Search for: All records

We developed a physics-informed machine learning platform that predicts stress–strain curves of 3D-printed thermoplastics from ink formulations, enabling virtual experimentation and rapid identification of optimal materials in vast chemical spaces. 

De novo design of molecules with targeted properties represents a new frontier in molecule development. Despite enormous progress, two main challenges remain: (i) generating novel molecules conditioned on targeted, continuous property values; (ii) obtaining molecules with property values beyond the range in the training data. To tackle these challenges, we propose a reinforced regressional and conditional generative adversarial network (RRCGAN) to generate chemically valid molecules with targeted HOMO–LUMO energy gap (ΔEH–L) as a proof-of-concept study. As validated by density functional theory (DFT) calculation, 75% of the generated molecules have a relative error (RE) of <20% of the targeted ΔEH–L values. To bias the generation toward the ΔEH–L values beyond the range of the original training molecules, transfer learning was applied to iteratively retrain the RRCGAN model. After just two iterations, the mean ΔEH–L of the generated molecules increases to 8.7 eV from the mean value of 5.9 eV shown in the initial training dataset. Qualitative and quantitative analyses reveal that the model has successfully captured the underlying structure–property relationship, which agrees well with the established physical and chemical rules. These results present a trustworthy, purely data-driven methodology for the highly efficient generation of novel molecules with different targeted properties. 

Rapid analysis of materials characterization spectra is pivotal for preventing the accumulation of unwieldy datasets, thus accelerating subsequent decision-making. However, current methods heavily rely on experience and domain knowledge, which not only proves tedious but also makes it hard to keep up with the pace of data acquisition. In this context, we introduce a transferable Vision Transformer (ViT) model for the identification of materials from their spectra, including XRD and FTIR. First, an optimal ViT model was trained to predict metal organic frameworks (MOFs) from their XRD spectra. It attains prediction accuracies of 70%, 93%, and 94.9% for Top-1, Top-3, and Top-5, respectively, and a shorter training time of 269 seconds (∼30% faster) in comparison to a convolutional neural network model. The dimension reduction and attention weight map underline its adeptness at capturing relevant features in the XRD spectra for determining the prediction outcome. Moreover, the model can be transferred to a new one for prediction of organic molecules from their FTIR spectra, attaining remarkable Top-1, Top-3, and Top-5 prediction accuracies of 84%, 94.1%, and 96.7%, respectively. The introduced ViT-based model would set a new avenue for handling diverse types of spectroscopic data, thus expediting the materials characterization processes. 

The printing outcome of vat photopolymerization (VPP) of thermoplastics largely depends on physicochemical properties of monomers and their compositions in resins, which also greatly determine the material properties, e.g., tensile strength (σT) and toughness (UT)and phase transition temperature (Tg). A methodology for optimizing the resin formulation is of paramount importance in realizing highly printable thermoplastics with balanced σT/UT and target Tg while remaining largely underexplored. Herein, we introduce a multi-objective Bayesian optimization (MOBO) algorithm under two physics informed constraints (printability and Tg) to optimize two conflicting properties: σT and UT. The two constraints are formulated as two machine learning (ML) models, which are trained with weight ratios of the six monomers and physics informed (PI) descriptors derived from their physiochemical parameters. Dimensional reduction analysis reveals that the algorithm avoids recommendation of the monomer ratios that do not pass the two constraints. The printing failure rate is reduced from 16% in the background experiments to 3% in the recommended experiments. Within only 36 iterations (72 samples), the MOBO algorithm successfully identifies five sets of ratios leading to Pareto optimal of σT and UT. Due to the constraint in Tg they show appropriate Tg for shape memory application. The partial dependence analysis indicates that σT and UT depend on both the ratios and physiochemical features of the monomers. These results underscore capability of such a smart decision-making algorithm—with constraints that are not fully understood but can be informed by prior knowledge—in planning the experiments from the vast design space, thus holding a great promise for broader applications in materials design and manufacturing. 

The SML model was trained on both direct experimental and indirect physics-informed features to predict graphene quality synthesized from Flash Joule heating. With anR²of 0.81, the model performs better compared to 0.73 without indirect features.