Instrument-Independent CAD Spectral Databases: Absolute Cross-Section Measurements In QQQ Instruments

The energy dependence of the cross section, σ(E), for the symmetric (resonant) charge transfer reaction Ar+(Ar,Ar)Ar+ was measured in our triple quadrupole (QQQ) tandem mass spectrometer.1 Our σ(E), for P≃0.04–0.43 mtorr and E≃5 – 60 eV (LAB) [the range of collision energies used for collisionally activated dissociation (CAD)], agrees to within 10% with the Rapp-Francis theory (impact parameter method in the two-state approximation), as corrected by Dewangan. We measured identical σ(E) from both the rate of reactant ion decay and the rate of product ion formation; i.e., our instrument is kinetically well behaved. The measurement of these σ(E) in other QQQ instruments can be used to validate whether or not a QQQ instrument has been properly designed to be kinetically well behaved. This is essential if generic, instrument-independent CAD spectral databases are to be developed on the basis of the absolute cross sections for the CAD of known ionic substructures. That is, since tandem mass spectrometry (MS/MS) exploits the ion fragmentation patterns characteristic of ionic substructures, the characteristic profiles [“breakdown curves”] of ion abundance versus target thickness (or collision energy) correspond uniquely to the sequence: (parent)i σij→ (daughter), σjk→ (granddaughter)k, etc. Hence, computer simulation of experimentally observed breakdown curves enables the structure of an unknown species to be assigned on the basis of the absolute cross sections σij, σjk, etc. for CAD of known ionic substructures i, j, k, etc. Thus, if the calculated and experimental breakdown curves agree, the structure would be characterized.


Introduction
used for collisionally activated dissociation (CAD)], agrees to within 10% with the Rapp-Francis theory (impact parameter method in the two-state approximation), as corrected by Dewangan. We measured identical cr(E) from both the rate of reactant ion decay and the rate of product ion formation; i.e., our instrument is kinetically well behaved. The measurement of these cr(E) in other QQQ instruments can be used to validate whether or not a QQQ instrument has been properly designed to be kinetically well behaved. This is essential if generic, instrument-independent CAD spectral databases are to be developed on the basis of the absolute cross sections for the CAD of known ionic substructures. That is, since tandem mass spectrometry (MS/MS) exploits the ion fragmentation patterns characteristic of May-June 1987 ionic substructures, the characteristic profiles ["breakdown curves"] of ion abundance versus target thickness (or collision energy) correspond uniquely to the sequence: (parent), ~ (daughter), ~ (granddaughter)~, etc. Hence, computer simulation of experimentally observed breakdown curves enables the structure of an unknown species to be assigned on the basis of the absolute cross sections cr,}, cr,~, etc. for CAD of known ionic substructures i, j, k, etc. Thus, if the calculated and experimental breakdown curves agree, the structure would be characterized.
Key words: calibration; cross sections; tandem mass spectrometry; target thickness.
Accepted: February 5, 1987. Triple quadrupole (QQQ) tandem mass spectrometry (MS/MS) is an analytical tool which can be used for rapid, direct speciation of complex multicomponent mixtures [1].2 The analysis makes use of the collisionally-activated dissociation (CAD) of "parent" ions. 3 A "parent" ion selected by the first quadrupole (Ql) is interacted with a target gas within the second quadrupole (Q2). Q2 channels un dissociated "parent" ions and "progeny" fragment ions into the third quadrupole (Q3) for mass analysis. The instrument thus produces a CAD spectrum of each initially-se1ected "parent" ion. IStandard physics notations: A +(B,C)D+ represents the reaction A + +B~C+D+; crE represents the value of the reaction cross section when measured at a particular interaction (collision) energy E; cr(E) is the functional form of the energy dependence observed when crE values are plotted versus their respective E values.
2Figures in brackets indicate literature references. .1A "parent" ion may be a molecular radical cation, a protonated molecule, or a "progeny" fragment ion (daughter, granddaughter, etc.) produced by the CAD of a larger precursor parent ion.

