Boris Kozinsky

The demand for high-energy and high-power energy storage devices motivates the search for electrode materials with both high capacity and fast ion transport. One class of materials that could achieve such performance are oxides containing crystallographic shear (CS) planes. Here, we compare the structural dynamics of tungsten trioxide (WO 3 ) and its oxygen deficient, CS Magnéli phase (WO 2.9 ) during electrochemical insertion of H + and Li + ions using a combined experimental and computational study. We found that WO 3 inserts more H + per formula unit than WO 2.9 yet operando electrochemical atomic force microscopy shows more deformation in WO 2.9 than WO 3 per inserted H + . In contrast, WO 2.9 accommodates ∼0.2 more Li + per formula unit than WO 3 and has higher Li + diffusion and better rate capability. Operando electrochemical X-ray diffraction shows that Li + insertion into WO 2.9 leads to lattice contraction and 5 % volume change up to Li 0.6 WO 2.9 followed by a zero-strain region up to Li 1.4 WO 2.9 . We find that the presence of CS planes, and its effect on octahedral tilting, lead to different outcomes depending on the inserting ion: while octahedral tilting and lack of CS planes promote H + insertion into WO 3 , their absence in WO 2.9 favor Li + insertion. We propose that the presence of CS planes impart structural rigidity, enabling higher capacity, improved rate capability, and enhanced cyclability during Li + insertion but remove favorable bridging oxygen sites for H + insertion.

The recent capability to use THz pulses to control the nuclear quantum degrees of freedom in crystals has opened promising avenues for the advanced manipulation of material properties. While numerical approaches exist for studying the time evolution of the quantum nuclear density matrix, an interpretable analytical framework to explicitly analyze the influence of quantum fluctuations on nuclear dynamics remains lacking. In this work, we present an analytical quantum theory of nonlinear phononics. This framework is a basis for deriving models of realistic materials, allowing for exact solutions of the nuclear time evolution with full consideration of quantum fluctuations. This is accomplished by treating for all possible third- and fourth-order phonon couplings and expressing forces as analytic functions of such fluctuations. We provide an analytic proof that, in general, a strong pulse displacing a phonon mode from equilibrium induces the quenching, or squeezing, of its quantum lattice fluctuations. This finding, which establishes a systematization of the mechanism observed in Ref. 1, introduces a new paradigm in nonlinear phononics, harnessing this cooling effect to drive symmetry breaking in quantum paraelectric materials.

IL-doped alkali-metal salts, commonly known as salt-in-ionic liquids (SiILs), have drawn attention over recent years as electrolytes in batteries. SiILs are a class of highly concentrated, strongly correlated, and asymmetric electrolytes, with low volatility, low flammability, and extraordinary thermal and chemical stability. It has been reported that the transference numbers of alkali metals are negative in Li-based SiILs when the mole fraction of Li+ is low. This behavior can be explained by the formation of negatively charged ionic clusters composed of alkali metal cations and anions. On the other hand, MD simulations have also suggested a different ionic arrangement under high concentrations of salts, where a percolated ionic network can form, leading to positive transference numbers of metal cations. In this work, we have focused on the interfacial structure and behavior of Na-based SiILs on charged surfaces. Sodium trifluoromethanesulfonimide (NaTFSI) and 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TFSI]) have been selected as the alkali-metal salt and the IL, respectively, and systematically mixed at different molar ratios. An extended Surface Forces Apparatus (eSFA) and Atomic Force Microscopy (AFM) were employed for probing structural features such as the arrangement and layering of ions at charged surfaces, while Wide Angle X-ray Scattering (WAXS) was used to explore the bulk structure of the SiILs. The effect of water as an additive to SiILs, so called water-in-salt-in-ionic liquids (WiSILs) has also been investigated, as an additional component to tune nanostructure ionic mobility, and interfacial interactions. By understanding the properties of electric double layer (EDL) and the formation of solid electrolyte interface (SEI), we aim to provide insights to design materials for the next-generation batteries beyond Li-based chemistries to improve energy resilience.

Light-matter dynamics in topological quantum materials enables ultralow-power, ultrafast devices. A challenge is simulating multiple field and particle equations for light, electrons, and atoms over vast spatiotemporal scales on Exaflop/s computers with increased heterogeneity and low-precision focus. We present a paradigm shift that solves the multiscale/multiphysics/heterogeneity challenge harnessing hardware heterogeneity and low-precision arithmetic. Divide-conquer-recombine algorithms divide the problem into not only spatial but also physical subproblems of small dynamic ranges and minimal mutual information, which are mapped onto best-characteristics-matching hardware units, while metamodel-space algebra minimizes communication and precision requirements. Using 60,000 GPUs of Aurora, DC-MESH (divide-and-conquer Maxwell-Ehrenfest-surface hopping) and XS-NNQMD (excited-state neural-network quantum molecular dynamics) modules of MLMD (multiscale light-matter dynamics) software were 152- and 3,780-times faster than the state-of-the-art for 15.4 million-electron and 1.23 trillion-atom PbTiO3 material, achieving 1.87 EFLOP/s for the former. This enabled the first study of light-induced switching of topological superlattices for future ferroelectric ‘topotronics’.

