日耳曼
材料科学
杰纳斯
带隙
密度泛函理论
异质结
半导体
光电子学
计算机科学
纳米技术
计算化学
硅
硅烯
化学
作者
Yuliang Mao,Zihua Yao,Jianmei Yuan,Xinxin Chang
标识
DOI:10.1016/j.apsusc.2022.153633
摘要
The high-throughput calculation of two-dimensional (2D) materials can provide new candidate materials for semiconductor devices in many new structures, optimize their electrical properties, and break through the performance bottleneck of original devices. The surface functionalization of the germanene structure significantly affects its stability and photoelectric properties. However, due to the diversity and complexity, the design of a potential new Janus germanene-based structure will necessitate costly computational resources and a lengthy research period. In this paper, a framework of the classification algorithm (random forests classification), regression algorithm (ridge regression and support vector regression), and fitting method was established using a combination of machine learning and density functional theory. Using our designed framework, the stability (binding energy) and electrical properties (band gap, valence band maximum) of 2D germanene-based structures Ge8HnX8−n (n = 0–8, X = F, Cl, Br, and I) can be predicted effectively by 33 simple features. In unexplored germanene-based structures, we successfully designed ten novel single structures and three new heterojunctions with potential photovoltaic or photocatalytic applications by the designed framework and density functional theory verification, in which Ge8H6ClBr/Ge8H5Br2I has a high photoelectric conversion efficiency of 23.24%.
科研通智能强力驱动
Strongly Powered by AbleSci AI