ReaxFF Parameters Research Articles

Detailed Reaction Kinetics for Hydrocarbon Fuels: The Development and Application of the ReaxFFCHO-S22 Force Field for C/H/O Systems with Enhanced Accuracy.

Efficient and accurate reactive force fields (e.g., ReaxFF) are pivotal for large-scale atomistic simulations to comprehend microscopic combustion processes. ReaxFF has been extensively utilized to describe chemical reactions in condensed phases, but most existing ReaxFF models rely on quantum mechanical (QM) data nearly two decades old, particularly in hydrocarbon systems, constraining their accuracy and applicability. Addressing this gap, we introduce a reparametrized ReaxFFCHO-S22 for C/H/O systems, tailored for studying the pyrolysis and combustion of hydrocarbon fuel. Our approach involves high-level QM benchmarks and large database construction for C/H/O systems, global ReaxFF parameter optimization, and molecular dynamics simulations of typical hydrocarbon fuels. Density functional theory (DFT) computations utilized the M06-2X functional at the meta-generalized gradient approximation (meta-GGA) level with a large basis set (6-311++G**). Our new ReaxFFCHO-S22 model exhibits a minimum 10% enhancement in accuracy compared to the previous ReaxFF models for a large variety of hydrocarbon molecules. This advanced ReaxFFCHO-S22 not only enables efficient large-scale studies on the microscopic chemical reactions of more complex hydrocarbon fuel but also can extend to biofuels, energetic materials, polymers, and other pertinent systems, thus serving as a valuable tool for studying chemical reaction dynamics of the large-scale hydrocarbon condensed phase at an atomistic level.

Managing Expectations and Imbalanced Training Data in Reactive Force Field Development: An Application to Water Adsorption on Alumina.

ReaxFF is a computationally efficient model for reactive molecular dynamics simulations that has been applied to a wide variety of chemical systems. When ReaxFF parameters are not yet available for a chemistry of interest, they must be (re)optimized, for which one defines a set of training data that the new ReaxFF parameters should reproduce. ReaxFF training sets typically contain diverse properties with different units, some of which are more abundant (by orders of magnitude) than others. To find the best parameters, one conventionally minimizes a weighted sum of squared errors over all of the data in the training set. One of the challenges in such numerical optimizations is to assign weights so that the optimized parameters represent a good compromise among all the requirements defined in the training set. This work introduces a new loss function, called Balanced Loss, and a workflow that replaces weight assignment with a more manageable procedure. The training data are divided into categories with corresponding "tolerances", i.e., acceptable root-mean-square errors for the categories, which define the expectations for the optimized ReaxFF parameters. Through the Log-Sum-Exp form of Balanced Loss, the parameter optimization is also a validation of one's expectations, providing meaningful feedback that can be used to reconfigure the tolerances if needed. The new methodology is demonstrated with a nontrivial parametrization of ReaxFF for water adsorption on alumina. This results in a new force field that reproduces both the rare and frequent properties of a validation set not used for training. We also demonstrate the robustness of the new force field with a molecular dynamics simulation of water desorption from a γ-Al2O3 slab model.

Development of a ReaxFF reactive force field for ternary phosphate-based bioactive glasses.

Phosphate-based glasses (PBGs) in the CaO-Na2O-P2O5 system have diverse applications as biomaterials due to their unique dissolution properties. A reactive force field (ReaxFF) has been developed to simulate these materials at the atomic level. The ReaxFF parameters of PBGs, including the interaction between phosphorus and calcium atoms, have been developed using a published code based on genetic algorithms. The training data, including the atomic charges, atomic forces, bond and angle parameters, and different differential energies, are chosen and measured from static quantum-mechanical calculations and abinitio molecular dynamics of PBGs. We did different short- and medium-range structural analyses on the bulk simulated PBGs with different compositions to validate the developed potential. Radial and angular distribution functions and coordination numbers of network formers and modifiers, as well as the network connectivity of the glass, are in agreement with experimental and previous simulations using both shell-model classical force fields and abinitio simulated data; for example, the coordination number of phosphorus is 4.0. This successful development of ReaxFF parameters being able to describe the bulk PBGs enables us to work on the dissolution behavior of the glasses, including the interaction of water molecules with PBGs, in future works.

