Machine learned force fields typically require manual construction of training sets consisting of thousands of first principles calculations, which can result in low training efficiency and unpredictable errors when applied to structures not represented in the training set of the model. This severely limits the practical application of these models in systems with dynamics governed by important rare events, such as chemical reactions and diffusion. We present an adaptive Bayesian inference method for automating the training of interpretable, low-dimensional, and multi-element interatomic force fields using structures drawn on the fly from molecular dynamics simulations.Within an active learning framework, the internal uncertainty of a Gaussian process regression model is used to decide whether to accept the model prediction or to perform a first principles calculation to augment the training set of the model. The method is applied to a range of single-and multi-element systems and shown to achieve a favorable balance of accuracy and computational efficiency, while requiring a minimal amount of ab initio training data. We provide a fully open-source implementation of our method, as well as a procedure to map trained models to computationally efficient tabulated force fields.

Photoaffinity labeling of the Torpedo nicotinic acetylcholine receptor (nAChR) with [3H]d-tubocurarine (dTC) has identified a residue within the gamma-subunit which, along with the analogous residue in delta-subunit, confers selectivity in binding affinities between the two agonist sites for dTC and alpha-conotoxin (alpha Ctx) MI. nAChR gamma-subunit, isolated from nAChR-rich membranes photolabeled with [3H]dTC, was digested with Staphylococcus aureus V8 protease, and a 3H-labeled fragment was purified by reversed-phase high-performance liquid chromatography. Amino-terminal sequence analysis of this fragment identified 3H incorporation in gamma Tyr-111 and gamma Tyr-117 at about 5% and 1% of the efficiency of [3H]dTC photoincorporation at gamma Trp-55, the primary site of [3H]dTC photoincorporation within gamma-subunit [Chiara, D. C., and Cohen, J. B. (1997) J. Biol. Chem 272, 32940-32950]. The Torpedo nAChR delta-subunit residue corresponding to gamma Tyr-111 (delta Arg-113) contains a positive charge which could confer the lower binding affinity seen for some competitive antagonists at the alpha-delta agonist site. To test this hypothesis, we examined by voltage-clamp analysis and/or by [125I]alpha-bungarotoxin competition binding assays the interactions of acetylcholine (ACh), dTC, and alpha Ctx MI with nAChRs containing gamma Y111R or delta R113Y mutant subunits expressed in Xenopus oocytes. While these mutations affected neither ACh equilibrium binding affinity nor the concentration dependence of channel activation, the gamma Y111R mutation decreased by 10-fold dTC affinity and inhibition potency. Additionally, each mutation conferred a 1000-fold change in the equilibrium binding of alpha Ctx MI, with delta R113Y enhancing and gamma Y111R weakening affinity. Comparison of these results with previous results for mouse nAChR reveals that, while the same regions of gamma- (or delta-) subunit primary structure contribute to the agonist-binding sites, the particular amino acids that serve as antagonist affinity determinants are species-dependent.

Deciphering the relationship between the active-site structure and CO 2 methanation mechanism over Ni-based catalysts faces great challenges. Herein, different distributions of frustrated Lewis pair (FLP) structures were precisely fabricated over Ni/ CeO 2 -nanorods, Ni/CeO 2 -nanocubes, and Ni/CeO 2 -nanooctahedra to make progress in this issue. Ni/CeO 2 -nanorods presented the highest possibility for FLP construction among these catalysts due to their CeO 2 (110) nature and the steric hindrance between the oxygen vacancy (O V ) and hydroxyl species (OH). Compared to other samples with fewer FLPs, FLPs-enriched Ni/CeO 2nanorods showed a significantly higher CO 2 conversion (84.2%) and a CH 4 productivity of up to 147.1 mmol g cat −1 h −1 with a higher CH 4 selectivity (97.8%) even at a temperature as low as 225 °C. As evidenced from systematical ex situ and in situ surface analysis results, this better low-temperature activity along with its acceptable stability was closely associated with the construction of catalytically active FLPs, which could effectively activate and convert CO 2 via the cooperation of O V and OH. Also, the in situ (Raman and diffuse-reflectance infrared Fourier transform spectroscopy) analysis combined with density functional theory calculations further demonstrated that the copromotion of the emerged CO* route and formate pathway was responsible for the promising low-temperature (≤225 °C) methanation performance over the FLP-enriched Ni/CeO 2 -nanorods. Such CO 2 activation by FLPs will potentially guide the design of CO 2 hydrogenation catalysts.

Restructuring of interfaces plays a crucial role in materials science and heterogeneous catalysis. Bimetallic systems, in particular, often adopt very different composition and morphology at surfaces compared to the bulk. For the first time, we reveal a detailed atomistic picture of long-timescale restructuring of Pd deposited on Ag, using microscopy, spectroscopy, and novel simulation methods. By developing and performing accelerated machine-learning molecular dynamics followed by an automated analysis method, we discover and characterize previously unidentified surface restructuring mechanisms in an unbiased fashion, including Pd-Ag place exchange and Ag pop-out, as well as step ascent and descent. Remarkably, layer-by-layer dissolution of Pd into Ag is always preceded by an encapsulation of Pd islands by Ag, resulting in a significant migration of Ag out of the surface and a formation of extensive vacancy pits within a period of microseconds. These metastable structures are of vital catalytic importance, as Ag-encapsulated Pd remains much more accessible to reactants than bulk-dissolved Pd. Our approach is broadly applicable to complex multimetallic systems and enables the previously intractable mechanistic investigation of restructuring dynamics at atomic resolution. File list (2) download file view on ChemRxiv 061220_PdAg_ESI_v5.pdf (13.63 MiB) download file view on ChemRxiv 061220_PdAg_Main_v5.pdf (11.39 MiB)

We present a way to dramatically accelerate Gaussian process models for interatomic force fields based on many-body kernels by mapping both forces and uncertainties onto functions of low-dimensional features. This allows for automated active learning of models combining near-quantum accuracy, built-in uncertainty, and constant cost of evaluation that is comparable to classical analytical models, capable of simulating millions of atoms. Using this approach, we perform large-scale molecular dynamics simulations of the stability of the stanene monolayer. We discover an unusual phase transformation mechanism of 2D stanene, where ripples lead to nucleation of bilayer defects, densification into a disordered multilayer structure, followed by formation of bulk liquid at high temperature or nucleation and growth of the 3D bcc crystal at low temperature. The presented method opens possibilities for rapid development of fast accurate uncertainty-aware models for simulating long-time large-scale dynamics of complex materials.