Development and application of a ReaxFF reactive force field for hydrogen combustion

To investigate the reaction kinetics of hydrogen combustion at high-pressure and high-temperature conditions, we constructed a ReaxFF training set to include reaction energies and transition states relevant to hydrogen combustion and optimized the ReaxFF force field parameters against training data obtained from quantum mechanical calculations and experimental values. The optimized ReaxFF potential functions were used to run NVT MD (i.e., molecular dynamics simulation with fixed number of atoms, volume, and temperature) simulations for various H(2)/O(2) mixtures. We observed that the hydroperoxyl (HO(2)) radical plays a key role in the reaction kinetics at our input conditions (T ≥ 3000 K, P > 400 atm). The reaction mechanism observed is in good agreement with predictions of existing continuum-scale kinetic models for hydrogen combustion, and a transition of reaction mechanism is observed as we move from high pressure, low temperature to low pressure, high temperature. Since ReaxFF derives its parameters from quantum mechanical data and can simulate reaction pathways without any preconditioning, we believe that atomistic simulations through ReaxFF could be a useful tool in enhancing existing continuum-scale kinetic models for prediction of hydrogen combustion kinetics at high-pressure and high-temperature conditions, which otherwise is difficult to attain through experiments.     

A polarizable force field for water using an artificial neural network

A force field for liquid water including polarization effects has been constructed using an artificial neural network (ANN). It is essential to include a many-body polarization effect explicitly into a potential energy function in order to treat liquid water which is dense and highly polar. The new potential energy function is a combination of empirical and nonempirical potentials. The TIP4P model was used for the empirical part of the potential. For the nonempirical part, an ANN with a back-propagation of error algorithm (BPNN) was introduced to reproduce the complicated many-body interaction energy surface from ab initio quantum mechanical calculations. BPNN, described in terms of a matrix, provides enough flexibility to describe the complex potential energy surface (PES). The structural and thermodynamic properties, calculated by isobaric-isothermal (constant-NPT) Monte Carlo simulations with the new polarizable force field for water, are compatible with experimental results. Thus, the simulation establishes the validity of using our estimated PES with a polarization effect for accurate predictions of liquid state properties. Applications of this approach are simple and systematic so that it can easily be applied to the development of other force fields besides the water-water system.     

Development of interatomic ReaxFF potentials for Au-S-C-H systems

We present fully reactive interatomic potentials for systems containing gold, sulfur, carbon, and hydrogen, employing the ReaxFF formalism. The potential is designed especially for simulating gold-thiol systems and has been used for studying cluster deposition on self-assembled monolayers. Additionally, a large number of density functional theory calculations are reported, including molecules containing the aforementioned elements and adsorption energetics of molecules and atoms on gold.     

ReaxFF reactive force field for the Y-doped BaZrO3 proton conductor with applications to diffusion rates for multigranular systems

Proton-conducting perovskites such as Y-doped BaZrO 3 (BYZ) are promising candidates as electrolytes for a proton ceramic fuel cell (PCFC) that might permit much lower temperatures (from 400 to 600 degrees C). However, these materials lead to relatively poor total conductivity ( approximately 10 (-4) S/cm) because of extremely high grain boundary resistance. In order to provide the basis for improving these materials, we developed the ReaxFF reactive force field to enable molecular dynamics (MD) simulations of proton diffusion in the bulk phase and across grain boundaries of BYZ. This allows us to elucidate the atomistic structural details underlying the origin of this poor grain boundary conductivity and how it is related to the orientation of the grains. The parameters in ReaxFF were based entirely on the results of quantum mechanics (QM) calculations for systems related to BYZ. We apply here the ReaxFF to describe the proton diffusion in crystalline BYZ and across grain boundaries in BYZ. The results are in excellent agreement with experiment, validating the use of ReaxFF for studying the transport properties of these membranes. Having atomistic structures for the grain boundaries from simulations that explain the overall effect of the grain boundaries on diffusion opens the door to in silico optimization of these materials. That is, we can now use theory and simulation to examine the effect of alloying on both the interfacial structures and on the overall diffusion. As an example, these calculations suggest that the reduced diffusion of protons across the grain boundary results from the increased average distances between oxygen atoms in the interface, which necessarily leads to larger barriers for proton hopping. Assuming that this is the critical issue in grain boundary diffusion, the performance of BYZ for multigranular systems might be improved using additives that would tend to precipitate to the grain boundary and which would tend to pull the oxygens atoms together. Possibilities might be to use a small amount of larger trivalent ions, such as La or Lu or of tetravalent ions such as Hf or Th. Since ReaxFF can also be used to describe the chemical processes on the anode and cathode and the migration of ions across the electrode-membrane interface, ReaxFF opens the door to the possibility of atomistic first principles predictions on models of a complete fuel cell.     

