Simple Force Field Research Articles

Modeling solvation dynamics of transition metal redox ion through on-the-fly multi-objective Bayesian-optimized force field.

Modeling solvation dynamics and properties is crucial for developing electrolytes for electrochemical energy storage and conversion devices. This work reports an on-the-fly multi-objective Bayesian optimization (OTF-MOBO) method to parameterize force fields for modeling ionic solvation structures, thermodynamics, and transport properties using molecular dynamics simulations. By leveraging solvation-free energy and solvation radii as training data, we employ the data-driven OTF-MOBO algorithm to actively optimize the force field parameters. The modeling accuracy was evaluated in molecular dynamics simulations until the Pareto front in the parameter space was reached through minimized prediction errors in both solvation-free energy and solvation radii. Using transition metal redox ions (Fe3+/Fe2+, Cr3+/Cr2+, and Cu2+/Cu+) in aqueous solution as examples, we demonstrate that simple force fields combining the Lenard-Jones potential and Coulombic potential can achieve relative error below 2% in both solvation free energy and solvation radii. The optimized force fields can be further extrapolated to predict solvation entropy and diffusivities with relative error below 10% compared with experiments.

Getting the intermolecular forces correct: introducing the ASTA strategy for a water model.

Having a force field for water providing good bulk properties is paramount for modern studies of most biological systems. Some of the most common three-site force fields are TIP3, SPC/ε or OPC3, providing a decent range of bulk properties. That does not mean though, that they have realistic inter-atomic forces. These force fields have been parameterized with a top-down approach, meaning, by fitting the force field parameters to the experimental bulk properties. This approach has been the governing strategy also for many variants of four- and more-site models. We test a bottom-up approach, in which the force field is parameterized by optimizing the non-bonded inter-atomic forces. Our philosophy is that correct inter-atomic forces lead to correct geometrical and dynamical properties. The first system we try to optimize with the accurately system tailored atomic (ASTA) approach is water, but we aim to eventually probe other systems in the future as well. We applied our ASTA strategy to find a good set of parameters providing accurate bulk properties for the simple three-site force field forms, and also for AMOEBA, a more detailed and polarizable force field. Even though our bottom-up approach did not provide satisfactory results for the simple three-site force fields (with fixed charges), for the case of the AMOEBA force field it led to a modification of the original strategy, giving very good intra- and inter-molecular forces, as compared to accurate quantum chemically calculated reference forces. At the same time, important bulk properties, in this study restricted to the density and diffusion, were accurately reproduced with respect to the experimental values.

Transport Parameters for Combustion Species Based on cAMOEBA Polarizable Force Field.

In this study, we report a simplified yet accurate general AMOEBA polarizable force field for combustion-interested molecular species, denoted as Combustion-AMOEBA or cAMOEBA. By eliminating the permanent atomic dipoles and quadrupoles, retaining the explicit polarization and defining the general atom types of each molecule species, including alkanes, alkenes, alkynes, alcohols, peroxides, and aldehydes, a simplified and general cAMOEBA force field was constructed and validated using the benchmark results obtained at the QCISD(T)/CBS level of theory. In this way, the tedious parametrization step for permanent atomic multipoles of each new molecule in the original AMOEBA (Poltype/MP2) force field could be avoided, hence providing the capability of accurate high-throughput calculation for a large number of molecules at lower computational cost. The averaged difference between the calculated transport parameters, σ and ε, for approximately 100 different molecules and four bath gases (He, Ne, Ar, and N2) using cAMOEBA and AMOEBA (Poltype/MP2) are of 0.09% and 1.27%, respectively, showing a good consistence of the general cAMOEBA force field with the original AMOEBA (Poltype/MP2) force field where the multipole force field parameters were obtained from quantum mechanical calculation for each small molecule. Our results also indicated that the Lorentz-Berthelot combination rule was more applicable than Waldman-Hagler for obtaining the molecular Lennard-Jones parameters of pure gases from one bath gas, while the Waldman-Hagler combination rule was better for obtaining such parameters from all four bath gases. The pure gas parameters obtained using cAMOEBA can be applied to develop high quality transport property database for combustion modeling.

