Interpretable Performance Models for Energetic Materials using Parsimonious Neural Networks.

Predictive models for the performance of explosives and propellants are important for their design, optimization, and safety. Thermochemical codes can predict some of these properties from fundamental quantities such as density and formation energies that can be obtained from first principles. Models that are simpler to evaluate are desirable for efficient, rapid screening of material screening. In addition, interpretable models can provide insight into the physics and chemistry of these materials that could be useful to direct new synthesis. Current state-of-the-art performance models are based on either the parametrization of physics-based expressions or data-driven approaches with minimal interpretability. We use parsimonious neural networks (PNNs) to discover interpretable models for the specific impulse of propellants and detonation velocity and pressure for explosives using data collected from the open literature. A combination of evolutionary optimization with custom neural networks explores and trains models with objective functions that balance accuracy and complexity. For all three properties of interest, we find interpretable models that are Pareto optimal in the accuracy and simplicity space.

Effects of carbon concentration on the local atomic structure of amorphous GST.

Ge-Sb-Te (GST) alloys are leading phase-change materials for data storage due to the fast phase transition between amorphous and crystalline states. Ongoing research aims at improving the stability of the amorphous phase to improve retention. This can be accomplished by the introduction of carbon as a dopant to Ge2Sb2Te5, which is known to alter the short- and mid-range structure of the amorphous phase and form covalently bonded C clusters, both of which hinder crystallization. The relative importance of these processes as a function of C concentration is not known. We used molecular dynamics simulation based on density functional theory to study how carbon doping affects the atomic structure of GST-C. Carbon doping results in an increase in tetrahedral coordination, especially of Ge atoms, and this is known to stabilize the amorphous phase. We observe an unexpected, non-monotonous trend in the number of tetrahedral bonded Ge with the amount of carbon doping. Our simulations show an increase in the number of tetrahedral bonded Ge up to 5 at.% C, after which the number saturates and begins to decrease above 14 at.% C. The carbon atoms aggregate into clusters, mostly in the form of chains and graphene flakes, leaving less carbon to disrupt the GST matrix at higher carbon concentrations. Different degrees of carbon clustering can explain divergent experimental results for recrystallization temperature for carbon doped GST.

Design of Atomic Ordering in Mo2Nb2C3Tx MXenes for Hydrogen Evolution Electrocatalysis.

The need for novel materials for energy storage and generation calls for chemical control at the atomic scale in nanomaterials. Ordered double-transition-metal MXenes expanded the chemical diversity of the family of atomically layered 2D materials since their discovery in 2015. However, atomistic tunability of ordered MXenes to achieve ideal composition-property relationships has not been yet possible. In this study, we demonstrate the synthesis of Mo2+αNb2-αAlC3 MAX phases (0 ≤ α ≤ 0.3) and confirm the preferential ordering behavior of Mo and Nb in the outer and inner M layers, respectively, using density functional theory, Rietveld refinement, and electron microscopy methods. We also synthesize their 2D derivative Mo2+αNb2-αC3Tx MXenes and exemplify the effect of preferential ordering on their hydrogen evolution reaction electrocatalytic behavior. This study seeks to inspire further exploration of the ordered double-transition-metal MXene family and contribute composition-behavior tools toward application-driven design of 2D materials.

High-pressure and temperature neural network reactive force field for energetic materials.

Reactive force fields for molecular dynamics have enabled a wide range of studies in numerous material classes. These force fields are computationally inexpensive compared with electronic structure calculations and allow for simulations of millions of atoms. However, the accuracy of traditional force fields is limited by their functional forms, preventing continual refinement and improvement. Therefore, we develop a neural network-based reactive interatomic potential for the prediction of the mechanical, thermal, and chemical responses of energetic materials at extreme conditions. The training set is expanded in an automatic iterative approach and consists of various CHNO materials and their reactions under ambient and shock-loading conditions. This new potential shows improved accuracy over the current state-of-the-art force fields for a wide range of properties such as detonation performance, decomposition product formation, and vibrational spectra under ambient and shock-loading conditions.

