Cluster reactivity experiments: Employing mass spectrometry to investigate the molecular level details of catalytic oxidation reactions

Abstract

Mass spectrometry is the most widely used tool in the study of the properties and reactivity of clusters in the gas phase. In this article, we demonstrate its use in investigating the molecular-level details of oxidation reactions occurring on the surfaces of heterogeneous catalysts via cluster reactivity experiments. Guided ion beam mass spectrometry (GIB-MS) employing a quadrupole–octopole–quadrupole (Q–O–Q) configuration enables mass-selected cluster ions to be reacted with various chemicals, providing insight into the effect of size, stoichiometry, and ionic charge state on the reactivity of catalyst materials. For positively charged tungsten oxide clusters, it is shown that species having the same stoichiometry as the bulk, WO₃⁺, W₂O₆⁺, and W₃O₉⁺, exhibit enhanced activity and selectivity for the transfer of a single oxygen atom to propylene (C₃H₆), suggesting the formation of propylene oxide (C₃H₆O), an important monomer used, for example, in the industrial production of plastics. Furthermore, the same stoichiometric clusters are demonstrated to be active for the oxidation of CO to CO₂, a reaction of significance to environmental pollution abatement. The findings reported herein suggest that the enhanced oxidation reactivity of these stoichiometric clusters may be due to the presence of radical oxygen centers (W–O●) with elongated metal–oxygen bonds. The unique insights gained into bulk-phase oxidation catalysis through the application of mass spectrometry to cluster reactivity experiments are discussed.

Mass spectrometry has been crucial to the field of cluster research since its inception (1–9). A variety of cluster formation methods, primarily including laser vaporization (LaVa) (10–12), pulsed-arc discharge (PACIS) (13–15), electrospray ionization (ESI) (16), gas aggregation (17–18), and inert gas sputtering (CORDIS) (18) have enabled the creation of both positively and negatively charged as well as neutral gas-phase clusters across a large size range and with diverse elemental composition. The thermodynamic properties of clusters, including bond-dissociation energies (19–21), endothermic-reaction barriers (22), heat capacities (23), and the enthalpy, entropy, and free-energy changes associated with clustering reactions (24, 25), including those composed of hydrogen-bonded and van der Waals systems, have been widely studied, employing both GIB-MS and flow-tube experiments. Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR) (26) along with other linear ion trapping techniques (27) have been used to investigate the time-dependent kinetics of cluster molecule reactions, providing insight into reaction mechanisms, activation energies, and reaction intermediates. The structural properties of clusters, including bonding motifs and collision cross-sections have been examined through collision-induced dissociation (CID) (28) and high-pressure drift cell experiments (29). Time-of-flight (TOF) mass spectrometry has been applied extensively to separate specific cluster ions from an overall cluster distribution, thereby allowing sophisticated spectroscopic studies of mass-selected species. For example, the electronic structures of metal and metal oxide clusters have been comprehensively examined through magnetic bottle photoelectron spectroscopy (PES) (30, 31) and more recently, velocity map electron imaging (VMI) (32). The structural properties of clusters, derived from infrared spectra, have also been characterized through infrared multiphoton ionization (IR-REMPI) (33) and dissociation (IR-MPD) (34) spectroscopy. Although the above is not an exhaustive summary of all of the applications of mass-spectrometry to cluster research, it does reveal the degree to which the field depends on the ability to create, select, and interrogate individual cluster ions in the gas-phase environment of a mass spectrometer.

A particularly fruitful area of cluster research that depends entirely on mass spectrometry is the application of gas-phase clusters to model the reactions taking place on the surface of heterogeneous catalysts (35, 36). Gas-phase cluster experiments allow the fundamental reactive behavior of catalytic materials to be studied in an environment that avoids the complications present in condensed-phase research (36). Furthermore, mass-selected cluster studies, employing tandem mass spectrometry, enable the influence of factors such as size, stoichiometry, and ionic charge state on cluster reactivity to be determined with atomic-level precision (35). For nanocatalysis, this is extremely relevant because the reactive properties of clusters have been observed to change dramatically with the addition or removal of single atoms (37). Charging effects have also been shown to exert a pronounced influence on the activity of supported catalyst particles (38, 39). Comparison of the reactivity of anionic and cationic clusters can, therefore, provide insight into the importance of an accumulation or deficiency of electron density to the reactivity of catalytic materials.