A new reactive Mo-S-C force field to explore structure and properties of MoS2-C 2D materials and films

We employed a Monte-Carlo optimization scheme to successfully create a novel ReaxFF parameter set tailored for the exploration of the structures and energetics of Mo-S-C materials. This force field underwent rigorous testing against Density Functional Theory (DFT) data and exhibited its efficacy in modeling a broad spectrum of structures, ranging from composite heterolayer 2D structures to fully amorphous coatings. Subsequently, we applied this force field to get insights into the structural formation of amorphous MoS2-C films. Our simulations yielded promising results, revealing, in agreement with the experimental findings, preferential formation of Mo-S and C-C bonds along with rapid phase segregation.

Adaptive Force Field Parameter Optimization for Expanding Reaction Simulations within Wide-Ranged Temperature.

Large-scale and long-term simulation of chemical reactions are key research topics in computational chemistry. However, there are still difficulties in simulating high-temperature reactions, such as polymer thermal decomposition. Herein, we introduce an adaptive potential parameter optimization framework designed to automatically fine-tune parameters, and the application of it to optimize ReaxFF parameters enhances the accuracy of chemical reaction simulations conducted at experimental temperatures. To achieve this, we leverage the power of Random Forests and interpretable machine learning techniques that enable the identification and selection of parameters that exert a substantial influence on the target attribute. By training deep neural network (NN) models, we established optimized parameter associations with reference properties. We train deep neural network (NN) models to establish the relationship between the optimized parameters and reference properties. We employ a Genetic Algorithm (GA) to utilize the surrogate NN model and the quantum mechanical targets to speed up the search for the optimal parameters. Our simulation results of resin pyrolysis show that the adaptive optimized ReaxFF can predict the peak temperature more accurately and obtain reasonable product composition under conditions that more closely resemble experimental scenarios. This work facilitates advances in force field parameter optimization for more accurate and universal reaction simulations.

A reactive force field approach to modeling corrosion of NiCr alloys in molten FLiNaK salts

The interface between NiCr alloys and FLiNaK molten salt exhibits complex corrosion behavior, mainly driven by intricate chemical interactions involving Cr and F−ions. Understanding these dynamic reactions is crucial for developing effective corrosion mitigation strategies to ensure the long-term durability of Ni-based alloy components in molten salt technologies. However, obtaining molecular-level understanding through experiments is challenging. To address this, we utilize reactive molecular dynamics simulations enabled by a reactive force field, ReaxFF, to investigate detailed reaction dynamics at the atomic level. We first present the development of the ReaxFF parameter set for Ni/Cr/F/Li/Na/K based on extensive first-principles calculations. With this force field, we achieve a strong agreement for the structure of FLiNaK molten salt by comparing the pair distribution functions with experimental and simulation results. Furthermore, our simulations successfully reproduce the experimental phenomenon of Cr dissolution in fluoride salt, with the corrosion rate depending on the alloy and salt compositions. Particularly, our simulations reveal that increasing the concentration of Li can enhance the formation of a compact double layer, mitigating Cr dissolution. This work enables a fundamental understanding of the interfacial behavior between fluoride salt and NiCr alloys.

Computationally Predicted New Solid-State Electrolyte (Li5+x PS4+x Cl2-x : 0 ≤ x ≤ 2) and Poly Sulfide Cathodes (Li3+y PS9 or Li5+y PS9Cl2: 0 ≤ y ≤ 9) for High Performance Li Metal Anode Batteries

