Enhanced Reactivity of Highly Vibrationally Excited Molecules on Metal Surfaces

See allHide authors and affiliations

Science  04 Jun 1999:
Vol. 284, Issue 5420, pp. 1647-1650
DOI: 10.1126/science.284.5420.1647


The chemical dynamics of highly vibrationally excited molecules have been studied by measuring the quantum state–resolved scattering probabilities of nitric oxide (NO) molecules on clean and oxygen-covered copper (111) surfaces, where the incident NO was prepared in single quantum states with vibrational energies of as much as 300 kilojoules per mole. The dependence of vibrationally elastic and inelastic scattering on oxygen coverage strongly suggests that highly excited NO (v = 13 and 15) reacts on clean copper (111) with a probability of 0.87 ± 0.05, more than three orders of magnitude greater than the reaction probability of ground-state NO. Vibrational promotion of surface chemistry on metals (up to near-unit reaction probability) is possible despite the expected efficient relaxation of vibrational energy at metal surfaces.

Excited states of molecules are often much more reactive than ground-state species, in part because internal energy can help overcome barriers to reaction. Reactions of electronically excited molecules have been studied extensively; indeed, most of the field of photochemistry involves such states. The preparation and study of vibrationally excited reactants has proven more difficult. In polyatomic molecules, vibrational excitation may not remain localized in particular vibrations, and even in diatomic species, such states can prove difficult to prepare in large fluxes (1). These difficulties are particularly great for highly vibrationally excited molecules, where the vibrational excitation is a substantial fraction of the bond energy. The motivation to study the chemistry of such species is strong. The atoms in such molecules have large kinetic energies, which may enhance reaction rates. Furthermore, large amplitude vibration can distort the molecular electronic wave function, altering physical and chemical properties.

A few measurements of the reactivity of highly vibrationally excited molecules in the gas phase have been performed (2) and have shown that such excitation can produce reaction cross sections approaching the gas kinetic limit. Studies of the interactions of vibrationally excited molecules with surfaces have been restricted to low-lying vibrational states, populated either thermally (3) or with optical excitation (4). The reaction probabilities observed, although strongly influenced by vibration, were low (10–2).

Unlike gas-phase reactions, it has not been clear that increasing the reactant vibrational excitation for reactions at surfaces can ever yield high reactivity. Particularly for metal surfaces, vibrational relaxation is potentially an important competitive channel. Indeed, theory predicts that vibrational relaxation rates will increase with the vibrational energy of the initial state (5, 6) and thus could effectively thwart attempts to reach large reaction probabilities by using highly vibrationally excited molecules.

We present an experimental method that can reveal the chemical dynamics of highly vibrationally excited molecules at solid surfaces. We illustrate our approach using the model system of NO molecules interacting with a Cu(111) surface, where we observe a thousandfold vibrational enhancement of surface reactivity. In contrast, neither reagent translational energy nor surface temperature has a strong effect on the reaction probability over the range we have probed. The results are consistent with a “late” transition state (7) and illustrate the potential of this experimental approach for detailed studies of the transition-state dynamics of surface chemical reactions. We studied the scattering of NO molecules with vibrational states from v = 13 to v = 22 (vibrational energies of 290 and 390 kJ/mol). The narrow bandwidths of the lasers made it possible to populate a single rotation-vibration state with defined spin-orbit (Ω) and parity quantum numbers, so that the internal energy of NO before scattering was accurately known. Because the experimenter is freed from the limitations of the Boltzmann factor on the degree of vibrational energy available, a much broader range of surface chemical dynamics becomes accessible to experimental investigation. These results show that it is possible to promote surface reactivity to near-unit probability with vibrational excitation.

We prepared highly vibrationally excited NO with the use of stimulated emission pumping (SEP) (1, 7), as shown schematically in Fig. 1A. One laser, tuned to ωpump, excites ground-state molecules to an intermediate electronically excited “stepping-stone” state, while a second laser, tuned to ωdump, induces emission back down into a specific vibrationally excited level of the ground electronic state (8). The large structure change upon electronic excitation in many molecules allows good Franck-Condon overlap to high vibrational states in the two-photon transition. For example, the 24th excited vibrational state of NO, with vibrational energy of >400 kJ/mol, can be easily prepared by this method with efficiencies exceeding 1% (7, 9).

