Generative design of stable semiconductor materials using deep learning and density functional theory
Generative design of stable semiconductor materials using deep learning and density functional theory
复制标题
DOI：
10.1038/s41524-022-00850-3
复制
发表时间：
2022-08
影响因子：
9.7
通讯作者：
E. M. Siriwardane;Yong Zhao;Indika Perera;Jianjun Hu
中科院分区：
文献类型：
--
作者：
E. M. Siriwardane;Yong Zhao;Indika Perera;Jianjun Hu
自从发现双极晶体管以来，半导体器件技术在复杂性方面有了很大的发展。在这项工作中，我们开发了一个计算管道，通过结合生成对抗网络（GAN），分类器和高通量第一性原理计算来发现稳定的半导体。我们使用CubicGAN，一种基于GAN的算法来生成立方体材料，并开发了一个分类器来筛选半导体，并使用第一性原理研究它们的稳定性。在F-43 m空间群中发现了12种稳定的AAMH 6半导体，包括BaNaRhH 6、BaSrZnH 6、BaCsAlH 6、SrTlIrH 6、KNaNiH 6、NaYRuH 6、CsKSiH 6、CaScMnH 6、YZnMnH 6、NaZrMnH 6、AgZrMnH 6和ScZnMnH 6。先前的研究报道了五个具有相同空间群的AAIrH 6半导体的合成。我们的研究表明，AAMnH 6和NaYRuH 6半导体与其他AAMH 6半导体相比具有相当不同的性质。基于精确的杂化泛函计算，发现AAMH 6半导体是宽带隙半导体。此外，BaSrZnH 6和KNaNiH 6是直接带隙半导体，而其他表现出间接带隙。
Semiconductor device technology has greatly developed in complexity since discovering the bipolar transistor. In this work, we developed a computational pipeline to discover stable semiconductors by combining generative adversarial networks (GAN), classifiers, and high-throughput first-principles calculations. We used CubicGAN, a GAN-based algorithm for generating cubic materials and developed a classifier to screen the semiconductors and studied their stability using first principles. We found 12 stable AAMH6semiconductors in the F-43m space group including BaNaRhH6, BaSrZnH6, BaCsAlH6, SrTlIrH6, KNaNiH6, NaYRuH6, CsKSiH6, CaScMnH6, YZnMnH6, NaZrMnH6, AgZrMnH6, and ScZnMnH6. Previous research reported that five AAIrH6 semiconductors with the same space group were synthesized. Our research shows that AAMnH6and NaYRuH6semiconductors have considerably different properties compared to the rest of the AAMH6semiconductors. Based on the accurate hybrid functional calculations, AAMH6semiconductors are found to be wide-bandgap semiconductors. Moreover, BaSrZnH6and KNaNiH6are direct-bandgap semiconductors, whereas others exhibit indirect bandgaps.