We used Molecular Dynamics (MD) simulations to study the structures and conductivity of sulfide-based ceramics from the superionic conductor argyrodite family for applications as solid-state electrolyte and cathode materials. These studies combined quantum mechanics (QM) and ReaxFF reactive force field based MD, where the ReaxFF parameters were developed from QM MD.[1]Currently Li10GeP2S12 (LGPS) demonstrates one of the highest Li-ion conductivity for solid state electrolyte, 12 mS/cm at 300K, but it is highly reactive with the Li-metal anode, and it is expensive due to the presence of Ge. We studied the Li ion diffusion mechanism and ionic conductivity of the promising Li6(PS4)SCl solid state electrolyte, predicted to have a conductivity of s = 6 mS/cm, close to experiment (4 mS/cm). We find that Li migration in this electrolyte occurs via conjugated substitutional type diffusion involving rearrangements of three (or more) Li-ions and ~ 3 vacant sites in a 3D matrix of anions that are essentially stationary at 298K (over 20 ns of simulation). We carried out 10 ns of MD simulation to predict a Li-ion conductivity for a single phase Li6(PS4)SCl of 5.9 mS/cm in agreement with solid state NMR measurements of 3.9 mS/cm. Our calculated activation energy of 0.24 eV is within experimentally reported range.[2]We report here the computationally predicted structure of Li5PS4Cl2, a new sulfide Li-superionic conductor with the highest Li-ion conductivity at solid state, which we predict to have Li-ion conductivity of 20 mS/cm at 298 K. We also predicted the directional ionic conductivity and Li-migration barrier in it. This Li5PS4Cl2 has not yet been synthesized and studied experimentally.We also report our results on QM and ReaxFF MD for new polysulfide cathodes, Li3+y PS4+n and Li5+y PS4+nCl2 , both based on Li3PS4. Here, n is the number of excess sulfur atoms in the fully charged polysulfide cathode per formula unit with y=0 and y is the excess Li added to the cathode during discharge processes. We evaluated the performance of these cathode electrolyte systems in terms of interfacial stability and discharge voltages.Li3PS4+5 is a potential cathode for lithium-sulfur batteries. We predicted the structures of Li3PS4+5 finding extra S3 and S7 chains attached to one S atom of each PS4 group. This leads to a density of 2.2 g/cm3. As the cathode is discharged, the lithium atoms fill up all the gaps between S atoms, leading to Li3+9 PS4+5, with 1.7 g/cm3. We studied the dynamics of the discharge as lithium ions from the metal move into the material and react with the S-S bonds to make Li2S. Our predicted discharge curve agrees well with the experimental data (Figure 1).[3]We extended these studies to design a new polysulfide cathode Li5PS4+5Cl2 to be used with our newly predicted Li-superionic conductor electrolytes. The interfacial stability of the Solid electrolyte with S-based cathode and with Li-anode were studied (Figure 2).Our work presents a new strategy for designing materials and processes for solid-state batteries using computational screening and chemical substitution. This demonstrates the potential of using lithium PS4 based superionic conductors with PS4 based cathodes for applications in solid-state batteries. We also addresses some of the key challenges in accelerating solid-state battery development, such as phase stability, transport properties, electrode compatibility etc. This work demonstrates how computational studies can guide new avenues for experimental developments to advance of solid-state battery technology for applications ranging from grid storage to electric vehicles.Acknowledgements:We acknowledge the support from Hong Kong Quantum AI Lab, AIR@InnoHK of Hong Kong Government.

Atomistic-Scale Simulations Toward the Understanding of the Conversion–Alloying Mechanism in Li-Ion Battery Anodes