Ultrafast spectroscopy paved the way for probing transient states of matter produced through photoexcitation. The microscopic processes governing the formation of these states remain largely unknown, due to the inherent challenges in accessing the microscopic behavior of materials, which is strongly influenced by nuclear quantum effects. Here, we perform simulations of quantum nuclear dynamics in the nonequilibrium regime, extending beyond the current state of the art. By combining first-principles simulations with machine learning, we unveil the complex quantum dynamics of SrTiO 3 emerging after terahertz laser pumping. We disclose the microscopic origin of the phonon upconversion, observed experimentally but not fully understood, and quantify the lifetime of the out-of-equilibrium motion, which is beyond the reach of the state-of-the-art simplified models. Crucially, our simulations predict that terahertz pump pulses can generate persistent out-of-equilibrium stress capable of inducing polar order. This work lays the foundation for systematic explorations of complex quantum materials sensitive to photoexcitation.

Light-matter dynamics in topological quantum materials enables ultralow-power, ultrafast devices. A challenge is simulating multiple field and particle equations for light, electrons, and atoms over vast spatiotemporal scales on Exaflop/s computers with increased heterogeneity and low-precision focus. We present a paradigm shift that solves the multiscale/multiphysics/heterogeneity challenge harnessing hardware heterogeneity and low-precision arithmetic. Divide-conquer-recombine algorithms divide the problem into not only spatial but also physical subproblems of small dynamic ranges and minimal mutual information, which are mapped onto best-characteristics-matching hardware units, while metamodel-space algebra minimizes communication and precision requirements. Using 60,000 GPUs of Aurora, DC-MESH (divide-and-conquer Maxwell-Ehrenfest-surface hopping) and XS-NNQMD (excited-state neural-network quantum molecular dynamics) modules of MLMD (multiscale light-matter dynamics) software were 152- and 3,780-times faster than the state-of-the-art for 15.4 million-electron and 1.23 trillion-atom PbTiO3 material, achieving 1.87 EFLOP/s for the former. This enabled the first study of light-induced switching of topological superlattices for future ferroelectric 'topotronics'.

The performance of bimetallic dilute alloy catalysts is largely determined by the size of minority metal ensembles on the nanoparticle surface. By analyzing the synthesis of catalysts comprising Pd8Au92 nanoparticles supported on silica using surface-sensitive techniques, we report that whether Pd overgrowth occurs before or after Au nanoparticle deposition onto the support controls the surface Pd ensemble size and abundance. These differences in Pd ensembles influence catalytic reactivity in H2–D2 isotope exchange and benzaldehyde hydrogenation, which, in correlation with theoretical calculations, is used to elucidate the active site(s) in each reaction. To clarify how the synthetic sequence controls the formation of Pd ensembles, we combine numerical wetting calculations and molecular dynamics simulations (with a machine-learned force field) to visualize Pd deposition and migration on the nanoparticle surface, respectively. Our results suggest that the nanoparticle–support interface restricts nanoparticle accessibility to Pd deposition, which consequently controls the Pd ensemble size, illustrating the critical role of nanoscale wetting phenomena during bimetallic catalyst preparation. The catalytic performance of dilute Pd-in-Au alloys depends on the Pd ensemble size on the bimetallic nanoparticle surface. Here the authors reveal how Pd ensemble formation on Au nanoparticles depends on the deposition sequence and nanoparticle–support wetting interactions, consequently affecting reactivity.

Understanding and controlling the atomistic-level reactions governing the formation of the solid-electrolyte interphase (SEI) is crucial for the viability of next-generation solid state batteries. However, challenges persist due to difficulties in experimentally characterizing buried interfaces and limits in simulation speed and accuracy. We conduct large-scale explicit reactive simulations with quantum accuracy for a symmetric battery cell, {\symcell}, enabled by active learning and deep equivariant neural network interatomic potentials. To automatically characterize the coupled reactions and interdiffusion at the interface, we formulate and use unsupervised classification techniques based on clustering in the space of local atomic environments. Our analysis reveals the formation of a previously unreported crystalline disordered phase, Li$_2$S$_{0.72}$P$_{0.14}$Cl$_{0.14}$, in the SEI, that evaded previous predictions based purely on thermodynamics, underscoring the importance of explicit modeling of full reaction and transport kinetics. Our simulations agree with and explain experimental observations of the SEI formations and elucidate the Li creep mechanisms, critical to dendrite initiation, characterized by significant Li motion along the interface. Our approach is to crease a digital twin from first principles, without adjustable parameters fitted to experiment. As such, it offers capabilities to gain insights into atomistic dynamics governing complex heterogeneous processes in solid-state synthesis and electrochemistry.