Global optimization of parameters in the reactive force field ReaxFF for SiOH

We have used unbiased global optimization to fit a reactive force field to a given set of reference data. Specifically, we have employed genetic algorithms (GA) to fit ReaxFF to SiOH data, using an in‐house GA code that is parallelized across reference data items via the message‐passing interface (MPI). Details of GA tuning turn‐ed out to be far less important for global optimization efficiency than using suitable ranges within which the parameters are varied. To establish these ranges, either prior knowledge can be used or successive stages of GA optimizations, each building upon the best parameter vectors and ranges found in the previous stage. We have finally arrive‐ed at optimized force fields with smaller error measures than those published previously. Hence, this optimization approach will contribute to converting force‐field fitting from a specialist task to an everyday commodity, even for the more difficult case of reactive force fields. © 2013 Wiley Periodicals, Inc.     

Conformational Analysis of 2,2′-bifuran: Correlated High-level Ab initio and DFT Results

The torsional potential for inter-ring rotation in 2,2′-bifuran has been systematically tackled using highly accurate ab initio calculations as well as cost-effective DFT methods. The successful convergence of the ab initio results allowed to confirm the presence of a shallow gauche minimum in the torsional potential curve. The standard DFT methods failed to capture such a tiny energy barrier but, interestingly, the results could be remarkably improved by a mixture of wavefunction and DFT energies in a multi-coefficient fashion; thus, accurate DFT-based and ab initio reference data also become available. Since the experimental evaluation of torsional potentials faces quantitative problems, the outcome of high-level theoretical calculations is expected to be reliably used in further investigation on structure and conformational distribution of this system.     

First principle and ReaxFF molecular dynamics investigations of formaldehyde dissociation on Fe(100) surface

Detailed formaldehyde adsorption and dissociation reactions on Fe(100) surface were studied using first principle calculations and molecular dynamics (MD) simulations, and results were compared with available experimental data. The study includes formaldehyde, formyl radical (HCO), and CO adsorption and dissociation energy calculations on the surface, adsorbate vibrational frequency calculations, density of states analysis of clean and adsorbed surfaces, complete potential energy diagram construction from formaldehyde to atomic carbon (C), hydrogen (H), and oxygen (O), simulation of formaldehyde adsorption and dissociation reaction on the surface using reactive force field, ReaxFF MD, and reaction rate calculations of adsorbates using transition state theory (TST). Formaldehyde and HCO were adsorbed most strongly at the hollow (fourfold) site. Adsorption energies ranged from ?22.9 to ?33.9 kcal/mol for formaldehyde, and from ?44.3 to ?66.3 kcal/mol for HCO, depending on adsorption sites and molecular direction. The dissociation energies were investigated for the dissociation paths: formaldehyde → HCO + H, HCO → H + CO, and CO → C + O, and the calculated energies were 11.0, 4.1, and 26.3 kcal/mol, respectively. ReaxFF MD simulation results were compared with experimental surface analysis using high resolution electron energy loss spectrometry (HREELS) and TST based reaction rates. ReaxFF simulation showed less reactivity than HREELS observation at 310 and 523 K. ReaxFF simulation showed more reactivity than the TST based rate for formaldehyde dissociation and less reactivity than TST based rate for HCO dissociation at 523 K. TST‐based rates are consistent with HREELS observation. © 2013 Wiley Periodicals, Inc.     

