We present a new type of large-scale computational screening approach for identifying promising candidate materials for solid state electrolytes for lithium ion batteries that is capable of screening all known lithium containing solids. To be useful for batteries, high performance solid state electrolyte materials must satisfy many requirements at once, an optimization that is difficult to perform experimentally or with computationally expensive ab initio techniques. We first screen 12 831 lithium containing crystalline solids for those with high structural and chemical stability, low electronic conductivity, and low cost. We then develop a data-driven ionic conductivity classification model using logistic regression for identifying which candidate structures are likely to exhibit fast lithium conduction based on experimental measurements reported in the literature. The screening reduces the list of candidate materials from 12 831 down to 21 structures that show promise as electrolytes, few of which have been examined experimentally. We discover that none of our simple atomistic descriptor functions alone provide predictive power for ionic conductivity, but a multi-descriptor model can exhibit a useful degree of predictive power. We also find that screening for structural stability, chemical stability and low electronic conductivity eliminates 92.2% of all Li-containing materials and screening for high ionic conductivity eliminates a further 93.3% of the remainder. Our screening utilizes structures and electronic information contained in the Materials Project database.

We propose a novel statistical learning framework for automatically and efficiently building reduced kinetic Monte Carlo (KMC) models of large-scale elementary reaction networks from data generated by a single or few molecular dynamics simulations (MD). Existing approaches for identifying species and reactions from molecular dynamics typically use bond length and duration criteria, where bond duration is a fixed parameter motivated by an understanding of bond vibrational frequencies. In contrast, we show that for highly reactive systems, bond duration should be a model parameter that is chosen to maximize the predictive power of the resulting statistical model. We demonstrate our method on a high temperature, high pressure system of reacting liquid methane, and show that the learned KMC model is able to extrapolate more than an order of magnitude in time for key molecules. Additionally, our KMC model of elementary reactions enables us to isolate the most important set of reactions governing the behavior of key molecules found in the MD simulation. We develop a new data-driven algorithm to reduce the chemical reaction network which can be solved either as an integer program or efficiently using L1 regularization, and compare our results with simple count-based reduction. For our liquid methane system, we discover that rare reactions do not play a significant role in the system, and find that less than 7% of the approximately 2000 reactions observed from molecular dynamics are necessary to reproduce the molecular concentration over time of methane. The framework described in this work paves the way towards a genomic approach to studying complex chemical systems, where expensive MD simulation data can be reused to contribute to an increasingly large and accurate genome of elementary reactions and rates.

Layered materials held together by weak interactions including van der Waals forces, such as graphite, have attracted interest for both technological applications and fundamental physics in their layered form and as an isolated single-layer. Only a few dozen single-layer van der Waals solids have been subject to considerable research focus, although there are likely to be many more that could have superior properties. To identify a broad spectrum of layered materials, we present a novel data mining algorithm that determines the dimensionality of weakly bonded subcomponents based on the atomic positions of bulk, three-dimensional crystal structures. By applying this algorithm to the Materials Project database of over 50,000 inorganic crystals, we identify 1173 two-dimensional layered materials and 487 materials that consist of weakly bonded one-dimensional molecular chains. This is an order of magnitude increase in the number of identified materials with most materials not known as two- or one-dimensional materials. Moreover, we discover 98 weakly bonded heterostructures of two-dimensional and one-dimensional subcomponents that are found within bulk materials, opening new possibilities for much-studied assembly of van der Waals heterostructures. Chemical families of materials, band gaps, and point groups for the materials identified in this work are presented. Point group and piezoelectricity in layered materials are also evaluated in single-layer forms. Three hundred and twenty-five of these materials are expected to have piezoelectric monolayers with a variety of forms of the piezoelectric tensor. This work significantly extends the scope of potential low-dimensional weakly bonded solids to be investigated.Keywords: data mining; Layered materials; piezoelectricity; two-dimensional materials; van der Waals heterostructures; van der Waals solids;

Quantum nuclear effects include zero-point energy for each vibrational mode and potentially significant deviations of the heat capacity from classical values. While these effects play important roles in shock-induced chemical reactions and phase transitions, they are absent in classical atomistic shock simulations. To address this shortcoming, we have studied the quantum nuclear effects for stishovite crystallization in shock-compressed fused silica by employing the quantum bath multiscale shock technique, which couples the shock simulation with a colored-noise Langevin thermostat. We find that this semiclassical approach gives shock temperatures as much as 7% higher than classical simulations near the onset of stishovite crystallization in silica. We have also studied the impact of this approach on the kinetics of stishovite crystallization and the position of high-pressure silica melt line. We further describe a systematic way of setting up the parameters for the quantum thermal bath and quantum bath multiscale shock techniques.

