Coarse-Grained Crystal Graph Neural Networks for Reticular Materials Design.

Reticular materials, including metal-organic frameworks and covalent organic frameworks, combine the relative ease of synthesis and an impressive range of applications in various fields from gas storage to biomedicine. Diverse properties arise from the variation of building units─metal centers and organic linkers─in almost infinite chemical space. Such variation substantially complicates the experimental design and promotes the use of computational methods. In particular, the most successful artificial intelligence algorithms for predicting the properties of reticular materials are atomic-level graph neural networks, which optionally incorporate domain knowledge. Nonetheless, the data-driven inverse design involving these models suffers from the incorporation of irrelevant and redundant features such as a full atomistic graph and network topology. In this study, we propose a new way of representing materials, aiming to overcome the limitations of existing methods; the message passing is performed on a coarse-grained crystal graph that comprises molecular building units. To highlight the merits of our approach, we assessed the predictive performance and energy efficiency of neural networks built on different materials representations, including composition-based and crystal-structure-aware models. Coarse-grained crystal graph neural networks showed decent accuracy at low computational costs, making them a valuable alternative to omnipresent atomic-level algorithms. Moreover, the presented models can be successfully integrated into an inverse materials design pipeline as estimators of the objective function. Overall, the coarse-grained crystal graph framework is aimed at challenging the prevailing atom-centric perspective on reticular materials design.

Parametrization of Nonbonded Force Field Terms for Metal-Organic Frameworks Using Machine Learning Approach.

The enormous structural and chemical diversity of metal-organic frameworks (MOFs) forces researchers to actively use simulation techniques as often as experiments. MOFs are widely known for their outstanding adsorption properties, so a precise description of the host-guest interactions is essential for high-throughput screening aimed at ranking the most promising candidates. However, highly accurate ab initio calculations cannot be routinely applied to model thousands of structures due to the demanding computational costs. Furthermore, methods based on force field (FF) parametrization suffer from low transferability. To resolve this accuracy-efficiency dilemma, we applied a machine learning (ML) approach: extreme gradient boosting. The trained models reproduced the atom-in-material quantities, including partial charges, polarizabilities, dispersion coefficients, quantum Drude oscillator, and electron cloud parameters, with accuracy similar to the reference data set. The aforementioned FF precursors make it possible to thoroughly describe noncovalent interactions typical for MOF-adsorbate systems: electrostatic, dispersion, polarization, and short-range repulsion. The presented approach can also readily facilitate hybrid atomistic simulation/ML workflows.

A search for a DFT functional for actinide compounds.

Actinide chemistry often lies beyond the applicability domain of the majority of modern theoretical tools due to high computational costs, relativistic effects, or just the absence of actinide data for semiempirical method fitting. On the other hand, radioactivity pushes the usage of computational methods instead of experimental ones. Here, we would like to present a novel relPBE functional as an actinide-fitted version of the PBE0 functional.

Graph Convolutional Neural Networks as "General-Purpose" Property Predictors: The Universality and Limits of Applicability.

Nowadays the development of new functional materials/chemical compounds using machine learning (ML) techniques is a hot topic and includes several crucial steps, one of which is the choice of chemical structure representation. The classical approach of rigorous feature engineering in ML typically improves the performance of the predictive model, but at the same time, it narrows down the scope of applicability and decreases the physical interpretability of predicted results. In this study, we present graph convolutional neural networks (GCNNs) as an architecture that allows for successfully predicting the properties of compounds from diverse domains of chemical space, using a minimal set of meaningful descriptors. The applicability of GCNN models has been demonstrated by a wide range of chemical domain-specific properties. Their performance is comparable to state-of-the-art techniques; however, this architecture exempts from the need to carry out precise feature engineering.

Simple Automatized Tool for Exchange-Correlation Functional Fitting.

Density functional theory is, perhaps, the most popular and convenient tool in computational chemistry. DFT methods allow solving different chemical tasks with a good balance of accuracy and computational time. Dozens of existing functionals cover a majority of possible systems, and the development of new ones is still ongoing. However, despite the existence of different databases with accurate quantum-chemical data, the functional design remains a complicated and time-demanding task. Here, we propose a novel approach for simplifying and accelerating this process. The approach is based on a Bayesian search with stochastic sub-sampling that allows considering the 'history' of fitting steps, reduces the computational time for each step, and avoids overfitting to training data. Besides the general testing of the approach efficiency, we also showed an example of training specialized DFT functionals, outperforming the popular ones. The approach is presented as a free code with built-in analysis tools. Using the code with an appropriate reference database can help in constructing a DFT approximation for a highly specialized task.

Effect of wettability on sessile drop freezing: when superhydrophobicity stimulates an extreme freezing delay.

