Accelerating Formulation Design via Machine Learning: Generating a High-throughput Shampoo Formulations Dataset.

Liquid formulations are ubiquitous yet have lengthy product development cycles owing to the complex physical interactions between ingredients making it difficult to tune formulations to customer-defined property targets. Interpolative ML models can accelerate liquid formulations design but are typically trained on limited sets of ingredients and without any structural information, which limits their out-of-training predictive capacity. To address this challenge, we selected eighteen formulation ingredients covering a diverse chemical space to prepare an open experimental dataset for training ML models for rinse-off formulations development. The resulting design space has an over 50-fold increase in dimensionality compared to our previous work. Here, we present a dataset of 812 formulations, including 294 stable samples, which cover the entire design space, with phase stability, turbidity, and high-fidelity rheology measurements generated on our semi-automated, ML-driven liquid formulations workflow. Our dataset has the unique attribute of sample-specific uncertainty measurements to train predictive surrogate models.

Experimentally validated inverse design of multi-property Fe-Co-Ni alloys.

This study presents a machine learning (ML) framework aimed at accelerating the discovery of multi-property optimized Fe-Ni-Co alloys, addressing the time-consuming, expensive, and inefficient nature of traditional methods of material discovery, development, and deployment. We compiled a detailed heterogeneous database of the magnetic, electrical, and mechanical properties of Fe-Co-Ni alloys, employing a novel ML-based imputation strategy to address gaps in property data. Leveraging this comprehensive database, we developed predictive ML models using tree-based and neural network approaches for optimizing multiple properties simultaneously. An inverse design strategy, utilizing multi-objective Bayesian optimization (MOBO), enabled the identification of promising alloy compositions. This approach was experimentally validated using high-throughput methodology, highlighting alloys such as Fe66.8Co28Ni5.2 and Fe61.9Co22.8Ni15.3, which demonstrated superior properties. The predicted properties data closely matched experimental data within 14% accuracy. Our approach can be extended to a broad range of materials systems to predict novel materials with an optimized set of properties.

Solvent-Templated Methylammonium-Based Ruddlesden-Popper Perovskites with Short Interlayer Distances.

Two-dimensional (2D) halide perovskites are exquisite semiconductors with great structural tunability. They can incorporate a rich variety of organic species that not only template their layered structures but also add new functionalities to their optoelectronic characteristics. Here, we present a series of new methylammonium (CH3NH3+ or MA)-based 2D Ruddlesden-Popper perovskites templated by dimethyl carbonate (CH3OCOOCH3 or DMC) solvent molecules. We report the synthesis, detailed structural analysis, and characterization of four new compounds: MA2(DMC)PbI4 (n = 1), MA3(DMC)Pb2I7 (n = 2), MA4(DMC)Pb3I10 (n = 3), and MA3(DMC)Pb2Br7 (n = 2). Notably, these compounds represent unique structures with MA as the sole organic cation both within and between the perovskite sheets, while DMC molecules occupy a tight space between the MA cations in the interlayer. They form hydrogen-bonded [MA···DMC···MA]2+ complexes that act as spacers, preventing the perovskite sheets from condensing into each other. We report one of the shortest interlayer distances (∼5.7-5.9 Å) in solvent-incorporated 2D halide perovskites. Furthermore, the synthesized crystals exhibit similar optical characteristics to other 2D perovskite systems, including narrow photoluminescence (PL) signals. The density functional theory (DFT) calculations confirm their direct-band-gap nature. Meanwhile, the phase stability of these systems was found to correlate with the H-bond distances and their strengths, decreasing in the order MA3(DMC)Pb2I7 > MA4(DMC)Pb3I10 > MA2(DMC)PbI4 ∼ MA3(DMC)Pb2Br7. The relatively loosely bound nature of DMC molecules enables us to design a thermochromic cell that can withstand 25 cycles of switching between two colored states. This work exemplifies the unconventional role of the noncharged solvent molecule in templating the 2D perovskite structure.

Constructing custom thermodynamics using deep learning.

One of the most exciting applications of artificial intelligence is automated scientific discovery based on previously amassed data, coupled with restrictions provided by known physical principles, including symmetries and conservation laws. Such automated hypothesis creation and verification can assist scientists in studying complex phenomena, where traditional physical intuition may fail. Here we develop a platform based on a generalized Onsager principle to learn macroscopic dynamical descriptions of arbitrary stochastic dissipative systems directly from observations of their microscopic trajectories. Our method simultaneously constructs reduced thermodynamic coordinates and interprets the dynamics on these coordinates. We demonstrate its effectiveness by studying theoretically and validating experimentally the stretching of long polymer chains in an externally applied field. Specifically, we learn three interpretable thermodynamic coordinates and build a dynamical landscape of polymer stretching, including the identification of stable and transition states and the control of the stretching rate. Our general methodology can be used to address a wide range of scientific and technological applications.

