Search

Hybrid molecular ferroelectrics with orientationally disordered mesophases offer significant promise as lead-free alternatives to traditional inorganic ferroelectrics owing to properties such as room temperature ferroelectricity, low-energy synthesis, malleability, and potential for multiaxial polarization. The ferroelectric molecular salt HdabcoClO 4 is of particular interest due to its ultrafast ferroelectric room-temperature switching. However, so far, there is limited understanding of the nature of dynamical disorder arising in these compounds. Here, we employ the neural network NeuralIL to train a machine-learned force field (MLFF) with training data generated using density functional theory. The resulting MLFF-MD simulations exhibit phase transitions and thermal expansion in line with earlier reported experimental results, for both a low-temperature phase transition coinciding with the orientational disorder of ClO 4 - and the onset of rotation of both Hdabco + and ClO 4 - in a high-temperature phase transition. We also find proton transfer even in the low-temperature phase, which increases with temperature and leads to associated proton disorder as well as the onset of disorder in the direction of the hydrogen-bonded chains., Competing Interests: The authors declare no competing financial interest., (© 2024 The Authors. Published by American Chemical Society.)

Determining the local coordination of the active site is a prerequisite for the reliable modeling of single-atom catalysts (SACs). Obtaining such information is difficult on powder-based systems and much emphasis is placed on density functional theory computations based on idealized low-index surfaces of the support. In this work, we investigate how Pt atoms bind to the (11̅02) facet of α-Fe 2 O 3 ; a common support material in SACs. Using a combination of scanning tunneling microscopy, X-ray photoelectron spectroscopy, and an extensive computational evolutionary search, we find that Pt atoms significantly reconfigure the support lattice to facilitate a pseudolinear coordination to surface oxygen atoms. Despite breaking three surface Fe-O bonds, this geometry is favored by 0.84 eV over the best configuration involving an unperturbed support. We suggest that the linear O-Pt-O configuration is common in reactive Pt-based SAC systems because it balances thermal stability with the ability to adsorb reactants from the gas phase. Moreover, we conclude that extensive structural searches are necessary to determine realistic active site geometries in single-atom catalysis.

The investigation of inhomogeneous surfaces, where various local structures coexist, is crucial for understanding interfaces of technological interest, yet it presents significant challenges. Here, we study the atomic configurations of the (2 × m ) Ti-rich surfaces at (110)-oriented SrTiO 3 by bringing together scanning tunneling microscopy and transferable neural-network force fields combined with evolutionary exploration. We leverage an active learning methodology to iteratively extend the training data as needed for different configurations. Training on only small well-known reconstructions, we are able to extrapolate to the complicated and diverse overlayers encountered in different regions of the inhomogeneous SrTiO 3 (110)-(2 × m ) surface. Our machine-learning-backed approach generates several new candidate structures, in good agreement with experiment and verified using density functional theory. The approach could be extended to other complex metal oxides featuring large coexisting surface reconstructions., Competing Interests: There are no conflicts to declare., (This journal is © The Royal Society of Chemistry.)

Machine learning potentials have become an essential tool for atomistic simulations, yielding results close to ab initio simulations at a fraction of computational cost. With recent improvements on the achievable accuracies, the focus has now shifted on the data set composition itself. The reliable identification of erroneously predicted configurations to extend a given data set is therefore of high priority. Yet, uncertainty estimation techniques have achieved mixed results for machine learning potentials. Consequently, a general and versatile method to correlate energy or atomic force uncertainties with the model error has remained elusive to date. In the current work, we show that epistemic uncertainty cannot correlate with model error by definition but can be aggregated over groups of atoms to yield a strong correlation. We demonstrate that our method correctly estimates prediction errors both globally per structure and locally resolved per atom. The direct correlation of local uncertainty and local error is used to design an active learning framework based on identifying local subregions of a large simulation cell and performing ab initio calculations only for the subregion subsequently. We successfully utilized this method to perform active learning in the low-data regime for liquid water.