An increasing number of studies directed at supercooling water droplets on surfaces with different wettabilities have appeared in recent years. This activity has been stimulated by the recognition that water supercooling phenomena can be effectively used to develop methods for protecting outdoor equipment and infrastructure elements against icing and snow accretion. In this article, we discuss the nucleation kinetics of supercooled sessile water droplets on hydrophilic, hydrophobic, and superhydrophobic surfaces under isothermal conditions at temperatures of -8, -10, and -15 °C and a saturated water vapor atmosphere. The statistics of nucleation events for the ensembles of freezing sessile droplets is completed by the detailed analysis of the contact angle temperature dependence and freezing of individual droplets in a saturated vapor atmosphere. We have demonstrated that the most essential freezing delay is characteristic of the superhydrophobic coating on aluminum, with the texture resistant to contact with ice and water. This delay can reach many hours at T = -8 °C and a few minutes at -23 °C. The observed behavior is analyzed on the basis of different nucleation mechanisms. The dissimilarity in the total nucleation rate, detected for two superhydrophobic substrates having the same apparent contact angle of the water drop but different resistivities of surface texture to the contact with water/ice, is associated with the contribution of heterogeneous nucleation on external centers located at the water droplet/air interface.

The carbon footprint of predicting CO2 storage capacity in metal-organic frameworks within neural networks.

While artificial intelligence drives remarkable progress in natural sciences, its broader societal implications are mostly disregarded. In this study, we evaluate environmental impacts of deep learning in materials science through extensive benchmarking. In particular, a set of diverse neural networks is trained for a given supervised learning task to assess greenhouse gas (GHG) emissions during training and inference phases. A chronological perspective showed diminishing returns, manifesting themselves as a 28% decrease in mean absolute error and nearly a 15,000% increase in the carbon footprint of model training in 2016-2022. By means of up-to-date graphics processing units, it is possible to partially offset the immense growth of GHG emissions. Nonetheless, the practice of employing energy-efficient hardware is overlooked by the materials informatics community, as follows from a literature analysis in the field. On the basis of our findings, we encourage researchers to report GHG emissions together with standard performance metrics.

Accurate, interpretable predictions of materials properties within transformer language models.

Property prediction accuracy has long been a key parameter of machine learning in materials informatics. Accordingly, advanced models showing state-of-the-art performance turn into highly parameterized black boxes missing interpretability. Here, we present an elegant way to make their reasoning transparent. Human-readable text-based descriptions automatically generated within a suite of open-source tools are proposed as materials representation. Transformer language models pretrained on 2 million peer-reviewed articles take as input well-known terms such as chemical composition, crystal symmetry, and site geometry. Our approach outperforms crystal graph networks by classifying four out of five analyzed properties if one considers all available reference data. Moreover, fine-tuned text-based models show high accuracy in the ultra-small data limit. Explanations of their internal machinery are produced using local interpretability techniques and are faithful and consistent with domain expert rationales. This language-centric framework makes accurate property predictions accessible to people without artificial-intelligence expertise.

A universal similarity based approach for predictive uncertainty quantification in materials science.

Immense effort has been exerted in the materials informatics community towards enhancing the accuracy of machine learning (ML) models; however, the uncertainty quantification (UQ) of state-of-the-art algorithms also demands further development. Most prominent UQ methods are model-specific or are related to the ensembles of models; therefore, there is a need to develop a universal technique that can be readily applied to a single model from a diverse set of ML algorithms. In this study, we suggest a new UQ measure known as the Δ-metric to address this issue. The presented quantitative criterion was inspired by the k-nearest neighbor approach adopted for applicability domain estimation in chemoinformatics. It surpasses several UQ methods in accurately ranking the predictive errors and could be considered a low-cost option for a more advanced deep ensemble strategy. We also evaluated the performance of the presented UQ measure on various classes of materials, ML algorithms, and types of input features, thus demonstrating its universality.

Size Doesn't Matter: Predicting Physico- or Biochemical Properties Based on Dozens of Molecules.

The use of machine learning in chemistry has become a common practice. At the same time, despite the success of modern machine learning methods, the lack of data limits their use. Using a transfer learning methodology can help solve this problem. This methodology assumes that a model built on a sufficient amount of data captures general features of the chemical compound structure on which it was trained and that the further reuse of these features on a data set with a lack of data will greatly improve the quality of the new model. In this paper, we develop this approach for small organic molecules, implementing transfer learning with graph convolutional neural networks. The paper shows a significant improvement in the performance of the models for target properties with a lack of data. The effects of the data set composition on the model's quality and the applicability domain of the resulting models are also considered.

Aroma Molecules as Dynamic Volatile Surfactants: Functionality beyond the Scent.

Understanding of nonequilibrium processes at dynamic interfaces is indispensable for advancing design and fabrication of solid-state and soft materials. The research presented here unveils specific interfacial behavior of aroma molecules and justifies their usage as multifunctional volatile surfactants. As nonconventional volatile amphiphiles, we study commercially available poorly water-soluble compounds from the classes of synthetic and essential flavor oils. Their disclosed distinctive feature is a high dynamic interfacial activity, so that they decrease the surface tension of aqueous solutions on a time scale of milliseconds. Another potentially useful property of such amphiphiles is their volatility, so that they notably evaporate from interfaces on a time scale of seconds. This behavior allows for control of wetting and spreading processes. A revealed synergetic interfacial behavior of mixtures of conventional and volatile surfactants is attributed to a decrease of the activation barrier as a result of high statistical availability of new sites at the surface upon evaporation of the volatile component. Our results offer promising advantages in manufacturing technologies which involve newly creating interfaces, such as spraying, coating technologies, ink-jet printing, microfluidics, laundry, and stabilization of emulsions in cosmetic and food industry, as well as in geosciences for controlling aerosol formation.