A novel approach for modeling of quantum dynamics and spectroscopy in large polyatomic systems is presented. The recently developed classical separable potential (CSP) method propagates a multidimensional separable quantum wavepacket using effective time-dependent single mode potentials guided by classical trajectories. In this way, couplings between individual degrees of freedom are taken into account in a mean-field approximation. Comparison with ``numerically exact'' calculations for small systems and with experiments on large systems show that the CSP scheme is accurate and suitable for the description of short timescale processes in the context of dynamical and spectroscopic applications. Moreover, calculations for systems having up to 10(3) coupled modes on scalar and up to 10(4) degrees of freedom on parallel computers are not computationally extremely demanding. This paper presents the program package QDYN which implements the CSP method. Copyright (C) 1998 Elsevier Science Ltd.

A nonseparable method for time-dependent quantum simulations of large polyatomic systems is presented and applied to the dynamics of the I2Ar17 cluster, following electronic excitation of the iodine molecule. The new method is an extension of the classical separable potential (CSP) approximation, in which the evolution of each mode is governed by a time-dependent mean potential due to the other modes and the total wave packet is a product of single mode wave functions. The computational effectiveness of the CSP approach stems from the use of classical molecular dynamics (MD) trajectories, carried out at the outset of the procedure, for obtaining the effective single-mode potentials. The present method generalizes the CSP scheme by a configuration interaction (CI) treatment, in which the total wave packet is represented as a linear combination of separable terms with coefficients determined from the time-dependent Schrodinger equation. The single mode wave functions for each configuration are propagated along effective potentials that are generated using individual classical trajectories. The classical MD simulation is also used for simplifying the dynamical equations for the CI coefficients. Thus, the selection of correlations that are included quantum mechanically is guided by classical mechanics, which is the basis for the computational efficiency of this approach. The CI wave packet for the I2Ar17 system with 51 vibrational degrees of freedom was propagated for 500 fs following I-2 (B<–X) excitation. About 1500 configurations proved sufficient for convergence of the CI series. The separable approximation to the wave function holds for 60 fs and begins to break down upon the first collision of the iodine atoms with argons. After the second iodine-argon collision this breakdown is almost complete, and at t = 500 fs the CSP term represents less than 5% of the correlated wave packet. Both absorption and resonance Raman spectra are, however, well described by the separable CSP method, since they are determined within the first 60 fs. The CI-CSP method offers very good accuracy due to inclusion of important correlation effects between different modes, while remaining computationally feasible for systems up to 100 degrees of freedom and more. (C) 1997 American Institute of Physics.

A time-dependent self-consistent field approach is used to simulate a He atom colliding with an Ar-13 cluster. Direct energy transfer during the collision, and energy redistribution among the vibrational degrees of freedom of the anharmonic cluster following the collision, are studied. An important advantage of the method used is that quantum state-to-state transition cross sections can be computed for large systems. The following main results are obtained: (1) The process can be interpreted in terms of a direct collision, followed by post-collision energy redistribution in Ar-13, a description that appears only when the cluster vibrations are not described by the eigenstates of this system. A time scale of one picosecond is found for the post-collision intracluster energy distribution. (2) The long-time final state distribution of Ar-13 is less state selective than the distribution immediately after the impact, but it is also not completely statistical. (3) There are state-to-state transitions having cross sections of observable magnitude. (4) The dominant transitions are those involving zero, one, and two `'phonon'' excitations. Some of the `'two phonon'' excitations have cross sections comparable to strong `'single phonon'' transitions. (5) Different types of modes show different propensities for excitations in the collision, in close relation to the geometric character of the modes. The results show that the TDSCF approximation is a powerful tool for treating both direct collision dynamics and collision-induced dynamics in scattering of large anharmonic systems. (C) 1997 American Institute of Physics.

The dynamical properties of a hydrogen atom chemisorbed at a site of trigonal symmetry on a (110) metal surface are studied by classical trajectory calculations. A model potential is used which is highly anharmonic, and which includes many realistic features present in the case of H on W(110). The calculations show that periodic, quasi-periodic and chaotic trajectories all play significant roles in the dynamical properties of the system at the temperatures of interest. The power spectrum is computed, and the resulting frequency structure is in semi-quantitative accord with recent experimental data on the parallel vibrations of H on W(110). Implications of the results as to the properties of the H/metal surface potential and of the related frequency spectrum are discussed. (C) Elsevier Science B.V.