${\mathrm{YBa}}_{2}{\mathrm{Cu}}_{3}{\mathrm{O}}_{7\ensuremath{-}\ensuremath{\delta}}$ (YBCO) has favorable macroscopic superconducting properties of ${T}_{\mathrm{c}}$ up to 93 K and ${H}_{c2}$ up to 150 T. However, its nanoscale electronic structure remains mysterious because bulk-like electronic properties are not preserved near the surface of cleaved samples for easy access by local or surface-sensitive probes. It has been hypothesized that Ca-doping at the Y site could induce an alternate cleavage plane that mitigates this issue. We use scanning tunneling microscopy (STM) to study both Ca-free and 10% Ca-doped YBCO . We provide experimental evidence, supported by density functional theory (DFT) calculations, that the Ca-doped samples do indeed cleave on an alternate plane, yielding a spatially disordered partial (Y, Ca) surface. On this surface, we image a superconducting gap with an average value of 26 meV $\ifmmode\pm\else\textpm\fi{}$ 4 meV and characteristic length scale of around 1 nm, similar to Bi-based high-${T}_{\mathrm{c}}$ cuprates, and the first map of gap inhomogeneity in YBCO.

Similar to other perovskites in its family, $\mathrm{SrTiO_{3}}$ exhibits a significant softening of the ferroelectric mode with decreasing temperature, a behavior that typically heralds the onset of a ferroelectric transition. However, this material remains paraelectric down to 0K due to quantum fluctuations that prevent stabilization of the ferroelectric minimum. This work shows that in the strong out-of-equilibrium regime induced by resonant mid-IR pulses, quantum fluctuations can be suppressed, inducing a ferroelectric transition in $\mathrm{SrTiO_{3}}$ that is otherwise impossible at equilibrium. The appearance of a metastable state, that is distinct from the conventional ground state, is the first demonstration of how it is possible to leverage and control quantum fluctuations with pulsed light to qualitatively alter the free energy landscape of a quantum system. We predict the conditions and system parameters under which the induced non-equilibrium state can be long-lived and metastable. In providing a quantitative description, based on first principles machine learned potential energy surface, we explain recent experimental observations of light-induced ferroelectric transition in this material. Our results indicate a general nonequilibrium route to light-induced ferroelectric order in oxide perovskites near a ferroelectric instability.

In this article, in Equations (18) and (20), open curly brackets were inserted within round brackets and has been removed. The original article has been corrected.

Predicting response of materials to external stimuli is a primary objective of computational materials science. However, current methods are limited to small-scale simulations due to the unfavorable scaling of computational costs. Here, we implement an equivariant machine-learning framework where response properties stem from exact differential relationships between a generalized potential function and applied external fields. Focusing on responses to electric fields, the method predicts electric enthalpy, forces, polarization, Born charges, and polarizability within a unified model enforcing the full set of exact physical constraints, symmetries and conservation laws. Through application to α−SiO2, we demonstrate that our approach can be used for predicting vibrational and dielectric properties of materials, and for conducting large-scale dynamics under arbitrary electric fields at unprecedented accuracy and scale. We apply our method to ferroelectric BaTiO3 and capture the temperature dependence, frequency dependence, and time evolution of the ferroelectric hysteresis, revealing the underlying intrinsic mechanisms of nucleation and growth that govern ferroelectric domain switching. The authors introduce a machine-learning framework that predicts how materials respond to electric fields with quantum-level accuracy, capturing vibrational, dielectric, and ferroelectric behaviors up to the million-atom scale.

Machine learning interatomic potentials, particularly those based on deep equivariant neural networks, have demonstrated state-of-the-art accuracy and computational efficiency in atomistic modeling tasks like molecular dynamics and high-throughput screening. The size of datasets and demands of downstream workflows are growing rapidly, making robust and scalable software essential. This work presents a major overhaul of the NequIP framework focusing on multi-node parallelism, computational performance, and extensibility. The redesigned framework supports distributed training on large datasets and removes barriers preventing full utilization of the PyTorch 2.0 compiler at train time. We demonstrate this acceleration in a case study by training Allegro models on the SPICE 2 dataset of organic molecular systems. For inference, we introduce the first end-to-end infrastructure that uses the PyTorch Ahead-of-Time Inductor compiler for machine learning interatomic potentials. Additionally, we implement a custom kernel for the Allegro model's most expensive operation, the tensor product. Together, these advancements speed up molecular dynamics calculations on system sizes of practical relevance by up to a factor of 18.