A coarse-grain reactive model of RDX: Molecular resolution at the µm scale.

Predictive models for the thermal, chemical, and mechanical response of high explosives at extreme conditions are important for investigating their performance and safety. We introduce a particle-based, reactive model of 1,3,5-trinitro-1,3,5-triazinane (RDX) with molecular resolution utilizing generalized energy-conserving dissipative particle dynamics with reactions. The model is parameterized with respect to the data from atomistic molecular dynamics simulations as well as from quantum mechanical calculations, thus bridging atomic processes to the mesoscales, including microstructures and defects. It accurately captures the response of RDX under a range of thermal loading conditions compared to atomistic simulations. In addition, the Hugoniot response of the CG model in the overdriven regime reasonably matches atomistic simulations and experiments. Exploiting the model's high computational efficiency, we investigate mesoscale systems involving millions of molecules and characterize size-dependent criticality of hotspots in RDX. The combination of accuracy and computational efficiency of our reactive model provides a tool for investigation of mesoscale phenomena, such as the role of microstructures and defects in the shock-to-deflagration transition, through particle-based simulation.

Modeling environment-dependent atomic-level properties in complex-concentrated alloys.

Complex-concentrated-alloys (CCAs) are of interest for a range of applications due to a host of desirable properties, including high-temperature strength and tolerance to radiation damage. Their multi-principal component nature results in a vast number of possible atomic environments with the associated variability in chemistry and structure. This atomic-level variability is central to the unique properties of these alloys but makes their modeling challenging. We combine atomistic simulations using many body potentials with machine learning to develop predictive models of various atomic properties of CrFeCoNiCu-based CCAs: relaxed vacancy formation energy, atomic-level cohesive energy, pressure, and volume. A fingerprint of the local atomic environments is obtained combining invariants associated with the local atomic geometry and periodic-table information of the atoms involved. Importantly, all descriptors are based on the unrelaxed atomic structure; thus, they are computationally inexpensive to compute. This enables the incorporation of these models into macroscopic simulations. The models show good accuracy and we explore their ability to extrapolate to compositions and elements not used during training.

Predicted Reaction Mechanisms, Product Speciation, Kinetics, and Detonation Properties of the Insensitive Explosive 2,6-Diamino-3,5-dinitropyrazine-1-oxide (LLM-105).

2,6-Diamino-3,5-dinitropyrazine-1-oxide (LLM-105) is a relatively new and promising insensitive high-explosive (IHE) material that remains only partially characterized. IHEs are of interest for a range of applications and from a fundamental science standpoint, as the root causes behind insensitivity are poorly understood. We adopt a multitheory approach based on reactive molecular dynamic simulations performed with density functional theory, density functional tight-binding, and reactive force fields to characterize the reaction pathways, product speciation, reaction kinetics, and detonation performance of LLM-105. We compare and contrast these predictions to 1,3,5-triamino-2,4,6-trinitrobenzene (TATB), a prototypical IHE, and 1,3,5,7-tetranitro-1,3,5,7-tetrazoctane (HMX), a more sensitive and higher performance material. The combination of different predictive models allows access to processes operative on progressively longer timescales while providing benchmarks for assessing uncertainties in the predictions. We find that the early reaction pathways of LLM-105 decomposition are extremely similar to TATB; they involve intra- and intermolecular hydrogen transfer. Additionally, the detonation performance of LLM-105 falls between that of TATB and HMX. We find agreement between predictive models for first-step reaction pathways but significant differences in final product formations. Predictions of detonation performance result in a wide range of values, and one-step kinetic parameters show the similar reaction rates at high temperatures for three out of four models considered.

Nitromethane Decomposition via Automated Reaction Discovery and an Ab Initio Corrected Kinetic Model.

