Reactions of platinum clusters 195Ptn±, n = 1–24, with N2O studied with isotopically enriched platinum

Cationic and anionic clusters 195Ptn±, 1 ≤ n ≤ 24, were generated by laser vaporization using an isotopically enriched platinum sample.

The oxidation of the clusters by nitrous oxide under binary collision conditions was investigated as a function of charge and size in a Fourier transform ion cyclotron resonance mass spectrometer.

The only reaction observed is sequential addition of an oxygen atom, with the loss of molecular N2.

The reaction rates as well as the number of reaction steps exhibit a strong and irregular dependence upon the number of atoms in the cluster and its charge.

These results show that isotopically enriched samples allow for the investigation of large cluster sizes together with an unambiguous assignment of reaction products and quantitative determination of rate constants.

Transition metals and their compounds, with their multitude of oxidation states, are efficient and frequently used catalysts.

The reactions of ionic metal clusters, in particular those involving transition metals, with small gas phase molecules represent convenient, relatively simple models for heterogeneous catalysis, and for these reasons they were in the last two decades extensively investigated by mass spectrometry.1–5

Platinum, palladium and rhodium are useful dehydrogenation catalysts,6,7 but are also used extensively in removing toxic oxides such as carbon monoxide or nitrogen oxides from automotive exhaust.8,9

Several groups have investigated dehydrogenation of simple hydrocarbons, such as methane, ethane, ethylene or acetylene on platinum cluster ions, and their reactions with various simple molecules, like CO, N2O or CH4.7,9–22

We have previously studied methane activation by gas phase Ptn± cluster ions in the n = 1–9 size range.7

Schwarz and coworkers succeeded, by mass-selecting the most intense peak, and then thermalizing the ions in argon collisions, in studying the reactions of 195Ptn+ with N2O and other small molecules up to n = .59

Larger clusters up to n = 30 have also been investigated in flow reactors,11,21 albeit without resolving the individual isotopomers, which made interpretation of the data somewhat difficult.

Heiz et al. succeeded in depositing size selected platinum clusters with up to 20 atoms on surfaces and investigated their chemical reactivity.23,24

While gas phase reactivity studies on metals like niobium or rhodium can easily be done for clusters of sizes ranging up to thirty atoms,6,25,26 large clusters of platinum and palladium have been studied in less detail.

The reason for this is that unlike rhodium or niobium, which are monoisotopic elements, both palladium and platinum have six stable isotopes each.

For larger clusters, this very quickly leads to dilution of the signal among many isotopomers, and overlapping of the products with the reactants.

Both effects together so far have limited the cluster sizes accessible in reactivity studies.

All works in which the reactivity was probed in the gas phase by mass spectrometry with at least unit mass resolution, have therefore been restricted to smaller clusters, with ten atoms or less.7,9,14,19,20,22

In the present study we use isotopically enriched platinum (97.28% 195Pt, Oak Ridge National Laboratories) to investigate the oxidation of 195Pt+/−n clusters, n = 1–24, with N2O under binary collision conditions in a Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer.

The platinum sponge was pressed into a disk and subjected to zonal melting.

The resulting pearl was rolled into a foil which was used as a target for our home-built laser vaporization source.26–29

In this source, the firing of the laser is synchronized with a helium pulse from a home-built piezo-electric valve.

From the laser generated plasma, cold clusters form in the supersonic expansion of the high pressure gas into high vacuum.

Ionic clusters are guided by a series of electrostatic lenses through several stages of differential pumping, and trapped and stored inside the ICR cell in the ultrahigh vacuum of the FT-ICR mass spectrometer (Bruker/Spectrospin CMS47X, 4.7 T, APEX III data station).

For each mass spectrum, clusters generated in 20 laser shots over a period of 2 s are accumulated in the cell.

To study ion-molecule reactions, the pressure inside the ICR cell was raised from its base value of about 2 × 10−10 mbar to a constant value of 2.0 × 10−8 mbar by controlled admission of N2O (99.5%, Messer Griesheim) at room temperature.

Mass spectra were acquired after a series of varying reaction delays.

A crude selection of a cluster size range was accomplished by adjusting the timing of the cell trapping, effectively selecting a fraction of ions within a certain time of flight window from the source to the ICR cell.

Relative rate constants were obtained by fitting the experimental data to pseudo-first order reaction kinetics, and converted to absolute rate constants and reaction efficiencies using the average dipole orientation (ADO) theory.30–32

The advantage of using isotopically enriched platinum is exemplified by the n = 7 cluster in Fig. 1.

In the top trace, obtained with a non-enriched sample, the intensity is distributed over at least seventeen isotopic peaks.

In the bottom panel, obtained with the isotopically enriched target, a strong 195Pt7+ peak is present.

The weak side peaks corresponding to 194Pt195Pt6+ and 195Pt6196Pt+ clusters exhibit intensities that are consistent with the specified enrichment grade of the sample.

A typical mass spectrum of cationic clusters 195Ptn+ with n = 10–21 after a reaction delay of 5 s is shown in Fig. 2.

The reactions occurring are quite simple, and similar to those observed previously for smaller clusters:9,14 Oxide clusters, 195PtnO+ are formed with loss of N2.

