First-Principles Nonadiabatic Dynamics of Molecules at Metal Surfaces with Vibrationally Coupled Electron Transfer.

Accurate description of nonadiabatic dynamics of molecules at metal surfaces involving electron transfer has been a long-standing challenge for theory. Here, we tackle this problem by first constructing high-dimensional neural network diabatic potentials including state crossings determined by constrained density functional theory, then applying mixed quantum-classical surface hopping simulations to evolve coupled electron-nuclear motion. Our approach accurately describes the nonadiabatic effects in CO scattering from Au(111) without empirical parameters and yields results agreeing well with experiments under various conditions for this benchmark system. We find that both adiabatic and nonadiabatic energy loss channels have important contributions to the vibrational relaxation of highly vibrationally excited CO(v_{i}=17), whereas relaxation of low vibrationally excited states of CO(v_{i}=2) is weak and dominated by nonadiabatic energy loss. The presented approach paves the way for accurate first-principles simulations of electron transfer mediated nonadiabatic dynamics at metal surfaces.

Hydrogen atom collisions with a semiconductor efficiently promote electrons to the conduction band.

The Born-Oppenheimer approximation is the keystone of modern computational chemistry and there is wide interest in understanding under what conditions it remains valid. Hydrogen atom scattering from insulator, semi-metal and metal surfaces has helped provide such information. The approximation is adequate for insulators and for metals it fails, but not severely. Here we present hydrogen atom scattering from a semiconductor surface: Ge(111)c(2 × 8). Experiments show bimodal energy-loss distributions revealing two channels. Molecular dynamics trajectories within the Born-Oppenheimer approximation reproduce one channel quantitatively. The second channel transfers much more energy and is absent in simulations. It grows with hydrogen atom incidence energy and exhibits an energy-loss onset equal to the Ge surface bandgap. This leads us to conclude that hydrogen atom collisions at the surface of a semiconductor are capable of promoting electrons from the valence to the conduction band with high efficiency. Our current understanding fails to explain these observations.

Electronically Nonadiabatic H Atom Scattering from Low Miller Index Surfaces of Silver.

The reactivity of a surface depends strongly on the surface structure. To study the influence of surface structure on H atom adsorption, we performed inelastic scattering experiments and complementary electronically nonadiabatic molecular dynamics (MD) simulations for H atoms colliding with the three low Miller index surface facets of silver. Experiment reveals very similar energy loss distributions for all three investigated facets. However, for the (100) facet a dependence on the surface orientation is observed that is absent for the other two facets. The nonadiabatic MD simulations manage to describe the experiments well. Despite the observed insignificant influence of the surface geometry on the energy loss distributions, our simulations predict that the capability of the H atoms to penetrate the surface critically depends on the surface structure. The observed crystal orientation dependence of the energy loss distributions in the experiment for Ag(100) cannot be explained with our simulations, and we provide a discussion for a better theoretical description of this system to stimulate future computational investigations.

Adsorption and Absorption Energies of Hydrogen with Palladium.

Thermal recombinative desorption rates of HD on Pd(111) and Pd(332) are reported from transient kinetic experiments performed between 523 and 1023 K. A detailed kinetic model accurately describes the competition between recombination of surface-adsorbed hydrogen and deuterium atoms and their diffusion into the bulk. By fitting the model to observed rates, we derive the dissociative adsorption energies (E 0, ads H2 = 0.98 eV; E 0, ads D2 = 1.00 eV; E 0, ads HD = 0.99 eV) as well as the classical dissociative binding energy ϵads = 1.02 ± 0.03 eV, which provides a benchmark for electronic structure theory. In a similar way, we obtain the classical energy required to move an H or D atom from the surface to the bulk (ϵsb = 0.46 ± 0.01 eV) and the isotope specific energies, E 0, sb H = 0.41 eV and E 0, sb D = 0.43 eV. Detailed insights into the process of transient bulk diffusion are obtained from kinetic Monte Carlo simulations.

H atom scattering from W(110): A benchmark for molecular dynamics with electronic friction.

Molecular dynamics with electronic friction (MDEF) at the level of the local density friction approximation (LDFA) has been applied to describe electronically non-adiabatic energy transfer accompanying H atom collisions with many solid metal surfaces. When implemented with full dimensional potential energy and electron density functions, excellent agreement with experiment is found. Here, we compare the performance of a reduced dimensional MDEF approach involving a simplified description of H atom coupling to phonons to that of full dimensional MDEF calculations known to yield accurate results. Both approaches give remarkably similar results for H atom energy loss distributions with a 300 K W(110) surface. At low surface temperature differences are seen; but, quantities like average energy loss are still accurately reproduced. Both models predict similar conditions under which H atoms that have penetrated into the subsurface regions could be observed in scattering experiments.

