Giovanni Garberoglio

Potential energy surface for interactions between two hydrogen molecules

Nonrelativistic clamped-nuclei energies of interaction between two ground-state hydrogen molecules with intramolecular distances fixed at their average value in the lowest rovibrational state have been computed. The calculations applied the supermolecular coupled-cluster method with single, double, and noniterative triple excitations [CCSD(T)] and very large orbital basis sets-up to augmented quintuple zeta size supplemented with bond functions. The same basis sets were used in symmetry-adapted perturbation theory calculations performed mainly for larger separations to provide an independent check of the supermolecular approach. The contributions beyond CCSD(T) were computed using the full configuration interaction method and basis sets up to augmented triple zeta plus midbond size. All the calculations were followed by extrapolations to complete basis set limits. For two representative points, calculations were also performed using basis sets with the cardinal number increased by one or two. For the same two points, we have also solved the Schrodinger equation directly using four-electron explicitly correlated Gaussian (ECG) functions. These additional calculations allowed us to estimate the uncertainty in the interaction energies used to fit the potential to be about 0.15 K or 0.3% at the minimum of the potential well. This accuracy is about an order of magnitude better than that achieved by earlier potentials for this system. For a near-minimum T-shaped configuration with the center-of-mass distance R=6.4 bohrs, the ECG calculations give the interaction energy of -56.91+/-0.06 K, whereas the orbital calculations in the basis set used for all the points give -56.96+/-0.16 K. The computed points were fitted by an analytic four-dimensional potential function. The uncertainties in the fit relative to the ab initio energies are almost always smaller than the estimated uncertainty in the latter energies. The global minimum of the fit is -57.12 K for the T-shaped configuration at R=6.34 bohrs. The fit was applied to compute the second virial coefficient using a path-integral Monte Carlo approach. The achieved agreement with experiment is substantially better than in any previous work.

Quantitative protein dynamics from dominant folding pathways

We develop a theoretical approach to the protein-folding problem based on out-of-equilibrium stochastic dynamics. Within this framework, the computational difficulties related to the existence of large time scale gaps are removed, and simulating the entire reaction in atomistic details using existing computers becomes feasible. We discuss how to determine the most probable folding pathway, identify configurations representative of the transition state, and compute the most probable transition time. We perform an illustrative application of these ideas, studying the conformational evolution of alanine dipeptide, within an all-atom model based on the empiric GROMOS96 force field.

Hydrogen isotope separation in carbon nanotubes: calculation of coupled rotational and translational States at high densities.

The effect of the quantized rotational degrees of freedom of hydrogen on the adsorption and sieving properties in carbon nanotubes is studied using computer simulations. We have developed a highly efficient multiple timestep algorithm for hybrid Monte Carlo sampling of quantized rotor configurations and extended the grand canonical Boltzmann bias method to rigid linear molecules. These new computational tools allow us to calculate accurately the quantum sieving selectivities for cases of extreme two-dimensional confinement as a function of pressure. The para-T2/para-H2 selectivity at 20 K is analyzed as a function of the tube diameter and the density of adsorbed hydrogen. Extraordinarily high selectivities, up to 2.6 x 10(8), are observed in the narrowest nanotube. The quantized nature of the rotational degrees of freedom is found to dramatically affect adsorption and selectivity for hydrogen isotopes adsorbed in very narrow nanotubes. The T2/H2 zero-pressure selectivity increases from 2.4 x 10(4) to 1.7 x 10(8) in the (3,6) nanotube at 20 K when quantum rotations are accounted for. The isotopic selectivity is found to increase with pressure, tending to a constant value at saturation. A simplified mean-field model is used to discuss the origin of this behavior.

Water in contact with magnetite nanoparticles, as seen from experiments and computer simulations.

The adsorption of water vapor at the surface of magnetite nanoparticles has been investigated both by experimental and by computer simulation methods. The water vapor adsorption/desorption isotherm has been measured on freshly prepared magnetite nanocrystals of the size below 10 nm. The change of the isosteric heat of adsorption with the surface coverage has been determined from the temperature dependence of this isotherm using the isosteric method. The adsorption isotherm has also been determined by performing a set of grand canonical Monte Carlo simulations at 300 K. X-ray photoelectron spectroscopy results as well as the temperature and coverage dependence of the isosteric heat of adsorption clearly indicates that dissociative chemisorption of the water molecules in the first adsorption layer occurs at the bare magnetite surface, resulting in a high density of surface hydroxyl groups. This dissociative chemisorption is followed by a multilayer physisorption of water at higher pressures. Computer simulation results can reproduce excellently both the adsorption isotherm and the isosteric heat of adsorption beyond the first chemisorbed layer of water. Results of the computer simulations reveal that physisorbed water forms several well-distinguished molecular layers on the magnetite surface; however, these layers are not built up sequentially. Instead, the building up of several molecular layers occurs simultaneously. The adsorption of the water molecules in this range appears to be a nucleation-like process, resulting in a rather rough external surface of the adsorption layer.