Tungsten oxides are useful as both catalyst and catalyst-support materials because of their mechanical strength and resistance to thermal degradation (40). Bulk-phase studies have shown tungsten oxides to be catalytically active for the selective oxidation of sulfides to sulfoxides (41), the combustion of toluene (42), the skeletal isomerization of 1-butene (43, 44), and the decomposition of simple alcohols (45). Monoclinic, stoichiometric WO₃ particles in the size range of 2–5 nm have been found to be the active species for the selective oxidation of sulfides to sulfoxides (41). In contrast, for the catalytic combustion of toluene to CO₂, oxygen vacancies resulting in nonstoichiometric tungsten oxide were found to be critical to the observed activity (42). Moreover, the isomerization of 1-butene to isobutene was observed to be most selective over oxygen-rich tungsten oxides containing WO₄²⁻ surface sites (43, 44). These bulk-phase findings suggest that the stoichiometry of tungsten oxide particles directly influences both their activity and selectivity for catalytic reactions. Mass-selected cluster studies are uniquely capable of investigating the effect of stoichiometry on reactivity and may provide insight into the specific active sites responsible for catalytic activity.

Previous gas-phase studies employing PES have enabled an examination of the electronic structure and bonding in tungsten oxide clusters containing one (46), two (47), three (48), and four (49) tungsten atoms. For clusters containing one tungsten atom, it has been shown that WO₃⁻ exhibits a large energy gap between the first and second PES features, indicating that the WO₃ neutral cluster is a stable closed-shell species (46). Large increases in electron-binding energy were observed for the more oxygen-saturated clusters WO₄⁻ and WO₅⁻ (46). Oxygen was determined to be bound only in the atomic form in WO_3–4^−/0 and both atomically and molecularly in WO₅^−/0 (46). Furthermore, radical centers (W–O● and O₂⁻●) were observed in WO₄ and WO₅, respectively, suggesting that these clusters may exhibit enhanced oxidation reactivity (46). Clusters with two and three tungsten atoms (W₂O₈ and W₃O₁₁) were also found to contain molecular oxygen bonded in a highly activated superoxide (O₂⁻) state (47). Because of the elongation of the O–O bond in these superoxide subunits (1.33 Å), it is expected that these clusters would also be effective oxidizing agents. Clusters containing two tungsten atoms were shown to have structures with two oxygen atoms bridging the tungsten sites, whereas clusters with three tungsten atoms formed six-membered rings with a single oxygen atom bridging the tungsten centers (47, 48). For clusters with both two and three tungsten atoms, the terminal oxygen atoms were found to bind atomically at low oxygen coverage and molecularly at higher oxygen saturation (47, 48). The W₃O₁₀ cluster was shown to contain a molecularly bound O₂ unit with an O–O bond distance similar to that found in a peroxo (O₂²⁻) group (1.49 Å) (48). The single-bond character of this peroxo subunit may result in W₃O₁₀ being a strong oxidizer. In the case of clusters containing four tungsten atoms, it has been found that a transition from metallic to semiconductor-like properties occurs at a content of five oxygen atoms (49). Similar structures to those observed in the PES studies were found in a separate theoretical work (50) that also showed that (WO₃)₂ has features of bulk tungsten oxide, whereas W₄O₁₂ can be considered an embryonic form of the bulk material. Studies of small tungsten oxide clusters, therefore, should provide direct insight into the behavior of the bulk-phase material. The structures and acidic properties of (WO₃)_n (n = 1–6) were examined computationally, revealing that these clusters are weak Brønsted bases and strong Lewis acids (51). The high Lewis acidity of tungsten oxide clusters should facilitate the strong bonding of nucleophilic reactants such as alcohols and CO. Finally, a recent theoretical study established the existence of two nearly degenerate low lying electronic states in W₃O₉⁻ (52). The different spin distributions of these two states could exert significant effects in reactions involving spin transitions (53). Mass-spectrometry techniques enable the reactivity of specific tungsten oxide clusters containing these various potential active sites to be examined, thereby providing valuable experimental confirmation of theoretical predictions.

In this study, we demonstrate that positively charged tungsten oxide clusters having the same stoichiometry as the bulk WO₃⁺, W₂O₆⁺, and W₃O₉⁺ are highly active and selective for the transfer of a single oxygen atom to C₃H₆, suggesting the formation of propylene oxide. The same stoichiometric clusters are also shown to be active for the transfer of a single oxygen atom to CO, forming CO₂. We propose that the enhanced oxidation reactivity of these clusters may be due to the presence of oxygen radicals (W–O●) (54–57) with elongated W–O bonds.