Abstract The rapid advancements in ultrafast laser technology have paved the way for pumping and probing the out-of-equilibrium dynamics of nuclei in crystals. However, interpreting these experiments is extremely challenging due to the complex nonlinear responses in systems where lattice excitations interact, particularly in crystals composed of light atoms or at low temperatures where the quantum nature of ions becomes significant. In this work, we address the nonequilibrium quantum ionic dynamics from first principles. Our approach is general and can be applied to simulate any crystal, in combination with a first-principles treatment of electrons or external machine-learning potentials. It is implemented by leveraging the nonequilibrium time-dependent self-consistent harmonic approximation (TD-SCHA), with a stable, energy-conserving, correlated stochastic integration scheme that achieves an accuracy of $${\mathcal{O}}(d{t}^{3})$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mi>O</mml:mi> <mml:mrow> <mml:mo>(</mml:mo> <mml:mrow> <mml:mi>d</mml:mi> <mml:msup> <mml:mrow> <mml:mi>t</mml:mi> </mml:mrow> <mml:mrow> <mml:mn>3</mml:mn> </mml:mrow> </mml:msup> </mml:mrow> <mml:mo>)</mml:mo> </mml:mrow> </mml:mrow> </mml:math> . We benchmark the method with both a simple one-dimensional model to test its accuracy and a realistic 40-atom cell of SrTiO 3 under THz laser pump, paving the way for simulations of ultrafast THz-Xray pump-probe spectroscopy like those performed in synchrotron facilities.

In solid acid solid electrolytes CsH$_2$PO$_4$ and CsHSO$_4$, mechanisms of fast proton conduction have long been debated and attributed to either local proton hopping or polyanion rotation. However, the precise role of polyanion rotation and its interplay with proton hopping remained unclear. Nanosecond-scale molecular dynamics simulations, driven by equivariant neural network force fields, reveal a nuanced proton slingshot mechanism: protons are initially carried by rotating polyanions, followed by O$-$H bond reorientation, and the combined motion enables long-range jumps. This challenges the conventional revolving paddlewheel model and reveals significant independent proton motion that is assisted by limited rotations. Despite structural similarities, we identify qualitative differences in transport mechanisms between CsH$_2$PO$_4$ and CsHSO$_4$, caused by different proton concentrations. CsH$_2$PO$_4$ exhibits two distinct rates of rotational motions with different activation energies, contrasting with CsHSO$_4$'s single-rate behavior. The higher proton concentration in CsH$_2$PO$_4$ correlates with frustrated PO$_4$ polyanion orientations and slower rotations compared to SO$_4$ in CsHSO$_4$. Additionally, we reveal a correlation between O-sharing and proton transport in CsH$_2$PO$_4$, a unique feature due to extra proton per polyanion compared to CsHSO$_4$. Our findings suggest that reducing proton concentration could accelerate rotations and enhance conductivity. This work provides a unified framework for understanding and optimizing ionic mobility in solid-acid compounds, offering new insights into the interplay between proton hopping and disordered dynamics in polyanion rotation.

Environmentally benign, nontoxic electrolytes with combinatorial design spaces are excellent candidates for green solvents, green leaching agents, and carbon capture sources. We examine ethaline, a 2:1 molar ratio of ethylene glycol and choline chloride. Despite its touted green credentials, we find partial decomposition of ethaline into toxic chloromethane and dimethylaminoethanol at room temperature, limiting its sustainable advantage. We experimentally characterize these decomposition products and computationally develop a general, quantum-chemically accurate workflow to understand its decomposition. We find that fluctuations in the hydrogen bonds bind chloride near reaction sites, initiating the reaction between choline cations and chloride anions. The strong hydrogen bonds formed in ethaline are resistant to thermal perturbations, entrapping Cl in high-energy states and promoting the uphill reaction. In the design of stable green solvents, we recommend detailed evaluation of the hydrogen-bonding potential energy landscape as a key consideration for generating stable solvent mixtures.

Silicon carbide (SiC) is an important technological material, but its high-temperature phase diagram has remained unclear due to conflicting experimental results about congruent versus incongruent melting. Here, we employ large-scale machine learning molecular dynamics (MLMD) simulations to gain insights into SiC decomposition and phase transitions. Our approach relies on a Bayesian active learning workflow to efficiently train an accurate machine learning force field on density functional theory data. Our large-scale simulations provide direct indication that melting of SiC proceeds incongruently via decomposition into silicon-rich and carbon phases at high temperature and pressure. During cooling at high pressures, carbon nanoclusters nucleate and grow within the homogeneous molten liquid. During heating, the decomposed mixture reversibly transitions back into a homogeneous SiC liquid. The full pressure-temperature phase diagram of SiC is systematically constructed using MLMD simulations, providing new understanding of the nature of phases, resolving long-standing inconsistencies from previous experiments and yielding technologically relevant implications for processing and deposition of this material.

