Stochastic machine learning via sigma profiles to build a digital chemical space.

This work establishes a different paradigm on digital molecular spaces and their efficient navigation by exploiting sigma profiles. To do so, the remarkable capability of Gaussian processes (GPs), a type of stochastic machine learning model, to correlate and predict physicochemical properties from sigma profiles is demonstrated, outperforming state-of-the-art neural networks previously published. The amount of chemical information encoded in sigma profiles eases the learning burden of machine learning models, permitting the training of GPs on small datasets which, due to their negligible computational cost and ease of implementation, are ideal models to be combined with optimization tools such as gradient search or Bayesian optimization (BO). Gradient search is used to efficiently navigate the sigma profile digital space, quickly converging to local extrema of target physicochemical properties. While this requires the availability of pretrained GP models on existing datasets, such limitations are eliminated with the implementation of BO, which can find global extrema with a limited number of iterations. A remarkable example of this is that of BO toward boiling temperature optimization. Holding no knowledge of chemistry except for the sigma profile and boiling temperature of carbon monoxide (the worst possible initial guess), BO finds the global maximum of the available boiling temperature dataset (over 1,000 molecules encompassing more than 40 families of organic and inorganic compounds) in just 15 iterations (i.e., 15 property measurements), cementing sigma profiles as a powerful digital chemical space for molecular optimization and discovery, particularly when little to no experimental data is initially available.

Ionic Liquids for the Separation of Fluorocarbon Refrigerant Mixtures.

This review discusses the research being performed on ionic liquids for the separation of fluorocarbon refrigerant mixtures. Fluorocarbon refrigerants, invented in 1928 by Thomas Midgley Jr., are a unique class of working fluids that are used in a variety of applications including refrigeration. Fluorocarbon refrigerants can be categorized into four generations: chlorofluorocarbons, hydrochlorofluorocarbons, hydrofluorocarbons, and hydrofluoroolefins. Each generation of refrigerants solved a key problem from the previous generation; however, each new generation has relied on more complex mixtures that are often zeotropic, near azeotropic, or azeotropic. The complexity of the refrigerants used and the fact that many refrigerants form azeotropes when mixed makes handling the refrigerants at end of life extremely difficult. Today, less than 3% of refrigerants that enter the market are recycled. This is due to a lack of technology in the refrigerant reclaim market that would allow for these complex, azeotropic refrigerant mixtures to be separated into their components in order to be effectively reused, recycled, and if needed repurposed. As the market for recovering and reclaiming refrigerants continues to grow, there is a strong need for separation technology. Ionic liquids show promise for separating azeotropic refrigerant mixtures as an entrainer in extractive distillation process. Ionic liquids have been investigated with refrigerants for this application since the early 2000s. This review will provide a comprehensive summary of the physical property measurements, equations of state modeling, molecular simulations, separation techniques, and unique materials unitizing ionic liquids for the development of an ionic-liquid-based separation process for azeotropic refrigerant mixtures.

Consistent and reproducible computation of the glass transition temperature from molecular dynamics simulations.

In many fields, from semiconductors for opto-electronic applications to ionic liquids (ILs) for separations, the glass transition temperature (Tg) of a material is a useful gauge for its potential use in practical settings. As a result, there is a great deal of interest in predicting Tg using molecular simulations. However, the uncertainty and variation in the trend shift method, a common approach in simulations to predict Tg, can be high. This is due to the need for human intervention in defining a fitting range for linear fits of density with temperature assumed for the liquid and glass phases across the simulated cooling. The definition of such fitting ranges then defines the estimate for the Tg as the intersection of linear fits. We eliminate this need for human intervention by leveraging the Shapiro-Wilk normality test and proposing an algorithm to define the fitting ranges and, consequently, Tg. Through this integration, we incorporate into our automated methodology that residuals must be normally distributed around zero for any fit, a requirement that must be met for any regression problem. Consequently, fitting ranges for realizing linear fits for each phase are statistically defined rather than visually inferred, obtaining an estimate for Tg without any human intervention. The method is also capable of finding multiple linear regimes across density vs temperature curves. We compare the predictions of our proposed method across multiple IL and semiconductor molecular dynamics simulation results from the literature and compare other proposed methods for automatically detecting Tg from density-temperature data. We believe that our proposed method would allow for more consistent predictions of Tg. We make this methodology available and open source through GitHub.

Adsorption of difluoromethane (HFC-32) and pentafluoroethane (HFC-125) and their mixtures in silicalite-1: An experimental and Monte Carlo simulation study.