Results and Discussion

Fig. 1 displays a typical distribution of tungsten oxide cation clusters produced by laser vaporization. Oxide clusters containing one to three tungsten atoms were reacted with C₃H₆ or CO to determine the influence of size and stoichiometry on cluster reactivity. Oxygen- rich clusters with an oxygen content greater than W_xO_3x+1⁺ will not be discussed because fragmentation experiments with inert N₂ produced intense collisional O₂-loss products from these species. The findings, therefore, show that the additional oxygen is very weakly molecularly bound onto the surface of these clusters and not sufficiently activated to participate in oxygen-transfer reactions with either C₃H₆ or CO. Clusters with an oxygen content of less than W_xO_3x⁺ exhibited no significant products other than minor peaks corresponding to the association of C₃H₆ onto the reactant ion.

Fig. 1.

A typical mass distribution of tungsten oxide cation clusters obtained through laser vaporization.

Tungsten oxide cation clusters with the same stoichiometry as the bulk WO₃⁺, W₂O₆⁺, and W₃O₉⁺ were found to be highly reactive and selective toward the transfer of a single oxygen atom to C₃H₆. We propose that this oxygen-transfer reaction may result in the formation of propylene oxide (C₃H₆O) according to Eq. 1.

The C₃H₆O product, however, may also correspond to propionaldehyde, acetone, or oxetane. Fig. 2 displays the relative intensity of the reactant and product ions with increasing pressure of C₃H₆. The reactant ions, W_xO_3x⁺, decrease in relative intensity with increasing pressure of C₃H₆, whereas the product ions, W_xO_3x−1⁺, become more pronounced. The most reactive cluster is WO₃⁺, with an oxygen-loss product, WO₂⁺, accounting for ≈30% of the total ion intensity at a pressure of 10 mTorr of C₃H₆. W₂O₆⁺ is less reactive, with a product ion intensity of ≈10% at the maximum reactant gas pressure. Interestingly, the reactivity of W₃O₉⁺ is higher than that of W₂O₆⁺. Close inspection of Figs. 2c and 3c reveals more scatter in the relative ion intensities of W₃O₉⁺ with increasing pressure of C₃H₆ or CO than for WO₃⁺ and W₂O₆⁺. This scatter results from the fact that the absolute intensity of the W₃O₉⁺ cluster is significantly lower than WO₃⁺ and W₂O₆⁺. Therefore, the integrated peak areas that are used to create the relative ion-intensity plots in Figs. 2c and 3c are more susceptible to instantaneous noise and variations in reactant-ion intensity. Nevertheless, the overall trends in reactivity for W₃O₉⁺ with both C₃H₆ and CO are reproducible. To further analyze the difference in relative reactivity among WO₃⁺, W₂O₆⁺, and W₃O₉⁺, the reaction rate constant was calculated for each cluster assuming pseudofirst-order kinetics according to Eq. 2.

In Eq. 2, I_r is the reactant-ion intensity with the addition of C₃H₆, I₀ is the reactant-ion intensity without C₃H₆, k is the rate constant, R is the concentration of C₃H₆ reactant gas, and t is the time it takes the reactant ion to pass through the octopole reaction cell. The reaction time may be calculated based on the length of the reaction cell that was determined by using a trapezoidal pressure falloff approximation to be 12.9 cm (58) and the velocity of the ions resulting from the supersonic expansion that was calculated by using the equations of Anderson and Fenn (59, 60). Ideally, all of the clusters exiting the supersonic expansion have the same initial kinetic energy. Therefore, more massive clusters will have a lower velocity and, consequently, spend more time in the reaction cell. Previous studies in our laboratory have shown that the pseudofirst-order rate constants obtained by using Eq. 2 agree well with the phenomenological rate constants calculated from zero-pressure cross-section data (58). The plots of ln[I_r/I₀] as a function of C₃H₆ concentration are displayed in supporting information (SI) Fig. S1. Assuming pseudofirst-order kinetics, the slopes of the plots are equal to −kt. The values of the slopes, when divided by the reaction time, reveal rate constants on the order of 3.0 × 10¹² cm³s¹ for WO₃⁺, 3.6 × 10¹³ cm³s¹ for W₂O₆⁺ and 5.9 × 10¹³ cm³s¹ for W₃O₉⁺ that are three to four orders of magnitude below collision rates. The pseudofirst-order rate constants reported herein allow for a more rigorous comparison of the relative reactivity of WO₃⁺, W₂O₆⁺, and W₃O₉⁺ with C₃H₆ or CO, thereby providing insight into how cluster size influences the oxidation of these molecules. We do not intend the reported values to be interpreted as absolute quantitative rate constants. This kinetic analysis, nevertheless, confirms that WO₃⁺ is the most reactive species, followed by W₃O₉⁺, and finally by W₂O₆⁺.

