Report

Steric Effects in the Chemisorption of Vibrationally Excited Methane on Ni(100)

See allHide authors and affiliations

Science  30 Jul 2010:
Vol. 329, Issue 5991, pp. 553-556
DOI: 10.1126/science.1191751

Abstract

Newly available, powerful infrared laser sources enable the preparation of intense molecular beams of quantum-state prepared and aligned molecules for gas/surface reaction dynamics experiments. We present a stereodynamics study of the chemisorption of vibrationally excited methane on the (100) surface of nickel. Using linearly polarized infrared excitation of the C-H stretch modes of two methane isotopologues [CH43) and CD3H(ν1)], we aligned methane’s angular momentum and vibrational transition dipole moment in the laboratory frame. An increase in methane reactivity of as much as 60% is observed when the laser polarization is parallel rather than normal to the surface. The dependence of the alignment effect on the rotational branch used for excitation indicates that alignment of the vibrational transition dipole moment of methane is responsible for the steric effect. Potential explanations for the steric effect in terms of an alignment-dependent reaction barrier height or electronically nonadiabatic effects are discussed.

The need to understand the dynamics of gas/surface reactions is driven by the critical role that these reactions play in many industrial processes, such as heterogeneous catalysis for molecular synthesis and chemical vapor deposition of thin films. Highly detailed state-resolved measurements are required to reveal the underlying microscopic dynamics and reaction mechanisms. Quantum-state–resolved data also enable stringent tests of theoretical models of gas/surface reactivity. The choice of methane chemisorption on a Ni surface as a model system is pragmatic, because C-H bond activation of this system is the rate-limiting step in steam reforming of CH4 to produce industrial hydrogen and syngas, the starting material for the synthesis of many commercial compounds.

In the late 1970s, the use of molecular beams for reactivity measurements on single-crystal surfaces enabled the preparation of reactant molecules with well-defined translational energy and incident angle. Molecular beam studies clearly demonstrated that the translational energy of the incident CH4 activates its chemisorption (1). Laser excitation of methane isotopologues to specific rovibrational quantum states (24) further refined the measurements, providing unequivocal evidence that the chemisorption is also activated by vibrational excitation of the incident methane molecule in a mode- (5) and bond-specific (6) manner that statistical models (7) fail to capture.

In the experiments described here, we move a step beyond quantum-state resolution, by exerting steric control over this gas/surface reaction. Excitation by linearly polarized infrared radiation prepares CH4 in a single rovibrationally excited state with aligned angular momentum J and vibrational transition dipole moment μif in the laboratory frame (8). The resulting aligned, state-prepared reactants are used to detect and quantify steric effects in the reaction of vibrationally excited methane isotopologues with a Ni(100) surface. Here, alignment refers to an anisotropic spatial distribution of J and μif vectors, which is either preferentially parallel or perpendicular to a laboratory fixed axis such as the electric field vector E of the excitation laser. In contrast, orientation describes a preference in vectorial direction of J and μif (i.e., parallel or antiparallel) relative to E.

Polanyi highlighted the opportunity to study stereodynamics in surface reactions, given that a single-crystal surface represents a perfectly oriented reaction partner (9). What is needed to completely define the collision geometry in a gas/surface reaction is a method to orient or align incident molecules relative to the surface plane. Several techniques have been used to study alignment and orientation effects in gas/surface reactions, such as hexapole state-selection followed by reactant orientation in a static electric field (10, 11), reactant alignment by resonance-enhanced multiphoton ionization (REMPI) by linearly polarized light (12), and collisional rotational alignment in supersonic jet expansions (13). Another method to obtain information about gas/surface stereodynamics is linearly polarized REMPI of molecules desorbing from a surface, coupled with analysis using the principle of detailed balance (14, 15). Unfortunately, none of these techniques is applicable to CH4.

In contrast to previous stereodynamical studies of gas/surface reactions, our experiments align molecules without permanent dipole moments and probe specifically the alignment-dependent reactivity of vibrationally excited neutral reactant molecules. We explore whether the reactivity of vibrationally excited CH4 depends on the alignment of J and/or μif relative to the surface plane, with the aim of elucidating the mechanism of the vibrationally mode-specific reactivity (5), which is not fully understood.

