A 213-nm microwave-driven far-UVC light source and its optimization based on backpropagation artificial neural network

Abstract

Mercury lamps emitting 253.7 nm are well-known UV light devices but cause problems to environment. Here, a mercury-free light source is shown which is enabled by ZnI2 microwave-driven light-emitting plasma (ZnI2 MLEP), emitting 213-nm far-UVC. Its condition (pressure, ZnI2 mass and microwave power) was optimized using a backpropagation artificial neural network (ANN) to obtained the maximum 213-nm intensity (I213). The ANN was trained with experiment data and constructed the relationship between the condition and I213 with error less than 5 %. The I213 over all possible conditions allowed by the experimental accuracy was predicted. And the optimal condition was obtained and tested experimentally with the error of 5.6 %. After optimized by this method, the I213 of ZnI2 MLEP increases 53 % compared to conventional method. Based on the optimal condition, the ZnI2 MLEP was made into a standalone ZnI2 lamp with a recorded-high 213 nm intensity of 20.74 mW/cm2/nm. The comparison of spectra shows ZnI2 lamp have shorter wavelengths and higher UV radiant power density than Hg lamp. Coupled with its properties of environmentally friendly and electrodeless, the ZnI2 lamp have a potential to replace the Hg based UV lamps.