Fig. 2.

Relative ion intensity of WO₃⁺ (a), W₂O₆⁺ (b), and W₃O₉⁺ (c) with increasing pressure of C₃H₆. Note the decrease in the reactant ion intensity and the increase in the product corresponding to the oxidation of C₃H₆. The relative reactant ion intensity is plotted on the left y axis and the product ion intensity on the right y axis.

Fig. 3.

Relative intensity of WO₃⁺ (a), W₂O₆⁺ (b), and W₃O₉⁺ (c) with increasing pressure of CO. Note the decrease in the reactant ion intensity and the increase in the product corresponding to the oxidation of CO. The relative reactant ion intensity is plotted on the left y axis and the product ion intensity on the right y axis.

Typically, the relative reactivity observed in ion-molecule reactions decreases with increasing cluster size as the single positive charge of the cation is distributed over a larger number of atoms resulting in a smaller partial positive charge at each individual metal center. Here, however, it appears that cation oxides containing an odd number of tungsten atoms are more reactive with C₃H₆ than W₂O₆⁺. Nevertheless, because the rate constant for W₃O₉⁺ is an order of magnitude lower than for WO₃⁺, it also appears that the reactivity is decreasing with increasing cluster size despite the odd–even oscillation.

In a recent study, it was shown that a silver oxide cation cluster with the same stoichiometry as the bulk Ag₂O⁺ was reactive for the epoxidation of C₃H₆, albeit with competing C–H activation reactions (61). Another theoretical work also demonstrated that positively charged silver oxide clusters, when reacted with C₃H₆, favored propylene oxide formation (62). A bulk-phase study of the catalytic activity of molybdenum oxide nanoparticles supported on silica revealed high selectivity toward the formation of propylene oxide (63). Because molybdenum is the lighter congener of tungsten, it is reasonable that tungsten oxides may also exhibit similar selectivity. Based on these previous findings, we propose that the C₃H₆O product resulting from an oxygen transfer from cationic, stoichiometric tungsten oxide clusters is probably propylene oxide, although we do not have experimental confirmation of this product assignment. Previous studies conducted by our group (54) and by others (56, 57) have shown that stoichiometric metal oxide clusters, for example V₂O₅, V₄O₁₀, and MgO, can form oxygen-centered radicals (M–O●, with M representing metal) with elongated metal–oxygen bonds when a single electron is removed from the neutral species. Indeed, comparison of the results of structural calculations on neutral and cationic vanadium oxides show that for the neutral V₂O₅ and V₄O₁₀ clusters, all of the terminal V–O bond lengths are ≈1.61 Å (64). For the cationic species, in contrast, one terminal V–O bond is found to elongate significantly to 1.78 Å in V₂O₅⁺ and to 1.75 Å in V₄O₁₀⁺ (54). Spin-density calculations also reveal that the single unpaired electron is localized at the terminal oxygen atom with the elongated bond (57). Metal oxide clusters containing oxygen radicals have been found to be reactive for the selective oxidation of ethylene (54) and the activation of methane (56). Structural calculations on neutral (50) and anionic (46–48) tungsten oxide clusters illustrate that the stoichiometric species all contain terminal oxygen atoms that are bound exclusively in the atomic form. Furthermore, each of the terminal oxygen atoms in a given cluster have approximately the same W–O bond length. PES experiments have established that with increasing oxygen content, there is a shift in the detachment features from both W 5d and O 2p features to solely O 2p type orbitals (48). Plots of the highest occupied molecular orbitals (HOMO) clearly show the orbitals to be highly localized on the terminal oxygen atoms (50). Removal of a single electron to form the cationic stoichiometric cluster, therefore, would likely result in the localization of a single unpaired electron on one terminal oxygen atom. Based on these previous findings for both transition metal and alkali earth metal oxides, we propose that the enhanced oxidation reactivity of WO₃⁺, W₂O₆⁺, and W₃O₉⁺ may be explained by the presence of radical oxygen centers with elongated W–O bonds. Theoretical calculations to address this proposal would be valuable.

The same clusters that are active for the oxidation of C₃H₆ were also found to be reactive for the transfer of a single oxygen atom to CO, forming CO₂ according to Eq. 3.

