Fully Automated Quantum Cascade Laser Design Aided by Machine Learning with up to 100X Design Cycle Time Reduction

A Quantum Cascade Laser's (QCL) core material is a series of nanometer scale layers of conduction band barrier and well materials designed to induce lasing electron energy levels. The key design feature of a QCL is the ability to repeat the laser core superlattice design to cascade electrons through the superlattice repetitions by repeated stimulated emission. Such designs are generally carefully created manually by well trained scientists with a strong intuition for quantum behavior and quantum well simulation programs. A method is proposed to design and evaluate superlattice designs by training a neural network to replicate the intuition of the superlattice designer and the quantitative results of the superlattice simulation. Training such a neural network with thousands of example designs and evaluations will allow the network to produce usable QCL superlattice designs without human assistance.