Vapour–liquid coexistence of natural phenolic monoterpenoid, thymol: comparison with structural isomer, carvacrol

Thymol, a naturally occurring structural isomer of carvacrol, is known to exhibit medicinal properties. Since vapour–liquid equilibria data are critical for the design of efficient separation processes, theoretical methods and molecular simulations in conjunction with a simple united atom force field are employed to predict the coexistence properties; also at high temperatures where experimental data are scarce. Process design simulators used to model separation processes employ the equation of state approach which requires as input the critical state properties and acentric factor. Here, these inputs are computed using structure–property relations for use with the theoretical approach and are also determined from the more fundamental molecular simulations to be Tc = 70216 K, Pc = 336 bar, Vc = 48637 cc/mol and ω = 0.459. The structure of the coexisting liquid phase has been investigated using the radial distribution functions and number integrals to determine the extent of association present. The simple united atom force field captures the differences in the phase behaviour of thymol with that of carvacrol that arise due to steric effects of the relative placement of the hydroxyl group with respect to the methyl and isopropyl groups on the aromatic ring.

A Refined Molecular Dynamics Approach to Predict the Thermophysical Properties of Positively Charged Alumina Nanoparticles Suspended in Water

An enhanced Molecular Dynamics (MD) approach, synergistically combined with a fine-tuned Force Field (FF) model, is introduced to predict the behavior and thermophysical properties of cationic alumina (γ-Al 2 O 3 ) nanoparticles (NPs) dispersed in water at 4 different volumetric concentrations (1%, 2%, 3%, 4%) and over a temperature range of 5 to 40°C. Prior MD models have been limited to simplified force field models which fail in the representation of the actual chemical bonding, surface chemistry, and interfacial interactions of nanoparticles in aqueous solutions. In this contribution, a hybrid potential field model along with pH-resolved surface structure for alumina nanoparticles has been integrated with the MD model to represent the intermolecular interactions, dynamics and ultimately the thermophysical properties of the nanofluid that is in excellent quantitative agreement with the experimental data. In fact, our enhanced model performs better than the MD models found in the literature. The analysis of the numerically obtained properties emphasizes that at lower φ, the system shows a higher propensity for enhancement, especially at high temperatures. On the other hand, for volume fractions higher than 2%, the system's thermal performance is expected to exhibit a critical condition of aggregation and deteriorates due to higher dynamic viscosity. With all the above findings, we establish that the MD framework presented in this paper represents an improved step towards a precise and computationally balanced modelling that bridges the relation between the molecular signatures of the nanoparticles with pH resolved surface and the macroscopic behavior and thermal features of alumina-water nanofluids at different volumetric concentrations and temperatures.

Differentiable molecular simulation can learn all the parameters in a coarse-grained force field for proteins.

Finding optimal parameters for force fields used in molecular simulation is a challenging and time-consuming task, partly due to the difficulty of tuning multiple parameters at once. Automatic differentiation presents a general solution: run a simulation, obtain gradients of a loss function with respect to all the parameters, and use these to improve the force field. This approach takes advantage of the deep learning revolution whilst retaining the interpretability and efficiency of existing force fields. We demonstrate that this is possible by parameterising a simple coarse-grained force field for proteins, based on training simulations of up to 2,000 steps learning to keep the native structure stable. The learned potential matches chemical knowledge and PDB data, can fold and reproduce the dynamics of small proteins, and shows ability in protein design and model scoring applications. Problems in applying differentiable molecular simulation to all-atom models of proteins are discussed along with possible solutions and the variety of available loss functions. The learned potential, simulation scripts and training code are made available at https://github.com/psipred/cgdms.

Performing Molecular Dynamics Simulations and Computing Hydration Free Energies on the B3LYP-D3(BJ) Potential Energy Surface with Adaptive Force Matching: A Benchmark Study with Seven Alcohols and One Amine.

The potential energy surfaces at the B3LYP-D3(BJ) level for eight solutes in dilute aqueous solutions were mapped into simple pairwise additive force field expressions using the adaptive force matching (AFM) method. The quality of the fits was validated by computing the hydration free energy (HFE), enthalpy of hydration, and diffusion constant for each solute. By force matching B3LYP-D3(BJ), the predictions from the models agree with the closest experimental HFE and enthalpy of hydration within chemical accuracy. The diffusion constants from the models are also in good agreement with experimental references. The good agreement provides confidence on the quality of B3LYP-D3(BJ) in producing potential energy surfaces for thermodynamic property calculations through AFM for the molecules studied. Accurate computational predictions could potentially provide validations to experimental measurements in cases where experimental measurements from different sources do not agree.