Fig. 3 displays the relative ion intensities with increasing pressure of CO. For each reaction the intensity of the stoichiometric W_xO_3x⁺ cluster is observed to decrease with increasing CO pressure, whereas the oxygen-loss product, W_xO_3x−1⁺, becomes more pronounced. WO₃⁺ is by far the most reactive cluster, with the product intensity accounting for ≈20% of the total ion intensity at a pressure of 20 mTorr of CO. W₂O₆⁺ and W₃O₉⁺ are less reactive, with product intensities constituting ≈5% and 8% of the total ion intensity, respectively, at the maximum gas pressure. To quantify the change in relative CO oxidation reactivity with increasing cluster size, the reaction rate constants were calculated as described above for C₃H₆. The plots of ln[I_r/I₀] as a function of CO concentration are displayed in Fig. S2. The calculations establish rate constants on the order of 4.5 × 10¹³ cm³s¹ for WO₃⁺, 9.0 × 10¹⁴ cm³s¹ for W₂O₆⁺, and 1.1 × 10¹³ cm³s¹ for W₃O₉⁺, also far below the collision rates. Similar to the trend in relative reactivity observed for C₃H₆, WO₃⁺ is again the most reactive species with CO, followed by W₃O₉⁺, and last by W₂O₆⁺. Therefore, clusters with an odd number of tungsten atoms are found to be relatively more reactive with CO than W₂O₆⁺. Because the rate constant for W₃O₉⁺ is less than WO₃⁺, it also appears that the reactivity is decreasing with increasing cluster size.

This decrease in relative reactivity with increasing cluster size can likely be understood by the decreasing partial positive charge at the tungsten centers and the increasing number of internal modes with larger cluster size. CO binds strongly to positively charged metal centers through donation of its lone pair of electrons (65). Therefore, small oxide clusters with a large partial positive charge on each tungsten atom would result in a more exothermic adsorption of CO. Because all of the energy is retained by the gas-phase cluster, this may enable any subsequent barriers to oxidation to be easily overcome (66). However, because the partial positive charge of each metal center decreases with increasing cluster size (67), the CO-binding energy also decreases. The cluster, therefore, gains less energy through CO adsorption, and fewer initial encounter complexes have sufficient energy to overcome the barriers leading to CO₂ formation. A recent theoretical study of the Lewis acidity of neutral tungsten oxide clusters showed that WO₃ had the highest acidity (136.3 kcal/mol), whereas larger clusters were less acidic (W₂O₆ = 125.7 kcal/mol and W₃O₉ = 113 kcal/mol) (51). Furthermore, with increasing cluster size, it becomes less probable that the energy gained through the adsorption of CO onto the cluster will partition into the internal mode leading to CO oxidation. Last, in larger clusters, which contain more tungsten atoms, it is likely that reactant molecules may bind to a tungsten site that is further away from the proposed radical oxygen centers. It is more unlikely, therefore, that the reactant molecules will migrate to the active center and be oxidized. Small positively charged tungsten oxide centers, however, may still be the most active sites for CO oxidation. The oxygen radical centers that we propose as potential active sites for the oxidation of C₃H₆ are also suggested to be the active centers for CO oxidation.

Comparison of the relative reactivity of WO₃⁺, W₂O₆⁺, and W₃O₉⁺ with CO and C₃H₆ reveals that the rate constants for C₃H₆ are generally larger than for CO. It is possible that the higher reactivity observed with C₃H₆ is the result of the larger collision cross-section and higher polarizability of this molecule compared with CO. To investigate this possibility, the Langevin cross-sections (σ_L) for ion-molecule association were calculated for the reactions of both CO and C₃H₆ with WO₃⁺, W₂O₆⁺, and W₃O₉⁺ by using Eq. 4.

In Eq. 4, e is the charge of the ion, ε_o is the vacuum permittivity, α is the polarizability of CO or C₃H₆, and E_CM is the center of mass collision energy between the cluster ions and the reactant molecules. The Langevin cross-sections for the interaction of WO₃⁺, W₂O₆⁺, and W₃O₉⁺ with CO are ≈72 Ų, 99 Ų, and 120 Ų, respectively. For C₃H₆, the cross-sections are ≈108 Ų, 146 Ų, and 177 Ų for WO₃⁺, W₂O₆⁺, and W₃O₉⁺, respectively. Although the larger polarizability of C₃H₆ results in a larger Langevin cross-section, it cannot completely account for the higher reactivity observed between stoichiometric tungsten oxide cations and C₃H₆ compared with CO. It is likely that the energy barriers separating the initial-encounter complexes from the oxidation products may also be lower for C₃H₆ than for CO. Theoretical calculations would be able to address this possibility.