We present detailed methodology and results for the approximate calculation of the anharmonic vibrational wave functions for a given structure of a protein. The ground state and the fundamental vibrationally excited states of hydrated BPTI are computed with an approach that is of very good accuracy for low-lying states. The eigenfunctions are used to predict quantum properties such as vibrational excitation frequencies and IR intensities as well as atomic mean square displacements at T = 0 K. The method treats diagonal anharmonic effects exactly up to fourth order in normal mode coordinates, while using a mean-field approximation, the vibrational self-consistent field (SCF) approximation for the mode-mode couplings. The inclusion of the diagonal effects (exact in fourth order) makes the potentials stiffer than quadratic. When the mode-mode coupling is included as a mean field, the system becomes even stiffer. These effects produce significant modifications of such observables as the vibrational spectrum and Debye-Waller factors. The results presented here should be considered essentially exact except for potentially important approximations: (a) that the potential energy used bears some resemblance to reality, (b) that the system is restricted to a single minimum in the potential energy surface, and (c) that the quartic expansion in normal modes, when truncated to fourth order, provides a good description of the potential. These issues will be addressed in the main text. The results show that the vibrational absorption spectrum is strongly affected by anharmonic and mode-mode coupling effects. Deviations from the corresponding harmonic absorption intensities are very large. Unlike our previous study that found only weak mode-mode coupling effects for the lowest 100 modes, the SCF corrections calculated here for the intermediate frequency modes are seen to be very significant. The results have important implications for the vibrational spectroscopy and other properties of proteins at low temperatures.

Quantum calculations are reported for vibrational states and energy levels of several peptide and peptide-water complexes, to provide insight into the spectroscopic properties of such systems and their dependency on presently available anharmonic force fields. A blocked di-L-serine-H2O complex and trialanine in an antiparallel beta sheet configuration are the main systems treated. The calculations are for variants of the Amber force field and use the vibrational self-consistent field (VSCF) method, which includes effects of anharmonicity as well as interactions between modes. The main findings are as follows: (1) Distinct isomers of a single peptide-H2O complex corresponding to different hydrogen binding sites of the H2O are observed. (2). The H2O induces shifts of up to 50 cm(-1) in the frequencies of fundamental transitions associated with the peptide modes. (3). Some of the `'intermolecular'' modes of the peptide-H2O cluster are of frequencies and geometry that suggest effective peptide-to-water energy transfer. (5). Changing the TIP3 water potential in the Amber force field into the Coker-Watts potential produces significant spectroscopic changes. (5). The peptide-H2O cluster is found to have several tunneling states, which are assigned in detail. At least some of these states should be spectroscopically observable by the magnitude of the splitting. (6) The trialanine calculations are compared with experimental data available for part of the transitions. For many of the `'stiff'' transitions good agreement is found, but the remaining differences suggest that the force field should be revised. It is argued on the basis of these results that high-resolution spectroscopy in jets should be an excellent tool for improving and developing biomolecular force fields.

We propose a novel method to measure the fractal dimension of a submonolayer metal adatom system grown under conditions of limited diffusivity on a surface. The method is based on measuring the specular peak attenuation of He atoms scattered from the surface, as a function of incidence energy. The (Minkowski) fractal dimension thus obtained is that of contours of constant electron density of the adatom system. Simulation results are presented, based on experimental data. A coverage dependent fractal dimension is found from a two-decade wide scaling regime.

Quantum simulations are reported for the dynamics following the photoexcitation Ba(S-1)–>Ba(P-1) in Ba(Ar)(10) and Ba(Ar)(20) clusters. The evolution in time is studied in a framework that treats quantum-mechanically all the coupled degrees of freedom. The focus is on the role of nonadiabatic transitions between the three adiabatic surfaces corresponding to the P states of the Ba atom. The time scales of electronic relaxation and of electronic depolarization (orbital reorientation) are computed, and the competition between adiabatic and nonadiabatic effects is assessed. The calculations are carried out by a new scheme that extends the recent classically based separable potential method. Semiclassical surface-hopping simulations are used to produce effective single-mode potentials on which nuclear `'orbitals'' are then generated. The full wave packet is constructed from the electronic states involved, and from these nuclear wave functions. Among the main results we find that nonadiabatic transitions become appreciable around 1 ps after photoexcitation, and they are stronger in the smaller cluster. Comparing Tully's semiclassical method with the quantum simulations, good qualitative agreement is found. Quantitatively, the semiclassical predictions for the electronic states branching rations deviate from the quantum results roughly by a factor of 2 after 1 ps. In the smaller cluster direct dissociation of the Ba atom dominates over energy redistribution within the cluster, the opposite being true for the large system. This example demonstrates the feasibility of quantum simulations of nonadiabatic processes in large systems with the new method. (C) 1996 American Institute of Physics.