The increasing demands for energy storage call for improved capacity and cycling stability of modern Li-ion batteries (LIBs). However, the application of the state-of-the-art LIB anode material – graphite is limited as graphite underperforms as a high-rate/high-storage-capacity anode. Alternatively, materials capable of alloying with Li-ions have attracted attention as a promising pathway to solve high-rate and high-capacity challenges. Furthermore, when alloying reaction in a material is coupled with irreversible conversion reaction, the materials operating under such mechanism exhibit a high Li-ion storage capacity and long-term stability during electrochemical cycling. However, despite the optimistic performance of these materials, a number of parameters need to be optimized and fundamentally understood.Among the unknowns for the conversion/alloying materials are the atomistic details of the reaction mechanism - understanding the atomistic structure during transformations is a pre-requisite for the future rational optimization of the materials. Furthermore, amorphization, which is prevalent in these materials, severely limits the conventional characterization methods essentially impeding further development and understanding the degradation mechanisms. Hence, to investigate the atomistic structure changes, we proposed a computational methodology for these conversion-alloying materials and their reaction mechanism by taking amorphous substoichiometric silicon nitride as a showcase system [1]. The molecular dynamics (MD) simulations utilized the ReaxFF parameter set which was specifically developed for the investigation. Subsequently, the experimental pair distribution functions (PDF) were used to verify the modeling results.While the reactive molecular dynamics was coupled with the PDF simulation to complement the experimental measurements, the effective and practical approach guided us to the atomistic structure evolution understanding for LIBs in a conversion–alloying process, revealing the bond coordinations around the Si atoms in the active material. Consequently, the analysis opened an in-depth understanding of the cycling stability of substoichiometric a-SiNx anode materials by revealing the atomistic features relevant to the changes this material undergoes during lithiation and delithiation processes in a LIB. Furthermore, the proposed approach allows modeling of the changes at the atomistic scale for the amorphous materials and predicting the experimental PDF at different cycling stages, verifying the modeling outcome. The topic also addresses the applications of the approach for other anode materials.

Atomistic study on the origins of the anisotropic lithiation behaviors of the silicon anode using the reactive force field based molecular dynamics simulations

We employed the Reactive-Force-field (ReaxFF) based molecular dynamics simulations to investigate the lithiation mechanisms and anisotropic lithiation behaviors of the Si-based anode during the lithiation processes. In order to accurately predict the relative lithiation kinetics on different surface facets of c-Si, we have developed new ReaxFF parameters for the Li-Si system based on first-principles calculations. Our results demonstrated that the lithiation kinetics on the Si(110) and (112) surfaces manifestly outperformed those occurred on the Si(100) and (111) surfaces. The lithiation reactions on the Si(110) and (112) surfaces tend to proceed through a ledge mechanism involving the layer-by-layer peeling of the {111} atomic facets. However, the lithiation reaction on the Si(100) surface is more likely to occur via the layer-by-layer etching of the {100} atomic facets. The discrepancy in the lithiation kinetics among the c-Si(100), (110) and (112) substrates was mainly ascribed to the variation in the compressive stresses that developed within the phase boundary upon lithiation. Regarding the Si(111) surface, the significantly slower lithiation kinetics than other surface facets were primarily attributed to the presence of a considerably larger barrier for Li insertion into the subsurface region to initiate the lithiation reaction. Furthermore, the order of the energy barriers for Li insertion into different c-Si surfaces was determined as follows: Si(111)>>(100)>(110)>(112), indicating that Li atoms tend to migrate into c-Si predominantly through the (110) and (112) surfaces upon lithiation. Consequently, lithiation reactions would occur majorly on the (110) and (112) surfaces due to the much lower compressive stresses that developed within the phase boundary and smaller Li insertion barriers than other surface facets, resulting in the observed anisotropic lithiation behaviors of the c-Si nanomaterials.

Effect of Fe-O ReaxFF on Liquid Iron Oxide Properties Derived from Reactive Molecular Dynamics.

As iron powder nowadays attracts research attention as a carbon-free, circular energy carrier, molecular dynamics (MD) simulations can be used to better understand the mechanisms of liquid iron oxidation at elevated temperatures. However, prudence must be practiced in the selection of a reactive force field. This work investigates the influence of currently available reactive force fields (ReaxFFs) on a number of properties of the liquid iron-oxygen (Fe-O) system derived (or resulting) from MD simulations. Liquid Fe-O systems are considered over a range of oxidation degrees ZO, which represents the molar ratio of O/(O + Fe), with 0 < ZO < 0.6 and at a constant temperature of 2000 K, which is representative of the combustion temperature of micrometric iron particles burning in air. The investigated properties include the minimum energy path, system structure, (im)miscibility, transport properties, and the mass and thermal accommodation coefficients. The properties are compared to experimental values and thermodynamic calculation results if available. Results show that there are significant differences in the properties obtained with MD using the various ReaxFF parameter sets. Based on the available experimental data and equilibrium calculation results, an improved ReaxFF is required to better capture the properties of a liquid Fe-O system.