Apparent is a size-specific variation in reactivity.

While n = 11, 12, 15 and 20 are very reactive, n = 10, 13, 14, 19 and 21 show very little to no reactivity.

One can also see that in the case of n = 18, and to some extent n = 16, reaction with a second N2O occurs, resulting in the loss of another N2 molecule, and formation of dioxide clusters, 195PtnO2+.

More detailed insight and quantitative reaction rates are obtained by actually fitting the time-intensity profile to pseudo-first-order kinetics, as exemplified in Fig. 3. 195Pt8+ is shown in the top panel.

In this case, one can follow an exponential decay of the reactant cluster, and the oxide product, 195Pt8O+ reacts further with a second molecule of N2O, resulting in a dioxide cluster 195Pt8O2+.

The figure also reveals that already at a nominal time t = 0 some product is present.

This is due to reactions occurring during the cluster accumulation period.

As a second example, in the bottom panel the exponential decay of the 195Pt20+ cluster is shown, and concurrent growth of the only product, 195Pt20O+.

The reaction proceeds faster for n = 20 than for n = 8, but only one reaction step is observed.

Overall, the chemistry pattern found for all of the clusters, both anions and cations, within the range studied is quite simple.

We do not observe any evidence for loss of a platinum atom from the cluster, which would in the present analysis lead to a curvature of the reactant ion decay in the semi-logarithmic plots of Fig. 3.

When reaction upon collision with N2O does occur, the N2 molecule is lost, and the single oxygen atom adds to the cluster.

While in some cases, 195PtnO± appears unreactive, in others it reacts further to form the respective dioxide, 195PtnO2±.

In a few cases, also a third step resulting in 195PtnO3+ is detected, so that the reactions can be described by the simple equation pattern:Fig. 4 shows a graphical comparison of the observed rates, on a logarithmic scale, of the first reaction step, eqn. (1) for cations and anions as a function of cluster size.

Numerical data for the first reaction step are summarized in Table 1, and for the second, and in the few cases it is observed, third reaction step in Table 2.

One finds that the rates of the reaction, as well as the number of oxidation steps observed in the time frame of the experiment vary appreciably, and in a seemingly random manner from cluster to cluster, and the reactivity pattern is also strongly dependent on the cluster charge.

Among the cations, the n = 6 and 20 clusters exhibit the fastest reaction according to eqn. (1), with rates above the ADO collision efficiency.30–32

This indicates that the collision probability for large clusters is higher than predicted by the ADO theory.

In neither case does the primary product 195PtnO+ react further, and after about five seconds, these clusters are almost completely converted to the primary product.

The clusters n = 7, 8, 11, 12, 15 exhibit an appreciable reactivity, with the primary product reacting further (2).

On the other hand, no reaction could be established for clusters n = 10, 14, 23, and upper limits for their reactivity are derived.

Also rather unreactive are very small clusters 195Ptn+, n ≤ 5, with n = 1, 2, and 5 reacting only marginally, and no detectable reaction for n = 2–4.

For n = 15 and n = 18, also a third reaction step, according to (3), is observed, which is in both cases faster than the second step.

The negative cluster ions overall react somewhat slower than the cations, and the size dependence is considerably different.

Interestingly, the n = 6 cluster, which was the fastest reacting cluster among the cations, represents in fact a deep reactivity minimum among the anions, with the 195Pt6 cluster reacting twenty times slower than the neighboring n = 4,5,7,8 clusters.

The n = 4 cluster, which is completely unreactive among the cations, is in fact the fastest reacting anion, and also exhibits a fast second reaction step to form 195Pt4O2.

On the other hand, the n = 10 and 14 clusters, which appear completely inert as cations, are also as anions quite unreactive.

To summarize the observations, one can understand the reactions as a decomposition of the nitrous oxide on the metal cluster surface, yielding an oxygen atom and molecular N2.

While the oxygen atom oxidizes the platinum, the weakly bound nitrogen is released.

We do not observe stabilization of the undecomposed N2O on the surface, and see no trace of products containing nitrogen.

The nearly three orders of magnitude wide variation in the reaction rates between various clusters may reflect their structural differences, and the presence of differently efficient “sites” on the cluster surface.

The fact that anions and cations behave differently may suggest considerable differences between the structures of anions and cations.

We have previously succeeded in “soft-landing” otherwise reactive molecules such as benzene or methane on a cluster surface by exchanging them for weaker bound argon ligands,4 and it might be interesting to try a similar experiment with N2O on platinum clusters.

For small anions, n = 3–7, where our data overlap with earlier results of Hintz and Ervin,13,14 as well as for cations, n = 1–5, measured by Schwarz and coworkers,9 there is good quantitative agreement in the measured rates.

Over the entire range of sizes explored, addition or subtraction of a single atom can change the reactivity by orders of magnitude.

Isotopically enriched platinum allows the quantitative investigation of size-dependent reactivities of platinum clusters.

The present work demonstrates the feasibility and the potential of this approach to further substantiate the notion introduced by Schwarz and coworkers that small platinum clusters can serve as model system for surface catalytic reactions.9

They also indicate, however, that, as was previously shown for supported platinum clusters by Heiz et al.,24 also in the gas phase “each atom counts”.