A detailed reaction mechanism for hexamethyldisiloxane combustion via experiments and ReaxFF molecular dynamics simulations

Hexamethyldisiloxane (HMDSO) is one of the main impurities in the syngas produced from sewage and landfill plants. In order to utilize this syngas or control the characteristics of the generated silica particles, it is crucial to understand the chemical kinetics of HMDSO combustion. This study investigated the process of HMDSO combustion using synchrotron radiation mass spectrometry (SRMS), gas chromatography (GC), and ReaxFF molecular dynamics simulations. First, the force field used for ReaxFF simulation was validated by comparing the energies of different bond lengths, bond angles, and dihedral angles with the ones from DFT calculations. Good agreements were found. Then, ReaxFF simulations of HMDSO combustion with this force field were conducted under various conditions, which include different equivalence ratios (0.67, 1.0, and 1.5) and temperatures ranging from 2000 to 3500 K. The oxidation characteristics of HMDSO were analyzed, including the evolution of gas products and particle formation. Finally, based on the results from experiments and ReaxFF simulations, the reaction pathways, reaction lists, and reaction kinetics data during HMDSO combustion were obtained. A detailed reaction mechanism was proposed and validated by applying it in modeling the H₂/HMDSO/O₂ combustion systems. The temperature and part of the gas products such as CO and CO₂ as well as SiO could be well predicted.     

Recognition of the bouquet of Chinese spirits by artificial neural network analysis

Forty-six samples of Chinese spirits, whose bouquets were determined by sensory evaluations, and 17 compounds characteristic of the flavors determined by gas chromatography/gas chromatography-mass spectrometry (GC/GC-MS), were subjected to neural network analysis and their corresponding factor scores developed. To make the bouquet recognition more efficient, an improved artificial back-propagation neural network (BPNN) was applied. In each kind of data, the BPNN was trained repeatedly until the error rate was less than the predetermined threshold error; then the trained network was applied to the test set that was not involved in the training process to establish the validity of the network, and a correct prediction rate of 100% was obtained. The BPNN provided a correlation between the data offered from sensory evaluations and the data of chemical compositions determined by instrumental analysis. The BPNN approach is feasible regardless of whether the crude data or the factor scores are used; however, recognition results were better with the latter than with the former. In a comparison of all the results obtained by BPNN, cluster analysis, and discriminant analysis, the method of artificial neural network analysis appeared to be the optimal technique for recognizing the bouquet of Chinese spirits.     

Parameterization of a reactive force field using a Monte Carlo algorithm

Parameterization of a molecular dynamics force field is essential in realistically modeling the physicochemical processes involved in a molecular system. This step is often challenging when the equations involved in describing the force field are complicated as well as when the parameters are mostly empirical. ReaxFF is one such reactive force field which uses hundreds of parameters to describe the interactions between atoms. The optimization of the parameters in ReaxFF is done such that the properties predicted by ReaxFF matches with a set of quantum chemical or experimental data. Usually, the optimization of the parameters is done by an inefficient single‐parameter parabolic‐search algorithm. In this study, we use a robust metropolis Monte‐Carlo algorithm with simulated annealing to search for the optimum parameters for the ReaxFF force field in a high‐dimensional parameter space. The optimization is done against a set of quantum chemical data for MgSO₄ hydrates. The optimized force field reproduced the chemical structures, the equations of state, and the water binding curves of MgSO₄ hydrates. The transferability test of the ReaxFF force field shows the extend of transferability for a particular molecular system. This study points out that the ReaxFF force field is not indefinitely transferable. © 2013 Wiley Periodicals, Inc.     

