Application of Machine Learning to Quantum Cascade Laser Design

We propose to develop fully automated QCL design tools based on neural network machine learning (ML) codes. The current QCL design process requires human intervention at multiple points in the process. The primary goal of the Phase-I STTR program is to eliminate the need for human intervention, which will greatly accelerate the design process, and allow for the generation of a large ML training set. In addition, we will utilize these codes to further develop high-performance QCL designs. Successful completion of the program will enable breakthroughs in the production of powerful (multi-Watt range), single-mode MWIR sources for applications ranging from remote sensing to free-space communications.