Material discovery by combining stochastic surface walking global optimization with a neural network.

The prediction for the stability of glassy material is a key challenge in physical science. Here, we report a theoretical framework to predict the glass stability based on stochastic surface walking global optimization and reaction pathway sampling. This is demonstrated by revealing for the first time the global potential energy surface (PES) of two systems, CuZr binary metallic glass and nonglassy pure Cu systems, and establishing the lowest energy pathways linking glassy/amorphous structures with crystalline structures. The CuZr system has a significant number of glassy structures on PES that are $\ensuremath{\sim}0.045\phantom{\rule{0.16em}{0ex}}\mathrm{eV}/\mathrm{atom}$ above the crystal structure. Two clear trends are identified from global PES in the glass-to-crystal transition of the CuZr system: (i) the local Zr-Cu coordination (nearest neighbor) increases, and (ii) the local Zr bonding environment becomes homogeneous. This allows us to introduce quantitative structural and energetics conditions to distinguish the glassy structures from the crystalline structures. Because of the local Zr-Cu exchange in the glass-to-crystal transition, a high reaction barrier $(g0.048\phantom{\rule{0.16em}{0ex}}\mathrm{eV}/\mathrm{atom})$ is present to separate the glassy structures and the crystals in CuZr. By contrast, the Cu system, although it does possess amorphous structures that appear at much higher energy $(\ensuremath{\sim}0.075\phantom{\rule{0.16em}{0ex}}\mathrm{eV}/\mathrm{atom})$ with respect to the crystal structure, has very low reaction barriers for the crystallization of amorphous structures, i.e. $l0.011\phantom{\rule{0.16em}{0ex}}\mathrm{eV}/\mathrm{atom}$. The quantitative data on PES now available from global optimization techniques deepens our understanding on the microscopic nature of glassy material and might eventually facilitate the design of stable glassy materials.

Predicting the adsorption structure of molecules has long been a challenging topic given the coupled complexity of surface binding sites and molecule flexibility. Here, we develop AIMAP, an Artificial Intelligence Driven Molecule Adsorption Prediction tool, to achieve the general-purpose end-to-end prediction of molecule adsorption structures. AIMAP features efficient exploration of the global potential energy surface of the adsorption system based on global neural network (G-NN) potential, by rapidly screening qualified adsorption patterns and fine searching using stochastic surface walking (SSW) global optimization. We demonstrate the AIMAP efficiency in constructing the Cu-HCNO6 adsorption database, encompassing 1 182 351 distinct adsorption configurations of 9592 molecules on three copper surfaces. AIMAP is then utilized to identify the best adsorption structure for 18 amino acids (AAs) on achiral Cu surfaces and the chiral Cu(3,1,17) S surface. We find that AAs chemisorb on copper surfaces in their highest deprotonated state, through both the carboxylate-amino skeleton and side groups. The chiral recognition is identified for the d-preference of Asp, Glu, and Tyr, and l-preference for His. The physical origin for the enantiospecific adsorption is thus rationalized, pointing to the critical role of the competitive adsorption between functional side groups and the carboxylate-amino skeleton at surface low-coordination sites.

[1] A new publicly available daily gridded precipitation data set over mainland Portugal is presented. This data set is also combined with a recent Spanish data set to obtain a high resolution (0.2° × 0.2°) Iberian data set, labeled IB02. This data set covers the period from 1950 to 2003 and is based on a dense network, with more than 2000 and 400 quality-controlled stations over Spain and Portugal, respectively. The ordinary kriging method, applied over Portugal for consistency with the Spanish data set, performs slightly better than simpler interpolation techniques tested over Portugal. Additionally, this paper evaluates four global gridded data sets: two based on rain gauges (Climate Research Unit (CRU) and Global Precipitation Climate Center (GPCC)) and two European Centre for Medium-Range Weather Forecasts (ECMWF) reanalyses (ERA-40 and ERA-Interim), comparing them with the IB02 data set. The main features of the spatial distribution of IB02 mean annual precipitation are reasonably captured by the global data sets, despite their dry biases, mostly in mountainous regions. The four data sets perform better in western Iberia and are able to identify the major drought spells at the Iberian scale. Despite these similarities, GPCC outperforms CRU and ERA-Interim is superior to ERA-40 with respect to several aspects, such as annual cycle and drought detection. The performance of CRU is similar to that of ERA-Interim. The frequency of wet days is overestimated by reanalyses, mainly by ERA-Interim, while heavy precipitation events are underestimated, mostly by ERA-40. At 5 day scales, ECMWF reanalyses reveal difficulties in predicting the magnitude of precipitation, despite their greater ability to estimate the peak locations.

