Incident Angle Dependence of CHD3 Dissociation on the Stepped Pt(211) Surface.

The dissociation of methane on transition metal surfaces is not only of fundamental interest but also of industrial importance as it represents a rate-controlling step in the steam-reforming reaction used commercially to produce hydrogen. Recently, a specific reaction parameter functional (SRP32-vdW) has been developed, which describes the dissociative chemisorption of CHD3 at normal incidence on Ni(111), Pt(111), and Pt(211) within chemical accuracy (4.2 kJ/mol). Here, we further test the validity of this functional by comparing the initial sticking coefficients (S0), obtained from ab-initio molecular dynamics calculations run using this functional, with those measured with the King and Wells method at different angles of incidence for CHD3 dissociation on Pt(211). The two sets of data are in good agreement, demonstrating that the SRP32-vdW functional also accurately describes CHD3 dissociation at off-normal angles of incidence. When the direction of incidence is perpendicular to the step edges, an asymmetry is seen in the reactivity with respect to the surface normal, with S0 being higher when the molecule is directed toward the (100) step rather than the (111) terrace. Although there is a small shadowing effect, the trends in S0 can be attributed to different activation barriers for different surface sites, which in turn is related to the generalized co-ordination numbers of the surface atom to which the dissociating molecule is adsorbed in the transition state. Consequently, most reactivity is seen on the least co-ordinated step atoms at all angles of incidence.

Introduction

The dissociation of molecules on stepped and corrugated transition metal surfaces can be considered to model the reaction at defect sites on transition metal catalysts.[1,2] Calculations have shown that for the dissociation of methane, less co-ordinated surface atoms typically have lower activation barriers,[3−8] which can influence the dynamics of the collision of the molecule with the surface. For the dissociation of methane on “flat” low-index transition metal surfaces, the sticking coefficient, S0, typically increases with increasing incident energy because of the significant activation barrier,[9−14] although Utz et al. reported an increase in S0 with decreasing incident energy on Ir(111), which they attributed to precursor-mediated dissociation.[15] Normal energy scaling, where the reactivity is proportional to the incident translational energy directed normal to the surface, has been seen for methane dissociation on Ni(111),[16] Pt(111),[17] and Pd(111).[18] For the corrugated Pt(110)-(1 × 2) surface, deviations from normal energy scaling were observed,[19,20] both for molecules without vibrational energy and for molecules prepared in the antisymmetric stretch overtone state.[20] When the methane velocity was directed parallel to the rows in the Pt(110)-(1 × 2) surface, normal energy scaling was seen,[19,20] but when the incident energy was perpendicular to the rows, Madix et al.[19] reported a scaling of Eicos0.5θi, where Ei is the incidence energy and θi is the polar angle of incidence. Bisson et al.[20] attributed this slower decrease in S0 to a shadowing effect, as at larger values of θi, methane preferentially collided with a ridge atom where the activation barrier to the reaction is the lowest.

Previous experimental work by Gee et al. also reported that S0 for methane dissociation on stepped Pt(533) does not follow normal energy scaling.[21] They found that the sticking coefficients fell more slowly as the incident methane was directed toward the (100) step than the (111) terrace. By assuming that the reactivity on the (111) terrace is the same as on an extended Pt(111) surface, they could separate the total sticking coefficient into a contribution from the (100) step and the (111) terrace. The authors found that the reactivity on each facet of the surface fell faster than would be predicted by normal energy scaling.

In the current work, we present a combined experimental–theoretical study of the dependence of S0 on the angle of incidence of CHD3 with respect to a Pt(211) surface. The Pt(211) surface is stepped, consisting of three-atom wide (111) terraces separated by one-atom high (100) steps, as shown schematically in Figure A. There are three different types of atoms on this surface, which we refer to as step (red), terrace (blue), and corner (green) to be consistent with the notation used in previous studies.[8,22,23] The (111) terrace consists of red, blue, and green atoms, and the (100) step consists of adjacent red and green atoms, as shown by the shaded area in Figure B. The direction of incidence of the molecule is defined by a polar angle, θi, and an azimuthal angle, ϕi, as also shown in Figure B. For ϕi = 0°, changing θi corresponds to the molecule being directed toward the (111) terrace (θi < 0°) or the (100) step (θi > 0°). At θi ≈ −20°, the molecule’s velocity is perpendicular to the (111) terrace, and at θi ≈ 40°, it is perpendicular to the (100) step. The angle θi = 0° corresponds to incidence along the macroscopic surface normal. When ϕi = 90°, changing θi changes the component of the velocity parallel to the step edge. Due to the symmetry of the surface, if ϕi = 90°, |θi| and −|θi| correspond to the same incidence condition.