Hydrofluorocarbons are a class of fluorinated molecules used extensively in residential and industrial refrigeration systems. This study examines the potential of using adsorption processes with the silicalite-1 zeolite to separate a mixture of difluoromethane (CH2F2, HFC-32) and pentafluoroethane (CF3CF2H, HFC-125) at various concentrations. Pure adsorption data were measured using a XEMIS gravimetric microbalance, whereas binary data were determined using the Integral Mass Balance method. Grand canonical Monte Carlo molecular simulations were performed with the Cassandra package. We found that the results from molecular simulations are in satisfactory agreement with experimental loading measurements. Moreover, we show that ideal adsorbed solution theory could not quantitatively match the experimental or computational measurements of binary adsorption or selectivity. Molecular simulations show that refrigerant molecules do not have a uniform distribution in the zeolite framework.

Deep Eutectic Solvents: A Review of Fundamentals and Applications.

Deep eutectic solvents (DESs) are an emerging class of mixtures characterized by significant depressions in melting points compared to those of the neat constituent components. These materials are promising for applications as inexpensive "designer" solvents exhibiting a host of tunable physicochemical properties. A detailed review of the current literature reveals the lack of predictive understanding of the microscopic mechanisms that govern the structure-property relationships in this class of solvents. Complex hydrogen bonding is postulated as the root cause of their melting point depressions and physicochemical properties; to understand these hydrogen bonded networks, it is imperative to study these systems as dynamic entities using both simulations and experiments. This review emphasizes recent research efforts in order to elucidate the next steps needed to develop a fundamental framework needed for a deeper understanding of DESs. It covers recent developments in DES research, frames outstanding scientific questions, and identifies promising research thrusts aligned with the advancement of the field toward predictive models and fundamental understanding of these solvents.

Intermolecular interactions in clusters of ethylammonium nitrate and 1-amino-1,2,3-triazole.

The intermolecular interaction energies, including hydrogen bonds (H-bonds), of clusters of the ionic liquid ethylammonium nitrate (EAN) and 1-amino-1,2,3-triazole (1-AT) based deep eutectic propellants (DeEP) are examined. 1-AT is introduced as a neutral hydrogen bond donor (HBD) to EAN in order to form a eutectic mixture. The effective fragment potential (EFP) is used to examine the bonding interactions in the DeEP clusters. The resolution of the Identity (RI) approximated second order Møller-Plesset perturbation theory (RI-MP2) and coupled cluster theory (RI-CCSD(T)) are used to validate the EFP results. The EFP method predicts that there are significant polarization and charge transfer effects in the EAN:1-AT complexes, along with Coulombic, dispersion and exchange repulsion interactions. The EFP interaction energies are in good agreement with the RI-MP2 and RI-CCSD(T) results. The quasi-atomic orbital (QUAO) bonding and kinetic bond order (KBO) analyses are additionally used to develop a conceptual and semi-quantitative understanding of the H-bonding interactions as a function of the size of the system. The QUAO and KBO analyses suggest that the H-bonds in the examined clusters follow the characteristic hydrogen bonding three-center four electron interactions. The strongest H-bonding interactions between the (EAN)1:(1-AT)n and (EAN)2:(1-AT)n (n = 1-5) complexes are observed internally within EAN; that is, between the ethylammonium cation [EA]+ and the nitrate anion ([NO3]-). The weakest H-bonding interactions occur between [NO3]- and 1-AT. Consequently, the average strengths of the H-bonds within a given (EAN)x:(1-AT)n complex decrease as more 1-AT molecules are introduced into the EAN monomer and EAN dimer. The QUAO bonding analysis suggests that 1-AT in (EAN)x:(1-AT)n can act as both a HBD and a hydrogen bond acceptor simultaneously. It is observed that two 1-AT molecules can form H-bonds to each other. Although the KBOs that correspond to H-bonding interactions in [EA]+:1-AT, [NO3]-:1-AT and between two 1-AT molecules are weaker than the H-bonds in EAN, those weak H-bond networks with 1-AT could be important to form a stable DeEP.

Quantum Chemical Modeling of Propellant Degradation.

An ab initio quantum chemical approach for the modeling of propellant degradation is presented. Using state-of-the-art bonding analysis techniques and composite methods, a series of potential degradation reactions are devised for a sample hydroxyl-terminated-polybutadiene (HTPB) type solid fuel. By applying thermochemical procedures and isodesmic reactions, accurate thermochemical quantities are obtained using a modified G3 composite method based on the resolution of the identity. The calculated heats of formation for the different structures produced presents an ∼2 kcal/mol average error when compared against experimental values.

Revisiting the pseudo-supercritical path method: An improved formulation for the alchemical calculation of solid-liquid coexistence.