Molecular Dynamics simulations using a surface-hopping method for transitions between different electronic states are employed to study the dynamics following photoexcitation of the Ba(Ar)(125) cluster. The results are used to interpret spectroscopic experiments on large, size-distributed Ba(Ar)(n) clusters. The dynamics of the coupled electronic-nuclear motions in the cluster involves transitions between three potential energy surfaces, corresponding to the nearly-degenerate p-states of the excited Ba atom. Ejection of excited Ba atoms, adsorbed on the surface of the cluster, can take place. The focus in comparing theory and experiment is on the emission spectrum from the excited clusters, on the polarization of this radiation, and on the polarization of light emitted by excited Ba atoms ejected from the cluster. Based on the good agreement found between theory and experiment, a comprehensive picture of the excited state dynamics is given. It is found that upon excitation, energy is rapidly redistributed in the cluster and no direct ejection of Ba occurs. Electronic relaxation to the lowest P-state occurs, and the latter dominates the cluster emission spectrum and polarization. The electronic state relaxation is mostly complete within t less than or similar to 10 ps. Ejection of Ba atoms occurs as a rare and delayed event when a dynamical fluctuation creates a `'hot spot'' at the Ba site, with a non-adiabatic excitation to the highest electronic level. The results show the feasibility of near-quantitative understanding of non-adiabatic processes in large clusters. (C) 1996 American Institute of Physics.

The early quantum dynamics following the B((3) Pi(0u)+) <– X photoexcitation of I-2 in large rare gas clusters is studied and the resonance Raman spectrum of these systems is calculated by a novel time-dependent quantum mechanical simulation approach. The method used is the classically based separable potential (CSP) approximation, in which classical molecular dynamics simulations are used in a first step to determine an effective time-dependent separable potential for each mode, then followed by quantum wavepacket calculations using these potentials. In the simulations for I-2(Ar)(n) and I-2(Xe)(n), with n = 17, 47, all the modes are treated quantum mechanically. The Raman overtone intensities are computed from the multidimensional time-dependent wavepacket for each system, and the results are compared with experimental data on I-2 in Ar matrices and in liquid Xe. The main findings include: (i) Due to wavepacket dephasing effects the Raman spectra are determined well before the iodine atoms hit the rare gas `'wall'' at about 80 fs after photoexcitation. (ii) No recurrencies are found in the correlation functions for I-2(Ar)(n). A very weak recurrence event is found for I-2(Xe)(n). (iii) The simulations for I-2(Ar)(17) (first solvation layer) and for I-2(Ar)(47) (second solvation shell) show differences corresponding to moderate cluster size effects on the Raman spectra. (iv) It is estimated that coupling to the B `' ((1) Pi(1u)) state or to the a (1 g) state have a small effect on the Raman intensities. (v) For I-2(Ar)(47), the results are in very good quantitive agreement with I-2/Ar matrix experiments. The I-2(Xe)(n) results are in qualitative agreement with experiments on I-2 in liquid Xe. The reported calculations represent a first modeling of resonance Raman spectra by quantum dynamical simulations that include all degrees of freedom in large systems, and they demonstrate the power of the CSP method in this respect. (C) 1996 American Institute of Physics.

Quantum and mixed quantum/classical calculations of the photolysis of a HCl adsorbate on a MgO surface are reported. In the quantum calculation of the hydrogen dynamics (with rigid surface and chlorine atoms) a strong oscillatory structure is found in the angular distribution of the photofragmented hydrogen as well as in the absorption spectrum. These resonances are caused by temporary trapping of the hydrogen atom between the chlorine atom and the surface and reflect the initial perpendicular adsorption geometry. Corrugation of the surface potential leads to a significant modification of these interference patterns, which exist even for a flat surface. Within a mixed quantum/classical time-dependent self-consistent field (Q/C TDSCF) propagation the influence of surface degrees of freedom on the interference patterns is investigated. The thermal motion of the surface and inelastic collisions of the hydrogen atom with the surface and the chlorine atom washes out most of the oscillatory structure. In the fully angular and energy resolved spectra nevertheless clearly distinguishable peaks are seen. They can be used in practice to extract information about adsorption geometry and surface potential parameters. (C) 1996 American Institute of physics.