Figure 1

Panel A. Schematic depiction of the Pt(211) surface showing the step (red), terrace (blue), and corner (green) atoms. The three-atom wide (111) terraces consists of green, blue, and red atoms, and the one-atom high (100) step consists of the adjacent red and green atoms (the shaded area in panel B). The y-axis is parallel to the step edges, the x-axis along the direction of the corrugation, and the z-axis perpendicular to the macroscopic (211) plane. Panel B. Depiction of the polar angle θi and the azimuthal angle ϕi, which define the direction of the incoming CHD3.

In the calculations, we will make use of the SRP32-vdW specific reaction parameter exchange correlation functional, which was developed to give a chemically accurate (within 4.2 kJ/mol) description of CHD3 dissociation on Ni(111).[24] The same functional has been shown to also reproduce S0 for CHD3 dissociation on Pt(111) and Pt(211)[4] within chemical accuracy as well as to develop a 15-dimensional neural network potential energy surface for methane dissociation on Ni(111).[25] All this previous work with the SRP32-vdW functional has been done with the methane approaching the transition metal surfaces at normal incidence; this presents the first study where the angle of incidence is changed. The results from the ab-initio molecular dynamics (AIMD) calculations run with the SRP32-vdW functional will be used to explain the trends in the experimentally determined sticking coefficients.

The rest of the paper is organized as follows. In Sections and 3, the experimental and theoretical methods employed in the current study will be briefly described. Section presents the results and discussion, and Section summarizes our conclusions.

Experimental Methods

In the current work, we present different sets of experimental data referred to as 2016 and 2018 (A and B). These were done using the same apparatus and methods, but there are differences between the two sets of measurements that will be highlighted below. The experimental apparatus has been described in detail previously,[4,26] and only the most relevant details will be presented here. In brief, the molecular beam/surface science machine consists of a triply differentially pumped molecular beam source attached to an ultrahigh vacuum (UHV) surface science chamber. For the 2016 set of experiments, a 10 mm-diameter Pt(211) single-crystal surface was held in a tantalum support between two tungsten wires, whereas for the 2018 measurements, a 12 mm-diameter single-crystal surface was mounted directly between two tungsten wires. In each case, the surface could be heated resistively to over 1100 K and cooled to less than 100 K through thermal contact with a liquid nitrogen reservoir. The temperature was monitored using a K-type thermocouple that was spot welded to the tantalum mount in the 2016 experiments and directly to the Pt(211) crystal in the 2018 experiments. Although this may introduce a small difference in the surface temperature between the two sets of measurements, the sticking coefficient was found to be the same within error bars at surface temperatures between 500 and 800 K. Any small difference in the surface temperature because of the different positions of the thermocouple will therefore not significantly affect the data presented here.

In both sets of measurements, the surface was mounted on a manipulator that allows the surface to be both translated and rotated, with the step edge direction parallel to the axis of rotation; so, changing θi corresponded to changing the angle of incidence with respect to the (100) steps and the (111) terraces (ϕi = 0°, see Figure ). During the depositions, the surface was held at a temperature of 650 K using a proportional, integral, differential (PID) controller and was cleaned between measurements by Ar+ sputtering and annealing. The surface cleanliness was confirmed by Auger electron spectroscopy.

The molecular beam was formed by expansion of a 1.5% CHD3 seeded in H2 gas mix through a 50 μm-diameter hole in a stainless steel nozzle and a 2 mm-diameter skimmer. The nozzle was resistively heated to 500 K, and the resulting velocity was determined using time of flight methods, described in detail in Section S1 of the Supporting Information. For the 2016 experiments, this gave a normal incident energy of 96.8 kJ/mol, and for the 2018 experiments, it was 98.5 kJ/mol. Different-sized apertures in a chopper wheel (diameter 2, 1, and 0.5 mm) were used to collimate the molecular beam to ensure that all molecules hit the surface for all angles of incidence. (We direct the interested reader to Figure S7 in ref (4) for a schematic of the molecular beam path in the machine).