Alchemical free energy calculations via molecular dynamics have been applied to obtain thermodynamic properties related to solid-liquid equilibrium conditions, such as melting points. In recent years, the pseudo-supercritical path (PSCP) method has proved to be an important approach to melting point prediction due to its flexibility and applicability. In the present work, we propose improvements to the PSCP alchemical cycle to make it more compact and efficient through a concerted evaluation of different potential energies. The multistate Bennett acceptance ratio (MBAR) estimator was applied at all stages of the new cycle to provide greater accuracy and uniformity, which is essential concerning uncertainty calculations. In particular, for the multistate expansion stage from solid to liquid, we employed the MBAR estimator with a reduced energy function that allows affine transformations of coordinates. Free energy and mean derivative profiles were calculated at different cycle stages for argon, triazole, propenal, and the ionic liquid 1-ethyl-3-methyl-imidazolium hexafluorophosphate. Comparisons showed a better performance of the proposed method than the original PSCP cycle for systems with higher complexity, especially the ionic liquid. A detailed study of the expansion stage revealed that remapping the centers of mass of the molecules or ions is preferable to remapping the coordinates of each atom, yielding better overlap between adjacent states and improving the accuracy of the methodology.

Exchange-Mediated Transport in Battery Electrolytes: Ultrafast or Ultraslow?

Understanding the mechanisms of charge transport in batteries is important for the rational design of new electrolyte formulations. Persistent questions about ion transport mechanisms in battery electrolytes are often framed in terms of vehicular diffusion by persistent ion-solvent complexes versus structural diffusion through the breaking and reformation of ion-solvent contacts, i.e., solvent exchange events. Ultrafast two-dimensional (2D) IR spectroscopy can probe exchange processes directly via the evolution of the cross-peaks on picosecond time scales. However, vibrational energy transfer in the absence of solvent exchange gives rise to the same spectral signatures, hiding the desired processes. We employ 2D IR on solvent resonances of a mixture of acetonitrile isotopologues to differentiate chemical exchange and energy-transfer dynamics in a comprehensive series of Li+, Mg2+, Zn2+, Ca2+, and Ba2+ bis(trifluoromethylsulfonyl)imide electrolytes from the dilute to the superconcentrated regime. No exchange phenomena occur within at least 100 ps, regardless of the ion identity, salt concentration, and presence of water. All of the observed spectral dynamics originate from the intermolecular energy transfer. These results place the lower experimental boundary on the ion-solvent residence times to several hundred picoseconds, much slower than previously suggested. With the help of MD simulations and conductivity measurements on the Li+ and Zn2+ systems, we discuss these results as a continuum of vehicular and structural modalities that vary with concentration and emphasize the importance of collective electrolyte motions to ion transport. These results hold broadly applicable to many battery-relevant ions and solvents.

Structure of water-in-salt and water-in-bisalt electrolytes.

We report a systematic diffraction study of two "water-in-salt" electrolytes and a "water-in-bisalt" electrolyte combining high-energy X-ray diffraction (HEXRD) with polarized and unpolarized neutron diffraction (ND) on both H2O and D2O solutions. The measurements provide three independent combinations of correlations between the different pairs of atom types that reveal the short- and intermediate-range order in considerable detail. The ND interference functions show pronounced peaks around a scattering vector Q ∼ 0.5 Å-1 that change dramatically with composition, indicating significant rearrangements of the water network on a length scale around 12 Å. The experimental results are compared with two sets of Molecular Dynamics (MD) simulations, one including polarization effects and the other based on a non-polarizable force field. The two simulations reproduce the general shapes of the experimental structure factors and their changes with concentration, but differ in many detailed respects, suggesting ways in which their force fields might be modified to better represent the actual systems.

MoSDeF Cassandra: A complete Python interface for the Cassandra Monte Carlo software.

We introduce a new Python interface for the Cassandra Monte Carlo software, molecular simulation design framework (MoSDeF) Cassandra. MoSDeF Cassandra provides a simplified user interface, offers broader interoperability with other molecular simulation codes, enables the construction of programmatic and reproducible molecular simulation workflows, and builds the infrastructure necessary for high-throughput Monte Carlo studies. Many of the capabilities of MoSDeF Cassandra are enabled via tight integration with MoSDeF. We discuss the motivation and design of MoSDeF Cassandra and proceed to demonstrate both simple use-cases and more complex workflows, including adsorption in porous media and a combined molecular dynamics - Monte Carlo workflow for computing lateral diffusivity in graphene slit pores. The examples presented herein demonstrate how even relatively complex simulation workflows can be reduced to, at most, a few files of Python code that can be version-controlled and shared with other researchers. We believe this paradigm will enable more rapid research advances and represents the future of molecular simulations.

