Theoretical studies of surface reactions on metals and electronic materials. Final report, December 1, 1993--May 15, 1996 [report]

J.L. Whitten
1996 unpublished
DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government Neither the United States Government nor any agency thereof, nor any of their employees, makcs any warranty, u t p m or implied. or assumcs any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or proccy disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any spcific
more » ... rein to any spcific commercial product, process, or Knrjcc by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement. recornmendrtion, or favoring by tbe United States Government or any agency thereof. The views and opinions of authors e x p d hemn do not n d l y state or reflect thosc of the United States Government or any agency thereof. DISCLAIMER Portions of this document may be illegible in electronic image products. Images are produced from the best available original document. J I. Research Scope The objective of this research is the development and application of theoretical techniques that will provide a molecular level understanding of surface processes. These studies of surface phenomena involving adsorbate structure, energetics and reaction mechanisms relate to two different areas: electronic materials and transition metal catalysis. In order to achieve accurate energetics for surface reactions, fKst-principles calculations are performed using a cluster embedding theory. This permits an accurate many-electron treatment of the adsorbate/surface portion of the system while coupling this region with the bulk. Calculations are carried out for the full electrostatic Hamiltonian (except for core electron pseudopotentials), with wavefunctions constructed by self-consistent-field (SCF) and configuration interaction (CI) expansions. The projects dealing with hydrocyarbon reactions and CO adsorption and dissociation on nickel and iron address fundamental questions related to surface reactivity. The study of the dissociation of methane on a nickel surface containing an iron atom is relevant to the use of methane for synthesis of other hydrocarbons and to the general topic of alkane reactivity. Studies of oxygen and sulfur containing compounds are designed to aid in the interpretation of structural and spectroscopic experiments. Silicon surface studies focus on several questions related to chemical vapor deposition reactions and growth of electronic materials. In collaborative work with experimental groups at N.C. State developing diagnostic techniques for monitoring film growth, influences on second harmonic generation arising from chemical modifications of interfaces are explored. The project involving diamond growth, carried out in collaboration with an experimental group at the Research Triangle Institute, explores factors that influence the bonding of carbon to nickel surfaces and surface reactions on carbon. Studies of defects produced by H atom migration in amorphous siilicon are related to the degradation of the photovoltaic properties of this material on continued exposure to light. 2 Progress During Grant Period In order to obtain accurate energetics of surface reactions, we have developed an ab initio formulation of the cluster embedding problem that uses localized orbitals to define an electronic subspace encompassing the adsorbate and neighboring surface atoms. In this way it is possible to treat the adsorbate/surface portion of a system accurately while coupling this region to the bulk lattice. Calculations are carried out for the full electrostatic Hamiltonian of the system (except for core electron pseudopotentials), and wavefunctions are constructed by self-consistent-field (SCF) and multi-reference configuration interaction (CI) expansions. Details of the procedure have been described in previous reports. Numerous applications to chemisorption and reactions on transition metal surfaces have been reported and those completed during the present grant period are summarized in a subsequent section. Calculations on adsorbate/metal systems typically provide the following information as a function of surface site and adsorbate geometry: total energy (E) adsorption energy (AE relative to infinite separation from the surface) total wavefunction, electron densities ionization potentials and work function inner shell or core electron excitation energies vibrational frequencies dipole moments, dynamic dipoles Knowledge of the energy as a function of the coordinates of the species undergoing reaction allows calculation of the equilibrium geometry, minimum energy reaction pathway and transition state energy. One of the first objectives in our work was to treat the adsorption of simple adsorbates on metal surfaces in order to understand experiments in which adsorption sites were well characterized and to uncover the principal interactions that govern bonding between surfaces and adsorbed species. Several of the studies fall into this category. Although the electrons in the metal