The initial sticking coefficients were determined using the King and Wells (K&W) method.[4,27,28] An off-axis quadrupole mass spectrometer (QMS) was used to monitor the partial pressure of mass 19 in the UHV chamber, with a typical trace presented in Figure A. Initially, the separation valve between the molecular beam source and the UHV chamber was shut, and there is correspondingly no significant mass 19 QMS signal. The separation valve is then opened at t = −120 s. At this point, the molecular beam is scattered from an inert beam flag, and the QMS signal is a measure of the total number of molecules entering the UHV chamber. The beam flag is opened at t = 0 s, and the drop in the QMS signal corresponds to the number of molecules sticking to the surface. At t = 15 s, the beam flag is shut, and the separation valve is shut at t = 90 s. The time dependence of the sticking coefficient can then be found usingwhere ΔP is the change in partial pressure when the beam flag is open and P is the partial pressure increase when the separation valve is opened. Their values are shown in Figure A. S(t) is fit using a double exponential decay[28] to obtain the initial sticking coefficient S0, as shown in Figure B. The baseline of the K&W trace when the flag is shut (t < 0 s, t > 15 s) is not zero, as the QMS current was seen to increase when the beam flag is opened under conditions where no reactivity was observed. This has been accounted for in the analysis of both sets of experimental data, and the correction gives rise to the apparent nonzero baseline when the beam flag is closed.

Figure 2

Panel A. King and Wells QMS signal for the dissociative chemisorption of CHD3 on Pt(211) at a surface temperature of 650 K and an incident energy of 96.8 kJ/mol for θi = ϕi = 0°. The time axis has been shifted so that t = 0 s corresponds to the time that the King and Wells beam flag was opened. The inset shows a magnification for the 15 s that the flag is open for, and the red labels correspond to the quantities in eq . Panel B. Time dependence of the King and Wells trace (black) and the fit to the data (dashed red) for the data presented in panel A.

We present a comparison of the sticking coefficients measured at normal incidence (θi = 0°, ϕi = 0°) at a surface temperature of 650 K from the 2016 (red), 2018 A (blue), and 2018 B (black) experiments in Figure . The 2018 A angle of incidence data presented in Section were recorded in the same way as the unscaled 2018 A data shown as a blue-filled circle at an incident energy of 98.5 kJ/mol, which is larger than the sticking coefficients obtained from the 2016 experiments (the full unscaled data set is shown in the Supporting Information in Figure S5). After the data were recorded, a systematic error was found in the angular 2018 A data because of an unstable backing pressure in the molecular beam expansion. Once this was rectified, the data point at 98.5 kJ/mol was repeated, and the sticking coefficient that was obtained (black, 2018 B) is in agreement with the original 2016 data set. Rescaling the 2018 A data set so that the 98.5 kJ/mol S0 coincides with the 2018 B data point produced the open symbols in Figure , which are in agreement with the 2016 experiments. We therefore chose to rescale the 2018 A angle of incidence data set to bring it into agreement with the 2016 data set at θi = 0°, ϕi = 0°. This scaling then accounts for the systematic error in the acquisition of the 2018 A data, and the slightly different normal incident energies obtained at the 500 K nozzle temperature used to record the two sets of data. As will be shown in Section , this brings the two sets of experiments into excellent agreement for θi > 0°, ϕi = 0°, which were recorded in both 2016 and 2018.

Figure 3

Sticking coefficients obtained from different K&W measurements at normal incidence (θi = 0°, ϕi = 0°) to the Pt(211) plane at a surface temperature of 650 K. The first experimental data from 2018 (open blue triangles, A) have been scaled to agree with a second data point at 98.5 kJ/mol (black triangle, B), which are compared to the experiments from 2016[4] (red). See Section for more details.