Volume‐dependent potential approach for tungsten

A scheme to produce density‐of‐states‐(DOS)‐dependent potentials for d‐metals on the basis of the local density approximation calculations is suggested. As an example this scheme is applied to construct a DOS‐dependent potential for tungsten. The second moment of the tungsten DOS is calculated. We show that the obtained potentials give a good agreement of cohesive properties with the experimental data. ©1999 John Wiley & Sons, Inc. Int J Quant Chem 71: 343–348, 1999     

Distinct Electron Conductance Regimes in Bacterial Decaheme Cytochromes

Shewanella oneidensis MR‐1 gains energy by extracellular electron transfer to solid surfaces. They employ c‐type cytochromes in two Mtr transmembrane complexes, forming a multiheme wire for electron transport across the cellular outer membrane. We investigated electron‐ and hole‐transfer mechanisms in the external terminal of the two complexes, MtrC and MtrF. Comparison of computed redox potentials with previous voltammetry experiments in distinct environments (isolated and electrode‐bound conditions of PFV or in vivo) suggests that these systems function in different regimes depending on the environment. Analysis of redox potential shifts in different regimes indicates strong coupling between the hemes via an interplay between direct Coulomb and indirect interactions through local structural reorganization. The latter results in the screening of Coulomb interactions and explains poor correlation of the strength of the heme‐to‐heme interactions with the distance between the hemes.     

Multipole moments for embedding potentials: Exploring different atomic allocation algorithms

Polarizable quantum mechanical (QM)/molecular mechanics (MM)‐embedding methods are currently among the most promising methods for computationally feasible, yet reliable, production calculations of localized excitations and molecular response properties of large molecular complexes, such as proteins and RNA/DNA, and of molecules in solution. Our aim is to develop a computational methodology for distributed multipole moments and their associated multipole polarizabilities which is accurate, computationally efficient, and with smooth convergence with respect to multipole order. As the first step toward this goal, we herein investigate different ways of obtaining distributed atom‐centered multipole moments that are used in the construction of the electrostatic part of the embedding potential. Our objective is methods that not only are accurate and computationally efficient, but which can be consistently extended with site polarizabilities including internal charge transfer terms. We present a new way of dealing with well‐known problems in relation to the use of basis sets with diffuse functions in conventional atomic allocation algorithms, avoiding numerical integration schemes. Using this approach, we show that the classical embedding potential can be systematically improved, also when using basis sets with diffuse functions, and that very accurate embedding potentials suitable for QM/MM embedding calculations can be acquired. © 2016 Wiley Periodicals, Inc.     

Chemical Solutions in a Quantum Solvent: Anionic Electrolytes in 4He Nanodroplets

Variational and diffusion Monte Carlo (VMC and DMC) calculations are presented for anionic electrolytes solvated in ⁴He. The electrolytes have the general structure X^?(He)_N, with X=F, Cl, Br and I, and N varying up to 40 (41 for I^?). The overall interaction potential is obtained from accurate ab initio data for the two‐body components and then using the sum‐of‐potentials approximation. Our computational scheme is a robust procedure, giving us accurate trial wavefunctions that can be used to perform high‐quality DMC calculations. The results indicate very marked delocalization and permanence of the liquid‐like quantum features of the solvent adatoms surrounding the anionic impurities. This finding stands in contrast to the more structured, solid‐like behavior of the quantum solutions with alkali metal cations embedded in He nanodroplets. While other negatively charged species such as H^? have shown an overall repulsive interaction with He, the present calculations clearly indicate that the halogen anions remain solvated within liquid‐like solvent “bubbles” of species‐dependent size.     

Intermolecular potential of the O2-O2 dimer. An ab initio study and comparison with experiment

Accurate intermolecular potentials for the lowest three multiplet states of O2-O2 dimer have been produced on the basis of ab initio calculations. The quintet potential was taken from previous highly correlated CCSD(T) calculations. In this work, we perform MRCI calculations, with large basis sets including bond functions, of the singlet and triplet states, which are of multireference character. As expected the size inconsistency and lack of higher order excitations limit the accuracy of the MRCI potentials specifically in describing the long range interactions. We show that the Heisenberg Hamiltonian provides an accurate representation of the exchange interactions in this system and this enables us to combine the accurate CCSD(T) potentials with the MRCI spin-exchange parameter to obtain accurate singlet and triplet potentials. The reliability of these potentials is tested by computing integral cross sections and comparing them with the detailed experimental study of the Perugia group, with excellent results. More interestingly, comparison with the experimentally derived potential shows important discrepancies for some angular orientations including that corresponding with the global minima, indicating the need for further work, both theoretical and experimental, to clarify their origin.     