We first constructed six-dimensional potential energy surfaces (PESs) for the dissociative chemisorption of HCl on rigid Ag(111). Three global PESs were calculated using the neural network method based on extensive density functional theory (DFT) calculations with Perdew-Burke-Ernzerhof, Perdew-Wang91, and revised Perdew-Burke-Ernzerhof functionals, respectively. The resulting three PESs are all accurately fitted and smooth, based on the small fitting errors and good agreements between the fitted PESs and the direct DFT calculations. Time-dependent wave packet calculations show that the PESs are very well converged with respect to the fitting process, as well as the number of DFT data points. Comparisons of three PESs and time-dependent quantum dynamics results on the three PESs were given and discussed. The validity of site-averaging approximation holds for the title reaction, where the 25 site-averaged four-dimensional dissociation probability can accurately reproduce the six-dimensional dissociation probability on the current three PESs. We expect that the current theoretical investigations can stimulate the experimental work for this gas-surface reaction.

A global ground-state potential energy surface for the dissociation reactions of difluoroethylenes (DFEs) was computed by B3LYP and QCISD calculations, using the standard 6-311G(2d,2p) basis set. RRKM calculations were performed to compute relative abundances of hydrogen fluoride (HF) and molecular hydrogen produced from 1,1-DFE and from 1,2-DFE (cis and trans) at energies ranging from 110 to 180 kcal mol-1 relative to the zero point energy of 1,1-DFE. Thermal rate constants were also evaluated by the variational transition state theory for temperatures in the range 1250−1500 K. Both theoretical methods agree that, at the energies and temperatures studied, the main channel for HF elimination from 1,1-DFE is through a four-center transition state, whereas for 1,2-DFE the process occurs through a direct three-center elimination. At the energies studied, the RRKM method predicts that the main channel for molecular hydrogen elimination from the DFEs goes through a three-center transition state that connects 1,1-DFE with products.

Glucose pyrolysis, a model system in biomass utilization, is renowned for its great complexity, deep in reaction network hierarchy and rich in reaction patterns. The selectivity in glucose pyrolysis, e.g., the high yield of 5-hydroxymethylfurfural (HMF), a value-added platform product, remains an intriguing puzzle even after 60 years of experimental study. Here we resolve the whole reaction network of glucose pyrolysis using a global-to-global technique for reaction pathway sampling. This is achieved by establishing the first organic chemistry reaction database via stochastic surface walking (SSW) global optimization, building the global neural network (G-NN) potential via machine learning and extensively exploring the reaction network of glucose pyrolysis. In total, 6407 elementary reactions, screened out from more than 150 000 reaction pairs in glucose pyrolysis, are collected in our reaction database. The established reaction network from SSW-NN, further validated by first-principles calculations, reveals that for glucose to HMF, the lowest energy reaction pathway involves fructose and 3-deoxyglucos-2-ene (3-DGE) as key intermediates and a site-selective reaction type, retro-Michael-addition, for three consecutive dehydration steps. The overall barrier is determined to be 1.91 eV, being at least 0.19 eV lower than all previously proposed mechanisms, which assumes direct β-H elimination dehydration. The lowest pathways to the other two major products, furfural (FF) and hydroxyacetaldehyde (HAA), are also discovered with a similar barrier 1.95 eV, which exhibit a competing nature by sharing the same key intermediate, 3-ketohexose. Since chemical reactions occurring in fast glucose pyrolysis are generally present in biomass chemistry, containing essentially all reaction patterns of C-H-O elements, the methodology designed and the results presented would help to advance reaction design and mechanistic modeling in renewable fuels from biomass.

A global potential energy surface (PES) for the H+CH4↔H2+CH3 reaction has been constructed using the neural networks method based on 47783 high level ab initio geometry points. Extensive quasi-classical trajectories and quantum scattering calculations were carried out to check the convergence of the PES. This PES, fully converged with respect to the fitting procedure and the number of ab initio points, has a very small fitting error, and is much faster on evaluation than the modified Shepard interpolating PES, representing the best available PES for this benchmark polyatomic system.