Machine Learning Directed Optimization of Classical Molecular Modeling Force Fields.

Accurate force fields are necessary for predictive molecular simulations. However, developing force fields that accurately reproduce experimental properties is challenging. Here, we present a machine learning directed, multiobjective optimization workflow for force field parametrization that evaluates millions of prospective force field parameter sets while requiring only a small fraction of them to be tested with molecular simulations. We demonstrate the generality of the approach and identify multiple low-error parameter sets for two distinct test cases: simulations of hydrofluorocarbon (HFC) vapor-liquid equilibrium (VLE) and an ammonium perchlorate (AP) crystal phase. We discuss the challenges and implications of our force field optimization workflow.

Computing the Liquidus of Binary Monatomic Salt Mixtures with Direct Simulation and Alchemical Free Energy Methods.

We describe and validate a free-energy-based method for computing the liquidus for binary solid-liquid phase diagrams in molecular simulations of monatomic salts. The method is demonstrated by calculating the liquidus for LiCl-KCl and MgCl2-KCl salt mixtures with the polarizable ion model (PIM). The free-energy-based method is cross-validated with direct coexistence simulations. Both techniques show excellent agreement with one another. Though the predictions of the PIM disagree with experiments, we use our free-energy-based approach to decouple the contributions of liquid mixture nonidealities and pure component solid-liquid equilibrium to the phase diagram. In both mixtures, the PIM accurately reproduces the liquid phase nonidealities but fails to predict the liquidus because it does not accurately predict the pure component melting temperature of LiCl or MgCl2.

Functionalized Phosphonium Cations Enable Zinc Metal Reversibility in Aqueous Electrolytes.

Aqueous rechargeable zinc metal batteries promise attractive advantages including safety, high volumetric energy density, and low cost; however, such benefits cannot be unlocked unless Zn reversibility meets stringent commercial viability. Herein, we report remarkable improvements on Zn reversibility in aqueous electrolytes when phosphonium-based cations are used to reshape interfacial structures and interphasial chemistries, particularly when their ligands contain an ether linkage. This novel aqueous electrolyte supports unprecedented Zn reversibility by showing dendrite-free Zn plating/stripping for over 6400 h at 0.5 mA cm-2 , or over 280 h at 2.5 mA cm-2 , with coulombic efficiency above 99 % even with 20 % Zn utilization per cycle. Excellent full cell performance is demonstrated with Na2 V6 O16 ⋅1.63 H2 O cathode, which cycles for 2000 times at 300 mA g-1 . The microscopic characterization and modeling identify the mechanism of unique interphase chemistry from phosphonium and its functionalities as the key factors responsible for dictating reversible Zn chemistry.

Role of Molecular Modeling in the Development of CO2-Reactive Ionic Liquids.

Ionic liquids (ILs) have gained considerable attention in recent years as CO2-reactive solvents that could be used to improve the economic efficiency of industrial-scale CO2 separations. Researchers have demonstrated that IL physical and chemical properties can be optimized for a given application through chemical functionalization of both cations and anions. The tunability of ILs presents both a great potential and a significant challenge due to the complex chemistries and the many ways in which ILs can be made to react with CO2. However, computer simulations have demonstrated great potential in understanding the behavior of ILs from the underlying molecular interactions. In the present review, we examine how computer simulations have aided in the design of ILs that chemically bind CO2. We present the historical development of CO2-reactive ILs while highlighting the insights provided by molecular modeling which aided in understanding IL behavior to further experimental findings. We also provide a brief discussion of simulations focused on ionic liquids that physically dissolve CO2. We conclude with a discussion of areas where simulations can yet be used to advance the current understanding of these complex systems and an outlook on the use of computer simulations in the design of optimal ILs for CO2 separations.

The role of cations in uranyl nanocluster association: a molecular dynamics study.

Actinyl ions can self-assemble in aqueous solution to form closed cage clusters ranging from 1.5 to 4.0 nm in diameter. The self-assembly, stability, and behavior of the nanoclusters depend on the nature of the aqueous environment, such as the pH and cations present. In this work, a classical force field for [(UO2)20(O2)30]20- (U20) peroxide nanoclusters in aqueous solution was developed from quantum-mechanical calculations. Using molecular dynamics simulations, the preferred binding sites of six cations (Li+, Na+, K+, Rb+, Cs+, and Ca2+) to the nanocluster were determined. Replica exchange molecular dynamics was used to equilibrate the structure and determine the equilibrium distribution of cations and water with respect to the nanocluster cage. In addition, the free energy barriers associated with cations entering the cluster were computed. Finally, the association of two cages was investigated by computing the free energy as a function of intercage distance. The free energy profiles reveal that the nanoclusters prefer to be associated when neutralized with divalent cations, but do not associate when neutralized with monovalent cations. This could explain the formation of tertiary structures observed experimentally.