Figure 1

(A) Schematic representation of the stimulated emission pumping method. Shown are two diatomic electronic states, as well as the dependence of their energies on the displacement of the atomic separation, r, from the equilibrium value,r e. Molecules are excited out of ground-state levels through an allowed transition to an excited electronic state with a distorted structure. Laser-induced emission transfers population back to single highly vibrationally excited quantum states of the ground electronic state. Three realistic vibrational wave functions are shown. Energies accessible with this method can be large (>400 kJ/mol) and pumping can be efficient (∼10–1 to 10–2). (B) Experimental detail. Pump and dump lasers prepare highly vibrationally excited NO in the second differential pumping chamber downstream from a pulsed molecular beam of NO (seeded in He or H2). A photomultiplier tube is used for fluorescence dip spectroscopy, which helps to control the optical excitation step. Pump laser–induced fluorescence is also monitored by a photomultiplier for signal normalization. The excited molecules travel through one more region of differential pumping and collide with a well-characterized copper surface or oxygen adlayer on copper. Scattered NO is state-selectively ionized using 1+1 resonance-enhanced multiphoton ionization. The ions are extracted back along the molecular beam direction and deflected to a microchannel plate detector. By scanning the frequency of the probe laser, we could measure the vibrational and rotational state distribution of the products. Translational energy distributions of the products are measured by recording the intensity of the REMPI signal as a function of the time delay between the excitation and probe lasers. Angular distributions were recorded by translating the probe laser across the direction of the incident molecular beam.

The apparatus used for the studies of vibrationally excited molecules at solid surfaces is shown in Fig. 1B. A pulsed molecular beam is skimmed and excited by the pump (10) and dump (11) lasers. A photomultiplier views the region where the laser beams cross the molecular beam, allowing fluorescence dip measurements (12) to characterize the optical excitation efficiency. The resulting highly vibrationally excited NO molecules then travel through a set of differential pumping apertures and enter the ultrahigh vacuum scattering chamber, where they collide with a Cu(111) surface at normal incidence.

Although the Cu(111) surface could easily be prepared free of oxygen by several cycles of Ar+ sputtering and annealing, after the surface had been exposed to a molecular beam of ground-state NO for only a few minutes, a clear oxygen Auger electron spectroscopy (AES) signal (at 519 eV) was observed from the surface (13). Surface oxidation of copper by ground-state NO is well known and has been previously studied (14). At the surface temperatures used here (300 to 500 K), nitrogen is removed from the surface by subsequent surface reactions, and none was detected by AES (15). Stable and reproducible oxide surface overlayers of copper were readily prepared by exposure of the clean Cu(111) surface to NO2, followed by elevation of the surface temperature to 800 K for several seconds. This procedure ensures that only chemisorbed oxygen atoms remain on the surface (16).

With the use of such a stable O-atom adlayer, we carried out a series of experiments designed to examine the inelastic energy transfer of highly excited NO with an oxidized copper surface. In these experiments, specific vibration-rotation states of NO were prepared using SEP (9, 17). Molecules that scattered from the surface were detected by 1+1 resonance enhanced multiphoton ionization (REMPI) (18). In this way, we have measured the vibrational-rotational state distributions as well as angular and kinetic energy distributions of inelastically scattered NO (19). Figure 2 shows a portion of the REMPI spectrum obtained when NO is initially prepared in ∣v = 13, J = 2.5, Ω = 0.5〉 and the probe laser is tuned to detect molecules in thev = 13 state [probed through the A(4) ←X(13) γ-band]. The solid line is the result of a least squares fit to the data points (20) using an experimentally determined laser line shape and theoretical (nonadjustable) transition frequencies and line strengths (21). A Boltzmann distribution of the v = 13 rotational states was used to simulate the rotational state populations, and a single parameter (the rotational temperature) was used to optimize the fit. The excellent fit to the data shown in Fig. 3 is typical for the entire spectrum (which probes both Ω = 0.5 and 1.5 and rotational states up to J = 25.5) and provides an unambiguous fingerprint of the NO(v = 13) molecules redistributed among their rotational states after collision with the surface.

Figure 2

REMPI spectrum of NO(v = 13) scattered from an oxygen adlayer on Cu(111). The points are the data; the solid line is a least squares fit to the data. Only a portion of the spectrum is shown near the P 12 andP 22 band heads.

Figure 3

Survival probability,S(σ)/[S(0) + S(∞)], of NO(v = 13) as a function of surface exposure to the oxidizing action of ground-state NO. The increase in survival probability is explained by decreased reactivity of highly vibrationally excited NO on the O-covered copper surface.

We also obtained REMPI spectra of the scattered molecules at very early times after the beginning of the oxidation of the clean copper surface by the beam of ground-state NO. This experiment revealed that the survival of specific states of highly vibrationally excited NO is significantly reduced when the oxygen adlayer is removed from the surface. The dependence of the REMPI signal for molecules scattered into v = 13 (vibrationally elastic channel) upon exposure of the surface to ground-state NO could be fit to the theoretical (first-order kinetics) expression S(σ) =S(∞) × [1 – exp(–Kσ)] +S(0) (Fig. 3), where σ represents the surface exposure to NO in monolayers (ML) (22), S(σ) is the coverage-dependent REMPI signal, K is a rate constant, and S(0) is the REMPI signal obtained in the limit of zero exposure. The asymptotic magnitude of the REMPI signal is given byS(∞) + S(0). The survival probability as a function of surface exposure to NO is thereforeS(σ)/[S(0) + S(∞)].