We explore the systematic construction of kinetic models from in silico reaction data for the decomposition of nitromethane. Our models are constructed in a computationally affordable manner by using reactions discovered through accelerated molecular dynamics simulations using the ReaxFF reactive force field. The reaction paths are then optimized to determine reaction rate parameters. We introduce a reaction barrier correction scheme that combines accurate thermochemical data from density functional theory with ReaxFF minimal energy paths. We validate our models across different thermodynamic regimes, showing predictions of gas phase CO and NO concentrations and high-pressure induction times that are similar to experimental data. The kinetic models are analyzed to find fundamental decomposition reactions in different thermodynamic regimes.

Novel Mode of Noncrystallographic Branching in the Initial Stages of Polymer Fibril Growth.

Spherulites are the most ubiquitous of polycrystalline microstructure of polymers; they develop under a wide range of conditions by the subsequent branching of crystalline lamella that results in an overall spherical shape. Despite significant efforts over decades, the mechanisms behind branching remain unclear. Molecular dynamics simulations in polyethylene reveal the molecular-level origin of noncrystallographic branching and the initial formation of fibrils. We find that the growth of crystalline lamella by reeling in and folding of polymer chains causes surprisingly large local deformation which, in turn, aligns the chains in the neighboring undercooled liquid. Thus, subsidiary grains nucleate with preferred orientations resulting in fibril growth with branching at small angles, consistent with those observed experimentally.

Unsupervised Learning-Based Multiscale Model of Thermochemistry in 1,3,5-Trinitro-1,3,5-triazinane (RDX).

The response of high-energy-density materials to thermal or mechanical insults involves coupled thermal, mechanical, and chemical processes with disparate temporal and spatial scales that no single model can capture. Therefore, we developed a multiscale model for 1,3,5-trinitro-1,3,5-triazinane, RDX, where a continuum description is informed by reactive and nonreactive molecular dynamics (MD) simulations to describe chemical reactions and thermal transport. Reactive MD simulations under homogeneous isothermal and adiabatic conditions are used to develop a reduced-order chemical kinetics model. Coarse graining is done using unsupervised learning via non-negative matrix factorization. Importantly, the components resulting from the analysis can be interpreted as reactants, intermediates, and products, which allows us to write kinetics equations for their evolution. The kinetics parameters are obtained from isothermal MD simulations over a wide temperature range, 1200-3000 K, and the heat evolved is calibrated from adiabatic simulations. We validate the continuum model against MD simulations by comparing the evolution of a cylindrical hotspot 10 nm in diameter. We find excellent agreement in the time evolution of the hotspot temperature fields both in cases where quenching is observed and at higher temperatures for which the hotspot transitions into a deflagration wave. The validated continuum model is then used to assess the criticality of hotspots involving scales beyond the reach of atomistic simulations that are relevant to detonation initiation.

Investigation of structural ordering in network forming ionic liquids: A molecular dynamics study.

Molecular dynamics simulations reveal anomalous short- and medium-range ordering with increasing temperature in network-forming ionic liquids (NIL) consisting of alkyl-diammonium cations with long side chains of 6 carbon atoms and citrate anions (NIL 5-6). This effect is weaker, and only a short-range order is observed in equivalent systems with side chains shortened to 3 C atoms (NIL 5-3). The short-range ordering can be attributed to volume expansion during heating, but the intermediate range order requires volume expansion as well as an increase in temperature. We find that the cross (cation-anion) interactions are the major contributors to the observed trend and the development of complex 3D correlations in the two-particle correlation functions. The simulations suggest that the above phenomenon can be correlated to local trapping of cation molecules in a variety of configurations at lower temperatures where molecular shape distributions show great variability; as temperature increases, the distribution of molecular radii of gyration becomes narrower, enabling the increased ordering.

Role of electronic thermal transport in amorphous metal recrystallization: A molecular dynamics study.

Recrystallization of glasses is important in a wide range of applications including electronics and reactive materials. Molecular dynamics (MD) has been used to provide an atomic picture of this process, but prior work has neglected the thermal transport role of electrons, the dominant thermal carrier in metallic systems. We characterize the role of electronic thermal conductivity on the velocity of recrystallization in Ni using MD coupled to a continuum description of electronic thermal transport via a two-temperature model. Our simulations show that for strong enough coupling between electrons and ions, the increased thermal conductivity removes the heat from the exothermic recrystallization process more efficiently, leading to a lower effective temperature at the recrystallization front and, consequently, lower propagation velocity. We characterize how electron-phonon coupling strength and system size affect front propagation velocity. Interestingly, we find that initial recrystallization velocity increases with decreasing system size due to higher overall temperatures. Overall, we show that a more accurate description of thermal transport due to the incorporation of electrons results in better agreement with experiments.

