Library

Notes: We present a velocity reset procedure for the approximate description of the molecular dynamics of a tractable subset of the atoms composing a macroscopic solid which is subjected to collisions. The coupling of the subset to the remainder (the reservoir) is taken into account in a stochastic manner by periodically resetting the velocities of subset particles which interact with the reservoir. The Cartesian velocity components are reset to vnew =(1−θ)1/2vold +θ1/2vT, where vold is the previous velocity, vT is a random velocity chosen from a Maxwellian distribution at temperature T, and θ is a parameter which controls the strength of the reset. In the limit θ=1 and all subset particles are reset, the method is similar to Andersen's thermostat procedure [J. Chem. Phys. 72, 2384 (1980)]. In the double limit that θ→0 and the interval between resets Δtrs →0 such that β=θ/2Δtrs is fixed, the equations of motion for the subset reduce to Langevin form, where β is the frictional damping rate. This partial velocity reset method is a computational procedure allowing for (1) relaxation dynamics which are equivalent to the frictional damping theories, (2) inclusion of nonzero temperature effects on damping, (3) rapid generation of initial states selected from a canonical ensemble in preparation for individual transient scattering events, and, (4) simulations akin to molecular dynamics. We show that the velocity reset method reproduces previous calculations of the energy accommodation for the collision of an atom with a simple cubic lattice. Two new simulations of the Ag fcc 111 crystal face are done using a pairwise Lennard-Jones interaction. These involve thermostating to a fixed temperature and computation of spectral densities and autocorrelations.

Notes: The reduced-equations-of-motion (REOM) model for gas–solid energy transfer in molecule–surface collisions [D. J. Diestler and M. E. Riley, J. Chem. Phys. 89, 4137 (1989)] is extended to include effects of nonzero surface temperature by combining the REOM damping theory with a stochastic partial velocity reset (PVR) algorithm [M. E. Riley, M. E. Coltrin, and D. J. Diestler, J. Chem. Phys. 88, 5934 (1988)]. The REOM/PVR procedure, which involves integration of the gas–molecule EOM above the frozen solid with stochastic resetting of the molecule's velocities, is tested by comparing results for NO+LiF(001) scattering with those of a previous stochastic-trajectory study of this system [R. R. Lucchese and J. C. Tully, J. Chem. Phys. 80, 3451 (1984)]. The REOM/PVR results reproduce the trends in the stochastic-trajectory results very well for translational and rotational energy transfer as a function of the various system parameters. However, it is found that the coupling of the vibrational mode, whose frequency is greater than that of the Debye frequency of the solid, is not treated accurately by the REOM theory, which is based on adiabatic approximation to the solid's response function.

Notes: We report rigid-rotator close coupling calculations and quasiclassical trajectory calculations for HF–HF collisions with total angular momentum zero. The results are compared to test the trajectory method.

Notes: Quasiclassical trajectories are used to study rotational energy transfer in the collision of a rigid-rotor NH3 molecule with a flat, rigid gold surface. The anisotropic term in the long-range attractive potential causes the NH3 to preferentially reorient with its dipole moment normal to the surface plane as it approaches the surface. This reorientation decreases the rotational energy transfer and gives rise to a sharp rotational rainbow at zero rotational energy. Trajectory results predict that the molecule preferentially scatters into low K states (tumbling) rather than the J=K states (spinning). This prediction is in qualitative agreement with recent molecular beam/surface scattering experiments.

Notes: The perturbation−trajectory method (PTM) of Jezercak, Agrawal, Thompso and Raff for application to gas−surface collision processes is commented upon. The authors contend that the PTM does not function as a true thermal reservoir or heat bath. (AIP)

Notes: Molecular-beam mass spectroscopy was used to measure the gas composition near a growing diamond surface in a hot-filament-assisted chemical-vapor-deposition reactor. The dependencies of the gas composition on changes in (1) the carbon mole fraction in the reactor feed XC, (2) the identity of the inlet carbon source (CH4 versus C2H2), and (3) the surface temperature TS, were studied. For XC≤0.02, the gas composition appeared to be nearly independent of the identity of the inlet hydrocarbon source and depended only on the C/H ratio in the feed gas. At higher values of XC, catalytic poisoning of the hot filament resulted in different product distributions in these two systems. Increasing the surface temperature affected changes in the hydrocarbon composition; the dependencies of the CH3 and C2H2 mole fractions on TS can each be characterized as having an activation energy of 3±1 kcal/mol. Surprisingly, the H-atom mole fraction was independent of TS. These results suggest that reported temperature sensitivities of film growth properties are primarily due to changes in the kinetics of surface processes rather than changes in the gas composition near the surface. A numerical model of the process is presented. In the study of the compositional change as a function of XC, the code gives good prediction for the methane case but grossly underestimates the methane and methyl concentrations for the acetylene case. The H-atom mole fraction is predicted to increase by ×7 if the H destruction probability on the diamond surface is expected to have an activation energy of 7.3 kcal/mol. Good agreement with experimental data can be obtained, however, if H loss by lateral transport to the walls is taken into account. © 1994 American Institute of Physics.