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.

Closed-Loop Multi-Objective Optimization for Cu-Sb-S Photo-Electrocatalytic Materials' Discovery.

Copper antimony sulfides are regarded as promising catalysts for photo-electrochemical water splitting because of their earth abundance and broad light absorption. The unique photoactivity of copper antimony sulfides is dependent on their various crystalline structures and atomic compositions. Here, a closed-loop workflow is built, which explores Cu-Sb-S compositional space to optimize its photo-electrocatalytic hydrogen evolution from water, by integrating a high-throughput robotic platform, characterization techniques, and machine learning (ML) optimization workflow. The multi-objective optimization model discovers optimum experimental conditions after only nine cycles of integrated experiments-machine learning loop. Photocurrent testing at 0 V versus reversible hydrogen electrode (RHE) confirms the expected correlation between the materials' properties and photocurrent. An optimum photocurrent of -186 µA cm-2 is observed on Cu-Sb-S in the ratio of 9:45:46 in the form of single-layer coating on F-doped SnO2 (FTO) glass with a corresponding bandgap of 1.85 eV and 63.2% Cu1+ /Cu species content. The targeted intelligent search reveals a nonobvious CuSbS composition that exhibits 2.3 times greater activity than baseline results from random sampling.

Transfer Learning of Full Molecular Weight Distributions via High-Throughput Computer-Controlled Polymerization.

The skew and shape of the molecular weight distribution (MWD) of polymers have a significant impact on polymer physical properties. Standard summary metrics statistically derived from the MWD only provide an incomplete picture of the polymer MWD. Machine learning (ML) methods coupled with high-throughput experimentation (HTE) could potentially allow for the prediction of the entire polymer MWD without information loss. In our work, we demonstrate a computer-controlled HTE platform that is able to run up to 8 unique variable conditions in parallel for the free radical polymerization of styrene. The segmented-flow HTE system was equipped with an inline Raman spectrometer and offline size exclusion chromatography (SEC) to obtain time-dependent conversion and MWD, respectively. Using ML forward models, we first predict monomer conversion, intrinsically learning varying polymerization kinetics that change for each experimental condition. In addition, we predict entire MWDs including the skew and shape as well as SHAP analysis to interpret the dependence on reagent concentrations and reaction time. We then used a transfer learning approach to use the data from our high-throughput flow reactor to predict batch polymerization MWDs with only three additional data points. Overall, we demonstrate that the combination of HTE and ML provides a high level of predictive accuracy in determining polymerization outcomes. Transfer learning can allow exploration outside existing parameter spaces efficiently, providing polymer chemists with the ability to target the synthesis of polymers with desired properties.

Automated electrokinetic stretcher for manipulating nanomaterials.

In this work, we present an automated platform for trapping and stretching individual micro- and nanoscale objects in solution using electrokinetic forces. The platform can trap objects at the stagnation point of a planar elongational electrokinetic field for long time scales, as demonstrated by the trapping of <100 nm polystyrene beads and DNA molecules for minutes, with a standard deviation in displacement from the trap center <1 µm. This capability enables the stretching of deformable nanoscale objects in a high-throughput fashion, as illustrated by the stretching of more than 400 DNA molecules within ∼4 hours. The flexibility of the electrokinetic stretcher opens up numerous possibilities for complex manipulation, with sequential stretching of a molecule at different voltages and multiple stretch-relaxation cycles of the same molecule as examples. The platform described provides an automated, high-throughput method to track and manipulate objects for real-time studies of micro- and nanoscale systems.

Sub-10 nm Mixing and Alloying of Cu-Ag and Cu-Ni via Accelerated Solid Diffusion.