Molecular dynamics simulation is an important tool in computational materials science and chemistry, and in the past decade it has been revolutionized by machine learning. This rapid progress in machine learning interatomic potentials has produced a number of new architectures in just the past few years. Particularly notable among these are the atomic cluster expansion, which unified many of the earlier ideas around atom-density-based descriptors, and Neural Equivariant Interatomic Potentials (NequIP), a message-passing neural network with equivariant features that exhibited state-of-the-art accuracy at the time. Here we construct a mathematical framework that unifies these models: atomic cluster expansion is extended and recast as one layer of a multi-layer architecture, while the linearized version of NequIP is understood as a particular sparsification of a much larger polynomial model. Our framework also provides a practical tool for systematically probing different choices in this unified design space. An ablation study of NequIP, via a set of experiments looking at in- and out-of-domain accuracy and smooth extrapolation very far from the training data, sheds some light on which design choices are critical to achieving high accuracy. A much-simplified version of NequIP, which we call BOTnet (for body-ordered tensor network), has an interpretable architecture and maintains its accuracy on benchmark datasets.

Ionic liquids (ILs) are a promising class of electrolytes with a unique combination of properties, such as extremely low vapor pressures and nonflammability. Doping ILs with alkali metal salts creates an electrolyte that is of interest for battery technology. These salt-in-ionic liquids (SiILs) are a class of superconcentrated, strongly correlated, and asymmetric electrolytes. Notably, the transference numbers of the alkali metal cations have been found to be negative. Here, we investigate Na-based SiILs with a surface force apparatus, X-ray scattering, and atomic force microscopy. We find evidence of confinement-induced structural changes, giving rise to long-range interactions. Force curves also reveal an electrolyte structure consistent with our predictions from theory and simulations. The long-range steric interactions in SiILs reflect the high aspect ratio of compressible aggregates at the interfaces rather than the purely electrostatic origin predicted by the classical electrolyte theory. This conclusion is supported by the reported anomalous negative transference numbers, which can be explained within the same aggregation framework. The interfacial nanostructure should impact the formation of the solid electrolyte interphase in SiILs.

Ambidirectionality, which is the ability of structural elements to move beyond a reference state in two opposite directions, is common in nature. However, conventional soft materials are typically limited to a single, unidirectional deformation unless complex hybrid constructs are used. We exploited the combination of mesogen self-assembly, polymer chain elasticity, and polymerization-induced stress to design liquid crystalline elastomers that exhibit two mesophases: chevron smectic C (cSmC) and smectic A (SmA). Inducing the cSmC-SmA-isotropic phase transition led to an unusual inversion of the strain field in the microstructure, resulting in opposite deformation modes (e.g., consecutive shrinkage or expansion and right-handed or left-handed twisting and tilting in opposite directions) and high-frequency nonmonotonic oscillations. This ambidirectional movement is scalable and can be used to generate Gaussian transformations at the macroscale.

Environmentally-benign, non-toxic electrolytes with combinatorial design spaces are excellent candidates for green solvents, green leaching agents, and carbon capture sources. Here, we examine one particular green solvent, ethaline, a 2:1 molar ratio of ethylene glycol and choline chloride. Despite its touted green credentials, we find partial decomposition of ethaline into toxic chloromethane and dimethylaminoethanol at room temperature, limiting its sustainable advantage. We experimentally characterize these decomposition products and computationally develop a general, quantum chemically-accurate workflow to understand decomposition. We find that fluctuations of the hydrogen bonds bind chloride near reaction sites, initiating the reaction between choline cations and chloride anions. In summary, in the design of green solvents, we do not recommend the use of choline chloride due to its susceptibility to undergo decomposition in strongly hydrogen-bound mixtures.

The systematic underestimation of band gaps is one of the most fundamental challenges in semilocal density functional theory (DFT). In addition to hindering the application of DFT to predicting electronic properties, the band gap problem is intimately related to self-interaction and delocalization errors, which make the study of charge transfer mechanisms with DFT difficult. To expand the range of available tools for addressing the band gap problem, we design an approach for machine learning density functionals based on Gaussian processes to explicitly fit single-particle energy levels. We also introduce nonlocal features of the density matrix that are expressive enough to fit these single-particle levels. Combining these developments, we train a machine-learned functional for the exact exchange energy that predicts molecular energy gaps and reaction energies of a wide range of molecules in excellent agreement with reference hybrid DFT calculations. In addition, while being trained solely on molecular data, our model predicts reasonable formation energies of polarons in solids, showcasing its transferability and robustness. We discuss how this approach can be generalized to full exchange-correlation functionals, thus paving the way to the design of state-of-the-art functionals for the prediction of electronic properties of molecules and materials.

The xene family of topological insulators plays a key role in many proposals for topological electronic, spintronic, and valleytronic devices. These proposals rely on applying local perturbations, including electric fields and proximity magnetism, to induce topological phase transitions in xenes. However, these techniques lack control over the geometry of interfaces between topological regions, a critical aspect of engineering topological devices. We propose adatom decoration as a method for engineering atomically precise topological edge modes in xenes. Our first-principles calculations show that decorating stanene with Zn adatoms exclusively on one of two sublattices induces a topological phase transition from the quantum spin Hall (QSH) to quantum valley Hall (QVH) phase and confirm the existence of spin-valley polarized edge modes propagating at QSH/QVH interfaces. We conclude by discussing technological applications of these edge modes that are enabled by the atomic precision afforded by recent advances in adatom manipulation technology. The authors propose sublattice-selective decoration by Zn adatoms as a method to engineer precise topological edge modes in xenes. First-principles calculations on Zn decorated stanene reveal a quantum spin Hall (QSH) to quantum valley Hall (QVH) transition and spin-valley polarized modes propagating at the QSH/QVH interface.

