Feature Selection in Machine Learning for Perovskite Materials Design and Discovery

Feature mining for thermoelectric materials based on interpretable machine learning

In previous laboratory preparation processes, the selection and proportioning of specific experimental parameters often stemmed from the empirical experience of predecessors, necessitating the derivation of the optimal scheme for the target experiment through extensive trial and error. This process typically required the consumption of substantial resources and time. Thanks to the advancement of machine learning technologies, these have gradually become a powerful tool for addressing complex functional problems in material optimization. This study is based on a collected database of thermoelectric materials, aiming to clarify the correspondence between physical characteristics and the thermoelectric figure of merit, zT. The key features that can directly affect the experimental results are identified, and the features that indirectly affect the experimental results are also analyzed. By employing interpretable machine learning methods, this research analyzes critical molecular features in the characterized data, utilizing feature engineering techniques to construct and optimize machine learning models for the selected key features. Furthermore, in the subsequent model fitting and analysis phase, by comparing the efficiency of different feature combinations, the optimal feature descriptors for the current experimental system were determined. The adoption of the aforementioned improved methods endowed the related models with the potential for high-throughput screening, thereby further enhancing the efficiency of experimental optimization.

Machine learning-enhanced band gaps prediction for low-symmetry double and layered perovskites

Density functional theory (DFT) calculations are widely used for material property prediction, but their computational cost can hinder the discovery of novel perovskites. This work explores machine learning (ML) as a faster alternative for predicting band gaps in complex perovskites, focusing on low-symmetry double and layered structures. We employ Support Vector Regression (SVR), Random Forest Regression (RFR), Gradient Boosting Regression (GBR), and Extreme Gradient Boosting (XGBoost) to predict both direct and indirect band gaps. Model performance is evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared (R²) metrics. Our results reveal SVR as the most effective general model for predicting band gaps in both double and layered perovskites. Interestingly, for double perovskites specifically, XGBoost achieves even higher accuracy when incorporating derivative discontinuity as a feature. Feature importance analysis identifies the standard deviation of valence charges ("Valence (std)") as the most critical factor for band gap prediction across all studied perovskites. This research demonstrates the potential of ML for efficient and accurate band gap prediction in complex perovskites, accelerating material discovery efforts.

Plasmonics Meets Perovskite Photovoltaics: Innovations and Challenges in Boosting Efficiency

Perovskite solar cells (PSCs) have garnered immense attention in recent years due to their outstanding optoelectronic properties and cost-effective fabrication methods, establishing them as promising candidates for next-generation photovoltaic technologies. Among the diverse strategies aimed at enhancing the power conversion efficiency (PCE) of PSCs, the incorporation of plasmonic nanoparticles has emerged as a pioneering approach. This review summarizes the latest research advancements in the utilization of plasmonic nanoparticles to enhance the performance of PSCs. We delve into the fundamental principles of plasmonic resonance and its interaction with perovskite materials, highlighting how localized surface plasmons can effectively broaden light absorption, facilitate hot-electron transfer (HET), and optimize charge separation dynamics. Recent strategies, including the design of tailored metal nanoparticles (MNPs), gratings, and hybrid plasmonic-photonic architectures, are critically evaluated for their efficacy in enhancing light trapping, increasing photocurrent, and mitigating charge recombination. Additionally, this review addresses the challenges associated with the integration of plasmonic elements into PSCs, including issues of scalability, compatibility, and cost-effectiveness. Finally, the review provides insights into future research directions aimed at advancing the field, thereby paving the way for next-generation, high-performance perovskite-based photovoltaic technologies.

Predicting Yield Strength and Plastic Elongation in Body-Centered Cubic High-Entropy Alloys

We employ machine learning (ML) to predict the yield stress and plastic strain of body-centered cubic (BCC) high-entropy alloys (HEAs) in the compression test. Our machine learning model leverages currently available databases of BCC and BCC+B2 entropy alloys, using feature engineering to capture electronic factors, atomic ordering from mixing enthalpy, and the D parameter related to stacking fault energy. The model achieves low Root Mean Square Errors (RMSE). Utilizing Random Forest Regression (RFR) and Genetic Algorithms for feature selection, our model excels in both predictive accuracy and interpretability. Rigorous 10-fold cross-validation ensures robust generalization. Our discussion delves into feature importance, highlighting key predictors and their impact on mechanical properties. This work provides an important step toward designing high-performance structural high-entropy alloys, providing a powerful tool for predicting mechanical properties and identifying new alloys with superior strength and ductility.