A Reactive Force Field for Molecular Dynamics Simulations of Glucose in Aqueous Solution.

To expand the capabilities of reactive force field (ReaxFF) in simulations of biological processes involving glucose, in this work, using Metropolis Monte Carlo algorithm, new ReaxFF parameters for glucose have been developed to better describe the properties of glucose in water during molecular dynamics (MD) simulations. With the newly trained ReaxFF, the mutarotation of glucose in water can be better described, as suggested by our metadynamics simulations. In addition, the newly trained ReaxFF can better describe the distributions of the three stable conformers along the key dihedral angle of α-anomer and β-anomer. With better descriptions of hydration around glucose, the Raman and Raman optical activity spectra can be more accurately calculated. In addition, the infrared spectra obtained from simulations with the new glucose ReaxFF are more accurate than those obtained with the original ReaxFF. We note that although our trained ReaxFF performs better than the original ReaxFF, it is not generally applicable to all carbohydrates, which require further parametrization. We also find that the absence of explicit water molecules in the training sets may lead to inaccurate descriptions of water-water interactions around the glucose, implicating that it is necessary to optimize the water ReaxFF parameters together with the target molecule. The improved ReaxFF makes it possible to explore interesting biological processes involving glucose more accurately and efficiently.

Toward Atomistic Understanding of Materials with the Conversion–Alloying Mechanism in Li-Ion Batteries

The need for practical anodes for rechargeable batteries calls for new materials with improved storage capacity and cycling stability compared to the materials of today. Materials capable of alloying with alkali metals have been long viewed as a promising pathway to address this challenge. Meanwhile, severe drawbacks associated with the use of such materials led to the development of suitable alternatives, in which governing the reaction mechanism in a battery combines conversion and alloying reactions. These materials feature a high storage capacity and long-term stability during electrochemical cycling. However, the development of these materials is impeded by a lack of fundamental structural understanding due to the complexity of the chemical transformations taking place during battery cycling. Furthermore, amorphization, which is prevalent in these materials, also severely limits the conventional methods of characterization. In the present work, we propose a computational methodology for understanding the structure and reaction mechanism of these conversion–alloying materials by using amorphous substoichiometric silicon nitride as a model system. The developed ReaxFF parameter set allowed exploring the atomistic structure of this material and verify the resulting models through a comparison of calculated and experimentally obtained pair distribution functions (PDFs). The developed methodology also allowed exploring the structural changes occurring during lithiation and delithiation, i.e., operation mechanism in a battery. The analysis of the atomistic structure demonstrates that initial lithiation results in the formation of a Si network, which later participates in further lithiation process. Delithiation also influences the evolution of the amorphous Si network facilitating separation of nitrogen-rich and Si-rich regions and, therefore, confirming the conversion mechanism proposed earlier. Furthermore, the proposed approach allows modeling not only changes at the atomistic scale but also predicting the experimental PDF at different stages of cycling which allows experimentally verifying the outcome of modeling.

A Reactive Molecular Dynamics Study of Chlorinated Organic Compounds. Part II: A ChemTraYzer Study of Chlorinated Dibenzofuran Formation and Decomposition Processes.

In our two-paper series, we first present the development of ReaxFF CHOCl parameters using the recently published ParAMS parametrization tool. In this second part, we update the reactive Molecular Dynamics - Quantum Mechanics coupling scheme ChemTraYzer and combine it with our new ReaxFF parameters from Part I to study formation and decomposition processes of chlorinated dibenzofurans. We introduce a self-learning method for recovering failed transition-state searches that improves the overall ChemTraYzer transition-state search success rate by 10 percentage points to a total of 48 %. With ChemTraYzer, we automatically find and quantify more than 500 reactions using transition state theory and DFT. Among the discovered chlorinated dibenzofuran reactions are numerous reactions that are new to the literature. In three case studies, we discuss the set of reactions that are most relevant to the dibenzofuran literature: (i) bimolecular reactions of the chlorinated-dibenzofuran precursors phenoxy radical and 1,3,5-trichlorobenzene, (ii) dibenzofuran chlorination and pyrolysis, and (iii) oxidation of chlorinated dibenzofurans.