Improved First-Principles Calculation of the Third Virial Coefficient of Helium

We employ state-of-the-art pair and three-body potentials with path-integral Monte Carlo (PIMC) methods to calculate the third density virial coefficient C(T) for helium. The uncertainties are much smaller than those of the best experimental results, and approximately one-fourth the uncertainty of our previous work. We have extended our results in temperature down to 2.6 K, incorporating the effect of spin statistics that become important below approximately 7 K. Results are given for both the 3He and 4He isotopes. We have also performed PIMC calculations of the third acoustic virial coefficient γa; our calculated values compare well with the limited experimental data available. A correlating equation for C(T) of 4He is presented; differentiation of this equation provides a reliable and simpler way of calculating γa.

First-Principles Calculation of the Third Virial Coefficient of Helium

Knowledge of the pair and three-body potential-energy surfaces of helium is now sufficient to allow calculation of the third density virial coefficient, C(T), with significantly smaller uncertainty than that of existing experimental data. In this work, we employ the best available pair and three-body potentials for helium and calculate C(T) with path-integral Monte Carlo (PIMC) calculations supplemented by semiclassical calculations. The values of C(T) presented extend from 24.5561 K to 10 000 K. In the important metrological range of temperatures near 273.16 K, our uncertainties are smaller than the best experimental results by approximately an order of magnitude, and the reduction in uncertainty at other temperatures is at least as great. For convenience in calculation of C(T) and its derivatives, a simple correlating equation is presented.

Boltzmann bias grand canonical Monte Carlo.

We derive an efficient method for the insertion of structured particles in grand canonical Monte Carlo simulations of adsorption in very confining geometries. We extend this method to path integral simulations and use it to calculate the isotherm of adsorption of hydrogen isotopes in narrow carbon nanotubes (two-dimensional confinement) and slit pores (one-dimensional confinement) at the temperatures of 20 and 77 K, discussing its efficiency by comparison to the standard path integral grand canonical Monte Carlo algorithm. We use this algorithm to perform multicomponent simulations in order to calculate the hydrogen isotope selectivity for adsorption in narrow carbon nanotubes and slit pores at finite pressures. The algorithm described here can be applied to the study of adsorption of real oligomers and polymers in narrow pores and channels.

Instantaneous normal mode analysis of correlated cluster motions in hydrogen bonded liquids

We analyze the correlated motions of hydrogen bonded clusters in liquid hydrogen fluoride, methanol, and water using the Instantaneous Normal Mode approach. In the case of hydrogen fluoride and methanol, which form a topologically linear hydrogen bond network, the relevant cluster is a triplet formed by a molecule and its two neighbors. In the case of water, whose hydrogen bond structure has a local tetrahedral symmetry, the basic unit considered is the pentamer formed by a molecule and its four neighbors. For each of these clusters we identify, using symmetry arguments, the a priori modes describing the relative motions of the cluster molecules and introduce suitable projections in order to evaluate how much these modes contribute to the actual motions at different frequencies. In the case of hydrogen fluoride we confirm the assignment of a 50 rad/ps peak observed in the single and collective correlation function spectra to the stretching of the hydrogen bonded network. In the case of methanol the analysis of the correlated motions of the triplets shows that in the intermediate frequency range (around 25 rad/ps) a picture similar to what is observed in hydrogen fluoride applies, whereas the high frequency properties of the liquid (beyond 50 rad/ps) are mostly due to the asymmetric stretching motion. In the case of water we demonstrate that the a priori modes, based on the full tetrahedral symmetry of the water pentamer, do indeed mix strongly under the effect of the interaction with the neighbors. The results are related to the spectroscopic measurement.

Three-Body Nonadditive Potential for Argon with Estimated Uncertainties and Third Virial Coefficient

The three-body nonadditive interaction energy between argon atoms was calculated at 300 geometries using coupled cluster methods up to single, double, triple, and noniterative quadruple excitations [CCSDT(Q)], and including the core correlation and relativistic effects. The uncertainty of the calculated energy was estimated at each geometry. The analytic function fitted to the energies is currently the most accurate three-body argon potential. Values of the third virial coefficient C(T) with full account of quantum effects were computed from 80 to 10000 K by a path-integral Monte Carlo method. The calculation made use of an existing high-quality pair potential [Patkowski, K.; Szalewicz, K. J. Chem. Phys. 2010, 133, 094304] and of the three-body potential derived in the present work. Uncertainties in the potential were propagated to estimate uncertainties in C(T). The results were compared with available experimental data, including some values of C(T) newly derived in this work from previously published high-accuracy density measurements. Our results are generally consistent with the available experimental data in the limited range of temperatures where data exist, but at many conditions, especially at higher temperatures, the uncertainties of our calculated values are smaller than the uncertainties of the experimental values.

Second virial coefficients of H2 and its isotopologues from a six-dimensional potential

We employ path-integral Monte Carlo techniques to compute the second virial coefficient as a function of temperature for molecular hydrogen (H(2)), deuterium (D(2)), and tritium (T(2)), along with the mixed isotopologues HD, HT, and DT. The calculations utilize a new six-dimensional (6D) potential, which is derived by combining our previous high-quality ground-state 4D potential for the H(2) dimer with the 6D potential of Hinde. This new 6D potential is reduced to a set of 4D potentials by fixing the intramolecular coordinates at their expectation values for each temperature and isotopic combination. The results for H(2) are in good agreement with experimental data; the effect of the temperature dependence of the average bond length is only significant above approximately 1000 K. For D(2) and HD, the available experimental data are much more limited; our results agree with the data and provide reliable values at temperatures where no experimental data exist. For the species containing tritium, our results provide the only data available.