Molecular modeling of the microstructure evolution during carbon fiber processing.

The rational design of carbon fibers with desired properties requires quantitative relationships between the processing conditions, microstructure, and resulting properties. We developed a molecular model that combines kinetic Monte Carlo and molecular dynamics techniques to predict the microstructure evolution during the processes of carbonization and graphitization of polyacrylonitrile (PAN)-based carbon fibers. The model accurately predicts the cross-sectional microstructure of the fibers with the molecular structure of the stabilized PAN fibers and physics-based chemical reaction rates as the only inputs. The resulting structures exhibit key features observed in electron microcopy studies such as curved graphitic sheets and hairpin structures. In addition, computed X-ray diffraction patterns are in good agreement with experiments. We predict the transverse moduli of the resulting fibers between 1 GPa and 5 GPa, in good agreement with experimental results for high modulus fibers and slightly lower than those of high-strength fibers. The transverse modulus is governed by sliding between graphitic sheets, and the relatively low value for the predicted microstructures can be attributed to their perfect longitudinal texture. Finally, the simulations provide insight into the relationships between chemical kinetics and the final microstructure; we observe that high reaction rates result in porous structures with lower moduli.

Atomic origin of ultrafast resistance switching in nanoscale electrometallization cells.

Nanoscale resistance-switching cells that operate via the electrochemical formation and disruption of metallic filaments that bridge two electrodes are among the most promising devices for post-CMOS electronics. Despite their importance, the mechanisms that govern their remarkable properties are not fully understood, especially for nanoscale devices operating at ultrafast rates, limiting our ability to assess the ultimate performance and scalability of this technology. We present the first atomistic simulations of the operation of conductive bridging cells using reactive molecular dynamics with a charge equilibration method extended to describe electrochemical reactions. The simulations predict the ultrafast switching observed in these devices, with timescales ranging from hundreds of picoseconds to a few nanoseconds for devices consisting of Cu active electrodes and amorphous silica dielectrics and with dimensions corresponding to their scaling limit (cross-sections below 10 nm). We find that single-atom-chain bridges often form during device operation but that they are metastable, with lifetimes below a nanosecond. The formation of stable filaments involves the aggregation of ions into small metallic clusters, followed by a progressive chemical reduction as they become connected to the cathode. Contrary to observations in larger cells, the nanoscale conductive bridges often lack crystalline order. An atomic-level mechanistic understanding of the switching process provides guidelines for materials optimization for such applications and the quantitative predictions over an ensemble of devices provide insight into their ultimate scaling and performance.

The dynamics of copper intercalated molybdenum ditelluride.

Layered transition metal dichalcogenides are emerging as key materials in nanoelectronics and energy applications. Predictive models to understand their growth, thermomechanical properties, and interaction with metals are needed in order to accelerate their incorporation into commercial products. Interatomic potentials enable large-scale atomistic simulations connecting first principle methods and devices. We present a ReaxFF reactive force field to describe molybdenum ditelluride and its interactions with copper. We optimized the force field parameters to describe the energetics, atomic charges, and mechanical properties of (i) layered MoTe2, Mo, and Cu in various phases, (ii) the intercalation of Cu atoms and small clusters within the van der Waals gap of MoTe2, and (iii) bond dissociation curves. The training set consists of an extensive set of first principles calculations computed using density functional theory (DFT). We validate the force field via the prediction of the adhesion of a single layer MoTe2 on a Cu(111) surface and find good agreement with DFT results not used in the training set. We characterized the mobility of the Cu ions intercalated into MoTe2 under the presence of an external electric field via finite temperature molecular dynamics simulations. The results show a significant increase in drift velocity for electric fields of approximately 0.4 V/Å and that mobility increases with Cu ion concentration.

