In Pursuit of the Exceptional: Research Directions for Machine Learning in Chemical and Materials Science

Can Deep Learning Search for Exceptional Chiroptical Properties? The Halogenated [6]Helicene Case

The relationship between chemical structure and chiroptical properties is not always clearly understood. Nowadays, efforts to develop new systems with enhanced optical properties follow the trial-error method. A large number of data would allow us to obtain more robust conclusions and guide research toward molecules with practical applications. In this sense, in this work we predict the chiroptical properties of millions of halogenated [6]helicenes in terms of the rotatory strength (R). We have used DFT calculations to randomly create derivatives including from 1 to 16 halogen atoms, that were then used as a data set to train different deep neural network models. These models allow us to i) predict the Rmax for any halogenated [6]helicene with a very low computational cost, and ii) to understand the physical reasons that favour some substitutions over others. Finally, we synthesized derivatives with higher predicted Rmax obtaining excellent correlation among the values obtained experimentally and the predicted ones.

Embedding human knowledge in material screening pipeline as filters to identify novel synthesizable inorganic materials

How might one embed a chemist's knowledge into an automated materials-discovery pipeline? In generative design for inorganic crystalline materials, generating candidate compounds is no longer a bottleneck - there are now synthetic datasets of millions of compounds. However, weeding out unsynthesizable or difficult to synthesize compounds remains an outstanding challenge. Post-generation "filters" have been proposed as a means of embedding human domain knowledge, either in the form of scientific laws or rules of thumb. Examples include charge neutrality, electronegativity balance, and energy above hull. Some filters are "hard" and some are "soft" - for example, it is difficult to envision creating a stable compound while violating the rule of charge neutrality; however, several compounds break the Hume-Rothery rules. It is therefore natural to wonder: can one compile a comprehensive list of "filters" that embed domain knowledge, adopt a principled approach to classifying them as either non-conditional or conditional "filters," and envision a software environment to implement combinations of these in a systematic manner? In this commentary we explore such questions, "filters" for screening of novel inorganic compounds for synthesizability.

A Vision for the Future of Materials Innovation and How to Fast-Track It with Services

Today, we witness how our scientific ecosystem tries to accommodate a new form of intelligence, artificial intelligence (AI). To make the most of AI in materials science, we need to make the data from computational and laboratory experiments machine-readable, but while that works well for computational experiments, integrating laboratory hardware into a digital workflow seems to be a formidable barrier toward that goal. This paper explores measurement services as a way to lower this barrier. I envision the Entity for Multivariate Material Analysis (EMMA), a centralized service that offers measurement bundles tailored for common research needs. EMMA's true strength, however, lies in its software ecosystem to treat, simulate, and store the measured data. Its close integration of measurements and their simulation not only produces metadata-rich experimental data but also provides a self-consistent framework that links the sample with a snapshot of its digital twin. If EMMA was to materialize, its database of experimental data connected to digital twins could serve as the fuel for physics-informed machine learning and a trustworthy horizon of expectations for material properties. This drives material innovation since knowing the statistics helps find the exceptional. This is the EMMA approach: fast-tracking material innovation by integrated measurement and software services.

Dynamical Electron Correlation and the Chemical Bond. IV. Covalent Bonds in A2 Molecules (A = N-As and F-Br)

In a series of recent papers, we investigated the effect of dynamical electron correlation on the potential energy curves and spectroscopic constants of several diatomic molecules, including the simple diatomic hydrides (AH) and the more complex diatomic fluorides (AF) and homonuclear diatomic molecules (A2) with A = B-F (AF) or A = C-F (A2), respectively. Our goal was to understand the dependence of the dynamical electron correlation energy, EDEC, on the internuclear distance, R, and quantify how dynamical electron correlation influences the spectroscopic constants (De, Re, and ωe) of these molecules. At large R, we found that the magnitude of EDEC(R) had a simple dependence on R, with EDEC(R) increasing nearly exponentially with decreasing R. However, as R continued to decrease, there were significant variations in EDEC(R). These variations led to differing changes in the predicted spectroscopic constants of the molecules. In many molecules, the changes in EDEC(R) could be correlated with changes in the underlying spin-coupled generalized valence bond wave function, either in the orbitals or the spin-coupling coefficients. In the current paper, we extend these studies to higher main group elements, comparing the effects of EDEC(R) on P2 and As2 versus N2, and on Cl2 and Br2 versus F2. We find that there are significant differences between the effects of dynamical electron correlation on the molecules in the first and subsequent rows of the periodic table.