Development of nanoscale multicomponent solid inorganic materials is often hindered by slow solid diffusion kinetics and poor precursor mixing in conventional solid-state synthesis. These shortcomings can be alleviated by combining nanosized precursor mixtures and low temperature reaction, which could reduce crystal growth and accelerate the solid diffusion at the same time. However, high throughput production of nanoparticle mixtures with tunable composition via conventional synthesis is very challenging. In this work, we demonstrate that ∼10 nm homogeneous mixing of sub-10 nm nanoparticles can be achieved via spark nanomixing at room temperature and pressure. Kinetically driven Spark Plasma Discharge nanoparticle generation and ambient processing conditions limit particle coarsening and agglomeration, resulting in sub-10 nm primary particles of as-deposited films. The intimate mixing of these nanosized precursor particles enables intraparticle diffusion and formation of Cu/Ni nanoalloy during subsequent low temperature annealing at 100 °C. We also discovered that cross-particle diffusion is promoted during the low-temperature sulfurization of Cu/Ag which tends to phase-segregate, eventually leading to the growth of sulfide nanocrystals and improved homogeneity. High elemental homogeneity, small diffusion path lengths, and high diffusibility synergically contribute to faster diffusion kinetics of sub-10 nm nanoparticle mixtures. The combination of ∼10 nm homogeneous precursors via spark nanomixing, low-temperature annealing, and a wide range of potentially compatible materials makes our approach a good candidate as a general platform toward accelerated solid state synthesis of nanomaterials.

Ordering-Dependent Hydrogen Evolution and Oxygen Reduction Electrocatalysis of High-Entropy Intermetallic Pt4 FeCoCuNi.

Disordered solid-solution high-entropy alloys have attracted wide research attention as robust electrocatalysts. In comparison, ordered high-entropy intermetallics have been hardly explored and the effects of the degree of chemical ordering on catalytic activity remain unknown. In this study, a series of multicomponent intermetallic Pt4 FeCoCuNi nanoparticles with tunable ordering degrees is fabricated. The transformation mechanism of the multicomponent nanoparticles from disordered structure into ordered structure is revealed at the single-particle level, and it agrees with macroscopic analysis by selected-area electron diffraction and X-ray diffraction. The electrocatalytic performance of Pt4 FeCoCuNi nanoparticles correlates well with their crystal structure and electronic structure. It is found that increasing the degree of ordering promotes electrocatalytic performance. The highly ordered Pt4 FeCoCuNi achieves the highest mass activities toward both acidic oxygen reduction reaction (ORR) and alkaline hydrogen evolution reaction (HER) which are 18.9-fold and 5.6-fold higher than those of commercial Pt/C, respectively. The experiment also shows that this catalyst demonstrates better long-term stability than both partially ordered and disordered Pt4 FeCoCuNi as well as Pt/C when subject to both HER and ORR. This ordering-dependent structure-property relationship provides insight into the rational design of catalysts and stimulates the exploration of many other multicomponent intermetallic alloys.

Predicting Synthesizability using Machine Learning on Databases of Existing Inorganic Materials.

Defining the metric for synthesizability and predicting new compounds that can be experimentally realized in the realm of data-driven research is a pressing problem in contemporary materials science. The increasing computational power and advancements in machine learning (ML) algorithms provide a new avenue to solve the synthesizability challenge. In this work, using the Inorganic Crystal Structure Database (ICSD) and the Materials Project (MP) database, we represent crystal structures in Fourier-transformed crystal properties (FTCP) representation and use a deep learning model to predict synthesizability in the form of a synthesizability score (SC). Such an SC model, as a synthesizability filter for new materials, enables an efficient and accurate classification to identify promising material candidates. The SC prediction model achieved 82.6/80.6% (precision/recall) overall accuracy in predicting ternary crystal materials. We also trained the SC model by only considering compounds uploaded on the MP before 2015 as the training set and testing on multiple sets of materials uploaded after 2015. In the post-2019 test set, we obtain a high 88.60% true positive rate accuracy, coupled with 9.81% precision, indicating that newly added materials remain unexplored and have high synthesis potential. Further, we provide a list of 100 materials predicted to be synthesizable from this post-2019 dataset (highest SC) for future studies, and our SC model, as a validation filter, is beneficial for future material screening and discovery.

Surface charge as activity descriptors for electrochemical CO2 reduction to multi-carbon products on organic-functionalised Cu.

Intensive research in electrochemical CO2 reduction reaction has resulted in the discovery of numerous high-performance catalysts selective to multi-carbon products, with most of these catalysts still being purely transition metal based. Herein, we present high and stable multi-carbon products selectivity of up to 76.6% across a wide potential range of 1 V on histidine-functionalised Cu. In-situ Raman and density functional theory calculations revealed alternative reaction pathways that involve direct interactions between adsorbed histidine and CO2 reduction intermediates at more cathodic potentials. Strikingly, we found that the yield of multi-carbon products is closely correlated to the surface charge on the catalyst surface, quantified by a pulsed voltammetry-based technique which proved reliable even at very cathodic potentials. We ascribe the surface charge to the population density of adsorbed species on the catalyst surface, which may be exploited as a powerful tool to explain CO2 reduction activity and as a proxy for future catalyst discovery, including organic-inorganic hybrids.