Theoretical Methods

The methods used in the calculations have also been described in detail previously,[4,29] and so only the most relevant details will be presented here. In brief, either 500 or 1000 quasi-classical AIMD trajectories were run for CHD3 colliding with Pt(211) for ν1 = 1 or laser-off conditions, respectively, using the Vienna ab-initio simulation package (VASP) version 5.3.5.[30−33] We call the AIMD trajectories quasi-classical because zero point energy was imparted to the vibrational modes of CHD3. The first Brillouin zone was sampled using a 4 × 4 × 1 Γ-centered grid with a cutoff energy of 350 eV for the plane wave basis set. Projector augmented wave pseudopotentials[34,35] have been used to represent the core electrons. The Pt(211) surface has been modelled using a 4 layer (1 × 3) supercell slab,[4,29] with each slab separated from its first periodic replica by 13 Å of vacuum. To facilitate convergence, a Fermi smearing was used with a broadening parameter of 0.1 eV. Extensive tests of the parameters used in the calculations have been performed, the results of which can be found in the Supporting Information of ref (4).

The specific reaction parameter exchange correlation functional (SRP32-vdW) used in the present work is defined aswhere EXPBE and EXRPBE are, respectively, the PBE[36,37] and RPBE[38] exchange functionals and ECvdW is the van der Waals correlation functional of Dion et al.[39−41] Previous work[4] has shown that this weighted average produces chemically accurate results for the dissociation of CHD3 on Pt(211) under normal incidence conditions.

The initial conditions used for the trajectory calculations were sampled to replicate the molecular beam scattering experiments performed in 2016 with the velocity of the molecules sampled from the experimental time of flight measurements and rotated by ϕi and θi. For the “laser-off” trajectories, the vibrational populations of the molecules were sampled using a Boltzmann distribution at the 500 K nozzle temperature used to create the molecular beam expansion. Although it was not possible to perform state-resolved reactivity experiments, as the difference between the reactivity of the laser excited molecules could not be separated from the reactivity of the molecules without vibrational excitation, state-resolved calculations were performed where all the molecules were prepared with a single quantum of C–H stretch vibration in the J = 2, K = 1, υ1 = 1 rovibrational state.[4] The initial positions and velocities of the surface atoms were randomly sampled from calculations run to equilibrate the slab at a surface temperature of 650 K.

At the start of the trajectory, the CHD3 is positioned 6.5 Å above the surface with x and y chosen to randomly sample all positions on the Pt(211) slab. As in previous work,[4] the kinetic energy of the molecules was increased by 2 kJ/mol to compensate for the potential energy shift due to the unconverged vacuum space. The trajectories were propagated with a time step of 0.4 fs using the velocity-Verlet algorithm until the CHD3 dissociated on the Pt(211) surface, scattered back into the gas phase or was trapped on the Pt(211) surface. The molecule was considered to have reacted if one of the bonds in the molecule was greater than 3 Å, whereas if the center of mass (COM) of the molecule was 6.5 Å away from the surface, with the COM velocity directed away from the surface, it was considered to have been scattered. If neither outcome was reached within the maximum 1 ps timeframe that the trajectory was propagated for, the molecule was considered to be trapped on the surface.

The sticking coefficients were calculated from the AIMD calculations usingwhere Nreact is the number of trajectories that dissociate and Ntot is the total number of trajectories. The statistical error bars were found asand represent 68% confidence limits. The other probabilities and errors presented in Section are calculated with analogous expressions, unless the probability is 0 or 1, in which case the error is calculated as[42]which also represent 68% confidence limits.

