FMRG: Cyber: Manufacturing USA: Manufacturing of Next-Generation Perovskite Semiconductors at Scale

For the past century, semiconductor manufacturing has relied heavily on sequential deposition and removal of material layers to fabricate integrated devices. However, sequential layer-by-layer processing imposes restrictions on the processing parameters that can be used for the top layers in the device, which must maintain chemical and thermal compatibility with the underlying layers.

This limits the ability to engineer precise interfaces in many applications, especially when integrating emerging functional materials with new processing constraints. To overcome these limitations, this Future Manufacturing Research Grant (FMRG) explores a novel lamination approach for halide perovskite semiconductors that enables the engineering of functionality and new device architectures.

Potential applications of this technology are solar cells, LEDs, and other optoelectronic devices. By developing the cyberinfrastructure for distributed manufacturing, trained machine learning models are generated to optimize a specified objective function that is unique to each end-user, thereby supporting the ability of small-to-medium manufacturers (SMMs) to prototype their designs and scale-up their processes.

This research is closely integrated with education and workforce development activities, where partnerships with local workforce development organizations and an industry advisory board (IAB) are formed to identify the education and training needs of the next generation of cyber manufacturing workers, while ensuring a diverse manufacturing workforce. This project supports the national priorities of semiconductor manufacturing and renewable energy.

The research objective of this FMRG research is to understand, model, and control the process-structure-property relationships during halide perovskite (HP) semiconductor manufacturing using a novel lamination approach that enables new device architectures and material combinations that are currently inaccessible using traditional sequential deposition processing. In this approach, device half-stacks can be independently processed in parallel with relaxed process constraints, and subsequently integrated using a continuous lamination platform with controlled alignment.

By integrating in-line metrology with physics-informed data-driven models, the project develops a fundamental understanding of the thermo-chemo-mechanical mechanisms that guide the HP lamination process, which enables closed-loop process control. Algorithms are developed that bridge high-throughput, low-fidelity in-line metrology data streams with low-throughput, high-fidelity ex situ characterization methods, which are prohibitive in an industrial manufacturing setting.

Physics-informed process control is enabled through development of reduced-order models and process parameter optimization using federated learning approaches. The cyber manufacturing platform enables automated generation of a database of process-structure-property relationships under diverse (and non-idealized) manufacturing environments, which enables predictive modeling and process optimization through a shared cyberinfrastructure.

This Future Manufacturing award was supported by the Divisions of Civil, Mechanical and Manufacturing Innovation (CMMI) and Engineering Education and Centers (EEC) and by the National Nanotechnology Initiative (NNI) Special Studies Program.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.