Two-dimensional monolayer materials are a highly anomalous class of materials under vigorous exploration. Mo- and W-dichalcogenides are especially unusual two-dimensional materials because they exhibit at least three different monolayer crystal structures with strongly differing electronic properties. This intriguing yet poorly understood feature, which is not present in graphene, may support monolayer phase engineering, phase change memory and other applications. However, knowledge of the relevant phase boundaries and how to engineer them is lacking. Here we show using alloy models and state-of-the-art density functional theory calculations that alloyed MoTe2–WTe2 monolayers support structural phase transitions, with phase transition temperatures tunable over a large range from 0 to 933 K. We map temperature–composition phase diagrams of alloys between pure MoTe2 and pure WTe2, and benchmark our methods to analogous experiments on bulk materials. Our results suggest applications for two-dimensional materials as phase change materials that may provide scale, flexibility, and energy consumption advantages.Keywords: cluster expansion; molybdenum ditelluride; phase diagram; phase transition; two-dimensional material;

We study the adsorption of some common atoms and molecules onto monolayer MoTe2 and the potential for adsorption to induce a structural phase change between the semiconducting 2H-based and metallic 1T′-based crystal structures of the monolayer. Using density functional theory with spin–orbit and van der Waals energy contributions, we determined energetically favorable adsorption positions and orientations on the two crystalline phases of monolayer MoTe2. We then obtained the formation energies for these adsorption reactions and found that atomic adsorption generally favors 1T′ metallic phases while molecular adsorption favors semiconducting 2H phases. The phase sensitivity of this material is due to a relatively small energy difference, approximately 31 meV per MoTe2 formula unit. We further find that the monolayer alloy MoxW1–xTe2 can exhibit some degree of molecular selectivity in phase changes, potentially providing the basis for molecular sensing applications.

We utilize reactive empirical bond order (REBO)-based interatomic potentials to explore the potential for the engineering of strain in monolayer materials using lithographically or otherwise patterned adatom adsorption. In the context of graphene, we discover that the monolayer strain results from a competition between the in-plane elasticity and out-of-plane relaxation deformations. For hydrogen adatoms on graphene, the strain outside the adsorption region vanishes due to out-of-plane relaxation deformations. Under some circumstances, an annular adsorption pattern generates homogeneous tensile strains of approximately 2% in graphene inside the adsorption region, approximately 30% of the strain in the adsorbed region. We find that an elliptical adsorption pattern produces strains of as large as 5%, close to the strain in the adsorbed region. Also, nonzero maximum shear strain (∼4%) can be introduced by the elliptical adsorption pattern. We find that an elastic plane stress model provides qualitative guidance for strain magnitudes and conditions under which strain-diminishing buckling can be avoided. We identify geometric conditions under which this effect has potential to be scaled to larger areas. Our results elucidate a method for strain engineering at the nanoscale in monolayer devices.

The symmetry properties of atomically thin boron nitride (BN) monolayers endow them with piezoelectric properties, whereas the bulk parent crystal of stacked BN layers is not piezoelectric. This suggests potential for unusual electromechanical properties in the few layer regime. In this work, we explore this regime and discover that a bilayer consisting of two BN monolayers exhibits a strong mechanical coupling between curvature and electric fields. Using a mechanical model with parameters obtained from density functional theory, we find that these bilayers amplify in-plane piezoelectric displacements by exceedingly large factors on the order of 103–104. We find that this type of electromechanical coupling is an emergent nanoscale property that occurs only for the case of two stacked BN monolayers.

A systematic comparison of atomistic modeling methods including density functional theory (DFT), the self-consistent charge density-functional tight-binding (SCC-DFTB), and ReaxFF is presented for simulating the initial stages of phenolic polymer pyrolysis. A phenolic polymer system is simulated for several hundred picoseconds within a temperature range of 2500 to 3500 K. The time evolution of major pyrolysis products including small-molecule species and char is examined. Two temperature zones are observed which demark cross-linking versus fragmentation. The dominant chemical products for all methods are similar, but the yields for each product differ. At 3500 K, DFTB overestimates CO production (300–400%) and underestimates free H (∼30%) and small CmHnO molecules (∼70%) compared with DFT. At 3500 K, ReaxFF underestimates free H (∼60%) and fused carbon rings (∼70%) relative to DFT. Heterocyclic oxygen-containing five- and six-membered carbon rings are observed at 2500 K. Formation mechanisms for H2O, CO, and char are discussed. Additional calculations using a semiclassical method for incorporating quantum nuclear energies of molecules were also performed. These results suggest that chemical equilibrium can be affected by quantum nuclear effects at temperatures of 2500 K and below. Pyrolysis reaction mechanisms and energetics are examined in detail in a companion manuscript.