We performed the experiments in a surface science–molecular beam apparatus (4, 16). First, we exposed the Ni(100) surface in two different locations to an identical dose of vibrationally excited CH43), prepared with one quantum of the antisymmetric C-H stretch vibration ν3, via the R(0) transition at 3028.75 cm−1 but with different laser polarization directions (Fig. 1). Then, quantum-state–resolved reaction probabilities were measured using Auger electron spectroscopy detection of surface-bound C. Comparison of the detected amount of surface carbon (C) resulting from the two depositions indicated that the CH4 reactivity is up to 60% higher for ν3-R(0) excitation (Fig. 2, top) with the laser polarization parallel to the plane of the surface than for perpendicular polarization. The observed alignment effect was quantified by calculating the alignment contrast Δp Δp=S0S0S0+S0 (1)where S0 and S0 are state-resolved initial reaction probabilities for the excited state of CH4 prepared with laser polarization parallel and perpendicular to the surface plane, respectively. Based on nine repeated measurements, Δp = 0.216 ± 0.016 for excitation of the ν3-R(0) transition of CH4, with error limits of ±2σ from the standard deviation of replicate measurements.

Fig. 1

Schematic of the state-prepared and laser-aligned molecular beam deposition experiment. A molecular beam of CH4 with initially isotropic spatial distribution of angular momentum Embedded Image (indicated by the red spheres at top) impinges on a Ni(100) surface at normal incidence. Before surface impact, the molecules traverse a continuous, linearly polarized laser beam focused in the direction of the molecular beam. Incident CH4 is prepared in a specific rovibrationally excited state by rapid adiabatic passage (16) through the resonant laser beam. The resulting probability distributions for Embedded Image (red) and vibrational transition dipole moment Embedded Image (blue) are depicted for the case of R(0) excitation and two orthogonal polarization directions indicated by the double-headed arrows.

Fig. 2

Auger-detected C signal resulting from exposure of a clean Ni(100) surface to CH43) with translational energy of 34 kJ/mol, rovibrationally excited 1 mm upstream of the surface, irradiated with laser polarization parallel (II pol.) or perpendicular (⊥ pol.) to the surface plane. (Top) CH43) excited via R(0). (Middle) CH43) excited via P(1). (Bottom) CH43) excited via R(0) (left) and CH4 irradiated by laser output detuned from resonance (right). Surface temperature is 473 K in all cases.

In order to ascertain whether the observed reactivity difference is due solely to alignment effects of the vibrationally excited CH4, we performed the corresponding pair of depositions using excitation via the ν3-P(1) transition, which produces no alignment of the excited molecules (8). In agreement with our interpretation, we observed no significant difference between the CH43) reactivity for parallel and perpendicular laser polarizations (Δp = 0.016 ± 0.029, four measurements) for excitation via the ν3-P(1) transition (Fig. 2, middle). We verified that the detected C is formed exclusively by chemisorption of ν3-excited CH4 by repeating the deposition experiment with the excitation laser slightly detuned from the ν3 resonance. With the laser off-resonance, no C signal was detectable on the surface where the molecular beam had impinged (Fig. 2, bottom). Corresponding measurements for ν1-excited trideuteromethane, CD3H, yielded Δp = 0.178 ± 0.028 for the ν1-R(0) transition and Δp = 0.009 ± 0.015 for the ν1-P(1) transition, where ν1 is the C-H stretch normal mode of CD3H with a band origin at 2993 cm−1.

In our experimental setup, the homogeneous linewidths of the rovibrational transitions are dominated by transit time broadening to more than 1 MHz. Because the hyperfine splittings for CH4 and CD3H are in the range of 50 to 200 kHz, the laser coherently excites all hyperfine levels when the methane molecules pass through the excitation laser beam. After excitation at time t = 0, the interaction between methane’s nuclear spin and J leads to a dephasing of the hyperfine components, which scrambles alignment created at t = 0 on a time scale of the inverse of methane’s hyperfine splitting. By measuring Δp for excitation of the R(0) transition at distances between 1 and 30 mm from the target surface, we extracted a time scale for the CH43) hyperfine dephasing of ≈15 μs (fig. S2). Hyperfine depolarization was observed to be faster for CD3H (≈5 μs) than for CH4, which is in agreement with the larger splitting of CD3H hyperfine levels (8). Extrapolation of the measured hyperfine decay of Δp toward t = 0 predicts an insignificant reduction for excitation at 1 mm from the surface, compared to a hypothetical preparation directly on the surface.