Notes: Modulated-beam mass spectrometry and x-ray photoelectron spectroscopy (XPS) have been used to investigate the interaction of CH4, C2H4, C5H10, and H2 with carburized and uncarburized tungsten. It is shown that significant evaporation of C1, C2, and C3 occurs for carburized tungsten at temperatures above 1900 °C. The temperature dependence of the carbon evaporation rate was found to be similar to the temperature dependence of the diamond film deposition rate observed in chemical vapor deposition (CVD) reactors, similar to the temperature dependence for the carbon deposition rate observed in the present experiments, and similar to the expected evaporation rate of carbon from graphite and tungsten carbide. The desorption of hydrocarbon species (other than the incident gas) was not clearly observed under any conditions for methane or ethylene. In contrast, it is quite likely that cyclopentane decomposes at the surface to produce new species which are subsequently desorbed into the gas phase. The reaction of ethylene with tungsten is believed to result in complete decomposition with the hydrogen being desorbed as atoms or molecules while the carbon remains on the surface where there is competition between carburization and evaporation. The reaction probability of ethylene with tungsten was found to be close to unity while the reaction probability of methane was small. The removal of carbon from carburized tungsten via an etching reaction involving hydrogen was not observed.The production of hydrogen atoms from H2 was found to be largest on clean tungsten, less on carburized tungsten, and not observable on graphite. Evaporation of tungsten from carburized tungsten was seen at temperatures below 2500 °C but not below 2200 °C. XPS measurements indicated that slightly carburized tungsten contained some graphite in the surface region while heavily carburized tungsten contained much more graphite. The surface concentration of carbon was found to depend in a complicated manner on the balance between carbide and graphite growth and carbon evaporation. The reaction probability of the incident gas is also a determining factor. In addition, computer simulations were used to calculate the concentrations of various species in the gas phase under conditions which are typical of those used in diamond hot-filament CVD reactors. Calculated gas-phase species distributions near the substrate for carbon-atom/H2 mixtures are found to be similar for most species to those calculated for CH4/H2 mixtures. It appears that the fast H2 and H chemistry determines the equilibrium mixture and that it is nearly independent of the type of carbon containing species introduced near the filament. Literature results obtained in typical diamond hot-filament CVD reactors are compared and interpreted on the basis of the present data.

Notes: The growth of diamond in a hot-filament reactor has been modeled, and compared with existing experimental data. Studies have been carried out on non-growth systems containing only hydrogen, as well as on systems where the methane concentration at the inlet was varied between 0.4% and 7.2%. The one-dimensional stagnation flow model used here includes detailed gas-phase and surface kinetics. A simple model of filament poisoning has been implemented. The effect of the gas/filament temperature discontinuity on species distributions has also been examined. Gross errors between theory and experiment are obtained when filament poisoning is neglected, but good agreement is found using a simple linear poisoning model. A nonzero temperature discontinuity at the filament produces good overall agreement with experiment.

Notes: The growth of diamond in a subatmospheric dc-arc plasma-jet reactor has been studied theoretically. Full transport equations for this geometry, including gas-phase and surface chemistry, have been solved numerically. The surface-reaction mechanism includes pathways for the incorporation of CH3, C2H2, and C from the gas phase, as well as growth of graphite. The surface mechanism includes full reversibility for all reactions, based on estimates of the thermochemistry. Results are presented for degrees of dissociation of H2 in the plasma gun ranging from 2.6% to 90%, and inlet levels of CH4 spanning 0.1–5.0 mol %. It is seen that CH3 is the predominant growth species when there is little H2 dissociation within the plasma gun, but C becomes the dominant species at higher dissociation levels. The third growth species, C2H2, does not play a role in diamond growth under these conditions when there is less than 1% CH4 in the feed; but, at higher CH4 levels both C and CH3 addition rates drop to 50 times greater than C2H2.