher percentage of64%. The uncertainties of these two results areabout 8%. In both instances the population distributions are close to the statistical 2: 1mixture created with electron bombardment.

For a simple quick test we propose to meas-ure the effective cross-section at 0.4 (300 Ktarget temperature) and subtract from this value that for X-state formation using theeffective value

10 M . Schweizer et J . Mass Spectrom. lon Processes 135 (1994) 1-17

From the remaining value one can determinethe population distribution of the Arf ionbeam, using the two state-to-state cross-sections at =

(a)]= 0.065

The effective cross-sections measured in this work for the four REMPI schemes at

= are given in Table 1. Note thatthe fraction given in the last column ofthis table has been derived from the overallagreement with all measured data and not from the local values given in Eqs. (6)-(8).Applying this test to the cross-section,

= 1.2 reported in Ref. 14, it canbe concluded that these experiments have beenperformed with an ion ensemble very close tothe statistical mixture.

3.2 Comparison with other results

As already discussed above, only the PESstudies from Orlando et al. andgive any hints concerning the ionic statesformed by multiphoton ionization of Ar. Forionization via the 4s' intermediate these andour experiments lead consistently to the con-clusion that the large majority of ions areformed in the fine-structure ground state.There is, however a significant disagreement for ionization via the 4s state. While our meas-urements clearly indicate a mixture of bothspin-orbit states which is very close to thestatistical weight, mentions only onegroup of photoelectrons corresponding to the ground state This may be due to thelimited energy resolution 0.15 ofthe electrostatic energy analyzer whichprobably failed to separate the two peaks lying at nominal energies of 3.61 and 3.44A reinvestigation of the REMPI photoelectronenergy spectra at 320 nm may solve this discre- pancy. It would also be interesting to provide

the missing information for the other transi-tions in the 370 nm region.

Another possibility to cross check ourresults is to use published state-specific cross- sections for other + X collision pro-cesses, some of which have been discussed in arecent review by Ng However, a hugeproblem in that comparison is that our method of determining the state populationrelies on the precision of absolutesections and different experimental methodsoften lead to different values. In order tolimit these problems we have restricted ourcomparison predominantly to results whichhave been obtained with GIB machines since with this technique one can achieve a realcollection efficiency. But even then we werefaced with rather large discrepancies andtherefore we decided to further restrict thefollowing figures only to relative values of

In this way some of the errors canbe eliminated.

It should be emphasized that it is not the aimof the following section to discuss the reaction dynamics of the examined systems in detail. A few related hints only are given where the dynamics are important for understandingthe influence of the fine-structure energy.

State-selected cross-sections for the charge-transfer system + have been studiedalready over a wide energy range by Ng andco-workers and by Dutuit et al. Forcomparison their measured values are plotted in Fig. 6 as open symbols. Theerror bars of our data, shown as solidsymbols, include the uncertainty of the statepurity discussed above. The broken line repre-sents the ratio that has been calculated directly from a weighted linear combination of theanalytical functions (3)-(5). The weightsaccount for our slightly imperfect state prep-aration, are based on the experimentally

M . Schweizer J . Mass Spectrom. Processes 135 (1994) 1-17

Fig. 6 . Ratio of the cross-sections and for the charge-transfer system + as a function of the collision energy, measured by several groups: Ng and co-workers Dutuit (A); this work The broken and the solid lines have beencalculated from the corresponding curves shown in Fig. 5. The arrows mark those collision energies which are required to form incharge transfer with the given state.

determined and admixtures given inTable If one includes in the calculation alsothe thermal broadening caused by the targetmotion, the threshold onset is smeared outand the peak is flattened as can be seen fromthe solid line. Note that the experimentallydetermined ratios depend strongly on minor

admixtures in the ground-state beamand also on the contribution of the cross-section for forming given by Eq. (3).For well prepared beams and without target motion Eqs. (3)-(5) predict a maximumslightly above 8.