Moller-Plesset perturbation theory is employed to improve the accuracy of,static mean field computations in molecular vibration problems, This method is a simple and efficient way to get nearly exact frequencies for few-mode model potentials. For more realistic potentials representing the dynamics of water and formaldehyde, the Moller-Plesset treatment works equally as well, However, we find in general that MP2 level corrections give very accurate energies and additional corrections by higher level terms in the MP series are not substantial. Moreover, we find that for reference states on high energy manifolds degeneracies can result when higher level terms are included in the series, We discuss several ways to remove these degeneracies. (C) 1996 American Institute of Physics.

The dynamics of Cl(P-2) atoms in a solid Ar matrix is studied, with emphasis on electronic energy relaxation of excited states, and on p-orbital reorientation effects. The method used follows Tully's approach for nonadiabatic molecular dynamics simulations, which treats the electronic degrees of freedom quantum-mechanically, and the atomic motions classically, allowing for `'hopping'' of the atoms between different potential energy surfaces. We introduce an extended version of this method, to handle `'Berry Phase'' effects due to the doubly degenerate Kramers pairs of states present in this system. The role of both electrostatic and of spin-orbit interactions between different electronic states is incorporated in the treatment. The simulations yield a time scale of 13 ps for the energy relaxation of the highest excited electronic state of Cl(P-2). A time scale of similar magnitude is found for the depolarization of this state. However, the time scale for orbital reorientation at thermal conditions is only 0.7 ps. This is attributed to the fact that at thermal conditions, only the two lowest electronic states are populated. The physical mechanisms of these basic radiationless decay processes are discussed on the basis of the simulations. (C) 1996 American Institute of Physics.

Semiclassical molecular dynamics simulations are developed as a tool for studying anharmonic clusters and solids at energies near the zero point. The method employs the time-dependent self-consistent-held approximation, that describes each mode as moving in the mean dynamical field of all other modes. The method further describes each mode by a semiclassical Gaussian wave packet; The scheme is carried out in normal modes. The method is restricted to systems of moderate anharmonicity at low temperatures. It is, however, computationally efficient and practically applicable to large systems. It can be used for the dynamics of nonstationary states as well as for stationary ones. Structural, dynamical and a variety of spectroscopic properties can easily be evaluated. The method is tested for thermal equilibrium states of (Ne)(13), (Ar)(13) against `'numerically exact'' quantum Feynman path integral simulations. Excellent quantitative agreement is found for the atom-atom pair distribution functions. The method is also applied to (H2O)(n) clusters. Good agreement is found with experimentally available fundamental stretch-mode frequencies. (C) 1996 American Institute of Physics.

Diffusion quantum Monte Carlo (DQMC) simulations are used to study the vibrational ground state of the clusters Hg(H-2)(12) and Mg(H-2)(12). In these clusters, the metal atom is located inside, surrounded by a highly delocalized, liquid-like and, for Hg(H-2)(12), also nearly spherical envelope of hydrogens. However, the wavefunctions of the clusters exhibit traces of icosahedral behavior within the delocalized structure, related to the geometry of the corresponding classical systems. A simple model wavefunction is fitted, with excellent agreement to the numerical DQMC data. It provides a convenient interpretation of the vibrational ground state of these clusters.

