Quantum Monte Carlo Calculations ofH2Dissociation on Si(001)

Abstract

We present quantum Monte Carlo calculations for various reaction pathways of H2 with Si(001), using large model clusters of the surface. We obtain reaction energies and energy barriers noticeably higher than those from approximate exchange-correlation functionals. In improvement over previous studies, our adsorption barriers closely agree with experimental data. For desorption, the calculations give barriers for conventional pathways in excess of the presently accepted experimental value, and pinpoint the role of coverage effects and desorption from steps.The dissociative adsorption of molecular hydrogen on the Si(001) surface has become a paradigm in the study of adsorption systems. Despite its apparent simplicity, more than a decade of extensive experimental and theoretical investigations have not clarified fundamental aspects of the chemical reaction of H 2 with this surface.Many of the experimental observations are hard to reconcile in a unified picture: the sticking probability for dissociative adsorption of H 2 on the clean surface is very small at room temperature suggesting a high adsorption barrier; sticking increases dramatically with higher surface temperatures [1]. On the other hand, the nearly thermally distributed kinetic energy of desorbing molecules has lead researchers to the conclusion that the molecules have transversed almost no adsorption barrier [2]. Microscopically, these observations were originally interpreted in terms of an intra-dimer mechanism, where the hydrogen molecule interacts with one single dimer of the Si(001) surface [3]. However, very recent experiments have pointed to additional mechanisms involving not just a single dimer but nearby dimers [4,5]. The existence of highly reactive pathways was first demonstrated on steps [6] or H-precovered surfaces [7], and evidence that H 2 reacts with two adjacent dimers has also now been given for the clean surface [8]. In Fig. 1, the intra-dimer (H2*) and two inter-dimer pathways at different coverages (H2 and H4) are schematically shown.Theoretically, density-functional theory (DFT) calculations performed on intra-dimer and inter-dimer mechanisms have lead to limited agreement with experiments. While correctly predicting the existence of a barrier-less H4 inter-dimer reaction path at high coverages [9], previous DFT slab calculations yielded an adsorption barrier for the low-coverage H2* and H2 pathways too low to explain the small sticking coefficient observed at low temperatures [7]. Desorption barriers from DFT obtained within the generalized gradient approximation (GGA) were also generally lower than the experimental value (2.5 eV, Ref.[10]). Additional evidence for a possible inadequacy of DFT-GGA to describe this reaction comes from comparison with highly correlated quantum chemistry calculations for small cluster models of the surface: For the intra-dimer pathway, these methods obtain values for the desorption barrier that are at large variance with the DFT-GGA value [11][12][13]. In this Letter, we use quantum Monte Carlo (QMC)...

Theoretical and Experimental Studies of the Dechlorination Mechanism of Carbon Tetrachloride on a Vivianite Ferrous Phosphate Surface

“…Recently, large reaction mechanistic studies have been undertaken using computational calculations regarding sorption, 18 reductive dechlorination, 49 and oxidative decomposition, 50 among others. 15,51,52 We believe that the systematic investigation of the reaction of dechlorination taking place on a vivianite surface presented in this study may provide fundamental aid for the development of an adaptable application for other halogenated species and minerals. We believe that modeling systems and reaction mechanisms on the surfaces of structures serves not only as a supplement for experimental data but also can provide basic knowledge and optimal results through simple control of the studied environments, leading to promising alternative approaches.…”

Schottky defect formation energy in MgO calculated by diffusion Monte Carlo

“…[3][4][5][6][7] It is usually competitive in accuracy with high-level quantum chemistry methods, but it has the advantage of being practicable for large systems containing hundreds of atoms. Recently, efforts have been made to apply QMC to oxide systems, including transition metal oxides, 8,9 for which DFT gives poor predictions for magnetic properties, phonon frequencies, and other properties.…”

Quantum Monte Carlo calculations of the structural properties and the B1-B2 phase transition of MgO

“…Competitive in accuracy with high-level quantum chemistry methods, it has the enormous advantage of being practicable for large systems containing hundreds of atoms. The power of QMC in overcoming the deficiencies of density functional theory (DFT) has been amply demonstrated by recent applications, including the energetics of point defects in silicon [2] and carbon [3], the reconstruction of the Si (001) surface [4] and its interaction with H 2 [5], and the calculation of optical excitation energies [6]. Nevertheless, the classes of materials to which QMC has been applied have so far been rather limited.…”

“…The calculations are performed on periodically repeated cells, and system size errors need to be carefully treated. However, in many cases, the overall errors within QMC can be made much smaller than those within DFT, and QMC has already been important in revealing, quantifying and overcoming DFT errors in such quantities as defect formation energies, surface reaction energies and energy barriers [2,3,5].…”