It is obvious that there are significant differ- ences between the results of the three groups, concerning both the maximum and its posi- tion. In our measurements its value rises to amaximum of 5.5 1 at = 0.4 whileDutuit et al. found a value of 3 and Fleschet al. only 1.8 at higher collision energies. The discrepancies between the three experi-ments are even more serious if one compares the absolute values of the cross-sections. Forexample at = 0.4 and for ground-state

ions, Flesch et al. have reported a cross-2section as large as , Dutuit et al. have

measured 1.3 while our value is onlyWhat are the reasons for these huge

discrepancies? In all our experiments there isalmost no doubt about the purity of theprepared states. It is also impossible that theion beam energy widths are responsible forsmearing out the structure. The most prob-able reason we consider is based on the influ-ence of electrostatic potential barriers, whichare created by surface potential distortions and which have been found to reach heights up to 100 along the axis of the octopole

The following explanation is based on theexperimental observation that a small frac-tion of the primary ions can be trappedin the scattering between such barriers if theylose enough energy by inelastic collisions. It isnot easy to get a quantitative estimate of thiseffect since the fraction of trapped ions notonly depends on the potential distortions but also on the conditions in which the ionsare injected into the octopole. The energy

12 M . Schweizer et Mass Spectrom. Ion Processes 135 (1994) 1-17

distribution of the beam also plays a sig-nificant role; in particular, one has to ensurethat the distribution does not have a tailextending to very low energies. Since in mostexperiments the density in the scattering cell isbetween and and due to the

mass ratio which favors an efficientenergy exchange in an inelastic collision, onecan easily scatter and trap a few parts perthousand of ions. These thermalized Arf

ions can very efficiently charge transfer tosince at these energies the rate coefficient isclose to cm3 for both fine-structurestates and trapping times longer than milli-seconds are easily achieved. Under these con-ditions, trapping of only of the ions issufficient to explain the measuredsection of Dutuit et al. whereas in Ng'sapparatus, less than are required to reach

2an apparent cross-section of 3.9 .The proposed trapping mechanism can not

only explain the discrepancies in the size of thecross-section but also the differences in the

ratio depicted in Fig. 6 . Since thereactivity of the thermalized ions is inde- pendent of their fine-structure state, Fleschet al. who appear to have the largest contribu-tion from trapped ions, found the smallest

ratio, while the results obtained byDutuit et al. are in between ours and those ofFlesch.

To further corroborate these speculations we have performed experimental tests todemonstrate the influence of trapped ions byraising artificial barriers with ring electrodes atthe entrance and at the exit of the scatteringcell. The location and the electrostatic influ- ence of such electrodes have been describedin detail in Ref. 7. We have found that it ispossible to change the cross-section by oneorder of magnitude with potential barriers of

However we have also checked ouroctopole concerning unwanted potential dis-tortions, using the "ion reflection" methoddescribed in Ref. 7. These tests have proved

that the potential along the axis of ourpole is flat within Under these condi-tions trapping of ions by inelasticcollisions with target gas at 300K is veryunlikely and our measured cross-sections aretherefore not significantly affected by theseproblems.

There do not yet exist theoretical explana-tions for the measured fine-structure depen- dence of the system + Even thevarious potential surfaces which are involvedare not known or at least not known withsufficient accuracy. Therefore most experimen-tal results were explained simply by modelsbased on asymptotic energy levels and

factors or on unpublished Diatomics In Molecule (DIM)potential surfaces A more detailed discus- sion of the (Ar + collision system, includ- ing the reverse reaction + Ar, will be givenin Ref. 13.

A second system where fine-structure depen- dence has been studied is the charge transfer between and The overall shape of theenergy dependence of the cross-section for thisprocess is comparable to that for +except that the position of the minimum is shifted towards lower energies, namely to18 However, the origin of this struc-ture is quite different. From the fact that atsub-thermal energies the rate coefficientreaches only 10% of the Langevin limit, one can conclude that the charge transfer ishindered by an unknown mechanism,although it is exothermic. With increasingtemperature the reaction probability becomeseven smaller, which may be explained by theenergy dependence of the lifetime from thecollision complex. The increase in the ratecoefficient at collision energies above 18is due to the opening of a new reaction chan-nel. In contrast to the Arf + system where

M . Schweizer et J . Mass Spectrom. Ion Processes 135 (1994) 1-17

Fig. 7. Ratio of the cross-sections and for the charge-transfer reaction + + Ar as a function of the collision energy measured by several groups: Ng and co-workers Tobita et al. this work At low collision energies the excited state reacts faster whereas this trend is reversed above

a new electronic channel becomes accessible, here the first vibrational state, =plays an important role as already discussedby several authors (see Ref. 8).