There are two possible explanations for the observation shown in Fig. 3. First, vibrationally inelastic channels could be significantly more probable on the clean copper surface relative to the oxidized surface. If this is the case, the “missing v = 13 signal” at low exposures should appear as enhanced signal for other vibrational states. That is, we expect in this case to observe a decrease in signal with exposure for lower energy vibrational states that are populated by the supposed enhanced vibrationally inelastic energy transfer. The second possible explanation for the diminished signal at low exposures is that the highly vibrationally excited NO reacts, thereby sticking to the clean copper surface, but cannot react as rapidly (if at all) on the oxidized surface. In this case we would expect to see a growth in all of the inelastic channels with exposure as the surface is transformed from its reactive form (copper) to its inactive form (oxidized copper).

We failed to find any inelastic channels that exhibited a dependence to surface NO exposure different from that shown in Fig. 3, despite our sensitive detection of many inelastic channels (23). Indeed, for a fixed surface temperature, all of the growth curves were quantitatively identical, which suggests that they reflect the same kinetic process: reaction of highly vibrationally excited NO on a clean copper surface. Presumably, oxidation of the copper surface by ground-state NO removes the sites where reaction of highly excited NO can take place. Thus, for oxidized surfaces, inelastic channels dominate.

On the basis of this interpretation, we can experimentally derive the reaction probability of highly vibrationally excited NO with a Cu(111) surface, P react =S(∞)/[S(0) + S(∞)]. This analysis yields a value of 0.87 ± 0.05 for the data shown in Fig. 3 (24). That is, there is an 87% chance of reaction for NO(v = 13) on Cu(111). A similar value was obtained for NO(v = 15). Moreover, the reaction probability was found to be independent of the surface temperature and the translational energy of the molecular beam over the range studied (25).

To compare the rate of NO(v = 13, 15) reaction on Cu(111) with that for NO(v = 0), we measured the O AES signal versus exposure for ground-state NO. The oxygen coverage as a function of NO exposure exhibited first-order kinetics: I(σ) =I(∞) × [1 – exp(–kσ)]. Here,I(σ) is the intensity of the oxygen peak (at 510 V in AES) normalized to the intensity of the Cu peak (at 920 V). As before, σ stands for the NO exposure measured in ML. I(σ) is directly proportional to the oxygen coverage at the surface. Fitting the kinetic expression to the measurements, we obtained a value ofk ∼ 8 × 10–4 ML–1 for the rate of oxidation of Cu(111) by ground-state NO. TakingI(∞) = 0.3 ML as the saturation coverage of oxygen at Cu(111) (16), we could estimate the reaction probability at zero coverage for ground-state NO on Cu(111) [k × I(∞)] to be 2 × 10–4. This shows that vibrational excitation of NO enhances the probability of reaction by more than three orders of magnitude.

In principle, the observed vibrationally enhanced sticking of NO on Cu(111) could be nondissociative (trapping and desorption) or dissociative (forming O and N atoms bound to copper). If trapping and desorption were important, one might observe a reduced scattering signal at zero oxygen coverage because the desorbing molecules would be spread over a large range of scattering angles and speeds. Moreover, if the trapped molecules had a long lifetime on the surface, they might be emitted in vibrational states even lower than those probed in this work (up to Δv = –5). However, in light of the experimental evidence, the trapping and desorption scenario appears counterintuitive. In essence, trapping results from the rapid removal of energy from the gas-phase molecule during the surface collision. Thus, it is commonly observed that trapping is strongly inhibited by increased translational energy of the gas-phase molecule and to a more modest extent inhibited by increased surface temperature (26). Although there have been no studies of such an effect, it appears reasonable that the addition of vibrational energy to the gas-phase molecule would (like translation and rotation) also inhibit trapping.

In contrast, the sticking probability measured in these experiments is independent of both initial translation energy and surface temperature, and is enhanced by vibrational excitation by more than three orders of magnitude. This result indicates a direct chemical reaction where the well-defined motion of the reactant has not been overwhelmed by energy transfer to the surface. Vibrational excitation of NO enhancing the dissociation of the NO bond is also intuitively appealing. Indeed, the large enhancement by vibration and weak dependence on translation are consistent with a “late” transition state (27), where the transition state resembles more closely the adsorbed products of the reaction than it does the reactants. The results shown here lay the groundwork for detailed studies of the nature of the transition states of many surface reactions, coupling high-level experiment and theory.


Stay Connected to Science

Navigate This Article