Evaluating the Performance of Water Models with Host-Guest Force Fields in Binding Enthalpy Calculations for Cucurbit[7]uril-Guest Systems.

Computational prediction of thermodynamic components with computational methods has become increasingly routine in computer-aided drug design. Although there has been significant recent effort and improvements in the calculation of free energy, the prediction of enthalpy (and entropy) remains underexplored. Furthermore, there has been relatively little work reported so far that attempts to comparatively assess how well different force fields and water models perform in conjunction with each other. Here, we report a comprehensive assessment of force fields and water models using host-guest systems that mimic many features of protein-ligand systems. These systems are computationally inexpensive, possibly because of their small size compared to protein-ligand systems. We present absolute enthalpy calculations using the multibox approach on a set of 25 cucurbit[7]uril-guest pairs. Eight water models were considered (TIP3P, TIP4P, TIP4P-Ew, SPC, SPC/E, OPC, TIP5P, Bind3P), along with five force fields commonly used in the literature (GAFFv1, GAFFv2, CGenFF, Parsley, and SwissParam). We observe that host-guest binding enthalpies are strongly sensitive to the selection of force field and water model. In terms of water models, we find that TIP3P and its derivative Bind3P are the best performing models for this particular host-guest system. The performance is generally better for aliphatic compounds than for aromatic ones, suggesting that aromaticity remains a difficult property to include accurately in these simple force fields.

Exploring Free Energy Profiles of Enantioselective Organocatalytic Aldol Reactions under Full Solvent Influence.

We present a computational study on the enantioselectivity of organocatalytic proline-catalyzed aldol reactions between aldehydes in dimethylformamide (DMF). To explore the free energy surface of the reaction, we apply two-dimensional metadynamics on top of ab initio molecular dynamics (AIMD) simulations with explicit solvent description on the DFT level of theory. We avoid unwanted side reactions by utilizing our newly developed hybrid AIMD (HyAIMD) simulation scheme, which adds a simple force field to the AIMD simulation to prevent unwanted bond breaking and formation. Our condensed phase simulation results are able to nicely reproduce the experimental findings, including the main stereoisomer that is formed, and give a correct qualitative prediction of the change in syn:anti product ratio with different substituents. Furthermore, we give a microscopic explanation for the selectivity. We show that both the explicit description of the solvent and the inclusion of entropic effects are vital to a good outcome—metadynamics simulations in vacuum and static nudged elastic band (NEB) calculations yield significantly worse predictions when compared to the experiment. The approach described here can be applied to a plethora of other enantioselective or organocatalytic reactions, enabling us to tune the catalyst or determine the solvent with the highest stereoselectivity.

Force field for calcium sulfate minerals to predict structural, hydration, and interfacial properties

Calcium sulfates such as anhydrite, hemihydrate, and gypsum find widespread use in building materials, implants, and tissue healing. We introduce a simple and compatible atomistic force field for all calcium sulfate phases that reproduces a wide range of experimental data including lattice parameters, surface, hydration, mechanical, and thermal properties in 1% to 5% accuracy relative to experiments. The performance is several times better than prior force fields and DFT methods, which lead to errors in structures and energies up to 100%. We explain (hkl) cleavage energies, the dynamics of (hkl) water interfaces, and new insights into molecular origins of crystal-facet specific hydration and solubility. Impressive agreement of computed and experimentally measured hydration energies is shown. The models add to the Interface force field (IFF) and are compatible with multiple force fields (CHARMM, AMBER, GROMOS, CVFF, PCFF, OPLS-AA) for property predictions of sulfate-containing materials from atoms to the large nanometer scale.

Differentiable Optimization for the Prediction of Ground State Structures (DOGSS).

Ground state or relaxed inorganic structures are the starting point for most computational materials science or surface science analyses. Many of these structure relaxations represent systematic changes to the structure, but there are currently no general methods to improve the initial structure guess based on past calculations. Here we present a method to directly predict the ground state configuration using differentiable optimization and graph neural networks to learn the properties of a simple harmonic force field that approximates the ground state structure and properties. We demonstrate this flexible open source tool for improving the initial configurations for large datasets of inorganic multicomponent surface relaxations across 32 elements and the relaxation of adsorbates (H and CO) on these surfaces. Using these improved initial configurations reduces the expensive adsorbate-covered surface relaxations by approximately 50% and is complementary to other approaches to accelerate the relaxation process.