are delocalized, one of the early observations was that 3 I the strong interactions, typically on the order of 30 -100 kcal/mol, between nonmetal adsorbates and the surface split out a set of orbitals that could be regarded as localized in the vicinity of the adsorption site. The interaction between the metal s, p electrons and nonmetal adsorbates are found to be the principal contributor to the energetics of metaladsorbate bonding. Metal d orbitaladsorbate orbital interactions are also important in transition metal systems, but even for these systems most of the adsorption energy comes . from the s, p interactions. Studies of a variety of adsorbates on nickel have now been completed leading to a better understanding of surface sites at which reactions occur. Work was completed on the reaction of methane with a Ni( 11 1) surface containing a substitutional iron atom. The presence of the surface Fe atom substantially reduces the activation barrier for CH4 dissociation to -5 kcal/mol. Studies of the chemisorption H,C-0-and H3C-Son Ni ( 1 11) and Ni ( 100) and the interaction of methoxy with coadsorbed CO have been completed along with studies of the behavior of dynamic dipoles and adsorbate vibrations as a function of surface site. In order to illustrate the capabilities of the theoretical method, we shall report in this section results obtained for the methoxy / nickel system. References to studies of other adsorbatemetal systems are found at the conclusion of the section. For the adsorption of methoxy on Ni( 11 l), the embedded cluster calculations give adsorption energies of 90 kcaVmol (three-fold site), 87 kcal/mol (bridge site) and 61 kcdmol (atop site). At the three-fold site CH30 is nearly perpendicular to the surface, consistent with the H E E L S , UPS and FTIR experiments. The calculated energy increase of only 1.7 kcal/mol for a 30" tilt suggests a floppy CH3 motion that could easily be influenced by adsorbate-adsorbate interactions. Adsorption at the bridge site is only 3 kcdmol higher in energy and, in this case, the methoxy C-0 axis is tilted away from the surface normal by 20". The small energy difference means that both sites could be populated. For CH30 adsorbed at a three-fold site, the calculated C-0 stretching frequency is 1025 crn-l. This value is in good agreement with a value of 1027 cm-' recently obtained using IR and the value of 1040 cm-* observed using EELS. The calculated CH30-surface 4 J perpendicular stretching frequency of 495 cm-' is also consistent with a value of 500 cm-* from EELS measurements. ( 1 11) is similar to hydroxyl adsorption. For OH adsorbed at a three-fold site, the calculated adsorption energy is 87 kcal/mol with the 0-H axis inclined 10" away from the surface normal. The 10" tilt stabilizes the OH by only 2 kcal/mol. The calculated Ni-0 equilibrium distance is 1.51 A and the HO-surface stretching frequency is 476 cm-*. Methoxy adsorption on Ni Recent LEED studies of methoxy on Ni( 11 1) by Bradshaw et al. [Surf. Sci. 304 (1994) 74; 33 1/333 (1995) 2011 provide an opportunity to assess the accuracy of the quantum mechanical prediction of the geometry. Both theory and experiment agree on the preferred adsorption site and a geometry in which the C-0 axis is essentially normal to the surface. The calculated 0 -Ni distance between nuclei, 2.07 A, is longer than the 1.93 A from LEED, and this is likely due to a local reconstruction of the surface in which the LEED study shows the three nickel atoms bonded to the adsorbate are slightly lifted out of the surface by 0.1 A. The LEED results also show a relaxation horizontally away from the oxygen by 0.1 1 A. In the theoretical work, the Ni atoms are taken at their ideal positions corresponding to the bulk lattice. Methoxy on nickel has large ionic character, but, as with similar systems in which oxygen is directly bonded to the surface, the three-fold and bridge adsorption sites are preferred. At the bridge site, the C-0 axis tilts by 20" and this makes the site competitive, but still higher by 3 kcallmol compared to adsorption at the three-fold h e . The vibrational frequency associated with the methoxy-surface stretch varies. from 380 cm-' at the bridge adsorption site (tilted methoxy) to 495 crn-' at the three-fold adsorption site (untilted); the latter value is in agreement with EELS measurements. When CH,O bonds to nickel, charge is transferred from the metal to CH30; very small clusters cannot support the charge transfer and, consequently, the calculated CH30 adsorption energy is less for such models. Similar calculations for CH3S adsorbed on Ni( 1 11) show that both perpendicular and tilted geometries are possible. At the three-fold site, the C-S axis of CH3S is nearly perpendicular and the adsorption energy is 62 kcal /mol; at the bridge site, the C-S axis is 5 tilted 45" and the adsorption energy is unchanged, 62 kcal/mol. Thus, methane thiolate exhibits a dramatic two-geometry energy degeneracy. For the atop Ni site, the adsorption energy is higher and the C-S axis is tilted 55 O . Embedded