Ultrafast spectroscopy paved the way for probing transient states of matter produced through photoexcitation. Despite significant advances, the microscopic processes governing the formation of these states remain largely unknown. This study discloses the nuclear quantum dynamics of $\mathrm{SrTiO_3}$ when excited by THz laser pumping. We use a first-principles machine-learning approach accounting for all atomistic degrees of freedom to examine the time-resolved energy flow across phonon modes following the photoexcitation, revealing the mechanism underpinning the observed phonon upconversion and quantifying the lifetime of the out-of-equilibrium motion. Crucially, our simulations predict that THz pump pulses can generate persistent out-of-equilibrium stress capable of inducing polar order. We observe a correlation between the experimentally measured lifetime of the transient inversion-symmetry-broken state and the duration of the out-of-equilibrium nuclear state. This work not only explains the experimental results on $\mathrm{SrTiO_3}$ but also establishes a framework for simulating the photoexcited quantum dynamics of nuclei from first principles without any empirical input. It lays the groundwork for systematic explorations of complex materials sensitive to photoexcitation.

Here, the authors use machine learning and exact physical constraints to design a nonlocal exchange functional for both molecular and solid-state systems. The model is computationally efficient, achieves hybrid-DFT accuracy on molecular benchmarks, and improves the accuracy of band gap predictions over semilocal DFT. To demonstrate the efficiency and accuracy of the model, the authors compute charged point defect transition levels in silicon in good agreement with experiment, a task previously only possible with more expensive hybrid DFT calculations.

The rapid advancements in ultrafast laser technology have paved the way for pumping and probing the out-of-equilibrium dynamics of nuclei in crystals. However, interpreting these experiments is extremely challenging due to the complex nonlinear responses in systems where lattice excitations interact, particularly in crystals composed of light atoms or at low temperatures where the quantum nature of ions becomes significant. In this work, we address the nonequilibrium quantum ionic dynamics from first principles. Our approach is general and can be applied to simulate any crystal, in combination with a first-principles treatment of electrons or external machine-learning potentials. It is implemented by leveraging the nonequilibrium time-dependent self-consistent harmonic approximation (TD-SCHA), with a stable, energy-conserving, correlated stochastic integration scheme that achieves an accuracy of $\mathcal{O}(dt^3)$. We benchmark the method with both a simple one-dimensional model to test its accuracy and a realistic 40-atom cell of SrTiO3 under THz laser pump, paving the way for simulations of ultrafast THz-Xray pump-probe spectroscopy like those performed in synchrotron facilities.

Ionic liquids (ILs) are an exciting class of electrolytes finding applications in many areas from energy storage to solvents, where they have been touted as "designer solvents" as they can be mixed to precisely tailor the physiochemical properties. As using machine learning interatomic potentials (MLIPs) to simulate ILs is still relatively unexplored, several questions need to be answered to see if MLIPs can be transformative for ILs. Since ILs are often not pure, but are either mixed together or contain additives, we first demonstrate that a MLIP can be trained to be compositionally transferable; i.e., the MLIP can be applied to mixtures of ions not directly trained on, while only being trained on a few mixtures of the same ions. We also investigated the accuracy of MLIPs for a novel IL, which we experimentally synthesize and characterize. Our MLIP trained on ∼200 DFT frames is in reasonable agreement with our experiments and DFT.

Supported metal nanoparticle (NP) catalysts are vital for the sustainable production of chemicals, but their design and implementation are limited by the ability to identify and characterize their structures and atomic sites that are correlated with high catalytic activity. Identification of these ''active sites'' has relied heavily on extrapolation to supported NP systems from investigation of idealized surfaces, experimentally using single crystals or supported NPs which are always modelled computationally using slab or regular polyhedra models. However, the ability of these methods to predict catalytic activity remains qualitative at best, as the structure of metal NPs in reactive environments has only been speculated from indirect experimental observations, or otherwise remains wholly unknown. Here, by circumventing these limitations for highly accurate simulation methods, we provide direct atomistic insight into the dynamic restructuring of metal NPs by combining in situ spectroscopy with molecular dynamics simulations powered by a machine learned force field. We find that in reactive environments, NP surfaces evolve to a state with poorly defined atomic order, while the core of the NP may remain bulk-like. These insights prove that long-standing conceptual models based on idealized faceting for small metal NP systems are not representative of real systems under exposure to reactive environments. We show that the resultant structure can be elucidated by combining advanced spectroscopy and computational tools. This discovery exemplifies that to enable faithful quantitative predictions of catalyst function and stability, we must move beyond idealized-facet experimental and theoretical models and instead employ systems which include realistic surface structures that respond to relevant physical and chemical conditions.

