The electronic structure of a series perovskites ABX₃ (A = Cs; B = Ca, Sr, and Ba; X = F, Cl, Br, and I) in the presence and absence of antisite defect X_B were systematically investigated based on density-functional-theory calculations. Both cubic and orthorhombic perovskites were considered. It was observed that for certain perovskite compositions and crystal structure, presence of antisite point defect leads to the formation of electronic defect state(s) within the band gap. We showed that both the type of electronic defect states and their individual energy level location within the bandgap can be predicted based on easily available intrinsic properties of the constituent elements, such as the bond-dissociation energy of the B–X and X–X bond, the X–X covalent bond length, and the atomic size of halide (X) as well as structural characteristic such as B–X–B bond angle. Overall, this work provides a science-based generic principle to design the electronic states within the band structure in Cs-based perovskites in presence of point defects such as antisite defect.

Electrohydrodynamic (EHD) printing has been used in various applications (e.g., sensors, batteries, photonic crystals). Currently, research on studying the relationships between EHD jetting behaviors, material properties, and processing conditions is still challenging due to a large number of parameters, cost, time, and the complex nature of experiments. In this research, we investigated EHD printing behavior using a machine learning (ML)-guided approach to overcome limitations in the experiments. Specifically, we investigated two jetting modes and the size of printed material with a broader range of material properties and processing parameters. We used samples from both literature and our own experiment results with different type of materials. Different ML models have been developed and applied to the data. Our results have shown that ML can navigate a vast parameter search space to predict printing behavior with an accuracy of higher than 95% during EHD printing. Moreover, the results showed that ML models can be used to predict the printing behavior and feather size for new materials. The ML models can guide the investigation of EHD printing and helped us understand the printing behavior in a systematic manner with reduced time, cost, and required experiments.

Three-dimensional printing (3DP) holds significant potential for developing personalized pharmaceutical oral dosage forms (printlets). 3D printing has the advantage of fabricating complex geometric structures for versatile drug release profiles, enhancing patient preference, palatability and swallowability, reducing the pill burden, and increasing dose accuracy. Optimizing printing parameters is crucial in determining the quality of the printlets during dosage development. The integration of machine learning (ML) can reduce production costs and time through parameter optimization based on trained datasets. This research is focused on optimizing parameters for fused deposition modeling (FDM) based batch and continuous printing methods. The algorithm was trained using a three-level full factorial design, which generated data in the form of printlets with different parameters. Both defect and defect-free printlets were analyzed using image segmentation. Machine learning tools including Gaussian Process Regressor (GPR) and Efficient Global Optimization (EGO) were used to predict and select processing parameters for a targeted percentage surface defect. The final trained algorithm predicted new parameter sets for both batch (R ² -0.8783) and continuous (R ² -0.9364) printing methods to achieve zero defects, and the same was confirmed through printing and characterization of printlets which showed no defects. The algorithm was later adapted successfully to a variety of materials within the temperature range of 190–220 ℃ and predicted zero-defect printlets. Scanning electron microscopy (SEM) revealed the absence of defects on the surfaces of the materials. Results showed that flow rate (110 and 120 mm ³ /s) had a significant impact on printlet quality withwithout defects for both batch and continuous printing, compared to print speed, print temperature, and infill density. This research provides new insights into the development of optimized FDM printlets using batch and continuous printing with adaptive machine learning for pharmaceutical dosage manufacturing.

Deep learning tools have recently shown significant potential for accelerating the prediction of microstructure–property linkage in materials. While deep neural networks like convolution neural networks (CNNs) can extract physics information from 3D microstructure images, they often require a large network architecture and substantial training time. In this research, we trained a graph neural network (GNN) using phase field generated microstructures of Ni-Al alloys to predict the evolution of mechanical properties. We found that a single GNN is capable of accurately predicting the strengthening of Ni-Al alloys with microstructures of varying sizes and dimensions, which cannot otherwise be done with a CNN. Additionally, GNN requires significantly less GPU utilization than CNN and offers more interpretable explanation of predictions using saliency analysis as features are manually defined in the graph. We also utilize explainable artificial intelligence tool Bayesian Inference to determine the coefficients in the power law equation that governs coarsening of precipitates. Overall, our work demonstrates the ability of the GNN to accurately and efficiently extract relevant information from material microstructures without having restrictions on microstructure size or dimension and offers an interpretable explanation.

The atomic structure and chemistry at metal/oxide interfaces play a crucial role in determining their properties. However, studying semi-coherent metal/oxide interfaces that include misfit dislocations through density functional theory (DFT) is often computationally expensive due to the large number of atoms involved, ranging from hundreds to thousands. In this study, we explore solute segregation behavior at the Fe/Y₂O₃ interface—an important model interface for cladding applications in nuclear fission reactors—by combining DFT calculations with a machine learning (ML) approach. ML models are trained using DFT-calculated segregation energies ($E^{Seg}$) to identify the key chemical and geometric factors influencing solute segregation at metal/oxide interfaces, revealing the competition between these features in determining $E^{Seg}$. Moreover, the segregation behavior at a specific Fe/Y₂O₃ interface is predicted with high accuracy using ML models trained on data from this interface. Furthermore, it is found that the ML models could also predict solute segregation at a different Fe/Y₂O₃ interface with a new orientation relationship (OR), at a computational cost of less than 1/45 of that required for similar DFT calculations.