Abstract
Microstructure plays a critical role in determining the macroscopicproperties of materials, with applications spanning alloy design, MEMS devices,and tissue engineering, among many others. Computational frameworks have beendeveloped to capture the complex relationship between microstructure andmaterial behavior. However, despite these advancements, the steep learningcurve associated with domain-specific knowledge and complex algorithmsrestricts the broader application of these tools. To lower this barrier, wepropose a framework that integrates Natural Language Processing (NLP), LargeLanguage Models (LLMs), and Denoising Diffusion Probabilistic Models (DDPMs) toenable microstructure design using intuitive natural language commands. Ourframework employs contextual data augmentation, driven by a pretrained LLM, togenerate and expand a diverse dataset of microstructure descriptors. Aretrained NER model extracts relevant microstructure descriptors fromuser-provided natural language inputs, which are then used by the DDPM togenerate microstructures with targeted mechanical properties and topologicalfeatures. The NLP and DDPM components of the framework are modular, allowingfor separate training and validation, which ensures flexibility in adapting theframework to different datasets and use cases. A surrogate model system isemployed to rank and filter generated samples based on their alignment withtarget properties. Demonstrated on a database of nonlinear hyperelasticmicrostructures, this framework serves as a prototype for accessible inversedesign of microstructures, starting from intuitive natural language commands.