Voltage equilibration for reactive atomistic simulations of electrochemical processes.

We introduce electrochemical dynamics with implicit degrees of freedom (EChemDID), a model to describe electrochemical driving force in reactive molecular dynamics simulations. The method describes the equilibration of external electrochemical potentials (voltage) within metallic structures and their effect on the self-consistent partial atomic charges used in reactive molecular dynamics. An additional variable assigned to each atom denotes the local potential in its vicinity and we use fictitious, but computationally convenient, dynamics to describe its equilibration within connected metallic structures on-the-fly during the molecular dynamics simulation. This local electrostatic potential is used to dynamically modify the atomic electronegativities used to compute partial atomic changes via charge equilibration. Validation tests show that the method provides an accurate description of the electric fields generated by the applied voltage and the driving force for electrochemical reactions. We demonstrate EChemDID via simulations of the operation of electrochemical metallization cells. The simulations predict the switching of the device between a high-resistance to a low-resistance state as a conductive metallic bridge is formed and resistive currents that can be compared with experimental measurements. In addition to applications in nanoelectronics, EChemDID could be useful to model electrochemical energy conversion devices.

Mesoscale simulations of shockwave energy dissipation via chemical reactions.

We use a particle-based mesoscale model that incorporates chemical reactions at a coarse-grained level to study the response of materials that undergo volume-reducing chemical reactions under shockwave-loading conditions. We find that such chemical reactions can attenuate the shockwave and characterize how the parameters of the chemical model affect this behavior. The simulations show that the magnitude of the volume collapse and velocity at which the chemistry propagates are critical to weaken the shock, whereas the energetics in the reactions play only a minor role. Shock loading results in transient states where the material is away from local equilibrium and, interestingly, chemical reactions can nucleate under such non-equilibrium states. Thus, the timescales for equilibration between the various degrees of freedom in the material affect the shock-induced chemistry and its ability to attenuate the propagating shock.

Role of direct electron-phonon coupling across metal-semiconductor interfaces in thermal transport via molecular dynamics.

Motivated by significant interest in metal-semiconductor and metal-insulator interfaces and superlattices for energy conversion applications, we developed a molecular dynamics-based model that captures the thermal transport role of conduction electrons in metals and heat transport across these types of interface. Key features of our model, denoted eleDID (electronic version of dynamics with implicit degrees of freedom), are the natural description of interfaces and free surfaces and the ability to control the spatial extent of electron-phonon (e-ph) coupling. Non-local e-ph coupling enables the energy of conduction electrons to be transferred directly to the semiconductor/insulator phonons (as opposed to having to first couple to the phonons in the metal). We characterize the effect of the spatial e-ph coupling range on interface resistance by simulating heat transport through a metal-semiconductor interface to mimic the conditions of ultrafast laser heating experiments. Direct energy transfer from the conduction electrons to the semiconductor phonons not only decreases interfacial resistance but also increases the ballistic transport behavior in the semiconductor layer. These results provide new insight for experiments designed to characterize e-ph coupling and thermal transport at the metal-semiconductor/insulator interfaces.

Engineering curvature in graphene ribbons using ultrathin polymer films.

We propose a method to induce curvature in graphene nanoribbons in a controlled manner using an ultrathin thermoset polymer in a bimaterial strip setup and test it via molecular dynamics (MD) simulations. Continuum mechanics shows that curvature develops to release the residual stress caused by the chemical and thermal shrinkage of the polymer during processing and that this curvature increases with decreasing film thickness; however, significant deformation is only achieved for ultrathin polymer films. Quite surprisingly, explicit MD simulations of the curing and annealing processes show that the predicted trend not just continues down to film thicknesses of 1-2 nm but that the curvature development is enhanced significantly in such ultrathin films due to surface tension effects. This combination of effects leads to very large curvatures of over 0.14 nm(-1) that can be tuned via film thickness. This provides a new avenue to engineer curvature and, thus, electromagnetic properties of graphene.