229
In principle, standard CAD spectra of a variety of ions (fragment ions, molecular ions, protonated molecules, etc.) could be generated and collected as reference libraries, to be used for comparison About the Authors: Richard I. Martinez, a research chemist, is with the NBS Center for Chemical Physics. Seksan Dheandhanoo, a guest scientist at NBS from Georgetown University during the work described, is a physicist. The work was funded by the U.S. Air Force, Environics Division. against unknown spectra in a manner analogous to the use of reference libraries in the data handling systems of ordinary electron impact mass spectrometry. Further, it should be possible to infer the identity of an unknown complex molecule by identifying the ionic substructures of fragment ions generated in its CAD spectrum. However, to date reference libraries of CAD spectra have not been collected because of a lack of standardization of operating conditions of such instruments [2].
There are several instrument parameters which can cause significantly different CAD spectra to be observed for any given molecule. The key parameters are: 1) the number of collisions undergone by a "parent" ion within Q2, a parameter usually characterized in terms of "target thickness," which is defined as [(actual path length traversed by the ion through the gas target) X (effective number density of the CAD target gas)]; 2) the duration of the interaction between the "parent" ion and the target gas, which is determined by the collision energy for "parent" ions entering Q2; and 3) the energy level of the analyzing quadrupole Q3 relative to that of Q2 which, because of the translational energy distribution of the "progeny" ions, determines whether or not some progeny ions can enter Q3.
Results of a recent international round robin [2] demonstrated that the target thickness is not a wellcontrolled parameter, with estimated target thicknesses differing by factors of 2-4 from apparent actual values. The problem of determining target thickness is complicated in QQQ instruments because of the complex oscillatory trajectories of ions within a quadrupole mass filter [3-71; the actual path length traversed by the ion through the CAD gas can be significantly longer than the nominal gas target length [6]. Moreover, in QQQ instruments utilizing a molecular beam target (Type A configuration [2]) the problem is further complicated because of a lack of information about the extent of overlap of the projectile ion beam and the molecular beam target. On the other hand, in QQQ instruments utilizing a co1Jision chamber (Type B configuration [2]), the actual target thickness can be significantly greater than an estimated value based on the length of the Q2 colIision chamber and the pressure within it if the gas plume extends beyond the confines of the Q2 collision chamber into Q I and Q3.

Kinetic !\Iethod
In a recent study from this laboratory [8] it was suggested that these problems can be circumvented by using a kinetic method to measure the effective target thickness within a QQQ instrument. That is, 230 if a reaction can be identified for which the cross section (or rate coefficient) is well established as a function of collision energy, then a simple measurement of the intensity of the reactant ion and/or product ion in the absence and presence of CAD target gas at known collision energy leads to an experimental determination of the target thickness. For example, for the charge transfer reaction: where (jE is the value of the reaction cross section at a collision energy E, Leff   respectively) provides strong assurance that a QQQ instrument is kinetically well behaved. That is, it provides a very important test that the instrument parameters and the reaction kinetics are well ~ontrolled (no back reactions, no impurity reach(ms, no scattering losses, no fringing fields, wellconfined gas target, etc.). In our earlier study [8], the symmetric (resonant) charge transfer reaction Ne+(Ne,Ne)Ne+ was used as a calibrating reaction for the validation of the target thickness measurements in our QQQ instru-4S,tanda~~ kinetic notation: [A +]0 and [A +] are, respectively, the IntensItIes of the reactant ion A + when measured in the absence and presence of CAD target gas; [B+] is the intensity measured for the product ion B + when the target thickness is that,used for the [A +] measurement; [B+]", (=[A +]0) is the in-tensIty of the product ion at "infinite" reaction time and/or target thickness when all of A + has been converted to B+. ment. Abundant experimental and theoretical results had been previously reported for this reaction. Furthermore, because the NBS instrument had been constructed to incorporate the design considerations detailed by Dawson and coworkers [3][4][5][6][7], eq (4) [6] could be used to estimate Leff=R Lactual' (4) Here Lactual is the actual rectilinear pathlength for a well-confined CAD gas target; M = mass of projectile ion (in amu), E =axial ion energy (in eV), ro=field radius (in cm), F=rf frequency (in MHz). Equation (4) is based on operation of Q2 with the Mathieu parameters [3,4] at a2=O, Q2=O.28 [6].5 It was shown [8] that when the effective target thickness was estimated by using eq (4), values for the absolute reaction cross section derived from eq [2] were in excellent agreement with theoretical predictions, as well as with previous experimentallydetermined values. Furthermore, identical values for the reaction cross section were derived from 5It is important to note that eq (4) depends on MIE. Hence, for a given collision energy E, the effective target thickness Lcrr [B] will be different for different projectile ions, and must be corrected accordingly.
20cm reactant ion loss [eq (2)] and product ion formation [eq (3)] experiments, thus confirming that the NBS instrument is kineticaIIy weII behaved. This paper reports results of an analogous exercise carried out using the 4OAr+CAr,40 ArYAr+ reaction 6 for Ln Y measurements and the 36 Ar t (40Ar, 36 Ar)40Ar+ reaction for Ln W measurements. b The Ar+(Ar,Ar)Ar+ reaction is of special interest because argon is a target gas commonly used for CAD. Thus, this reaction may provide a convenient calibrant species for target thickness determinations in other laboratories. Since reference spectra for CAD libraries can be utilized only if they were obtained under conditions such that the target thickness is specified, the results reported here may permit the easy standardization of operating conditions for the determination of such reference spectra.