Quantum effects in thermal reaction rates at metal surfaces.

There is wide interest in developing accurate theories for predicting rates of chemical reactions that occur at metal surfaces, especially for applications in industrial catalysis. Conventional methods contain many approximations that lack experimental validation. In practice, there are few reactions where sufficiently accurate experimental data exist to even allow meaningful comparisons to theory. Here, we present experimentally derived thermal rate constants for hydrogen atom recombination on platinum single-crystal surfaces, which are accurate enough to test established theoretical approximations. A quantum rate model is also presented, making possible a direct evaluation of the accuracy of commonly used approximations to adsorbate entropy. We find that neglecting the wave nature of adsorbed hydrogen atoms and their electronic spin degeneracy leads to a 10× to 1000× overestimation of the rate constant for temperatures relevant to heterogeneous catalysis. These quantum effects are also found to be important for nanoparticle catalysts.

Effective medium theory for bcc metals: electronically non-adiabatic H atom scattering in full dimensions.

We report a newly derived Effective Medium Theory (EMT) formalism for bcc metals and apply it for the construction of a full-dimensional PES for H atoms interacting with molybdenum (Mo) and tungsten (W). We construct PESs for the (111) and (110) facets of both metals. The EMT-PESs have the advantage that they automatically provide the background electron density on the fly which makes incorporation of ehp excitation within the framework of electronic friction straightforward. Using molecular dynamics with electronic friction (MDEF) with these new PESs, we simulated 2.76 eV H atoms scattering and adsorption. The large energy losses at a surface temperature of 300 K is very similar those seen for H atom scattering from the late fcc metals and is dominated by ehp excitation. We see significant differences in the scattering from different surface facets of the same metal. For the (110) facet, we see strong evidence of sub-surface scattering, which should be observable in experiment and we predict the best conditions for observing this novel type of scattering process. At low temperatures the MD simulations predict that H atom scattering is surface specific due to the reduced influence of the random force.

Random Force in Molecular Dynamics with Electronic Friction.

Originally conceived to describe thermal diffusion, the Langevin equation includes both a frictional drag and a random force, the latter representing thermal fluctuations first seen as Brownian motion. The random force is crucial for the diffusion problem as it explains why friction does not simply bring the system to a standstill. When using the Langevin equation to describe ballistic motion, the importance of the random force is less obvious and it is often omitted, for example, in theoretical treatments of hot ions and atoms interacting with metals. Here, friction results from electronic nonadiabaticity (electronic friction), and the random force arises from thermal electron-hole pairs. We show the consequences of omitting the random force in the dynamics of H-atom scattering from metals. We compare molecular dynamics simulations based on the Langevin equation to experimentally derived energy loss distributions. Despite the fact that the incidence energy is much larger than the thermal energy and the scattering time is only about 25 fs, the energy loss distribution fails to reproduce the experiment if the random force is neglected. Neglecting the random force is an even more severe approximation than freezing the positions of the metal atoms or modelling the lattice vibrations as a generalized Langevin oscillator. This behavior can be understood by considering analytic solutions to the Ornstein-Uhlenbeck process, where a ballistic particle experiencing friction decelerates under the influence of thermal fluctuations.

Adsorbate modification of electronic nonadiabaticity: H atom scattering from p(2 × 2) O on Pt(111).

We report inelastic differential scattering experiments for energetic H and D atoms colliding at a Pt(111) surface with and without adsorbed O atoms. Dramatically, more energy loss is seen for scattering from the Pt(111) surface compared to p(2 × 2) O on Pt(111), indicating that O adsorption reduces the probability of electron-hole pair (EHP) excitation. We produced a new full-dimensional potential energy surface for H interaction with O/Pt that reproduces density functional theory energies accurately. We then attempted to model the EHP excitation in H/D scattering with molecular dynamics simulations employing the electronic density information from the Pt(111) to calculate electronic friction at the level of the local density friction approximation (LDFA). This approach, which assumes that O atoms simply block the Pt atom from the approaching H atom, fails to reproduce experiment due to the fact that the effective collision cross section of the O atom is only 10% of the area of the surface unit cell. An empirical adiabatic sphere model that reduces electronic nonadiabaticity within an O-Pt bonding length scale of 2.8 Å reproduces experiment well, suggesting that the electronic structure changes induced by chemisorption of O atoms nearly remove the H atom's ability to excite EHPs in the Pt. Alternatives to LDFA friction are needed to account for this adsorbate effect.