Development of a Mg/O ReaxFF Potential to describe the Passivation Processes in Magnesium-Ion Batteries.

One of the key challenges preventing the breakthrough of magnesium-ion batteries (MIB) is the formation of a passivating boundary layer at the Mg anode. To describe the initial steps of Mg anode degradation by O2 impurities, a Mg/O ReaxFF (force field for reactive systems) parameter set was developed capable of accurately modeling the bulk, surface, adsorption, and diffusion properties of metallic Mg and the salt MgO. It is shown that O2 immediately dissociates upon first contact with the Mg anode (modeled as Mg(0001), Mg(10 0)A, and Mg(10 1)), heating the surface to several 1000 K. The high temperature assists the further oxidation and forms a rock salt interphase intersected by several grain boundaries. Among the Mg surface terminations, Mg(10 0)A is the most reactive, forming an MgO layer with a thickness of up to 25 Å. The trained force field can be used to model the ongoing reactions in Mg-air batteries but also to study the oxidation of magnesium metal in general.

JAX-ReaxFF: A Gradient-Based Framework for Fast Optimization of Reactive Force Fields.

The reactive force field (ReaxFF) model bridges the gap between traditional classical models and quantum mechanical (QM) models by incorporating dynamic bonding and polarizability. To achieve realistic simulations using ReaxFF, model parameters must be optimized against high fidelity training data which typically come from QM calculations. Existing parameter optimization methods for ReaxFF consist of black box techniques using genetic algorithms or Monte Carlo methods. Due to the stochastic behavior of these methods, the optimization process oftentimes requires millions of error evaluations for complex parameter fitting tasks, thereby significantly hampering the rapid development of high quality parameter sets. Rapid optimization of the parameters is essential for developing and refining Reax force fields because producing a force field which exhibits empirical accuracy in terms of dynamics typically requires multiple refinements to the training data as well as to the parameters under optimization. In this work, we present JAX-ReaxFF, a novel software tool that leverages modern machine learning infrastructure to enable fast optimization of ReaxFF parameters. By calculating gradients of the loss function using the JAX library, JAX-ReaxFF utilizes highly effective local optimization methods that are initiated from multiple guesses in the high dimensional optimization space to obtain high quality results. Leveraging the architectural portability of the JAX framework, JAX-ReaxFF can execute efficiently on multicore CPUs, graphics processing units (GPUs), or even tensor processing units (TPUs). As a result of using the gradient information and modern hardware accelerators, we are able to decrease ReaxFF parameter optimization time from days to mere minutes. Furthermore, the JAX-ReaxFF framework can also serve as a sandbox environment for domain scientists to explore customizing the ReaxFF functional form for more accurate modeling.

New Reactive Force Field for Simulations of MoS2 Crystallization.

We present a new reactive force field (ReaxFF) parameter set for simulations of Mo–S structures. We compare our parameterization to the state-of-the-art ones in their performance against density functional theory (DFT) benchmarks and MoS2 crystallization simulations. Our new force field matches DFT data significantly better than any previously published force fields and provides a realistic layered MoS2 structure in crystallization simulations. It significantly improves the state-of-the-art force fields, which tend to crystallize in the experimentally unknown rock-salt MoS structure. Therefore, our new force field is a good candidate for further development and inclusion of other practically relevant elements, such as O, C, N, and H, which can be used to study the formation and tribological or catalytical properties of molybdenum disulfide.