Multi-component metal nanoparticles (NPs) are of paramount importance in the chemical industry, as most processes therein employ heterogeneous catalysts. While these multi-component systems have been shown to result in higher product yields, improved selectivities, and greater stability through catalytic cycling, the structural dynamics of these materials in response to various stimuli (e.g. temperature, adsorbates, etc.) are not understood with atomistic resolution. Here, we present a highly accurate equivariant machine-learned force field (MLFF), constructed from ab initio training data collected using Bayesian active learning, that is able to reliably simulate PdAu surfaces and NPs in response to thermal treatment as well as exposure to reactive H$_2$ atmospheres. We thus provide a single model that is able to reliably describe the full space of geometric and chemical complexity for such a heterogeneous catalytic system across single crystals, gas-phase interactions, and NPs reacting with H$_2$, including catalyst degradation and explicit reactivity. Ultimately, we provide direct atomistic evidence that verifies existing experimental hypotheses for bimetallic catalyst deactivation under reaction conditions, namely that Pd preferentially segregates into the Au bulk through aggressive catalytic cycling and that this degradation is site-selective, as well as the reactivity for hydrogen exchange as a function of Pd ensemble size. We demonstrate that understanding of the atomistic evolution of these active sites is of the utmost importance, as it allows for design and control of material structure and corresponding performance, which can be vetted in silico.

In density functional theory, charge density is the core attribute of atomic systems from which all chemical properties can be derived. Machine learning methods are promising in significantly accelerating charge density prediction, yet existing approaches either lack accuracy or scalability. We propose a recipe that can achieve both. In particular, we identify three key ingredients: (1) representing the charge density with atomic and virtual orbitals (spherical fields centered at atom/virtual coordinates); (2) using expressive and learnable orbital basis sets (basis function for the spherical fields); and (3) using high-capacity equivariant neural network architecture. Our method achieves state-of-the-art accuracy while being more than an order of magnitude faster than existing methods. Furthermore, our method enables flexible efficiency-accuracy trade-offs by adjusting the model/basis sizes.

We present a differentiable formalism for learning free energies that is capable of capturing arbitrarily complex model dependencies on coarse-grained coordinates and finite-temperature response to variation of general system parameters. This is done by endowing models with explicit dependence on temperature and parameters and by exploiting exact differential thermodynamic relationships between the free energy, ensemble averages, and response properties. Formally, we derive an approach for learning high-dimensional cumulant generating functions using statistical estimates of their derivatives, which are observable cumulants of the underlying random variable. The proposed formalism opens ways to resolve several outstanding challenges in bottom-up molecular coarse graining dealing with multiple minima and state dependence. This is realized by using additional differential relationships in the loss function to significantly improve the learning of free energies, while exactly preserving the Boltzmann distribution governing the corresponding fine-grain all-atom system. As an example, we go beyond the standard force-matching procedure to demonstrate how leveraging the thermodynamic relationship between free energy and values of ensemble averaged all-atom potential energy improves the learning efficiency and accuracy of the free energy model. The result is significantly better sampling statistics of structural distribution functions. The theoretical framework presented here is demonstrated via implementations in both kernel-based and neural network machine learning regression methods and opens new ways to train accurate machine learning models for studying thermodynamic and response properties of complex molecular systems.

Abstract This work examines challenges associated with the accuracy of machine-learned force fields (MLFFs) for bulk solid and liquid phases of d -block elements. In exhaustive detail, we contrast the performance of force, energy, and stress predictions across the transition metals for two leading MLFF models: a kernel-based atomic cluster expansion method implemented using sparse Gaussian processes (FLARE), and an equivariant message-passing neural network (NequIP). Early transition metals present higher relative errors and are more difficult to learn relative to late platinum- and coinage-group elements, and this trend persists across model architectures. Trends in complexity of interatomic interactions for different metals are revealed via comparison of the performance of representations with different many-body order and angular resolution. Using arguments based on perturbation theory on the occupied and unoccupied d states near the Fermi level, we determine that the large, sharp d density of states both above and below the Fermi level in early transition metals leads to a more complex, harder-to-learn potential energy surface for these metals. Increasing the fictitious electronic temperature (smearing) modifies the angular sensitivity of forces and makes the early transition metal forces easier to learn. This work illustrates challenges in capturing intricate properties of metallic bonding with current leading MLFFs and provides a reference data set for transition metals, aimed at benchmarking the accuracy and improving the development of emerging machine-learned approximations.