Multibounce and Subsurface Scattering of H Atoms Colliding with a van der Waals Solid.

We report the results of inelastic differential scattering experiments and full-dimensional molecular dynamics trajectory simulations for 2.76 eV H atoms colliding at a surface of solid xenon. The interaction potential is based on an effective medium theory (EMT) fit to density functional theory (DFT) energies. The translational energy-loss distributions derived from experiment and theory are in excellent agreement. By analyzing trajectories, we find that only a minority of the scattering results from simple single-bounce dynamics. The majority comes from multibounce collisions including subsurface scattering where the H atoms penetrate below the first layer of Xe atoms and subsequently re-emerge to the gas phase. This behavior leads to observable energy-losses as large as 0.5 eV, much larger than a prediction of the binary collision model (0.082 eV), which is often used to estimate the highest possible energy-loss in direct inelastic surface scattering. The sticking probability computed with the EMT-PES (0.15) is dramatically reduced (5 × 10-6) if we employ a full-dimensional potential energy surface (PES) based on Lennard-Jones (LJ) pairwise interactions. Although the LJ-PES accurately describes the interactions near the H-Xe and Xe-Xe energy minima, it drastically overestimates the effective size of the Xe atom seen by the colliding H atom at incidence energies above about 0.1 eV.

Hydrogen collisions with transition metal surfaces: Universal electronically nonadiabatic adsorption.

Inelastic scattering of H and D atoms from the (111) surfaces of six fcc transition metals (Au, Pt, Ag, Pd, Cu, and Ni) was investigated, and in each case, excitation of electron-hole pairs dominates the inelasticity. The results are very similar for all six metals. Differences in the average kinetic energy losses between metals can mainly be attributed to different efficiencies in the coupling to phonons due to the different masses of the metal atoms. The experimental observations can be reproduced by molecular dynamics simulations based on full-dimensional potential energy surfaces and including electronic excitations by using electronic friction in the local density friction approximation. The determining factors for the energy loss are the electron density at the surface, which is similar for all six metals, and the mass ratio between the impinging atoms and the surface atoms. Details of the electronic structure of the metal do not play a significant role. The experimentally validated simulations are used to explore sticking over a wide range of incidence conditions. We find that the sticking probability increases for H and D collisions near normal incidence-consistent with a previously reported penetration-resurfacing mechanism. The sticking probability for H or D on any of these metals may be represented as a simple function of the incidence energy, Ein, metal atom mass, M, and incidence angle, 𝜗in. S=(S0+a⋅Ein+b⋅M)*(1-h(𝜗in-c)(1-cos(𝜗in-c)d⋅h(Ein-e)(Ein-e))), where h is the Heaviside step function and for H, S0 = 1.081, a = -0.125 eV-1, b=-8.40⋅10-4 u-1, c = 28.88°, d = 1.166 eV-1, and e = 0.442 eV; whereas for D, S0 = 1.120, a = -0.124 eV-1, b=-1.20⋅10-3 u-1, c = 28.62°, d = 1.196 eV-1, and e = 0.474 eV.

Infrared detection of Criegee intermediates formed during the ozonolysis of ß-pinene and their reactivity towards sulfur dioxide.

Recently, direct kinetic experiments have shown that the oxidation of sulfur dioxide to sulfur trioxide by reaction with stabilized Criegee intermediates (CIs) is an important source of sulfuric acid in the atmosphere. So far, only small CIs, generated in photolysis experiments, have been directly detected. Herein, it is shown that large, stabilized CIs can be detected in the gas phase by FTIR spectroscopy during the ozonolysis of ß-pinene. Their transient absorption bands between 930 and 830 cm(-1) appear only in the initial phase of the ozonolysis reaction when the scavenging of stabilized CIs by the reaction products is slow. The large CIs react with sulfur dioxide to give sulfur trioxide and nopinone with a yield exceeding 80%. Reactant consumption and product formation in time-resolved ß-pinene ozonolysis experiments in the presence of sulfur dioxide have been kinetically modeled. The results suggest a fast reaction of sulfur dioxide with CIs arising from ß-pinene ozonolysis.