Towards Microstructure-Aware Autonomous Alloy Design

The focus of goal-oriented materials design is to find the necessary chemistry/processing conditions to achieve the desired properties. In this setting, a material's microstructure is either only used to carry out multiscale simulations to establish an invertible quantitative process-structure-property (PSP) relationship, or to rationalize a posteriori the underlying microstructural features responsible for the properties achieved. The materials design process itself, however, tends to be microstructure-agnostic: the microstructure only mediates the process-property (PP) connection and is---with some exceptions such as architected materials---seldom used for the optimization itself.<br/>While the existence of PSP relationships is the central paradigm of materials science, it would seem that for materials design, one only needs to focus on PP relationships. In this work, we attempt to resolve the issue of whether `PSP’ is a superior paradigm for materials design in cases where the microstructure itself cannot be (directly) manipulated to optimize materials’ properties. To this end, we formulate a novel microstructure-aware closed-loop multi-fidelity Bayesian optimization framework for materials design and rigorously demonstrate the importance of the microstructure information in the design process.<br/>We begin by introducing our reification-based multi-information source fusion batch Bayesian optimization approach r-MISBBO and show how it is capable of efficiently making optimal decisions on different information sources at the disposal of the optimizer as well as the location in the design space queried by the selected information source. We first show how our approach to information fusion arrives at optimal solutions at a cost that is much lower than what would be necessary if we only queried the (highest fidelity) ground truth. We then demonstrate how we can make the process even more efficient by carrying out the exploration/exploitation of the materials design space in parallel or batch mode. This last improvement of our framework makes it especially suited to modern platforms for (in silico or in vivo) high throughput materials discovery.<br/>Having set up the framework, we proceed to investigate whether a fully autonomous framework for in silico materials design arrives at an optimal solution faster when (micro)structural information is provided to the optimizer. Our results clearly show that an explicit incorporation of microstructure knowledge in the materials design framework significantly enhances the materials optimization process. We thus prove, in a computational setting, and for a particular representative problem where microstructure intervenes to influence properties of interest, that `PSP’ is superior to `PP’ when it comes to materials design. Finally, we provide some speculation as to the underlying reasons for why exploiting 'PSP' relationships is an optimal strategy when such information is available. These results have important consequences for autonomous materials discovery platforms, as they provide justification for microstructure characterization stages along the discovery workflow.