Large Language Models for Inorganic Synthesis Predictions

We evaluate the effectiveness of pretrained and fine-tuned large language models (LLMs) for predicting the synthesizability of inorganic compounds and the selection of precursors needed to perform inorganic synthesis. The predictions of fine-tuned LLMs are comparable to─and sometimes better than─recent bespoke machine learning models for these tasks but require only minimal user expertise, cost, and time to develop. Therefore, this strategy can serve both as an effective and strong baseline for future machine learning studies of various chemical applications and as a practical tool for experimental chemists.

Active learning streamlines development of high performance catalysts for higher alcohol synthesis

Developing efficient catalysts for syngas-based higher alcohol synthesis (HAS) remains a formidable research challenge. The chain growth and CO insertion requirements demand multicomponent materials, whose complex reaction dynamics and extensive chemical space defy catalyst design norms. We present an alternative strategy by integrating active learning into experimental workflows, exemplified via the FeCoCuZr catalyst family. Our data-aided framework streamlines navigation of the extensive composition and reaction condition space in 86 experiments, offering >90% reduction in environmental footprint and costs over traditional programs. It identifies the Fe65Co19Cu5Zr11 catalyst with optimized reaction conditions to attain higher alcohol productivities of 1.1 gHA h-1 gcat-1 under stable operation for 150 h on stream, a 5-fold improvement over typically reported yields. Characterization reveals catalytic properties linked to superior activities despite moderate higher alcohol selectivities. To better reflect catalyst demands, we devise multi-objective optimization to maximize higher alcohol productivity while minimizing undesired CO2 and CH4 selectivities. An intrinsic trade-off between these metrics is uncovered, identifying Pareto-optimal catalysts not readily discernible by human experts. Finally, based on feature-importance analysis, we formulate data-informed guidelines to develop performance-specific FeCoCuZr systems. This approach goes beyond existing HAS catalyst design strategies, is adaptable to broader catalytic transformations, and fosters laboratory sustainability.

Revisiting El-Sayed Synthesis: Bayesian Optimization for Revealing New Insights during the Growth of Gold Nanorods

In diverse fields, machine learning (ML) has sparked transformative changes, primarily driven by the wealth of big data. However, an alternative approach seeks to mine insights from "precious data", offering the possibility to reveal missed knowledge and escape potential knowledge traps. In this context, Bayesian optimization (BO) protocols have emerged as crucial tools for optimizing the synthesis and discovery of a broad spectrum of compounds including nanoparticles. In our work, we aimed to go beyond the commonly explored experimental conditions and showcase a workflow capable of unearthing fresh insights, even in well-studied research domains. The growth of AuNRs is a nonequilibrium process that remains poorly understood despite the presence of well-established seeded growth protocols. Traditional research aimed at understanding the mechanism of AuNR growth has primarily relied on altering one reaction condition at a time. While these studies are undeniably valuable, they often fail to capture the synergies between different reaction conditions, thus constraining the depth of insights they can offer. In the present study, we exploit BO, to identify diverse experimental conditions yielding AuNRs with similar spectroscopic characteristics. Notably, we identify viable and accelerated synthesis conditions involving elevated temperatures (36-40 °C) as well as high ascorbic acid concentrations. More importantly, we note that ascorbic acid and temperature can modulate each other's undesirable influences on the growth of AuNRs. Finally, by harnessing the power of interpretable ML algorithms, complemented by our deep chemical understanding, we revisited the established hierarchical relationships among reaction conditions that impact the El-Sayed-based growth of AuNRs.

Physics-Guided Dual Self-Supervised Learning for Structure-Based Material Property Prediction

Deep learning models have been widely used for high-performance material property prediction. However, training such models usually requires a large amount of labeled data, which are usually unavailable. Self-supervised learning (SSL) methods have been proposed to address this data scarcity issue. Herein, we present DSSL, a physics-guided dual SSL framework, for graph neural network-based material property prediction, which combines node masking-based generative SSL with atomic coordinate perturbation-based contrastive SSL strategies to capture local and global information about input crystals. Moreover, we achieve physics-guided pretraining by using the macroproperty (e.g., elasticity)-related microproperty prediction of atomic stiffness as an additional pretext task. We pretrain our DSSL model on the Materials Project database and fine-tune it with 10 material property data sets. The experimental results demonstrate that teaching neural networks some physics using the SSL strategy can afford ≤26.89% performance improvement compared to that of the baseline models.