Experimental
Our specially designed QQQ instrument can be configured to use either a molecular beam (Type A) or collision chamber (Type B) configuration (see schematic, fig. 1). All experiments reported here utilized the Type B configuration. 6 'Ar is the sum of 4OAr+·\~Ar+\bAr. The natural abundance of IOAr (O.336S±O.OOO6% [9]) is sufficient to permit the measurement, with good signal-to-noise ratios, of the product ion growth (Ln W measurements). An abbreviated description of the instrument follows (a detailed description will be published elsewhere [10]). The instrument was manufactured by Extrel, Inc. 7 to conform to the design considerations stipulated by Dawson and coworkers [3][4][5][6][7]. It consists of three standard 7-270-9 quadrupole rod assemblies (Ql, Q2, Q3) mounted in tandem on a special multipurpose track. Each mass filter assembly is operated at 1.2 MHz, controlled by a 300watt Model 150-QC quadrupole power supply and associated quadrupole control electronics. A C-50-IC controller regulates the standard Extrel electron impact ionizer mounted on the differential pumping wall. This ionizer has a filament perpendicular to the cylindrical quadrupole axis and has been modified to accommodate crossed molecular and laser beams. Each QQQ system parameter is computer controlled via its respective 16-bit DAC by the standard 8086-based Extrel Triple Quad Data System used for instrument control and data acquisition.
For the Type B configuration, Q2 is surrounded by a collision chamber enclosure while Ql and Q3 are completely nude (no housing), and are adequately pumped by four 1200 lis turbomolecular pumps, ensuring a well-confined collision region. The actual length of the collision region from the front face of the L4 aperture to the rear face of the L5 aperture is LaclUal=21.745±0.075 cm. All kinetic measurements were based on operation of Q2 with the Mathieu parameters [3,4] at a2=0, Q2=0.28 [6]. For our instrument, ro=field radius=0.684 cm (quadrupole rod diameter = 1.59 cm), F=rf frequency = 1.2 MHz, and the R correction factor from eq (4) is ca. 1.02 at E =60 eV and 1.18 at E = 5 e V. Furthermore, the diameter of our L4 and L5 inter-quadrupole lens apertures is 1.27 ±0.025 cm {> l.4ro [6]}, and thus conforms to the requirements for closely-coupled quadrupole fields [6]. Pressure measurements in the center of the collision chamber were made with a 1 torr MKS 310CA Baratron capacitance manometer [appropriate corrections were made for thermal transpiration (~3%) etc.]. Ar+ ions were generated by 70 eV electron impact [11], and the Ar+ projectiles were selected by Q 1 [19]. The energy spread of the projectiles entering Q2 was determined to be <; 1.8 eV for 90% of ~ Certain commercial equipment. instruments. and materials are identified in this paper in order to adequately specify the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards. nor does it imply that the material. instruments. or equipment identified is necessarily the best available for the purpose. 232 the ions [<; 3 e V for 99% of the ions] when measured by using the Q2 pole bias (rod offset) to generate a stopping potential curve (see fig. 2). E9O% is the Q2 potential required to stop 90% of the ions. The collision energy Ecoll was selected by setting the Q2 pole bias=E 90 <;t-E coll ' Projectile decay experiments (cf. fig. 3) were performed at each selected collision energy by setting the Q3 pole bias more positive relative to the Q2 pole bias (e.g., Q3-Q2~3 to 40 V for Ecoll~5 to 60 eV) to ensure only unreacted projectiles were able to enter Q3 [25]. Product growth experiments (cf. fig. 4) were performed by setting the Q3 pole bias sufficiently negative relative to the Q2 pole bias (e.g., Q2-Q3=110 to 140 V for Ecoll~40 to 10 eV) to ensure that all ions (products and unreacted projectiles) were drawn out of Q2 into Q3 [25].
The typical ion collection efficiency is > 97%; i.e., the total ion current for products+unreacted projectiles (i.e., with CAD gas on) > 97% of the initial projectile ion current (i.e., with CAD gas oft). This high ion collection efficiency allows one to set    where 13 is same as that of figure 3 and P4(J is partial pressure of 4(JAr.