Efficient Refinement and Free Energy Scoring of Predicted Protein-Protein Complexes Using Replica Exchange with Repulsive Scaling.

Accurate prediction and evaluation of protein-protein complex structures are of major importance to understand the cellular interactome. Typically, putative complexes are predicted based on docking methods, and simple force field or knowledge-based scoring functions are applied to evaluate single complex structures. We have extended a replica-exchange-based scheme employing different levels of a repulsive biasing between partners in each replica simulation (RS-REMD) to simultaneously refine and score protein-protein complexes. The bias acts specifically on the intermolecular interactions based on an increase in effective pairwise van der Waals radii (repulsive scaling (RS)-REMD) without affecting interactions within each protein or with the solvent. The method provides a free energy score that correlates quite well with experimental binding free energies on a set of 36 complexes with correlation coefficients of 0.77 and 0.55 in explicit and implicit solvent simulations, respectively. For a large set of docked decoy complexes, significant improvement of docked complexes was found in many cases with the starting structure in the vicinity (within 20 Å) of the native complex. In the majority of cases (14 out of 20 in explicit solvent), near native docking solutions were identified as the best scoring complexes. The approach is computational demanding but may offer a route for refinement and realistic ranking of predicted protein-protein docking geometries.

Simplified force field for molecular dynamics simulations of amorphous SiO2 for solar applications

Mixtures of molten-salts (e.g. NaNO3, KNO3) with α-SiO2 nanoparticles have potential applications in concentrated solar power generation both as a thermal energy storage material and heat transfer fluid. In this study, a set of Lennard-Jones parameters for amorphous silica is proposed, aimed at understanding the molten-salt/amorphous-silica interface at the nanoscale through molecular dynamics simulations. Current force fields used for the simulation of α-SiO2 do not have validated rules for cross-term interactions or are computationally expensive. In this work, a novel set of Lennard-Jones parameters capable of accurately simulating the structure of amorphous SiO2 is presented. Its advantage is that it has precise validated mixing rules, which can be used to model cross term (different species) interactions, which are dominant in simulations involving interfaces. The proposed potential is validated by experimental results and data from previous studies: all the calculated properties (i.e. radial distribution, angular distribution, mean bond length, coordination number, density, thermal expansion coefficient and thermal conductivity) are in good agreement with the data in the literature. Additionally, the molten salt-amorphous silica interface is investigated. The existence of a compressed liquid layer at the liquid-nanoparticle interface is observed. The density of the molten salt mixture is found to be significantly higher and the self-diffusivity lower, in the proximity of the interface. Finally, the Na+ and K+ ions appear to be ordered in the proximity of the interface, indicating the existence of an ordered double-layer.

The Role of Dimer Formation in the Nucleation of Superlattice Transformations and Its Impact on Disorder.

The formation of defect-free two-dimensional nanocrystal (NC) superstructures remains a challenge as persistent defects hinder charge delocalization and related device performance. Understanding defect formation is an important step toward developing strategies to mitigate their formation. However, specific mechanisms of defect formation are difficult to determine, as superlattice phase transformations that occur during fabrication are quite complex and there are a variety of factors influencing the disorder in the final structure. Here, we use Molecular Dynamics (MD) and electron microscopy in concert to investigate the nucleation of the epitaxial attachment of lead chalcogenide (PbX, where X = S, Se) NC assemblies. We use an updated implementation of an existing reactive force field in an MD framework to investigate how initial orientational (mis)alignment of the constituent building blocks impacts the final structure of the epitaxially connected superlattice. This Simple Molecular Reactive Force Field (SMRFF) captures both short-range covalent forces and long-range electrostatic forces and allows us to follow orientational and translational changes of NCs during superlattice transformation. Our simulations reveal how robust the oriented attachment is with regard to the initial configuration of the NCs, measuring its sensitivity to both in-plane and out-of-plane misorientation. We show that oriented attachment nucleates through the initial formation of dimers, which corroborate experimentally observed structures. We present high-resolution structural analysis of dimers at early stages of the superlattice transformation and rationalize their contribution to the formation of defects in the final superlattice. Collectively, the simulations and experiments presented in this paper provide insights into the nucleation of NC oriented attachment, the impact of the initial configuration of NCs on the structural fidelity of the final epitaxially connected superlattice, and the propensity to form commonly observed defects, such as missing bridges and atomic misalignment in the superlattice due to the formation of dimers. We present potential strategies to mitigate the formation of superlattice defects.