To explore the origin of the alignment dependence of the CH43) and CD3H(ν1) surface reactivity, we determined the alignment contrast Δp for excitation via R-, Q-, and P-branch transitions (fig. S3) and compared the experimental results to calculated alignment coefficients for angular momentum A0(2) and vibrational transition dipole moment βaxis, using expressions given by Zare and others (1719). Inspection of Fig. 3A shows a sign change for A0(2) when switching from R- to Q-branch excitation, whereas the βaxis coefficients remain positive (Fig. 3B). The fact that alignment contrasts Δp for CD3H(ν1) (Fig. 3) and CH43) (fig. S10) are positive for both R- and Q-branch transitions and scale with βaxis leads us to conclude that the observed reactivity enhancement is due to an alignment of the vibrational transition dipole moment μif rather than the angular momentum J, which is similar to the findings of Simpson et al. for the reaction of vibrationally excited CH4 and CD3H with gas-phase Cl atoms (8).

Fig. 3

Comparison of the observed alignment contrast Δp for CD3H(ν1 = 1) with (A) the angular momentum alignment coefficientEmbedded Image and (B) the vibrational alignment coefficientEmbedded Image. The vertical axes of the graphs are scaled so that their origins and the values associated with R(0) excitation coincide forEmbedded Image and Embedded Image in (A) and Embedded Image and Embedded Imagein (B). Error bars are ±2σ from replicate measurements.

To map out the polarization angle dependence of the alignment effect, we measured the reactivity for CH43) and CD3H(ν1) for R(0) excitation at several intermediate polarization angles, in addition to 0° (∥ pol) and 90° (⊥ pol). The results (figs. S11 and S12) show a continuous decrease in reactivity from the highest reactivity for parallel polarization to the lowest reactivity for perpendicular polarization for both CH43) and CD3H(ν1).

In dynamical stereochemistry, the preferred reactant bond alignment reflects the minimum energy path from reactants to products on the multidimensional potential energy surface (PES) via the reaction’s transition state. For example, the state-resolved experiments of Hou et al. (14) imply a much higher reactivity for broadside than for end-on collisions of D2 with Cu(111) at low collision energy, indicating an alignment-dependent barrier height for this reaction, with a minimum barrier occurring for a D-D bond alignment parallel to the surface. However, for CH4 dissociation on transition metals, calculations of the transition state structure for the dissociation of CH4 on Ni(100) (2022) predict the dissociating C-H bond to be elongated and oriented toward the surface at nearly 45° from the surface normal and barrier heights ranging from 60 to 90 kJ/mol.

Excitation of the infrared-active C-H stretch fundamentals CH43) and CD3H(ν1) by linearly polarized light aligns the vibrational transition dipole moment; i.e., the net C-H stretch, preferentially along the laser polarization axis. For CD3H(ν1), where μif is along the C-H bond axis, the excitation also aligns the C-H bond. Recent state-resolved experiments proved this reaction to be bond-specific by demonstrating that the excited C-H bond dissociates selectively in the chemisorption of CD3H(ν1) on Ni(111) (6).

In comparing the polarization direction leading to the highest reactivity of CD3H(ν1) on Ni(100) with the calculated transition state structure (2022), one must consider that the laser excitation defines alignment and not orientation of the C-H bond in CD3H. Excitation with ∥ polarization aligns the excited C-H bond preferentially parallel to the surface plane, whereas ⊥ polarization produces CD3H(ν1) with the C-H bond along the surface normal, but pointing either toward the surface or away from it. Our measurements therefore probe the average reactivity between the two orientations compatible with a given alignment direction. Such averaging could shift the maximum in the angle dependence away from the transition state angle and toward the parallel polarization direction if C-H orientations pointing away from the surface are nearly nonreactive and the variation in reactivity for C-H bond orientations toward the surface is small.

However, the observation of very similar polarization angle dependences for the reactivity of aligned CD3H(ν1) and CH43) with larger Δp values for CH43) is inconsistent with the idea that the polarization angle dependence reflects the transition state geometry of the reaction. CH43) has four identical C-H bonds with tetrahedral geometry that share the antisymmetric stretch excitation of the ν3 normal mode. The reactivity of CH43) on Ni(100) should show a different polarization angle dependence and smaller Δp values than what is observed for CD3H(ν1) if the orientation of the vibrating C-H bond relative to the transition state geometry were to determine the enhancement. In fact, R(0) excitation by linearly polarized light prepares CH43) in a state with J = 1, l = 1, and N = 0 (8), where J, l, and N designate the total, vibrational, and rotational angular momentum, respectively. This state of CH43) is rotationless (N = 0); i.e., it is characterized by a spatially isotropic rotational wave function for which any orientation of the H atoms is equally probable. Classically, CH4 in this state can be visualized as an ellipsoid vibrating along a principal axis, with the H atoms located on the ellipsoidal surface with unknown orientation (8). The direction of the laser polarization controls the alignment of the principal axis of the vibrating ellipsoid without specifying the C-H bond alignment of CH4. The fact that we observed the highest alignment effect (Δp = 0.216 ± 0.016) for this rotationless state of CH43) is a strong indication that the observed steric effects are due to the net C-H stretch alignment rather than C-H bond alignment.