Discussion
Together with our results for Ar+{Ar,Ar)Ar+, figure 5 also summarizes experimental [28][29][30][31][32][33][34][35][36][37][38][39][40][41] and theoretical [42][43][44][45][46][47] results for this reaction from the literature [48]. Prior to our work it was not clear which theoretical model one could or should use to obtain reliable estimates of UE values for use in target thickness calibrations in the 5-60 e V range of collision energies, the range typically used for CAD experiments. The results reported here for u(E) (see fig. 5) are in excellent agreement with the u(£) predicted by the Rapp-Francis theory (impact parameter method in the two-state approx· imation) [42] as corrected by Dewangan [43] (solid line D in fig. 5), as well as with the experimental u(E) of other workers (see fig. 5, data labeled HES [34), Z (28), H (35), KPS [37], DSEG [29], FS [36]). For the data labeled C [31], the UE values are sig. 234 nificantIy lower than those of the Dewangan line (labeled D) [43] and of other workers; however, the slope of his UE vs. E plot shows substantially the same u(E) as that of the Dewangan line. On the other hand, the u(E) of the data labeled HK [33] clearly differs from that of the Dewangan line and of other workers, even though some of the UE values labeled HK overlap some of the UE values of other workers. Hence, the data of figure 5 labeled C [31] and HK [33] are not considered further.
ments has also been observed in our Type B configuration for Ne+(Ne,Ne)Ne+ [8] and Ar+(N2,Ar)N2 + [49], further confirming that our instrument is kinetically well behaved. Thus we can use the (FE values measured in our Type B configuration to determine the effective target thickness of Ar in our Type A configuration. However, similar performance is expected only in kinetically well behaved QQQ instruments which incorporate Dawson's design considerations [3][4][5][6][7].

Conclusions
The kinetic method described in the introduction potentia]]y can provide a means whereby absolute target thicknesses for any gas can be accurately calibrated in-situ in kinetically well behaved QQQ instruments (in Type A or Type B configurations) for collision energies in the 5-60 e V range. Moreover, since the UE values for Ar+(Ar,Ar)Ar+ are not strongly dependent on E over the range of in-terest for CAD experiments {u5ev:::<1.3 U/ileV} , the kinetic method should provide fairly accurate target thickness calibrations even if the projectile energy distribution in other QQQ instruments is not as narrow as in the NBS instrument. R "Interlaboratory round-robin testing of our kinetic method in various types of QQQ instruments is essential to confirm its reliability as a generic target thickness standard. Moreover, it will provide much-needed information about which QQQ in"trument designs are not kinetically well behaved and therefore not well suited for the generation of standardized reference CAD spectra. The round-robin test will involve the experiment a.,,,ociated with figure l(a) of [2] after first having completed in-situ target thickness calibrations of the participanh' QQQ instruments by using our kinetic method with our u(£) for Ar' (Ar,Ar)Ar' (this work). A test protocol is being formu· lated. It will address how to set £".11 and q:. Several inve .. tiga· tors have agreed to participate. However. many more participants would be desirable to establish the degree of variability one encounters when using a standardized protocol with different operators on the same and/or different instruments of several types. Letters of inquiry from prospective participants may be sent to NBS.
The measurement of the u(E) for Ar+(Ar,Ar)Ar+ in other QQQ instruments can be used to validate whether or not a QQQ instrument has been properly designed to be kinetically well behaved. This is essential if generic, instrument-independent CAD spectral databases are to be developed on the basis of the absolute cross sections for the CAD of known ionic substructures. That is, since MS/MS exploits the ion fragmentation patterns characteristic of ionic substructures, the characteristic profiles ["breakdown curves"] of ion abundance versus target thickness (or collision energy) correspond uniquely to the sequence: (parent); a/I. (daughter)j ajle. (granddaughter)k, etc. Hence, computer simulation of experimentally observed breakdown curves should enable the structure of an unknown species to be assigned on the basis of the absolute cross sections u,). UjA, etc., for CAD of known ionic substructures i, j, k, etc. Thus, if the calculated and experimental breakdown curves agree, the structure would be characterized. Dawson, et a1. [50] demonstrated that computer simulation of breakdown curves is plausible. Hence, one can envision a CAD spectral database of critically-evaluated cross sections U;j. Ui/(, etc. for CAD of known ionic substructures measured in kinetically well-behaved instruments under standardized operating conditions. The advantages of such a database are: 1) the cross sections would uniquely characterize the CAD spectra of both known and unknown species (so long as the unknown species contain ionic substructures for which the CAD cross sections are known); 2) characterization of an unknown is not limited by the number of compounds in a "library"; 3) the format is compatible with its use in expert systems; and 4) end users are involved directly in its evolution by using critically-evaluated cross sections already in the database and by submitting new cross sections for inclusion in the database.