Search for: All records

Abstract We present a machine learning (ML) framework for predicting the structural dimensionality of hybrid metal halides (HMHs), including organic-inorganic perovskites, using a combination of chemically-informed feature engineering and advanced class-imbalance handling techniques. This study is motivated by the small and highly imbalanced nature of experimentally available HMH datasets, which limits the applicability and reliability of conventional ML approaches. The dataset, consisting of 494 HMH structures, is highly imbalanced across dimensionality classes (0D, 1D, 2D, 3D), posing significant challenges to predictive modeling. To mitigate this limitation, the dataset was augmented to 1336 samples using the synthetic minority oversampling technique, enabling improved learning of underrepresented dimensionality classes while preserving chemically meaningful feature relationships. We developed interaction-based descriptors designed to capture coupled steric and polarity effects relevant to dimensionality prediction, which are not readily captured by standard single-parameter or composition-only descriptors. These descriptors are integrated into a multi-stage workflow combining feature selection, ensemble stacking, and performance optimization. Our approach significantly improvesF1-scores for underrepresented classes, achieving robust cross-validation performance across all dimensionalities. This work demonstrates a generalizable strategy for extracting reliable and interpretable structure–dimensionality relationships from limited experimental data, enabling pre-synthesis screening of organic cations and providing a practical blueprint for small-data ML in hybrid materials systems. 

A Bayesian additive regression trees model predicts hybrid metal halide dimensionality, highlighting rules that govern low-dimensional structure formation. 

The melting properties and kinetics of glass formation in 2D perovskites can be finely tuned using isomeric organic cations bearing distinct substitutional functional group positions, resulting in enhancement of glass-crystalline switching speed. 

The rising atmospheric CO₂concentration is one of the biggest challenges human civilization faces. Direct air capture (DAC) that removes CO₂from the atmosphere provides great potential in carbon neutralization. However, the massive land use and capital investment of centralized DAC plants and the energy-intensive process of adsorbent regeneration limit its wide employment. We develop a distributed carbon nanofiber (CNF)–based DAC air filter capable of adsorbing CO₂downstream in ventilation systems. The DAC air filter not only has the potential to remove 596 MtCO₂year⁻¹globally but can also decrease energy consumption in existing building systems. The CNF-based adsorbent has a capacity of 4 mmol/g and can be regenerated via solar thermal or electrothermal methods with low carbon footprints. Through life cycle assessment, the CNF air filter shows a carbon removal efficiency of 92.1% from cradle to grave. Additionally, techno-economic analysis estimates a cost of $209 to 668 in capturing and storing 1 tonne of CO₂from direct air. 

Hybrid lead-halide perovskites have been studied extensively for their promising optoelectronic properties and prospective applications including photovoltaics, solid-state lighting, and radiation detection. Research into these materials has also been aided by the simple and low temperature synthetic conditions involved in solution-state deposition/crystallization or melt processing techniques. However, concern over lead toxicity has plagued the field since its early days. One of the most promising routes to mitigating toxicity in hybrid perovskite materials is substituting isoelectronic Bi(III) for Pb(II). Various methods have been developed to allow pnictide-based systems to capture properties of the Pb(II) analogs, but the ability to melt process extended hybrid pnictide-halide materials has not been investigated. In this work, we prepare a series of 1D antimony and bismuth-iodide hybrid materials employing tetramethylpiperazinium (TMPZ)-related cations. We observe, for the first time, the ability to melt hybrid pnictide-halide materials for both Sb(III) and Bi(III) systems. Additionally, we find that Sb(III) analogs melt at lower temperatures, and attribute this observation to structural changes induced by the increased stereochemical activity of the Sb(III) lone pair coupled with reduction in effective dimensionality due to steric interactions with the organic cations. Finally, we demonstrate the ability to melt process phase pure thin films of (S-MeTMPZ)SbI5.