Mixing ReaxFF parameters for transition metal oxides using force-matching method

We present the development of the method for the refitting the ReaxFF parameters for a system consisting of the mixed transition metal oxides. We have tested several methods allowing to calculate the differences between the vectors of the forces acting on atoms obtained from the reference DFT simulation and the parameters-dependent ReaxFF. We conclude that the footrule method yields the best parameters among the investigated options. We then validate the parameters using the system consisting of the hematite supported (TiO2)n clusters. The results indicate the refitted parameters allow to obtain acceptable geometries of the clusters upon MD simulation on the ReaxFF level, and despite the short timescale lead to the stable structures.

Development of Mg/Al/Si/O ReaxFF Parameters for Magnesium Aluminosilicate Glass Using an Artificial Neural Network-Assisted Genetic Algorithm

Commercial high-strength S-glass fiber used in structural composites mainly consists of SiO2, Al2O3, and MgO. There is no established reactive force field to characterize S-glass fiber. In this study, a newly developed artificial neural network (ANN)-assisted genetic algorithm (GA) is applied to optimize a new ReaxFF parameter set to describe Mg/Al/Si/O interactions in S-glass and other magnesium aluminosilicate (MAS) glass compositions. The training set includes the density functional theory data of the energy response of various Mg/Al/Si/O crystals during volumetric expansion and compression and Mg migration inside Mg/Al/Si/O crystals. Test molecular dynamics simulations showed the characteristics of tectosilicate MAS glasses. Different structural properties, including oxide coordination, density, structural factors, and mechanical properties, showed fair agreement with references from experiments and other simulations. A newly developed GA-ANN parametrization algorithm assisted the training process. This force field can be used for virtual composition mapping to develop new glass fiber materials. We also believe our force field would support computational studies of mechanical properties of amorphous materials used in geochemistry, construction, and protective material applications.

Molecular Dynamics Simulation of Silicon Dioxide Etching by Hydrogen Fluoride Using the Reactive Force Field.

In this study, we develop a reactive force field (ReaxFF) for a Si/O/H/F system to perform etching simulations of SiO2 with an HF etchant. Quantum mechanical (QM) training sets from density functional theory calculations, which contain structures of reactant/product and energies with bond dissociation, valence angle distortions, and reactions between SiO2 clusters and SiO2 slab with HF gases, are used to optimize the ReaxFF parameters. Structures and energies calculated using the ReaxFF match well with the QM training sets. Using the optimized ReaxFF, we conduct molecular dynamics simulations of the etching process of SiO2 substrates with active HF molecules. The etching yield and number of reaction products with different incident energies of the HF etchant are investigated. These simulations show that the developed ReaxFF offers insights into the atomistic surface reaction of the SiO2 etching process.

Development of ReaxFFSFOH Force Field for SF6‐H2O/O2 Hybrid System Based on Synergetic Optimization by CMA‐ES and MC Methodology

AbstractThe reactive force field (ReaxFF) based molecular dynamic (MD) simulation provides a new perspective for studying the microscopic decomposition mechanism of sulfur hexafluoride (SF6) under fault conditions. Since none of the published ReaxFF force fields contains proper interaction relationship between S, F, O and H elements, the development of ReaxFF parameters for SF6‐H2O/O2 hybrid system is urgently needed. In this work, a universal and efficient covariance matrix adaptation evolutionary strategy (CMA‐ES) and Monte‐Carlo FF (MCFF) synergetic (CMS) optimization strategy for ReaxFF parameter is proposed. The optimization results for two representative force fields (disulfide and silica) from literature show that under the same number of evaluations, the CMS optimization strategy significantly reduces the optimization error by 22.08 % (disulfide) and 10.3 % (silica) respectively. By applying CMS strategy, the ReaxFFSFOH force field for SF6‐H2O/O2 hybrid system is efficiently obtained. Through multi‐dimensional force field performance verification, we find that the bond length calculation error between ReaxFF and density functional theory (DFT) calculation results is mainly concentrated within ±0.05 Å, and the bond angle error is mostly within ±5°. In addition, the trends of relative energy changes during chemical reaction processes calculated by ReaxFF are basically consistent with the DFT results. This work lays a solid foundation for the in‐depth study of the microscopic influence mechanism of trace H2O and trace O2 in the decomposition process of SF6.