Tin halide perovskite's potential as a photovoltaic absorber has not been fully realized to date, largely due to its instability in ambient air. Here, we demonstrate by both experiments and simulations that the air instability of black-phase cesium tin iodide perovskite (γ-CsSnI3) could be greatly lessened by a controlled incorporation of bismuth (Bi) ions into the crystal lattice. Hall effect measurements on films of γ-CsSnI3 suggest the unwanted formation of a tin vacancy and p-type self-doping can be effectively suppressed by the Bi incorporation. Structural and optical results indicate that the Bi incorporation markedly enhances the air stability by impeding the direct conversion of γ-CsSnI3 to zero-dimensional Cs2SnI6. By using a stochastic surface walking (SSW) method integrating neural network (NN) potential and density functional theory (DFT), it is revealed that the remarkable enhanced stability could be attributed to a combination of factors originating from lattice-contraction-induced strain, a suppressed tin vacancy, and an increased energy barrier for the transformation of γ-CsSnI3 to Cs2SnI6. This study provides physical insights into the stabilization mechanism of tin perovskites by heterovalent B-site engineering, paving the way for realizing stable and efficient lead-free perovskite photovoltaics.

Abstract. Vegetation optical depth (VOD) from microwave satellite observations has received much attention in global vegetation studies in recent years due to its relationship to vegetation water content and biomass. We recently have shown that VOD is related to plant productivity, i.e., gross primary production (GPP). Based on this relationship between VOD and GPP, we developed a theory-based machine learning model to estimate global patterns of GPP from passive microwave VOD retrievals. The VOD-GPP model generally showed good agreement with site observations and other global data sets in temporal dynamic but tended to overestimate annual GPP across all latitudes. We hypothesized that the reason for the overestimation is the missing effect of temperature on autotrophic respiration in the theory-based machine learning model. Here we aim to further assess and enhance the robustness of the VOD-GPP model by including the effect of temperature on autotrophic respiration within the machine learning approach and by assessing the interannual variability of the model results with respect to water availability. We used X-band VOD from the VOD Climate Archive (VODCA) data set for estimating GPP and used global state-of-the-art GPP data sets from FLUXCOM and MODIS to assess residuals of the VOD-GPP model with respect to drought conditions as quantified by the Standardized Precipitation and Evaporation Index (SPEI). Our results reveal an improvement in model performance for correlation when including the temperature dependency of autotrophic respiration (average correlation increase of 0.18). This improvement in temporal dynamic is larger for temperate and cold regions than for the tropics. For unbiased root-mean-square error (ubRMSE) and bias, the results are regionally diverse and are compensated in the global average. Improvements are observed in temperate and cold regions, while decreases in performance are obtained mainly in the tropics. The overall improvement when adding temperature was less than expected and thus may only partly explain previously observed differences between the global GPP data sets. On interannual timescales, estimates of the VOD-GPP model agree well with GPP from FLUXCOM and MODIS. We further find that the residuals between VOD-based GPP estimates and the other data sets do not significantly correlate with SPEI, which demonstrates that the VOD-GPP model can capture responses of GPP to water availability even without including additional information on precipitation, soil moisture or evapotranspiration. Exceptions from this rule were found in some regions: significant negative correlations between VOD-GPP residuals and SPEI were observed in the US corn belt, Argentina, eastern Europe, Russia and China, while significant positive correlations were obtained in South America, Africa and Australia. In these regions, the significant correlations may indicate different plant strategies for dealing with variations in water availability. Overall, our findings support the robustness of global microwave-derived estimates of gross primary production for large-scale studies on climate–vegetation interactions.

The cropping intensity has received growing concern in the agriculture field in applications such as harvest area research. Notwithstanding the significant amount of existing literature on local cropping intensities, research considering global datasets appears to be limited in spatial resolution and precision. In this paper, we present an annual dynamic global cropping intensity dataset covering the period from 2001 to 2019 at a 250-m resolution with an average overall accuracy of 89%, exceeding the accuracy of the current annual dynamic global cropping intensity data at a 500-m resolution. We used the enhanced vegetation index (EVI) of MOD13Q1 as the database via a sixth-order polynomial function to calculate the cropping intensity. The global cropping intensity dataset was packaged in the GeoTIFF file type, with the quality control band in the same format. The dataset fills the vacancy of medium-resolution, global-scale annual cropping intensity data and provides an improved map for further global yield estimations and food security analyses.