The vibrational ground state and the fundamental excited states of (Ar)(13) were studied by vibrational self-consistent field (VSCF) calculations. These calculations treat the interaction between different modes through a mean potential approximation, and incorporate anharmonicity in full. The good accuracy of VSCF for such systems was demonstrated by test calculations for (AT)(3) and other clusters. The study of (Ar)(13) focused on the properties of the wave functions and the excitation energies, on the role of the coupling between the modes and on the deviation from the harmonic approximation. It was found that SCF excitation energies for the fundamental transitions differ from the harmonic values by about 25% for the softest modes, and by about 10% for the stiffest modes. Coupling between the modes, treated by SCF; was found to be much more important than the intrinsic anharmonicity of the individual modes. For the ground state, the harmonic wave function compares well with VSCF, but for the fundamental excited states appreciable differences were found. The results for a potential field expanded to fourth-order polynomial in the normal mode displacements are found to be valid, almost indentical with those for a more elaborate sixth-order polynomial expansion. The fundamental excitation frequencies computed using the Aziz-Slaman Ar-Ar pair potential are very similar, with some quantitative deviations, to the values obtained with a Lennard-Jones potential. The differences are larger for certain specific modes, and very small for the others. These calculations demonstrate the computational power of VSCF as a tool for quantum-mechanical calculations for large clusters, at the level of specific wave functions. (C) 1996 American Institute of Physics.

Vibrational energy levels, wave functions, and ir absorption intensities are computed for (H2O)(n) clusters with n = 2, 3, 4, and 5. The calculations were carried out by the vibrational self-consistent field (VSCF) approximation, with corrections for correlations between the modes by perturbation theory. This correlation corrected VSCF (CC-VSCF) is analogous to the familiar Moller-Plesset method in electronic structure theory. Test calculations indicate that this method is of very good accuracy also for very anharmonic systems. While the method is of highest relative accuracy for the stiffest modes, it works very well also for the soft ones. Some of the main results are (1) the frequencies calculated are in good but incomplete agreement with experimental data available for some of the intramolecular mode excitations. The deviations are attributed to the inaccuracy of the coupling between intramolecular and intermolecular modes for the potential function used. (2) Insight is gained into the pattern of blue- or redshifts from the corresponding harmonic excitation energies for the various modes. (3) Anharmonic coupling between the modes dominates in general over the intrinsic anharmonicity of individual modes in determining the spectrum. (4) The anharmonic corrections to the frequencies of some intermolecular modes (shearing, torsional) are extremely large, and exceed 100% or more in many cases. (5) An approximation of quartic potential field in the normal mode displacement is tested for the clusters. It works well for the high and intermediate frequency modes, but is in error for very soft shearing and torsional modes. (6) The relative errors of the VSCF approximation are found to decrease with the cluster size. This is extremely encouraging for calculations of large clusters, since the VSCF level is computationally simple. (C) 1996 American Institute of Physics.

The implementations of classical and quantum molecular dynamics simulations on distributed-memory massively parallel computers are presented. First, we discuss the implementation of large-scale classical molecular dynamics (MD) simulations on SIMD architecture parallel computers, and in particular, on the family of MasPar distributed-memory data parallel computers. We describe methods of mapping the problem onto the Processing Elements (PE's) of the SIMD architecture, and assess the performance of each strategy. The detailed implementations of this data parallel construct are illustrated for two case studies: classical MD simulation of a two-dimensional lattice and the photodissociation mechanisms of a diatomic iodine impurity in a three-dimensional argon lattice. We also present a study of quantum dynamics using the Time-Dependent Self-Consistent Field (TDSCF) method. These calculations demonstrate the potential of using massively parallel computers in MD simulations of thermodynamic properties and chemical reaction dynamics in condensed phases.

The angular intensity distribution of He beams scattered from compact clusters and from diffusion limited aggregates, epitaxially grown on metal surfaces, is investigated theoretically. The purpose is two-fold: to distinguish compact cluster structures from diffusion limited aggregates, and to find observable signatures that can characterize the compact clusters at the atomic level of detail. To simplify the collision dynamics, the study is carried out in the framework of the sudden approximation, which assumes that momentum changes perpendicular to the surface are targe compared with momentum transfer due to surface corrugation. The diffusion limited aggregates on which the scattering calculations were done, were generated by kinetic Monte Carlo simulations, It is demonstrated, by focusing on the example of compact Pt heptamers, that signatures of structure of compact clusters may indeed be extracted from the scattering distribution. These signatures enable both an experimental distinction between diffusion limited aggregates and compact clusters, and a determination of the cluster structure, The characteristics comprising the signatures are, to varying degrees, the rainbow, Fraunhofer, specular and constructive interference peaks, all seen in the intensity distribution, It is also shown, how the distribution of adsorbate heights above the metal surface can be obtained by an analysis of the specuIar peak attenuation. The results contribute to establishing He scattering as a powerful tool in the investigation of surface disorder and epitaxial growth on surfaces, alongside with STM.