Results and Discussion

Figure presents a comparison of S0 measured experimentally for ϕi = 0°, θi ≥ 0° (red circles) and for ϕi = 0° for both positive and negative θi (open blue circles) with those from AIMD calculations (black). The AIMD calculations were done sampling the velocity distribution used in the 2016 experiments (red), and the 2018 A experimental data (blue) have been scaled to agree with this set of data at θi = 0° (see Section ). We note that the agreement between the two sets of experimental data is excellent at θi > 0°, further justifying the scaling of the 2018 A data. There is good agreement between the experimental and calculated sticking coefficients, with the value only being significantly different at θi = −20°, which we attribute to statistics. The good agreement noted is additional proof of the accuracy of the SRP32-vdW density functional for CHD3 dissociation on Pt(211), as it seems to give a correct description of the angular dependence of the sticking coefficient. The dashed line shows S0(0°)cos2θi, the incident angle dependence expected for normal energy scaling on a flat surface,[16−18] which highlights the asymmetry of S0 with θi seen in both the calculated and experimental sticking coefficients. In both cases, S0 is seen to drop more quickly as the angle of incidence is changed from normal incidence to toward the (111) terrace (toward negative θi) compared to normal incidence to toward the (100) step (positive θi), as has been reported previously for methane dissociation on Pt(533).[21] Furthermore, the sticking coefficients are seen to follow normal energy scaling for θi > 0°, but not for θi < 0°.

Figure 4

Comparison of the sticking coefficients from the AIMD calculations (black circles) with those from K&W experiments at an incident energy of 96.8 kJ/mol (red circles) and scaled sticking coefficients from experiments at an incident energy of 98.5 kJ/mol (blue open circles) for ϕi = 0°. The dashed black line shows a cos2 θi distribution, and the arrows denote the angles of incidence perpendicular to the (100) step and the (111) terrace.

While no experimental data are available for ϕi = 90°, AIMD calculations were run with the results (squares) compared with those for ϕi = 0° (circles) in Figure . The sticking coefficients fall as quickly with θi for ϕi = 90° as is seen to occur rotating toward the (111) terrace (θi < 0°) for ϕi = 0° for molecules both under laser-off conditions (blue) and prepared in the υ1 = 1, J = 2, and K = 1 rovibrational state (red). This suggests that for angles where the molecules collide with the (111) terrace, only the polar angle of incidence (θi) appears to be important and not the azimuthal angle (ϕi). The dashed black line in Figure shows S0(0°)cos2θi scaled to the υ1 = 1, θi = 0°, ϕi = 90° sticking coefficient, illustrating that for ϕi = 90°, the reactivity drops more quickly than would be predicted by normal energy scaling. This is in contrast to the Pt(110)-(1 × 2) surface where the reactivity was found to obey normal energy scaling when the molecules were directed parallel to the ridge atoms.[20]

Figure 5

Sticking coefficients obtained from the AIMD calculations for molecules prepared in the υ1 = 1, J = 2, and K = 1 rovibrational state (red) and under laser-off conditions (blue) for ϕi = 0° (circles) and ϕi = 90° (squares). The dashed black line shows a cos2θi distribution for the υ1 = 1, ϕi = 90° data (red squares).

Figure shows the position of the COM of the molecules at the point where the dissociating bond becomes longer than the transition state value for C–H cleavage (red) and C–D cleavage (blue) for a range of θi for ϕi = 0° for the laser-off trajectories (left column) and for υ1 = 1 (right column). The dashed lines in each plot indicate the direction that the CHD3 approaches the surface. As the angle of incidence changes from normal to the (111) terrace (θi ≈ −20°) to normal to the (100) step (θi ≈ 40°), the reaction site shifts from the terrace and step atoms toward the (100) step, reflecting the change in position on the surface where the normal incidence energy is the highest. Most of the reactivity is seen to occur on top of the step atoms, which is the site with the lowest activation barrier for the dissociation of methane on Pt(211).[4,8]

Figure 6

xz plots showing the positions of the COM for C–H dissociation (red) and C–D dissociation (blue) under laser-off conditions (left column) and for the υ1 = 1 trajectories (right column) for the values of θi shown in the top left corner of the plots for ϕi = 0°. The dashed lines show the incident direction of the CHD3.

The fraction of molecules that dissociate on the step (red) and terrace (blue) atoms in the AIMD calculations are presented in Figure for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D) under laser-off conditions (panels A and C) and for molecules prepared with a quantum of C–H stretch vibration (panels B and D). The site of reaction was taken to be the surface atom closest to the COM of the CHD3 when the dissociating bond became larger than the transition state value. It should be noted that no dissociation was seen on the corner atoms at any angle of incidence. Dissociation on the step atoms dominates the reactivity at all angles of incidence under both laser-off conditions and for υ1 = 1, which is consistent with previous work at lower incident energies and a surface temperature of 120 K for CH4 dissociation on Pt(211) at θi = ϕi = 0°.[8] The highest reactivity observed on the terrace atoms is seen for θi < 0° at ϕi = 0°, which could be due to the kinetic energy of the incoming molecule normal to the (111) terrace being higher and the probability of the molecule hitting the terrace atoms being larger.