For the charge-transfer system +Fig. 7 shows a selection of results from severalgroups. Again we restrict the comparison to the ratio as a function of collisionenergy. Our data points (solid dots) are insatisfying agreement with the guided ionbeam results from Ng's group (open squares)

and, at thermal energies,also with the drift tube results (crosses) where they areusually rather reliable. Similar swarm

by Hamdan et al. have resulted insimilar values and are therefore omitted for clarity. Concerning the deviations in the energy dependence it should be noted thatthe drift field leads to a wide energy spreadand that collisions with the buffer gas mayalso affect the internal states of the ions. Inaddition the swarm experiments have notbeen performed with state-selected ions but with a statistical mixture and state-specificinformation has been obtained by applying

the attenuation method. This procedure requires very accurate data, especially in instances like the present one, where the ratecoefficients are almost equal. At collisionenergies above there exist many otherstate-selected experiments These arelikewise not reproduced in the figure since they all agree with the results of Ng and co-workers within the error bars.

Further, for the + system the massratio and the experimental conditions are com-parable to those of the + system and,in principle, trapping of primary ions due to inelastic scattering can lead to similar prob-lems. However, the charge-transfer rate coeffi-cient for thermalized ions is in this instance about a factor of ten smaller and thereforethe measured cross-sections do not depend socritically on potential distortions of thepole.

Our results clearly show that the spin-orbitstate effects the reaction differently at low andhigh collision energies. Below somestate-specific calculations from Clary andSonnenfroh are known; however, they

14 M . Schweizer J . Mass Ion Processes 135 (1994) 1-17

have examined only the reaction with thestate. Therefore we have only a speculativeexplanation for the measured low-energybehavior that the reaction is hindered by abarrier, and that fine-structure energy canhelp to surmount it. At energies abovetheoretical investigations have been per-formed by Parlant and Gislason Theyhave predicted the higher reactivity of thespin-orbit ground state as being due to anenergy resonance between the Ar +(v = 1) channel and Ourexperimental result, that the value of

drops below 1 with increasing energy, is consistent with the threshold of0.09 for formation of the (v = 1) molecule.

To introduce the + charge-transferreaction as a calibration standard would havethe advantages that the cross-sections are big-ger and that the quality of the data is muchbetter as can be seen from the small error bars in Fig. 7. Besides that the absolute values arenot so sensitive to the problem caused bytrapped primary ions. Nonetheless we havechosen the + charge-transfer systembecause the relative effect of the spin-orbitstates is much more pronounced, thecross-section ratio is 6 : 1 at =instead of only 2 : 3. In addition, the reactionwith is far superior if one intends to deter-mine small amounts of states in aground-state beam.

Among the systems we have studied, thestrongest fine-structure dependence has beenobserved for the + charge transfer. Figure shows the ratio for the

channel while Fig. depicts this ratiofor the hydrogen abstraction reaction. Atcollision energies below 100 the excitedion undergoes charge transfer with asection which is 8 1.5 times larger than that for the ground state. This pronounced fine-

Fig. 8. Ratio of the cross-sections and as a function ofthe collision energy for the charge-transfer reaction

+ Ar (upper panel) and the reaction+ + 2 + H (lower panel) measured by several

groups: Ng et et this workThe solid line shows the result from a theoretical investigation by Baer et al. The broken line represents thecalculated ratio determined by Tosi et al.

structure dependence has been explained by an energy resonance between thestate and the = 2) state Corre-sponding measurements which have been per-formed with deuterium as target gas aredepicted in Fig. 9. For the energy levelsare shifted such that there are no close reso-nances. Therefore the charge transfer is muchless dependent on the ionic state and the ratioof the cross-sections remains below three.Furthermore, the cross-sections for forming

and are only weakly dependenton the fine-structure state and, as can be seenin Figs. and the ratio liesonly slightly above one.

Since from the systems we have studied, the

M . Schweizer et J . Mass Spectrom. Ion Processes 135 (1994) 1-17 15

Fig. 9. Ratio of the cross-sections and as a function ofthe collision energy for the charge-transfer reaction

+Ar (upper panel) and the reaction+ + D (lower panel) measured by several

groups: et al. (0);this work The broken linerepresents a theoretical prediction by Tosi et al.