The underlying mechanism for the alignment-dependent reactivity of vibrationally excited methane on Ni(100) is not obvious. The higher dissociation probability of methane with the net C-H stretch aligned parallel to the surface could be due to either an alignment dependence of the dissociation probability of the vibrationally excited molecule or to an alignment dependence of the rate of vibrational energy transfer to the surface.

A canonical interpretation explains our results in terms of an alignment-dependent barrier height on a multidimensional reactive PES for the methane/surface system. Dynamical calculations of methane dissociation on a realistic PES with up to 15 degrees of freedom may be needed to understand which degrees of freedom play a decisive role in the observed steric effects. Such a high-dimensional PES [based on the Born-Oppenheimer (BO) approximation] and corresponding dynamics simulations are being developed (23) to arrive at a predictive understanding of this important reaction. Our state-resolved, alignment-dependent data can guide such calculations.

Alternatively, the alignment dependence could be due to steric effects in vibrational relaxation rate between the vibrationally excited, incident methane and electron-hole (e-h) pair excitations in the Ni surface (24). An early model for vibrational energy transfer between a vibrating molecule and a metal surface (25) consists of an oscillating dipole interacting with its electric image dipole induced in the conducting surface. A classical calculation (26), treating vibrational relaxation as ohmic dissipation of the induced image current, predicts a maximum dissipation rate for dipole alignment perpendicular to the surface plane, which is twice the minimum rate calculated for parallel dipole alignment. The higher predicted dissipation rate for the perpendicular dipole is consistent with the lower reactivity for this alignment direction if vibrational energy transfer to e-h pairs were significant during the approach of a vibrating methane molecule toward the metal surface.

The image dipole model could also explain the mode-specific reactivity previously observed in comparing the isoenergetic symmetric and antisymmetric C-H stretch normal modes ν3 and ν1 of CH4 (2, 27). In contrast to the infrared-active ν3mode, the ν1-normal mode of CH4 is only Raman-active and carries no vibrational transition dipole moment from the ground state. Therefore, CH41) cannot induce an image dipole in the metal surface, which excludes this potential pathway for vibrational energy transfer for ν1. A CH41) reactivity observed to be up to 10-fold higher (27) than for CH43) (2) is consistent with this model. However, there is currently no direct experimental evidence that e-h pair excitation occurs on the subpicosecond time scale of the reactive collisions of vibrationally excited CH4 with a metal surface. Theoretical modeling beyond the BO approximation, including the participation of electronically nonadiabatic channels (28), should be pursued to shed light on the important question of whether e-h pair excitation plays a significant role in CH4 chemisorption.

Irrespective of the mechanism, observation that the methane reactivity depends on the initial vibrational alignment, prepared far from the surface, implies the absence of significant steering effects under our experimental conditions. In other words, anisotropic molecule/surface interactions are unable to orient or steer the incident molecule into its lowest-energy reaction path on the subpicosecond time scale of the reactive collision in our experiments.

Furthermore, the observation of an alignment-dependent reactivity of methane on Ni(100) constitutes further evidence for a nonstatistical mechanism of methane chemisorption, beyond the previously reported mode specificity (5) and bond selectivity (6). Our results show that methane’s reaction probability is not simply controlled by the available (vibrational) energy but is state-specific and sensitive to C-H stretch alignment. These observations are incompatible with statistical rate theory, which assumes complete randomization of initial conditions in the collision complex and predicts reaction rates solely on the basis of energetics (7).

Supporting Online Material

www.sciencemag.org/cgi/content/full/329/5991/553/DC1

Materials and Methods

Figs. S1 to S12

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank A. C. Luntz for helpful discussions. Financial support was provided by the Swiss National Science Foundation (grant no. 124666) and the École Polytechnique Fédérale de Lausanne.
View Abstract

Stay Connected to Science

Navigate This Article