A global potential energy surface (PES) for the CH 2 (1 1 A′) was constructed using neural network method based on high‐level ab initio points. The topographic features of the PES were discussed and compared with available theoretical and experimental values. The results indicate that the present PES is accurate and reliable. To further clarify the reality of the PES, the dynamics calculations of the C( 1 D) + H 2 → H + CH(X 2 Π) reaction were performed using time‐dependent wave packet method. Some meaningful dynamic results including reaction probability, integral cross section, differential cross section, and rate constant were calculated and compared with previous theoretical and experimental values. The reasonable dynamics results indicate that the present PES can be used to perform any kind of dynamics studies.

As the most abundant molecule in the universe, collisions involving H2 have important implications in astrochemistry. Collisions between hydrogen molecules also represent a prototype for assessing various dynamic methods for understanding fundamental few-body processes. In this work, we develop a new and highly accurate full-dimensional potential energy surface (PES) covering all reactive channels of the H2 + H2 system, which extends our previously reported H2 + H2 nonreactive PES [J. Chem. Theory Comput., 2021, 17, 6747] by adding 39,538 additional ab initio points calculated at the MRCI/AV5Z level in the reactive channels. The global PES is represented with high fidelity (RMSE = 0.6 meV for a total of 79,000 points) by a permutation invariant polynomial neural network (PIP-NN) and is suitable for studying collision-induced dissociation, single-exchange, as well as four-center exchange reactions. Preliminary quasi-classical trajectory studies on the new PIP-NN PES reveal strong vibrational enhancement of all reaction channels.

A reconsideration on the resolution of phase stability analysis using stochastic global optimization methods: Proposal of a reliable set of benchmark problems

Accurate, global Potential Energy Surfaces (PESs) expressed in sum-of-products (SOP) form are a prerequisite for efficient high-dimensional quantum dynamics simulations using the multi-configuration time-dependent Hartree method. This work introduces a methodology for constructing such surfaces by combining hierarchical sparse grid sampling with a single-layer neural network using sinusoidal activation functions (sinNN). The sparse grid strategy provides a rigorous, unbiased discretization of the configuration space, enabling systematic improvement of the PES fidelity, where accuracy is strictly controlled by the refinement level, while successfully mitigating the curse of dimensionality. The sinNN fitting approach leverages a trigonometric factorization identity to maintain a compact SOP form, offering superior numerical stability compared to "standard" exponential-based networks for the molecular systems investigated. We validate this framework by refitting an analytical PES for nitrous acid (HONO). The flexibility of the sparse grid methodology is demonstrated through a dual-reference strategy, where grids centered on distinct isomers are merged to eliminate topological bias. This optimized sampling yields a global PES that reproduces fundamental vibrational transition energies for both trans- and cis-HONO with spectroscopic precision (<2.5 cm-1) and high data efficiency. Finally, the methodology is applied to fit potential energies computed via the AI-enhanced quantum mechanical method (AIQM2). The resulting AIQM2-based PES for HONO reproduces experimental vibrational frequencies with a root mean square deviation of ∼16 cm-1, a performance comparable to high-level abinitio methods. The robustness of the approach is further confirmed on larger molecules, formic acid (HCOOH) and carbamic acid (H2NCOOH), establishing the combination of sparse grid sampling and sinNN fitting as a powerful, automated tool for generating topologically sound, spectroscopic-quality potential energy surfaces.

A global adiabatic potential energy surface (PES) corresponding to the second singlet state 1 1A" (1 1B1) of CH2 has been computed in a similar way as the first singlet state 1 1A' in our previous work [B. Bussery-Honvault et al., J. Chem. Phys., 2001, 115, 10 701]. This PES has a calculated well depth of 79.9 kcal mol(-1) relative to the C(1D) + H2 asymptote and correlates to CH(2pi) + H. It presents large barriers in the C(1D) + H2 arrangement for both collinear and perpendicular geometries but no barrier for geometries about 60 degrees and leads to a sideways insertion mechanism for the reaction C(1D) + H2 --> CH(2pi) + H. The ab initio calculations were carried out for 4644 geometries and the resulting energies were fitted to a many-body expansion. Accurate three-dimensional quantum mechanical scattering calculations have been performed for the C(1D) + H2(v = 0, j = 0) reaction on this ab initio 1 1A" PES in the collision energy range [0-11.5 kcal mol(-1)]. The J = 0 reaction probabilities show dense resonance structures as those obtained with the 1 1A' PES. However some different dynamical features have been found.