Ab initio calculation of anion proton affinity and ionization potential for energetic ionic liquids

Developing a better understanding of the bulk properties of ionic liquids requires accurate measurements of the underlying molecular properties that help to determine the bulk behavior. Two computational methods are used in this work: second‐order perturbation theory (MP2) and completely renormalized coupled cluster theory [CR‐CC(2,3)], to calculate the proton affinity and ionization potential of a set of anions that are of interest for use in protic, energetic ionic liquids. Compared with experimental values, both methods predict similarly accurate proton affinities, but CR‐CC(2,3) predicts significantly more accurate ionization potentials. It is concluded that more time intensive methods like CR‐CC(2,3) are required in calculations involving open shell states like the ionization potential. © 2015 Wiley Periodicals, Inc.     

The classical molecular dynamics simulation of graphene on Ru(0001) using a fitted Tersoff interface potential

The accurate molecular dynamics simulation of weakly bound adhesive complexes, such as supported graphene (gr), is challenging because of the lack of an adequate interface potential. Instead of the widely used Lennard‐Jones potential for weak and long‐range interactions, we use a newly parameterized Tersoff potential for gr/Ru(0001) system. The new interfacial force field provides adequate moire superstructures in accordance with scanning tunneling microscopy images and with density functional theory (DFT) results. In particular, the corrugation of ξ ≈ 1.0 ± 0.2 Å is found that is somewhat smaller than found by DFT approaches (ξ ≈ 1.2 Å) and is close to scanning tunneling microscope measurements (ξ ≈ 0.8 ± 0.3 Å). The new potential could open the way toward large‐scale simulations of supported gr with adequate moire supercells in many fields of gr research. Moreover, the new interface potential might provide a new strategy in general for obtaining accurate interaction potentials for weakly bound adhesion in large‐scale systems in which atomic dynamics is inaccessible yet by accurate DFT calculations. Copyright © 2013 John Wiley & Sons, Ltd.     

Grid‐based density functional calculations of many‐electron systems

Exploratory variational pseudopotential density functional calculations are performed for the electronic properties of many‐electron systems in the 3D cartesian coordinate grid (CCG). The atom‐centered localized gaussian basis set, electronic density, and the two‐body potentials are set up in the 3D cubic box. The classical Hartree potential is calculated accurately and efficiently through a Fourier convolution technique. As a first step, simple local density functionals of homogeneous electron gas are used for the exchange‐correlation potential, while Hay‐Wadt‐type effective core potentials are employed to eliminate the core electrons. No auxiliary basis set is invoked. Preliminary illustrative calculations on total energies, individual energy components, eigenvalues, potential energy curves, ionization energies, and atomization energies of a set of 12 molecules show excellent agreement with the corresponding reference values of atom‐centered grid as well as the grid‐free calculation. Results for three atoms are also given. Combination of CCG and the convolution procedure used for classical Coulomb potential can provide reasonably accurate and reliable results for many‐electron systems. © 2007 Wiley Periodicals, Inc. Int J Quantum Chem, 2008     

Classification of Rat FTIR Colon Cancer Data Using Waveletsand BPNN

本文提出了一种新的基于水平衰减全反射-傅里叶变换红外光谱（HATR-FTIR）的小波特征提取与反向传播人工神经网络模式分类方法以提高FTIR对早期大鼠结肠癌的诊断准确率．对60只DMH诱导的SD大鼠,44只诱导鼠的第二代鼠,36只正常SD大鼠的结肠正常组织、异常增生、早癌及进展期癌组织所获得的的HATR-FTIR,利用连续小波多尺度分析法提取12个特征量,采用反向传播人工神经网络进行分类,识别准确率分别为100%、94%、97.5%及100%．实验结果表明此方法对早期结肠癌具有较高的诊断率．