Metal surfaces have long been known to reconstruct, significantly influencing their structural and catalytic properties. Many key mechanistic aspects of these subtle transformations remain poorly understood due to limitations of previous simulation approaches. Using active learning of Bayesian machine-learned force fields trained from ab initio calculations, we enable large-scale molecular dynamics simulations to describe the thermodynamics and time evolution of the low-index mesoscopic surface reconstructions of Au (e.g., the Au(111)-'Herringbone,' Au(110)-(1 × 2)-'Missing-Row,' and Au(100)-'Quasi-Hexagonal' reconstructions). This capability yields direct atomistic understanding of the dynamic emergence of these surface states from their initial facets, providing previously inaccessible information such as nucleation kinetics and a complete mechanistic interpretation of reconstruction under the effects of strain and local deviations from the original stoichiometry. We successfully reproduce previous experimental observations of reconstructions on pristine surfaces and provide quantitative predictions of the emergence of spinodal decomposition and localized reconstruction in response to strain at non-ideal stoichiometries. A unified mechanistic explanation is presented of the kinetic and thermodynamic factors driving surface reconstruction. Furthermore, we study surface reconstructions on Au nanoparticles, where characteristic (111) and (100) reconstructions spontaneously appear on a variety of high-symmetry particle morphologies.

Predicting response of materials to external stimuli is a primary objective of computational materials science. However, current methods are limited to small-scale simulations due to the unfavorable scaling of computational costs. Here, we implement an equivariant machine-learning framework where response properties stem from exact differential relationships between a generalized potential function and applied external fields. Focusing on responses to electric fields, the method predicts electric enthalpy, forces, polarization, Born charges, and polarizability within a unified model enforcing the full set of exact physical constraints, symmetries and conservation laws. Through application to $α$-SiO$_2$, we demonstrate that our approach can be used for predicting vibrational and dielectric properties of materials, and for conducting large-scale dynamics under arbitrary electric fields at unprecedented accuracy and scale. We apply our method to ferroelectric BaTiO$_3$ and capture the temperature-dependence and time evolution of hysteresis, revealing the underlying microscopic mechanisms of nucleation and growth that govern ferroelectric domain switching.

The systematic underestimation of band gaps is one of the most fundamental challenges in semilocal density functional theory (DFT). In addition to hindering the application of DFT to predicting electronic properties, the band gap problem is intimately related to self-interaction and delocalization errors, which make the study of charge transfer mechanisms with DFT difficult. In this work, we present two key innovations to address the band gap problem. First, we design an approach for machine learning density functionals based on Gaussian processes to explicitly fit single-particle energy levels. Second, we introduce novel nonlocal features of the density matrix that are expressive enough to fit these single-particle levels. Combining these developments, we train a machine-learned functional for the exact exchange energy that predicts molecular energy gaps and reaction energies of a wide range of molecules in excellent agreement with reference hybrid DFT calculations. In addition, while being trained solely on molecular data, our model predicts reasonable formation energies of polarons in solids, showcasing its transferability and robustness. Our approach generalizes straightforwardly to full exchange-correlation functionals, thus paving the way to the design of novel state-of-the-art functionals for the prediction of electronic properties of molecules and materials.

Ionic liquids (ILs) are a promising class of electrolytes owing to a unique combination of properties, such as extremely low vapour pressures, non-flammability and being universal solvents. Doping ILs with alkali metal salts creates an electrolyte that is of interest for batteries, among others. These salt-in-ionic liquids (SiILs) are a class of super-concentrated, strongly correlated and asymmetric electrolytes. The transference number of the alkali metal cations has been found to be negative, owing to the small but highly negatively charged aggregates which form between alkali metal ions and the anions. Here, we investigate Na-based SiILs with a surface forces apparatus and by atomic force microscopy. We find evidence of confinement induced structural changes, giving rise to unprecedented long-range (non-exponentially decaying) interactions. This observation is supported by the soft structure revealed by the force curves, and supplemented by theory and simulations. The long-ranged interactions in SiILs are reminiscent of polymer-like interactions, suggesting analogous high aspect ratio aggregates at the mica interfaces, rather than a purely electrostatic origin. Remarkably, our aggregation framework and conclusions can also explain the negative transference number, often observed in these systems by the battery community.

Ionic liquids (ILs) are an exciting class of electrolytes finding applications in many areas from energy storage to solvents, where they have been touted as ``designer solvents'' as they can be mixed to precisely tailor the physiochemical properties. As using machine learning interatomic potentials (MLIPs) to simulate ILs is still relatively unexplored, several questions need to be answered to see if MLIPs can be transformative for ILs. Since ILs are often not pure, but are either mixed together or contain additives, we first demonstrate that a MLIP can be trained to be compositionally transferable, i.e., the MLIP can be applied to mixtures of ions not directly trained on, whilst only being trained on a few mixtures of the same ions. We also investigate the accuracy of MLIPs for a novel IL, which we experimentally synthesize and characterize. Our MLIP trained on $\sim$200 DFT frames is in reasonable agreement with our experiments and DFT.