A data-driven QSPR model for screening organic corrosion inhibitors for carbon steel using machine learning techniques

Machine learning (ML) techniques have shown great potential for screening corrosion inhibitors. In this study, a data-driven quantitative structure-property relationship (QSPR) model using the gradient boosting decision tree (GB) algorithm combined with the permutation feature importance (PFI) technique was developed to predict the corrosion inhibition efficiency (IE) of organic compounds on carbon steel. The results showed that the PFI method effectively selected the molecular descriptors most relevant to the IE. Using these important molecular descriptors, an IE predictive model was trained on a dataset encompassing various categories of organic corrosion inhibitors for carbon steel, achieving RMSE, MAE, and R2 of 6.40%, 4.80%, and 0.72, respectively. The integration of GB with PFI within the ML workflow demonstrated significantly enhanced IE predictive capability compared to previously reported ML models. Subsequent assessments involved the application of the trained model to drug-based corrosion inhibitors. The model demonstrates robust predictive capability when validated on available and our own experimental results. Furthermore, the model has been employed to predict IE for more than 1500 drug compounds, suggesting five novel drug compounds with the highest predicted IE on carbon steel. The developed ML workflow and associated model will be useful in accelerating the development of next-generation corrosion inhibitors for carbon steel.

Room-Temperature Smart Sensor Based on Indium Acetate-Functionalized Perovskite CsPbBr3 Nanocrystals for Monitoring Electrolyte in Lithium-Ion Batteries

Monitoring electrolyte components is an effective means of determining the safety status of lithium-ion batteries. In this study, indium acetate was taken as a ligand to functionalize perovskite CsPbBr3 nanocrystals, and then the room-temperature electrolyte sensor based on CsPbBr3 nanocrystals with ligand indium acetate was prepared. The sensor offers high response, long-term stability (21 days), and low detection limits for ethyl methyl carbonate (10 ppm), diethyl carbonate (10 ppm), and ethyl butyrate (1 ppm) gases at room temperature and boasts a fast response/recovery time (1500 ppm, 58.27/103.82 s, 33.58/40.62 s, and 45.05/103.08 s, respectively). Density functional theory results show that the gas sensitivity comes from the adsorption of an electrolyte, which changes the density-of-state distribution so that the electrical response curve changes. And using computational fluid dynamics simulation, it was found that the time required for gas detection by the built-in sensor (3.1 s) was 8.7 times shorter than that of the implantable sensor. This work provides inspiration and rationale for embedding and integrating room-temperature sensors into lithium-ion batteries to monitor safety and health conditions.

Intermolecular Interactions as a Measure of Dapsone Solubility in Neat Solvents and Binary Solvent Mixtures

Dapsone is an effective antibacterial drug used to treat a variety of conditions. However, the aqueous solubility of this drug is limited, as is its permeability. This study expands the available solubility data pool for dapsone by measuring its solubility in several pure organic solvents: N-methyl-2-pyrrolidone (CAS: 872-50-4), dimethyl sulfoxide (CAS: 67-68-5), 4-formylmorpholine (CAS: 4394-85-8), tetraethylene pentamine (CAS: 112-57-2), and diethylene glycol bis(3-aminopropyl) ether (CAS: 4246-51-9). Furthermore, the study proposes the use of intermolecular interactions as molecular descriptors to predict the solubility of dapsone in neat solvents and binary mixtures using machine learning models. An ensemble of regressors was used, including support vector machines, random forests, gradient boosting, and neural networks. Affinities of dapsone to solvent molecules were calculated using COSMO-RS and used as input for model training. Due to the polymorphic nature of dapsone, fusion data are not available, which prohibits the direct use of COSMO-RS for solubility calculations. Therefore, a consonance solvent approach was tested, which allows an indirect estimation of the fusion properties. Unfortunately, the resulting accuracy is unsatisfactory. In contrast, the developed regressors showed high predictive potential. This work documents that intermolecular interactions characterized by solute-solvent contacts can be considered valuable molecular descriptors for solubility modeling and that the wealth of encoded information is sufficient for solubility predictions for new systems, including those for which experimental measurements of thermodynamic properties are unavailable.