+ charge-transfer system is the mostsensitive process concerning the influence ofspin-orbit energy it could be regarded as the best candidate for determining theratio; however, this reaction is one of the most problematic ones from an experimental pointof view. Firstly formation represents only a minor reaction channel since it competeswith hydrogen abstraction and the ArHf

products are formed roughly at the Langevinlimit. An experimental problem is that thesimultaneous confinement of productsand primary ions requires ofthe guiding octopole at sufficiently highfrequencies and amplitudes. Secondly, the

ions produced have small kinetic ener- gies, and thus potential distortions can easilylead to discrimination effects. More complica-

tions arise from the fact that both the and the products undergo secondary reac- tions with the target, and in both events thesame product, if formed. Our results have been corrected for this based on the knownthermal rate coefficients of these reactions

and accounting for the mean flight time ofthe products through the scattering cell Forthe hydrogen abstraction channel corrections ofonly a few percent are necessary whereas thecharge-transfer channel requires correctionsbetween 10 and 30% depending on theorbit state and the target.

Figures 8 and 9 show a selection of otherexperimental results and of theoretical predic- tions. In the overlapping energy range ourexperimental ratios are in good agreementwith the results from Liao et al. Thereare some minor discrepancies between oursand those from et al. For the

channel the three-dimensional quan-tum mechanical calculation from Baer et al.

is in fairly good agreement with the experi-ments whereas significant deviations areobvious in the case of the charge transfer(Fig. A very recent experimental andtheoretical investigation has been publishedby Tosi et al. These authors found, via ahigh-resolution beam experiment, an interest-ing structure in the energy dependence of thehydrogen abstraction cross-section which hasbeen explained using a theoretical model. There are, however, some controversies sincethe structure could not be reproduced in anindependent experiment 5,361. Beyond that,the theoretical results from Ref. 34 give only aqualitative description of the reaction since thecalculated absolute cross-sections are muchsmaller than the experimental results. Finally, also the theoretical ratios deviatesignificantly from the measurements as canbe seen from the broken line in Fig. 8. There-fore, more experimental and theoretical work is needed for this very fundamentalecule reaction.

16 M . Schweizer J . Mass Ion Processes 135 (1994) 1-17

The preceding comparisons and discussionshave revealed that it is not straightforward tofind an ideal reaction system for which theabsolute values of state-specific cross-sectionscan be used for characterizing quantitatively the states populated in the various photon ionization processes of Ar. Best agree-ments between results from different groupshave been found for the charge transfer withnitrogen, but this system has the disadvantagethat the cross-sections are only weakly depen-dent on the fine-structure state of the argon ion. More sensitive collision processes suchas charge transfer with hydrogen or oxygenare rather problematic from an experimentalpoint of view since they require an apparatuswhich is able to provide reliable informationon integral cross-sections. This condition can be fulfilled, at least in principle, by the GIBmethod which can provide real collectionefficiency. However, this feature is not suffi-cient and the results of this work, among others, clearly demonstrate that the operationof a guided ion beam machine requires many tests and rigorous checks, the majority ofwhich have been discussed in Ref. 7.

4. Conclusions

In this work, several REMPI schemes havebeen examined for preparing ions selec-tively in the two spin-orbit ground states. Incontrast to the commonly used photoelectronspectroscopy, state-specific reac-tions have been used to determine the resulting2 : population. The proposed proce-dure to calibrate this ratio is based on the

+ charge-transfer process whichshows a significant fine-structure dependencein the threshold region for forming the meta-stable ion. The use of a monitor reaction for characterizing reactants prepared in uncer-tain state distributions has the advantage thatthe ensemble is probed directly in the reaction

zone. It is evident that this method is limited tosituations where only a few states are popu-lated which is, however, often the case. Weare currently using this method for probingensembles which are thermalized at very low temperatures or photoionized with photonenergies just above the ionization limit, andin situations where we intend to distinguishbetween electronic states, vibrational modesor different structures.

5. Acknowledgments

The authors thank Professors Smith and Schlier for many contributions and helpfuldiscussions. Financial support of theDeutsche Forschungsgemeinschaft forschungsbereich 276) is gratefully acknowl-edged.

6. References

W.A. Chupka and M.E. Russell, J. Chem. Phys., 49(1968) 5426.

[2] K. and E.W. Schlag, Rev. Phys. Chem., 42 (1991) 109.

[3] F. Merkt, S.R. Mackenzie and T.P. Softley, J. Chem.Phys., 99 (1993) 4213.

[4] D. Smith and N.G. Adams, Phys. Rev. A, 23 (1981)2327.W. Lindinger, H. and F. Howorka, Int. J.Mass Spectrom. Ion Phys., 41 (1981) 89.