Haptic Coupling in Dyads Improves Motor Learning in a Simple Force Field.

In dyadic motor learning, pairs of people learn the same motion while their limbs are loosely coupled together using haptic devices. Such coupled learning has been shown to outperform solo learning (including robot-guided learning) for simple one-degree-of-freedom tasks. However, results from more complex tasks are limited and sometimes conflicting. We thus evaluated coupled learning in a two-degree-of-freedom tracking task where participants also had to compensate for a simple force field. Participant pairs were split into two groups: an experiment group that experienced a compliant haptic coupling between participants' hands and a control group that did not. The study protocol consisted of 70 repetitions of 18.9-second tracking trials: 10 initial solo trials with no coupling, 50 "learning" trials (where participants in the experiment group were coupled), and 10 final solo trials with no coupling. The experiment group (coupled) improved their solo tracking performance both in the presence (p = 0.008) and absence (p <; 0.001) of the force field; however, the control group (no coupling) only improved their solo performance in the absence of the force field (p <; 0.001) but not in the presence of the field (p = 0.81). This suggests that dyadic motor learning can outperform solo learning for two-dimensional tracking motions in the presence of a simple force field, though the mechanism by which learning is improved is not yet clear.Clinical Relevance-As motor learning is critical for applications such as motor rehabilitation, dyadic training could be used to achieve a better overall outcome and a faster learning speed in these applications compared to solo training.

Exploring the Correlation between Stability, Fluxionality, and Absorption Spectra of Ultrasmall CdSe Clusters: A Computational Study

Various properties of (CdSe)3n clusters (n = 1–9) were obtained using computational techniques. Using a simple force field, a simulated annealing technique was employed to explore the potential energy landscape and identify the isomers available at low temperatures. The lowest energy isomer was then optimized using density functional theory, and the quantities associated with their stability and optoelectronics were calculated. The possibility of fluxionality for a given range of temperatures and solvents was characterized by computing thermodynamic properties and inherent structure energies. All studied systems give rise to energetically stable hollow structures composed of six- and four-membered rings that changed from spherical to tubular as the number of atoms increased. For a solvent of dielectric constant equal to three, fluxionality was observed at 300 K for almost all clusters considered. The corresponding band gap was found to be ∼3 eV and relatively independent of the size of the clusters in this size range. The absorption coefficients (as calculated from the dielectric functions) obtained from fluxionality effects are in agreement with the experimental absorption profiles.

Reactive modeling of Mo3Si oxidation and resulting silica morphology

Oxidation and corrosion have a significant economic footprint. Mo-based alloys are a strong candidate for structural materials with oxidation resistance at high temperatures. However, understanding of the mechanisms remains limited as experimental techniques do not reach atomic-scale resolution. We examined the mechanism of oxidation of Mo3Si (A15 phase) in Mo–Si–B alloys, the emergence of a superficial silica scale, and explain available experimental data up to the large nanometer scale using chemically detailed reactive simulations. We introduce new simulation protocols for layer-by-layer oxidation and simple force fields for the reactants, intermediates, and products. Growth of thin superficial silica layers as a function of temperature and oxidation rate on the (001) surface involves the formation of silica clusters, rings, and chains with pore sizes of 0 to 2 nm. An increase in temperature from 800 to 1000 °C slightly decreased the pore size and lead to less accumulation of Mo oxides at the interface, consistent with observations by electron tomography and energy dispersive X-ray spectroscopy (EDS). The elimination of gaseous MoOx is essential to form open channels and much larger pores up to 100 nm size as observed by 3D tomography, in-situ transmission electron microscopy (TEM) and scanning electron microscopy (SEM) as the oxide phase grows. According to the simulation, these large pores would otherwise be closed. The rate of oxidation, represented by successive oxidation of layers of variable thickness per unit time, influences the structure and cohesion of silica layers. High rates of oxidation can destabilize and break apart the silica layer, supported by a very wide pore size distribution in electron tomography. Limitations of the simulations in time scale currently restrict the analysis to few-layer oxidation. Within these bounds, the proposed simulation protocols can provide insight into the oxidation of (hkl) surfaces, grain boundaries, and various alloys compositions up to the 100 nm scale in atomic-level detail.

