Polarization-Engineered High-Efficiency GaInN Light-Emitting Diodes Optimized by Genetic Algorithm

A genetic algorithm is employed to find an optimum epitaxial structure of multiple quantum wells (MQWs) and electron-blocking layer (EBL) for a GaInN-based light-emitting diode (LED). The optimized LED is composed of locally Si-doped quantum barriers (QBs) in the MQWs and a quaternary heterostructured AlGaInN EBL having a polarization-induced electric field directed oppositely to that of a conventional AlGaN EBL. The optimized LED shows 15.6% higher internal quantum efficiency, 24.6% smaller efficiency droop, and 0.21 V lower forward voltage at 200 A/cm² comparing to the reference LED, which has fully Si-doped QB and 20-nm-thick Al_0.19Ga_0.81N EBL. We find that local Si doping near the QB/QW interface compensates the negative polarization-induced sheet charge at the interface and reduces electric field in the QWs, thereby enhancing electron-hole wave function overlap. In addition, the inverted polarization field in the quaternary EBL provides a high barrier for electrons but a low barrier for holes, resulting in enhanced electron-blocking and hole-injection characteristics.