Energy gap estimation of zinc sulfide metal chalcogenide nanostructure semiconductor using genetically hybridized support vector regression

Abstract

Zinc sulfide is a metal chalcogenide semiconductor with promising potentials in environmental sensors, short wavelength light emitting diodes, biomedical imaging, display light sources, transistors, flat panel displays, optoelectronics, and photocatalysis. Adjusting the energy gap (EG) of zinc sulfide for light response enhancement that is suitable for desired applications involves foreign material incorporation through chemical doping or co-doping mechanisms with structural distortion and host symmetry breaking. This work optimizes support vector regression (SVR) parameters with a genetic algorithm to develop a hybrid genetically optimized SVR (HGSVR-EG) model with the precise capacity to estimate the EG of a doped zinc sulfide semiconductor using the crystal lattice constant and the crystallite size as descriptors. The precision of the developed HGSVR-EG model is compared with that of the stepwise regression based model for EG estimation (STR-EG) using different error metrics. The developed HGSVR-EG model outperforms the STR-EG model with a performance improvement of 64.47%, 74.52%, and 49.52% on the basis of correlation coefficient, mean squared error, and root mean square error, respectively. The developed HGSVR-EG model explores and investigates the zinc sulfide bandgap reduction effect of manganese and chromium nano-particle incorporation in the host semiconductor, and the obtained EGs agree well with the measured values. The developed HGSVR-EG model was further validated with an external set of data, and an excellent agreement between the measured and estimated EGs was obtained. The outstanding performance of the developed predictive models in this work would ultimately facilitate EG characterization of zinc sulfide without experimental stress.