Tuning Electronic Structure and Composition of FeNi Nanoalloys for Enhanced Oxygen Evolution Electrocatalysis via a General Synthesis Strategy.

Developing low-cost and efficient oxygen evolution electrocatalysts is key to decarbonization. A facile, surfactant-free, and gram-level biomass-assisted fast heating and cooling synthesis method is reported for synthesizing a series of carbon-encapsulated dense and uniform FeNi nanoalloys with a single-phase face-centered-cubic solid-solution crystalline structure and an average particle size of sub-5 nm. This method also enables precise control of both size and composition. Electrochemical measurements show that among Fex Ni(1- x ) nanoalloys, Fe0.5 Ni0.5 has the best performance. Density functional theory calculations support the experimental findings and reveal that the optimally positioned d-band center of O-covered Fe0.5 Ni0.5 renders a half-filled antibonding state, resulting in moderate binding energies of key reaction intermediates. By increasing the total metal content from 25 to 60 wt%, the 60% Fe0.5 Ni0.5 /40% C shows an extraordinarily low overpotential of 219 mV at 10 mA cm-2 with a small Tafel slope of 23.2 mV dec-1 for the oxygen evolution reaction, which are much lower than most other FeNi-based electrocatalysts and even the state-of-the-art RuO2 . It also shows robust durability in an alkaline environment for at least 50 h. The gram-level fast heating and cooling synthesis method is extendable to a wide range of binary, ternary, quaternary nanoalloys, as well as quinary and denary high-entropy-alloy nanoparticles.

Defect Passivation Using a Phosphonic Acid Surface Modifier for Efficient RP Perovskite Blue-Light-Emitting Diodes.

Defect management strategies are vital for enhancing the performance of perovskite-based optoelectronic devices, such as perovskite-based light-emitting diodes (PeLEDs). As additives can fucntion both as acrystallization modifier and/or defect passivator, a thorough study on the roles of additives is essential, especially for blue emissive Pe-LEDs, where the emission is strictly controlled by the n-domain distribution of the Ruddlesden-Popper (RP, L2An-1PbnX3n+1, where L refers to a bulky cation, while A and X are monovalent cation, and halide anion, respectively) perovskite films. Of the various additives that are available, octyl phosphonic acid (OPA) is of immense interest because of its ability to bind with uncoordinated Pb2+ ( notorious for nonradiative recombination) and therefore passivates them. Here, with the help of various spectroscopic techniques, such as X-ray photon-spectroscopy (XPS), Fourier-transform infrared spectroscopy (FTIR), and photoluminescence quantum yield (PLQY) measurements, we demonstrate the capability of OPA to bind and passivate unpaired Pb2+ defect sites. Modification to crystallization promoting higher n-domain formation is also observed from steady-state and transient absorption (TA) measurements. With OPA treatment, both the PLQY and EQE of the corresponding PeLED showed improvements up to 53% and 3.7% at peak emission wavelength of 485 nm, respectively.

Modulation of Spin Dynamics in 2D Transition-Metal Dichalcogenide via Strain-Driven Symmetry Breaking.

Transition metal dichalcogenides (TMDs) possess intrinsic spin-orbit interaction (SOI) with high potential to be exploited for various quantum phenomena. SOI allows the manipulation of spin degree of freedom by controlling the carrier's orbital motion via mechanical strain. Here, strain modulated spin dynamics in bilayer MoS2 field-effect transistors (FETs) fabricated on crested substrates are demonstrated. Weak antilocalization (WAL) is observed at moderate carrier concentrations, indicating additional spin relaxation path caused by strain fields arising from substrate crests. The spin lifetime is found to be inversely proportional to the momentum relaxation time, which follows the Dyakonov-Perel spin relaxation mechanism. Moreover, the spin-orbit splitting is obtained as 37.5 ± 1.4 meV, an order of magnitude larger than the theoretical prediction for monolayer MoS2 , suggesting the strain enhanced spin-lattice coupling. The work demonstrates strain engineering as a promising approach to manipulate spin degree of freedom toward new functional quantum devices.

Electronic transport descriptors for the rapid screening of thermoelectric materials.