We explore the phonon transport properties of defect-laden bilayer PtSTe using equilibrium molecular dynamics simulations based on a neural-network force field. Defects prove very efficient at depressing the thermal conductivity of the structure, and flower defects have a particularly powerful effect, comparable to that of double vacancies. Furthermore, the conductivity of the structure with flower defects exhibits an unusual temperature dependence due to structural instability at high temperatures. We look into the distortion to normal modes around the defect by means of the projected phonon density of states and find diverse phenomena including localized modes and blue shifts., Competing Interests: The authors declare no competing financial interest., (© 2024 The Authors. Published by American Chemical Society.)

Recent theoretical and experimental research suggests that θ-TaN is a semimetal with high thermal conductivity (κ), primarily due to the contribution of phonons (κ_{ph}). By using first-principles calculations, we show a nonmonotonic pressure dependence of the κ of θ-TaN. κ_{ph} first increases until it reaches a maximum at around 60 GPa, and then decreases. This anomalous behavior is a consequence of the competing pressure responses of phonon-phonon and phonon-electron interactions, in contrast to the known materials BAs and BP, where the nonmonotonic pressure dependence is caused by the interplay between different phonon-phonon scattering channels. Although TaN has phonon dispersion features similar to BAs at ambient pressure, its response to pressure is different and an overall stiffening of the phonon branches takes place. Consequently, the relevant phonon-phonon scattering weakens as pressure increases. However, the increased electronic density of states near the Fermi level, and specifically the emergence of additional pockets of the Fermi surface at the high-symmetry L point in the Brillouin zone, leads to a substantial increase in phonon-electron scattering at high pressures, driving a decrease in κ_{ph}. At intermediate pressures (∼20-70  GPa), the κ of TaN surpasses that of BAs. Our Letter provides deeper insight into phonon transport in semimetals and metals where phonon-electron scattering is relevant.

Transition metal dichalcogenides are investigated for various applications at the nanoscale because of their unique combination of properties and dimensionality. For many of the anticipated applications, heat conduction plays an important role. At the same time, these materials often contain relatively large amounts of point defects. Here, we provide a systematic analysis of the impact of intrinsic and selected extrinsic defects on the lattice thermal conductivity of MoS 2 and WS 2 monolayers. We combine Boltzmann transport theory and Green's function-based T -matrix approach for the calculation of scattering rates. The force constants for the defect configurations are obtained from density functional theory calculations via a regression approach, which allows us to sample a rather large number of defects at a moderate computational cost and to systematically enforce both the translational and rotational acoustic sum rules. The calculated lattice thermal conductivity is in quantitative agreement with the experimental data for heat transport and defect concentrations for both MoS 2 and WS 2 . Crucially, this demonstrates that the strong deviation from a 1/ T temperature dependence of the lattice thermal conductivity observed experimentally can be fully explained by the presence of point defects. We furthermore predict the scattering strengths of the intrinsic defects to decrease in the sequence V Mo ≈ V 2S = > V 2S ⊥ > V S > S ad in both materials, while the scattering rates for the extrinsic (adatom) defects decrease with increasing mass such that Li ad > Na ad > K ad . Compared with earlier work, we find that both intrinsic and extrinsic adatoms are relatively weak scatterers. We attribute this difference to the treatment of the translational and rotational acoustic sum rules, which, if not enforced, can lead to spurious contributions in the zero-frequency limit., Competing Interests: The authors declare no competing financial interest., (© 2024 The Authors. Published by American Chemical Society.)