cluster calculations on two other adsorbate-nickel systems, HCO on Ni( 100) and CN on Ni ( 1 1 l) , show a distinctly different behavior of these species, compared to others with unsaturated valence, in that the atop atom adsorption sites are comparable in stability to the higher symmetry sites. Studies of the charge distribution by examining the variation of the dipole moment of the system with changes in the carbonmetal surface distance show a linear variation in dipole moment with a slope approximating unity (in a.u.). This is indicative of adsorption with strong ionic contributions. In these molecules, the electron transfer to carbon from the surface offsets some of the directionality of the bonding. For formyl, the adsorption energies are essentially the same at atop, bridge and four-fold sites, namely 63.5 f 0.1 kcal/mol. For CN, the adsorption energies on Ni( 1 11) are 1 13, 114 and 115 kcdmol for atop, bridge and three-fold sites, respectively. Thus, these species follow the adsorption rules for lone pair adsorbates on nickel, and the flat interaction potential implies much greater surface mobility. One very interesting feature of the formyl system is the geometry change that takes place at different sites: for the four-fold and bridge sites, the C-0 bond is parallel to the surface, while for adsorption at the atop Ni site the C-0 axis tilts to give an angle of 110" with the surface normal. A large decrease in dipole moment accompanies the transition to the atop site with the C-0 axis tilt. Calculations on the methoxyhickel and methanethiolate/nickel systems, particularly dynamic dipole and core ionization calculations, were found to be of value in interpreting current experimental work. Some of these investigations were carried out in collaboration with experimentalists at other institutions. Coadsorption studies of CO and H3CO on Ni( 11 1) revealed large effects of CO on the methoxy geometry and adsorption energy. The results of these and other studies have allowed semiquantitative rules to be formulated that describe the behavior of a broad class of adsorbates on metals and the effect of coadsorbed species. 6 In mechanistic studies of silane decomposition on the Si(100)-2xl surface, a low energy pathway is found for the reaction SiH, = SiH3 + H. A multi-step mechanism for the desorption of H, from Si( 100)2xlH which involves a defect mechanism for forming a dihydride, SiH?, '-is proposed to explain experimental observations. Activation energy barriers were determined for the reaction of CH, and CH3F with Si( 1 1 1) and a new interpretation of supersonic beam experimental data is proposed based on the kinetic energy of two-atom components of the molecules colliding with the surface. Work on silicon systems also included studies of impurities and dopants at the Si/ SiO, interface, metastable defects, factors that affect optical second harmonic generation and H bonding linkages in amorphous silicon. In studies of the degradation of the photovoltaic properties of amorphous silicon alloys, an explanation of the Staebler-Wronski effect is proposed. The model involves a H-exchange reaction in which trapping of photo-generated holes promotes a transfer of the H atom from a -SiH group to a nearestneighbor -Si-NH-Si-creating a S: -dangling bond and a metastable (-Si-NH,-Si-) group. Calculations indicate that neutral (-Si-NH,-Si-) is unstable, so that relaxation of (-SiNH,-Si-)+ groups can occur by trapping of a thermally released (trapped) electron during a thermal anneal. + In studies of the possibility of nucleating diamond on nickel, subsurface Na is discovered to stabilize tetrahedrally bonded carbon whereas subsurface C facilitates the formation of planar configurations of CH3 and CgH6 on Ni( 11 1). Mechanistic studies of CVD reactions of CH, with C( 100) surfaces were carried out as part of an investigation of the potential use of carbon rich species in diamond growth. Factors that influence surface reconstruction and barriers to adsorption were examined. This is a field in which many semiquantitative calculations of reaction energetics have been reported in an attempt to elucidate factors that affect the quality of diamond films. The present ab initio configuration interaction calculations provide useful benchmarks for calibrating semiempirical methods as well as providing accurate results for elementary reactions on cluster models of the carbon surfaces. Results on current projects and abstracts of papers published or submitted during 7 the grant period are included in Section III. Abstract: Ab initio configuration interaction calculations are performed to study the dissociative adsorption of H, on a Ni( 11 1) surface. The lattice is modeled as an embedded three-layer, 41-atom cluster. Ni 3d orbitals are explicitly included on seven Ni atoms on the surface. H is preferentially chemisorbed at a 3-fold site on Ni( 1 1 1) and the calculated binding energy of 62 kcaUmo1, H-Ni distance of 1.86 A, and H vibrational frequency of
doi:10.2172/296693 fatcat:w42foxvls5da7evdrlmvz66ycm