The discovery of novel materials for thermoelectric energy conversion has potential to be accelerated by data-driven screening combined with high-throughput calculations. One way to increase the efficacy of successfully choosing a candidate material is through its evaluation using transport descriptors. Using a data-driven screening, we selected 12 potential candidates in the trigonal ABX2 family, followed by charge transport property simulations from first principles. The results suggest that carrier scattering processes in these materials are dominated by ionised impurities and polar optical phonons, contrary to the oft-assumed acoustic-phonon-dominated scattering. Using these data, we further derive ground-state transport descriptors for the carrier mobility and the thermoelectric powerfactor. In addition to low carrier mass, high dielectric constant was found to be an important factor towards high carrier mobility. A quadratic correlation between dielectric constant and transport performance was established and further validated with literature. Looking ahead, dielectric constant can potentially be exploited as an independent criterion towards improved thermoelectric performance. Combined with calculations of thermal conductivity including Peierls and inter-branch coherent contributions, we conclude that the trigonal ABX2 family has potential as high performance thermoelectrics in the intermediate temperature range for low grade waste heat harvesting.

Comparing data driven and physics inspired models for hopping transport in organic field effect transistors.

The past few decades have seen an uptick in the scope and range of device applications of organic semiconductors, such as organic field-effect transistors, organic photovoltaics and light-emitting diodes. Several researchers have studied electrical transport in these materials and proposed physical models to describe charge transport with different material parameters, with most disordered semiconductors exhibiting hopping transport. However, there exists a lack of a consensus among the different models to describe hopping transport accurately and uniformly. In this work, we first evaluate the efficacy of using a purely data-driven approach, i.e., symbolic regression, in unravelling the relationship between the measured field-effect mobility and the controllable inputs of temperature and gate voltage. While the regressor is able to capture the scaled mobility well with mean absolute error (MAE) ~ O(10-2), better than the traditionally used hopping transport model, it is unable to derive physically interpretable input-output relationships. We then examine a physics-inspired renormalization approach to describe the scaled mobility with respect to a scale-invariant reference temperature. We observe that the renormalization approach offers more generality and interpretability with a MAE of the ~ O(10-1), still better than the traditionally used hopping model, but less accurate as compared to the symbolic regression approach. Our work shows that physics-based approaches are powerful compared to purely data-driven modelling, providing an intuitive understanding of data with extrapolative ability.

High thermoelectric performance enabled by convergence of nested conduction bands in Pb7Bi4Se13 with low thermal conductivity.

Thermoelectrics enable waste heat recovery, holding promises in relieving energy and environmental crisis. Lillianite materials have been long-term ignored due to low thermoelectric efficiency. Herein we report the discovery of superior thermoelectric performance in Pb7Bi4Se13 based lillianites, with a peak figure of merit, zT of 1.35 at 800 K and a high average zT of 0.92 (450-800 K). A unique quality factor is established to predict and evaluate thermoelectric performances. It considers both band nonparabolicity and band gaps, commonly negligible in conventional quality factors. Such appealing performance is attributed to the convergence of effectively nested conduction bands, providing a high number of valley degeneracy, and a low thermal conductivity, stemming from large lattice anharmonicity, low-frequency localized Einstein modes and the coexistence of high-density moiré fringes and nanoscale defects. This work rekindles the vision that Pb7Bi4Se13 based lillianites are promising candidates for highly efficient thermoelectric energy conversion.

Metastable 1T'-phase group VIB transition metal dichalcogenide crystals.

Metastable 1T'-phase transition metal dichalcogenides (1T'-TMDs) with semi-metallic natures have attracted increasing interest owing to their uniquely distorted structures and fascinating phase-dependent physicochemical properties. However, the synthesis of high-quality metastable 1T'-TMD crystals, especially for the group VIB TMDs, remains a challenge. Here, we report a general synthetic method for the large-scale preparation of metastable 1T'-phase group VIB TMDs, including WS2, WSe2, MoS2, MoSe2, WS2xSe2(1-x) and MoS2xSe2(1-x). We solve the crystal structures of 1T'-WS2, -WSe2, -MoS2 and -MoSe2 with single-crystal X-ray diffraction. The as-prepared 1T'-WS2 exhibits thickness-dependent intrinsic superconductivity, showing critical transition temperatures of 8.6 K for the thickness of 90.1 nm and 5.7 K for the single layer, which we attribute to the high intrinsic carrier concentration and the semi-metallic nature of 1T'-WS2. This synthesis method will allow a more systematic investigation of the intrinsic properties of metastable TMDs.