Motivated by a search for electromechanical coupling in monolayer materials, we study graphene chemically modified by hydrogen adsorbed on one side and fluorine adsorbed on the other side. Such adsorption under experimental conditions can potentially lead to a variety of configurations of atoms on the surface. We perform an exhaustive evaluation of candidate configurations for two stoichiometries, C2HF and C4HF, and examine their electromechanical properties using density functional theory. While all configurations exhibit an e31 piezoelectric effect, the lowest energy configuration additionally exhibits an e11 effect. Therefore, both e31 and e11 piezoelectricity can potentially be engineered into nonpiezoelectric monolayer graphene, providing an avenue for monolithic integration of electronic and electromechanical devices in graphene monolayers for resonators, sensors, and nanoelectromechanical systems (NEMS).

We discovered that many of the commonly studied two-dimensional monolayer transition metal dichalcogenide (TMDC) nanoscale materials are piezoelectric, unlike their bulk parent crystals. On the macroscopic scale, piezoelectricity is widely used to achieve robust electromechanical coupling in a rich variety of sensors and actuators. Remarkably, our density-functional theory calculations of the piezoelectric coefficients of monolayer BN, MoS2, MoSe2, MoTe2, WS2, WSe2, and WTe2 reveal that some of these materials exhibit stronger piezoelectric coupling than traditionally employed bulk wurtzite structures. We find that the piezoelectric coefficients span more than 1 order of magnitude, and exhibit monotonic periodic trends. The discovery of this property in many two-dimensional materials enables active sensing, actuating, and new electronic components for nanoscale devices based on the familiar piezoelectric effect.Keywords: atomically thin materials; boron nitride; elastic coefficients; nanoelectromechanical systems (NEMS); piezoelectric coefficients; piezotronics; transition metal dichalcogenides (TMDCs);

The magnitude and role of electronic excitations in shocked energetic materials are studied theoretically using quantum molecular dynamics simulations. Focusing on the detonating primary explosive HN3 (hydrazoic acid), this work finds that the material transiently exhibits a high level of electronic excitation characterized by carrier densities in excess of 1021 cm–3, or one electronic excitation for every eight molecules. Electronic excitations enhance the kinetics of chemical decomposition by ∼30%. The electronic heat capacity has a minor impact on the temperatures exhibited, on the order of 100 K. These simulations are performed using the self-consistent charge density functional tight-binding method (SCC-DFTB) combined with a new modification of a multiscale computational scheme for simulation of the coupling between electrons and ions in shocked matter.

A methodology is described for atomistic simulations of shock-compressed materials that incorporates quantum nuclear effects on the fly. We introduce a modification of the multiscale shock technique (MSST) that couples to a quantum thermal bath described by a colored noise Langevin thermostat. The new approach, which we call QB-MSST, is of comparable computational cost to MSST and self-consistently incorporates quantum heat capacities and Bose–Einstein harmonic vibrational distributions. As a first test, we study shock-compressed methane using the ReaxFF potential. The Hugoniot curves predicted from the new approach are found comparable with existing experimental data. We find that the self-consistent nature of the method results in the onset of chemistry at 40% lower pressure on the shock Hugoniot than observed with classical molecular dynamics. The temperature shift associated with quantum heat capacity is determined to be the primary factor in this shift.

We discover that piezoelectric effects can be engineered into nonpiezoelectric graphene through the selective surface adsorption of atoms. Our calculations show that doping a single sheet of graphene with atoms on one side results in the generation of piezoelectricity by breaking inversion symmetry. Despite their 2D nature, piezoelectric magnitudes are found to be comparable to those in 3D piezoelectric materials. Our results elucidate a designer piezoelectric phenomenon, unique to the nanoscale, that has potential to bring dynamical control to nanoscale electromechanical devices.Keywords: adsorption; density functional theory; graphene; nanoelectromechanical systems (NEMS); piezoelectricity; two-dimensional materials

We propose a novel statistical learning framework for automatically and efficiently building reduced kinetic Monte Carlo (KMC) models of large-scale elementary reaction networks from data generated by a single or few molecular dynamics simulations (MD). Existing approaches for identifying species and reactions from molecular dynamics typically use bond length and duration criteria, where bond duration is a fixed parameter motivated by an understanding of bond vibrational frequencies. In contrast, we show that for highly reactive systems, bond duration should be a model parameter that is chosen to maximize the predictive power of the resulting statistical model. We demonstrate our method on a high temperature, high pressure system of reacting liquid methane, and show that the learned KMC model is able to extrapolate more than an order of magnitude in time for key molecules. Additionally, our KMC model of elementary reactions enables us to isolate the most important set of reactions governing the behavior of key molecules found in the MD simulation. We develop a new data-driven algorithm to reduce the chemical reaction network which can be solved either as an integer program or efficiently using L1 regularization, and compare our results with simple count-based reduction. For our liquid methane system, we discover that rare reactions do not play a significant role in the system, and find that less than 7% of the approximately 2000 reactions observed from molecular dynamics are necessary to reproduce the molecular concentration over time of methane. The framework described in this work paves the way towards a genomic approach to studying complex chemical systems, where expensive MD simulation data can be reused to contribute to an increasingly large and accurate genome of elementary reactions and rates.