Figure 7

Fraction of molecules that dissociate on the step atoms (red) and on the terrace atoms (blue) calculated for the laser-off (panels A and C) and υ1 = 1 (panels B and D) trajectories for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D). No dissociation was observed on the corner atoms.

To decide whether the differences in reactivity at different angles of incidence can be attributed to a shadowing effect, we identified the surface atom closest to the site of methane impact for both reactive and nonreactive trajectories. The results are shown in Figure for the step (red), terrace (blue), and corner (green) atoms for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D) for the trajectories run sampling laser-off conditions (panels A and C) and for υ1 = 1 (panels B and D). At nearly all θi, most trajectories collide with the more exposed step atoms, and very few hit the corner atoms where no reactivity is seen (see Figure ). Figure shows that for ϕi = 0°, with increasing θi, the number of terrace impacts changes compared to the step impacts from roughly equal for θi = −50° to less than half for θi = +50°, indicating a shadowing effect for the terrace sites caused by the step atoms with increasing θi (see also Figure ). There is no significant difference in this shadowing effect for molecules that are initially prepared in υ1 = 1 or under laser-off conditions. Note that even for θi = −20° and ϕi = 0° (incidence perpendicular to the terrace), more molecules hit the step atoms than the terrace atoms.

Figure 8

Fraction of all trajectories that impact nearest the step (red), terrace (blue), and corner (green) atoms for the laser-off (panels A and C) and υ1 = 1 (panels B and D) calculations for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D).

To disentangle the reactivity on each site of the surface from the shadowing effect, we calculated site-specific sticking coefficients for each site aswhere Nreact(site) is the number of reactive trajectories for step or terrace atoms and Nnear(site) is the number of trajectories for which that site is the site of impact. S0(site) are presented in Figure for the step (red) and terrace (blue) sites for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D) for laser-off conditions (panels A and C) and for υ1 = 1 (panels B and D). The site-specific sticking coefficients are higher for the step atoms than for the terrace atoms at all incident angles, with reactivity on the terrace atoms being at its highest when the normal incidence energy to the (111) terrace is higher. This is particularly apparent for the υ1 = 1 data for ϕi = 0° in Figure B, where the site-specific reaction probability on the terrace atoms is at its maximum at θi = −20°, which corresponds to the direction of incidence being normal to the (111) terrace. The same is not seen in the laser-off sticking coefficients in Figure A; but as the total reactivity is lower, the statistics are less good in this analysis. An asymmetry is seen in the reactivity around the angle where S0(site) is the largest for both the step and terrace atoms for ϕi = 0°. This asymmetry cannot be due to shadowing and is therefore likely to be due to different activation barriers for the dissociation at different positions on the Pt(211) surface.

Figure 9

Comparison of the site-specific sticking coefficients for each site calculated using eq for dissociation on the step atoms (red) and terrace atoms (blue) for the laser-off (panels A and C) and υ1 = 1 (panels B and D) trajectories for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D).