[6] J.D. Shao, Y.G. Li, G.D. Flesch and C.Y. Ng, Chem. Phys. Lett., 132 (1986) 58.

[7] D. Gerlich Adv. Chem. Phys., 82 (1992) 1.[8] C.Y. Ng, Adv. Chem. Phys., 82 (1992) 401.[9] S.L. Anderson Adv. Chem. Phys., 82 (1992) 177.

C.E. Moore, National Standard Reference Data Series,National Bureau of Standards, Washington, 1971.S.J. Thesis, Oak Ridge National Labora-tory Tennessee, TN, 1991.T.M. Orlando, B. Yangand and S.L. Anderson, J. Chem. Phys., 90 (1989) 1577. 0 . Dutuit, C. Alcaraz, D. Gerlich, P.M. J.Hepburn, C. J.B. Ozenne, M.Schweizer and T. Weng, in preparation.S. Scherbarth and D. Gerlich, J. Chem. Phys., 90 (1989) 1610.

M . Schweizer J . Mass Ion Processes 135 (1994) 1-17 17

I. and W. Lindinger, J. Chem. Phys., 76 (1982)4972.G.D. Flesch, S. Nourbakhsh and C.Y. Ng, J. Chem.Phys., 92 (1990) 3590.P.J. Chantry, J. Chem. Phys., 55 (1971) 2746.S.J. Bajic, R.N. Compton, X. Tang and P.poulos, Phys. Rev. A, 44 (1991) 2102.T. Kato, Chem. Phys., 80 (1984) 6105.T. Matsuo, N. Kobayashi and Y. Kaneko, J. Phys.Jpn., 55 (1986) 3045. C.L. Liao, J.D. Shao, R. Xu, G.D. Flesch, Y.G. Li andC.Y. Ng, Chem. Phys., 85 (1986) 3874. T. Tobita, N. Kobayashi and Y. Kaneko, in J. Eichler,I.V. Hertel and N. Stolterfoth (Eds.), ICPEAC,Int. Conference on the Physics of Electron and Atomic Collisions, 27 July-2 August 1983, Berlin. M. Hamdan, K. Birkenshaw and N.D. Twiddy, Int.Mass Spectrom. Ion Processes, 57 (1984) 225. T. Kato, K. and I. Koyano, J. Chem. Phys., 77(1982) 337.B.G. Lindsay and C.J. J. Phys. B, 21 (1988)1617.P.M. T.R. Govers and T. Baer, Z. Phys. D, 4(1986) 89.D.C. and D.M. Sonnenfroh, J. Chem. Phys., 90(1989) 1686. G. Parlant and E.A. Gislason, J. Chem. Phys., 86 (1987)6183.

C.L. Liao, R. Xu, S. Nourbaksh, G.D. Flesh, M. Baerand C.Y. Ng, J. Chem. Phys., 93 (1990) 4832. Y. Ikezoe, S. Matsuoka, M. and A. Viggiano, Gas Phase Ion-Molecule Reaction Rate Constants Through 1986, published by Ion Reaction Research Group of the Mass Spectrometry Society of Japan,Tokyo, 1987, distributed by Co.M. Schweizer, Thesis, Universitat Freiburg, (1993).K. Tanaka, J. Durup, T. Kato and I. Koyano, J. Chem.Phys., 74 (1981) 5561. M. Baer, C.L. Liao, R. Xu, S. Nourbahkah, G.D.Flesch, C.Y. Ng and D. Neuhauser, Chem. Phys.,93 (1990) 4845. P. Tosi, Dmitrijev,Y. D. Bassi, D. Cappelleti,F. and V. Aquilanti, Chem. Phys., 99 (1993)985.D. Gerlich, H.J. and 0 . Wick, in D. Bassi, M.Scotoni and P. Tosi (Eds.), Symposium on Atomic andSurface Physics, Pampeago, 1992.D. Gerlich, in T. Andersen, B. Fastrup, F. Folkmannand H. Knudsen and N. Andersen (eds.), Electronic and Atomic Collisions, invited paper on ICPEAC XVIII in Arhus, 1993.0. Dutuit, in K.R. Jennings, (ed.), Proc. NATOAdvanced Study Institutes on Fundamentals of GasPhase Ion Chemistry 1990, Kluwer, Dordrecht, 1991,p. 21.