Ultrasound and sonochemistry enhance education outcomes: From fundamentals and applied research to entrepreneurial potential

With this manuscript we aim to initiate a discussion specific to educational actions around ultrasonics sonochemistry. The importance of these actions does not just derive from a mere pedagogical significance, but they can be an exceptional tool for illustrating various concepts in other disciplines, such as process intensification and microfluidics. Sonochemistry is currently a far-reaching discipline extending across different scales of applicability, from the fundamental physics of tiny bubbles and molecules, up to process plants. This review is part of a special issue in Ultrasonics Sonochemistry, where several scholars have shared their experiences and highlighted opportunities regarding ultrasound as an education tool. The main outcome of our work is that teaching and mentorship in sonochemistry are highly needed, with a balanced technical and scientific knowledge to foster skills and implement safe protocols. Applied research typically features the use of ultrasound as ancillary, to merely enhance a given process and often leading to poorly conceived experiments and misunderstanding of the actual effects. Thus, our scientific community must build a consistent culture and monitor reproducible practices to rigorously generate new knowledge on sonochemistry. These practices can be implemented in teaching sonochemistry in classrooms and research laboratories. We highlight ways to collectively provide a potentially better training for scientists, invigorating academic and industry-oriented careers. A salient benefit for education efforts is that sonochemistry-based projects can serve multidisciplinary training, potentially gathering students from different disciplines, such as physics, chemistry and bioengineering. Herein, we discuss challenges, opportunities, and future avenues to assist in designing courses and research programs based on sonochemistry. Additionally, we suggest simple experiments suitable for teaching basic physicochemical principles at the undergraduatelevel. We also provide arguments and recommendations oriented towards graduate and postdoctoral students, in academia or industry to be more entrepreneurial. We have identified that sonochemistry is consistently seen as a 'green' or sustainable tool, which particular appeal to process intensification approaches, including microfluidics and materials science. We conclude that a globally aligned pedagogical initiative and constantly updated educational tools will help to sustain a virtuous cycle in STEM and industrial applications of sonochemistry.

Accelerated chemical science with AI

In light of the pressing need for practical materials and molecular solutions to renewable energy and health problems, to name just two examples, one wonders how to accelerate research and development in the chemical sciences, so as to address the time it takes to bring materials from initial discovery to commercialization. Artificial intelligence (AI)-based techniques, in particular, are having a transformative and accelerating impact on many if not most, technological domains. To shed light on these questions, the authors and participants gathered in person for the ASLLA Symposium on the theme of 'Accelerated Chemical Science with AI' at Gangneung, Republic of Korea. We present the findings, ideas, comments, and often contentious opinions expressed during four panel discussions related to the respective general topics: 'Data', 'New applications', 'Machine learning algorithms', and 'Education'. All discussions were recorded, transcribed into text using Open AI's Whisper, and summarized using LG AI Research's EXAONE LLM, followed by revision by all authors. For the broader benefit of current researchers, educators in higher education, and academic bodies such as associations, publishers, librarians, and companies, we provide chemistry-specific recommendations and summarize the resulting conclusions.

Predicting Poisson's Ratio: A Study of Semisupervised Anomaly Detection and Supervised Approaches

Auxetics are a rare class of materials that exhibit a negative Poisson's ratio. The existence of these auxetic materials is rare but has a large number of applications in the design of exotic materials. We build a complete machine learning framework to detect Auxetic materials as well as Poisson's ratio of non-auxetic materials. A semisupervised anomaly detection model is presented, which is capable of separating out the auxetics materials (treated as an anomaly) from an unknown database with an average precision of 0.64. Another regression model (supervised) is also created to predict the Poisson's ratio of non-auxetic materials with an R2 of 0.82. Additionally, this regression model helps us to find the optimal features for the anomaly detection model. This methodology can be generalized and used to discover materials with rare physical properties.