Activation barriers are typically found to scale linearly with adsorption energies for the dissociation of molecules on transition metal surfaces.[43−45] In turn, Calle-Vallejo et al. have shown that adsorption energies tend to scale linearly with the generalized co-ordination number of the surface atom to which the molecule adsorbs,[46,47] which, unlike co-ordination numbers, also takes into account the co-ordination number of the nearest neighbors of the atom of interest. It follows that activation barriers would be expected to scale linearly with the generalized co-ordination number. On Pt(211), the generalized co-ordination number follows the order, step atoms (5.58)[46] < terrace atoms (7.33) < corner atoms (8.75),[46] with the activation barriers following the same trend.[8] This would predict that most reactivity would occur on the step atoms and least on the corner atoms, as is observed in the AIMD calculations presented here, at all angles of incidence. It is also interesting to note that all atoms in the Pt(211) surface have a different generalized co-ordination number to those on an extended Pt(111) surface (7.50)[46] and Pt(100) surface (6.67)[46] despite the Pt(211) surface consisting of one-atom high (100) steps and three-atom wide (111) terraces. Previous work by Gee et al. has suggested that the dissociation of methane on a Pt(533) surface, which consists of one-atom high (100) steps and four-atom wide (111) terraces, can be accounted for by considering the Pt(533) surface as independent (100) and (111) facets.[21] As detailed in Section S4 of the Supporting Information, we followed their analysis for the laser-off ϕi = 0° AIMD calculations for CHD3 dissociation on Pt(211) and obtained similar qualitative results. However, the description is unlikely to be quantitatively correct, as there are a large number of adjustable parameters in the model, and to obtain good agreement between the model and the AIMD calculations, it is necessary to use unphysical values of the angles of the (100) step and the (111) terrace with respect to the (211) plane. This suggests that the structure of Pt(211) should not be considered as consisting of independent (100) and (111) facets for methane dissociation, as Juurlink et al. have previously shown is the case for the dissociation of H2, O2, and H2O on Pt(211).[48] This is reflected in the differences in the generalized co-ordination numbers of the atoms in the Pt(211), Pt(111), and Pt(100) surfaces.

The fraction of C–H cleavage seen in the AIMD calculations is presented in Figure for dissociation on the step (red) and terrace (blue) atoms for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D) for laser-off conditions (panels A and C) and for υ1 = 1 (panels B and D). For the laser-off trajectories, the fraction of C–H cleavage is found to be 0.25 for both sites within error bars, as would be expected for a statistical 3:1 branching ratio for C–D:C–H cleavage. More C–H bond cleavage is seen for the υ1 = 1 trajectories, with a slightly higher degree of bond selectivity being observed for dissociation on the terrace atoms than on the step atoms at all angles of incidence, although this difference is within the error bars of the calculations for individual incidence conditions. At a surface temperature of 150 K and at lower incident energies, the branching ratio for the dissociation of CHD3, CH2D2, and CH3D on Pt(111) has been shown to be statistical under laser-off conditions,[49] whereas when a quantum of C–H stretch was added to the molecule, only C–H cleavage was observed.[49,50] Increasing the surface temperature (to 650 K) lowers the effective activation barrier to the dissociative chemisorption due to the thermal motion of the atoms in the surface,[14,51,52] which when combined with a higher incident translational energy is likely to make CHD3 dissociation less bond selective for υ1 = 1, as is seen to be the case in the AIMD calculations.

Figure 10

Comparison of the fraction of C–H cleavage calculated for dissociation on the step atoms (red) and terrace atoms (blue) for the laser-off (panels A and C) and υ1 = 1 (panels B and D) trajectories for ϕi = 0° (panels A and B) and ϕi = 90° (panels C and D).

Summary

Sticking coefficients have been measured and calculated for CHD3 dissociation on a Pt(211) surface at a temperature of 650 K for different angles of incidence at a fixed incident energy (≈ 97 kJ/mol). The measured sticking coefficients, obtained by the K&W method, are in good agreement with those from AIMD calculations using the SRP32-vdW functional, further demonstrating the quality of the functional for describing methane dissociation on Pt(211). An asymmetry is seen in the polar incident angle distribution in both the calculated and experimental sticking coefficients, with a more rapid drop in reactivity for incidence toward the (111) terraces compared to toward the (100) steps. At all incident angles, the calculations show that preparing the CHD3 with one quantum of C–H stretch vibration increases the reactivity and favors C–H bond cleavage over C–D bond cleavage. A shadowing effect is seen, which favors impact on the step sites compared to the terrace sites as the polar angle of incidence is increased toward normal incidence to the steps, although this by itself does not account for the difference in reactivity seen at the two sites. The reactivity on the terrace atoms is seen to be the highest at angles of incidence where the energy normal to the (111) terrace is the highest, but reactivity on the step atoms dominates at all angles of incidence where the activation barrier for dissociation is the lowest. The site of dissociation is seen to shift around the step atoms as the angle of incidence is changed, reflecting the change of position where the normal energy is the highest and the difference in activation barrier heights at the different sites of the Pt(211) surface.