Impact of anion shape on Li+ solvation and on transport properties for lithium-air batteries: a molecular dynamics study.

Lithium-air batteries have emerged as an interesting alternative for advanced energy storage devices. The complexity of such systems imposes great challenges. One of them resides in the selection of the lithium salt/solvent pair. Many electrolyte properties affect the operation of the batteries. Among these, the transport properties and structural features have a special place. Via molecular dynamics simulations, we have calculated solution viscosity, ionic diffusivities and conductivities, as well as structural information, for two different salts in dimethyl sulfoxide (DMSO): lithium hexafluorophosphate - LiPF6, and lithium pyrrolide - LiPyr, at different temperatures and salt molalities. We show that, despite similar ionic transport properties, Li+ solvation in the different salts is significantly different. Therefore, solutions with different solvation properties, which impact the overall battery performance, might present analogous ionic dynamics.

Structure and dynamics of the molten alkali-chloride salts from an X-ray, simulation, and rate theory perspective.

Molten salts are of great interest as alternative solvents, electrolytes, and heat transfer fluids in many emerging technologies. The macroscopic properties of molten salts are ultimately controlled by their structure and ion dynamics at the microscopic level and it is therefore vital to develop an understanding of these at the atomistic scale. Herein, we present high-energy X-ray scattering experiments combined with classical and ab initio molecular dynamics simulations to elucidate structural and dynamical correlations across the family of alkali-chlorides. Computed structure functions and transport properties are in reasonably good agreement with experiments providing confidence in our analysis of microscopic properties based on simulations. For these systems, we also survey different rate theory models of anion exchange dynamics in order to gain a more sophisticated understanding of the short-time correlations that are likely to influence transport properties such as conductivity. The anion exchange process occurs on the picoseconds time scale at 1100 K and the rate increases in the order KCl < NaCl < LiCl, which is in stark contrast to the ion pair dissociation trend in aqueous solutions. Consistent with the trend we observe for conductivity, the cationic size/mass, as well as other factors specific to each type of rate theory, appear to play important roles in the anion exchange rate trend.

SEM-Drude Model for the Accurate and Efficient Simulation of MgCl2-KCl Mixtures in the Condensed Phase.

There is a long history of models that to different extents reproduce structural and dynamical properties of high-temperature molten salts. Whereas rigid ion models can work fairly well for some of the monovalent salts, polarizability is fundamentally important when small divalent or multivalent cations are combined with significantly polarizable anions such as Cl- to form networked liquids that display a first sharp diffraction peak. There are excellent polarizable ion models (PIMs) for these systems, but there has been little success with the less expensive Core-Shell type models, which are often described as unwieldy or difficult to fit. In this article, we present the Sharma-Emerson-Margulis (SEM)-Drude model for MgCl2/KCl mixtures that with the same ingredients used in the latest and most accurate PIM models overcome the aforementioned obstacles at significantly less computational cost; structural and dynamical properties are for all practical purposes very similar to what we obtain from the PIM but typical simulations can be more than 30 times faster. This has allowed us not only to expand our recent studies on the temperature and composition dependence of intermediate range order in MgCl2/KCl mixtures but also to access transport properties that were simply too costly to properly sample in our recently published studies.

Effect of alkyl-group flexibility on the melting point of imidazolium-based ionic liquids.

The low melting point of room temperature ionic liquids is usually explained in terms of the presence of bulky, low-symmetry, and flexible ions, with the first two factors related to the lattice energy while an entropic effect is attributed to the latter. By means of molecular dynamics simulations, the melting points of 1-ethyl-3-methyl-imidazolium hexafluorophosphate and 1-decyl-3-methyl-imidazolium hexafluorophosphate were determined, and the effect of the molecular flexibility over the melting point was explicitly computed by restraining the rotation of dihedral angles in both the solid and the liquid phases. The rotational flexibility over the bond between the ring and the alkyl chain affects the relative ordering of the anions around the cations and results in substantial effects over both the enthalpy and the entropy of melting. For the other dihedral angles of the alkyl group, the contributions are predominantly entropic and an alternating behavior was found. The flexibility of some dihedral angles has negligible effects on the melting point, while others can lead to differences in the melting point as large as 20 K. This alternating behavior is rationalized by the different probabilities of conformation defects in the crystal.