These gas-phase findings provide direct insight into the influence of size and stoichiometry on cluster reactivity. Out of a broad distribution of tungsten oxide clusters containing species with both lower and higher oxygen saturation than the bulk material, only stoichiometric clusters were observed to strongly oxidize both C₃H₆ and CO. Oxygen-deficient clusters were found to be completely unreactive or to adsorb an intact C₃H₆ molecule. Oxygen-rich clusters, in contrast, fragmented through the loss of weakly bound O₂ units. Furthermore, it was shown that increasing cluster size may result in lower CO-binding energies and, consequently, decreased oxidation reactivity. These unique insights into the reactive behavior of tungsten oxide catalysts were made possible through the application of tandem mass spectrometry.

Conclusion

In this study, we demonstrate the application of GIB tandem mass spectrometry to study the reactivity of cationic metal oxide clusters. Mass-selected studies provide insight into the influence of size, stoichiometry, and ionic-charge state on cluster reactivity. Positively charged tungsten oxide clusters having the same stoichiometry as the bulk WO₃⁺, W₂O₆⁺, and W₃O₉⁺ are shown to exhibit enhanced activity and selectivity for the transfer of a single oxygen atom to propylene (C₃H₆), suggesting the formation of propylene oxide (C₃H₆O). The same clusters are also demonstrated to be highly active for the oxidation of CO to CO₂. We propose that the enhanced oxidation reactivity of these species with both C₃H₆ and CO may result from the presence of oxygen-centered radicals (M–O●) with elongated metal–oxygen bonds.

Experimental Method

The reactivity of tungsten oxide cations with C₃H₆ or CO was studied by using a guided-ion-beam mass spectrometer described in detail in ref. 58. Briefly, tungsten oxide clusters were produced in a laser vaporization (LaVa) cluster source by pulsing oxygen seeded in helium (10%) into the plasma formed by ablating a tungsten rod with the second harmonic (532 nm) of a Nd:YAG laser. The clusters exit the source region through a 27-mm-long conical expansion nozzle and are cooled through supersonic expansion into vacuum. During supersonic expansion, the high-pressure (13.2 atm) expansion gas mixture passes through a narrow-diameter nozzle into vacuum. The random thermal energy of the clusters is thereby converted into directed kinetic energy of the molecular beam. Consequently, the internal vibrational and rotational energy of the clusters is lowered through collisions with the He carrier gas. The kinetic energy imparted to the cluster ions by the supersonic expansion was determined, employing a retarding potential analysis (58), to be ≈1.0 eV in the laboratory energy frame (E_LAB). Ideally, all clusters exiting the supersonic expansion source have the same initial kinetic energy. By using Eq. 5

the initial center-of-mass collision energy (E_CM) was calculated for WO₃⁺, W₂O₆⁺, and W₃O₉⁺ to be ≈0.11 eV, 0.06 eV, and 0.04 eV for CO and 0.15 eV, 0.08 eV, and 0.06 eV for C₃H₆, respectively. Because subsequent collisions are expected to dissipate the initial energy of a given cluster, the values reported above serve to establish an upper limit on the kinetic energy of the reactive collisions.

After exiting the source region, the clusters pass through a 3-mm skimmer, forming a collimated molecular beam and are then directed into a quadrupole mass filter employing a set of electrostatic lenses. The quadrupole mass filter isolates clusters of a desired mass that are then passed into an octopole collision cell. Variable pressures of C₃H₆ or CO are introduced into the octopole collision cell by employing a low-flow leak valve. The gas pressure is monitored by using a MKS Baratron capacitance manometer. Product ions formed in the collision cell are mass analyzed by a second quadrupole mass spectrometer. Finally, the ions are detected with a channeltron electron multiplier connected to a mutichannel scalar card. The experimental branching ratios presented in Results and Discussion illustrate the change in normalized ion intensity with increasing pressures of C₃H₆ or CO reactant gas. At higher gas pressures, therefore, the ratio of reactant ion intensity to total ion intensity becomes smaller, whereas the ratio of product ion intensity to total ion intensity becomes larger. Experiments were also conducted with inert N₂ to verify that the products observed with C₃H₆ or CO are the result of a chemical reaction and not the products of collisional fragmentation.