One of the well-known limitations of Kohn-Sham density functional theory is the tendency to strongly underestimate bandgaps. Meta-generalized gradient approximations (mGGAs), which include the kinetic energy density in the functional form, have been shown to significantly alleviate this deficiency. In this study, we explore the mechanisms responsible for this improvement from the angle of the underlying local densities. We find that the highest occupied and lowest unoccupied states are distinct in the space of the underlying descriptors. The gap opening is compared to a simple scaling of the local density approximation, and two mechanisms responsible for opening the mGGA gaps are identified. First of all, the relatively large negative derivative of the functional form with respect to reduced kinetic energy tends to elevate the lowest unoccupied state. Second, the curvature of functional, which ensures that it is bounded, tends to lower the highest occupied state. Remarkably, these two mechanisms are found to be transferable over a large and diverse database of compounds., (© 2023 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).)

Enzymatic reactions are an ecofriendly, selective, and versatile addition, sometimes even alternative to organic reactions for the synthesis of chemical compounds such as pharmaceuticals or fine chemicals. To identify suitable reactions, computational models to predict the activity of enzymes on non-native substrates, to perform retrosynthetic pathway searches, or to predict the outcomes of reactions including regio- and stereoselectivity are becoming increasingly important. However, current approaches are substantially hindered by the limited amount of available data, especially if balanced and atom mapped reactions are needed and if the models feature machine learning components. We therefore constructed a high-quality dataset (EnzymeMap) by developing a large set of correction and validation algorithms for recorded reactions in the literature and showcase its significant positive impact on machine learning models of retrosynthesis, forward prediction, and regioselectivity prediction, outperforming previous approaches by a large margin. Our dataset allows for deep learning models of enzymatic reactions with unprecedented accuracy, and is freely available online., Competing Interests: There are no conflicts to declare., (This journal is © The Royal Society of Chemistry.)

A reliable uncertainty estimator is a key ingredient in the successful use of machine-learning force fields for predictive calculations. Important considerations are correlation with error, overhead during training and inference, and efficient workflows to systematically improve the force field. However, in the case of neural-network force fields, simple committees are often the only option considered due to their easy implementation. Here, we present a generalization of the deep-ensemble design based on multiheaded neural networks and a heteroscedastic loss. It can efficiently deal with uncertainties in both energy and forces and take sources of aleatoric uncertainty affecting the training data into account. We compare uncertainty metrics based on deep ensembles, committees, and bootstrap-aggregation ensembles using data for an ionic liquid and a perovskite surface. We demonstrate an adversarial approach to active learning to efficiently and progressively refine the force fields. That active learning workflow is realistically possible thanks to exceptionally fast training enabled by residual learning and a nonlinear learned optimizer., (© 2023 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).)

The space of generalized gradient approximation (GGA) and meta-GGA (mGGA) exchange approximations is systematically explored by training 25 new functionals to produce accurate lattice parameter, cohesive energy, and bandgap predictions. The trained functionals are used to reproduce exact constraints in a data-driven way and to understand the accuracy trade-off between the mentioned properties. The functionals are compared to notable mGGA functionals to analyze how changes in the enhancement factor maps influence the accuracy of predictions. Some of the trained functionals are found to perform on par with specialized functionals for bandgaps, while outperforming them on the other two properties. The error surface of our trained functionals can serve as a soft-limit of what mGGA functionals can achieve.

The determination of atomic structures in surface reconstructions has typically relied on structural models derived from intuition and domain knowledge. Evolutionary algorithms have emerged as powerful tools for such structure searches. However, when density functional theory is used to evaluate the energy the computational cost of a thorough exploration of the potential energy landscape is prohibitive. Here, we drive the exploration of the rich phase diagram of TiO x overlayer structures on SrTiO 3 (110) by combining the covariance matrix adaptation evolution strategy (CMA-ES) and a neural-network force field (NNFF) as a surrogate energy model. By training solely on SrTiO 3 (110) 4×1 overlayer structures and performing CMA-ES runs on 3×1, 4×1 and 5×1 overlayers, we verify the transferability of the NNFF. The speedup due to the surrogate model allows taking advantage of the stochastic nature of the CMA-ES to perform exhaustive sets of explorations and identify both known and new low-energy reconstructions., Competing Interests: There are no conflicts to declare., (This journal is © The Royal Society of Chemistry.)