Adsorption of CO2, N2, CH4, and their mixtures on silicalite: A critical evaluation of force fields

The use of molecular simulation has been growing in the field of engineering, fueled not just by the advances in computational power but also on the availability of reliable software. One potential use of molecular simulation is related to the screening of materials for a specific application. The reliability of molecular simulation results depends on the trustworthiness of the force field used, which for engineering purposes should be as simple as possible. This work provides an evaluation of the potential accuracy cost of using simple generic force fields to predict the adsorption of CO2, CH4, N2 and their mixtures on MFI. We employed the GCMC technique for this investigation. Different models and force fields to describe adsorbates and adsorbent were tested. The force fields performances were estimated through comparison with available adsorption experimental data. Transferability was evaluated on the prediction of pure and mixtures adsorption on CHA, LTA and FER. The results showed that a simple force field presented similar performance when compared to a more sophisticated one.

Quantum Chemical Methods for Modeling Covalent Modification of Biological Thiols.

Targeted covalent inhibitor drugs require computational methods that go beyond simple molecular-mechanical force fields in order to model the chemical reactions that occur when they bind to their targets. Here, several semiempirical and density-functional theory (DFT) methods are assessed for their ability to describe the potential energy surface and reaction energies of the covalent modification of a thiol by an electrophile. Functionals such as PBE and B3LYP fail to predict a stable enolate intermediate. This is largely due to delocalization error, which spuriously stabilizes the prereaction complex, in which excess electron density is transferred from the thiolate to the electrophile. Functionals with a high-exact exchange component, range-separated DFT functionals, and variationally optimized exact exchange (i.e., the LC-B05minV functional) correct this issue to various degrees. The large gradient behavior of the exchange enhancement factor is also found to significantly affect the results, leading to the improved performance of PBE0. While ωB97X-D and M06-2X were reasonably accurate, no method provided quantitative accuracy for all three electrophiles, making this a very strenuous test of functional performance. Additionally, one drawback of M06-2X was that molecular dynamics (MD) simulations using this functional were only stable if a fine integration grid was used. The low-cost semiempirical methods, PM3, AM1, and PM7, provide a qualitatively correct description of the reaction mechanism, although the energetics is not quantitatively reliable. As a proof of concept, the potential of mean force for the addition of methylthiolate to methylvinyl ketone was calculated using quantum mechanical/molecular mechanical MD in an explicit polarizable aqueous solvent. © 2019 Wiley Periodicals, Inc.

Fick diffusion coefficients of binary fluid mixtures consisting of methane, carbon dioxide, and propane via molecular dynamics simulations based on simplified pair-specific ab initio-derived force fields

For the calculation of thermophysical properties of fluids with molecular dynamics (MD) simulations, effective force fields (FFs) which are optimized against experimental vapor-liquid equilibria are often used. Examples for this FF category are TraPPE, OPLS, and AMBER. An alternative are simplified pair-specific, ab initio-based FFs (AI-FFs), which are derived from quantum-chemical calculations in the limit of zero-density in combination with the kinetic theory of gases, and, thus, feature a predictive character. In the present study, we show with the help of selected binary fluid mixtures that the predictive power of MD simulations in calculating Fick diffusion coefficients can be improved using simplified pair-specific AI-FFs based on corresponding FFs for the pure substances. To evaluate the performance of the new AI-FFs in comparison with the TraPPE FFs, binary mixtures consisting of methane, carbon dioxide, and propane were investigated from the superheated vapor to the gas state and supercritical region up to the compressed liquid state. For the determination of the Fick diffusion coefficient at pressures between (0.1 and 12) MPa, temperatures between (293 and 355) K, and mole fractions between 0.05 and 0.95, separate simulations for the analysis of the Maxwell–Stefan diffusion coefficient and the thermodynamic factor were performed considering system size effects. With the exception of the compressed liquid state and regions in vicinity of the two-phase boundary, where the TraPPE FFs are generally superior, the AI-FFs show improved predictions for the Fick diffusion coefficient with average expanded statistical uncertainties of 12%. This could be demonstrated by comparison of the simulation results with the theoretical ab initio calculations and the available experimental data, resulting in average absolute deviations of 7% and 13% for the AI-FFs and TraPPE FFs.