光电半导体材料的理论设计

Theoretical design of optoelectronic semiconductors

光电半导体作为一种可以将光能和电能相互转换的材料,近几十年来,在能源和电子信息等领域得到广泛应用.随着计算机算力的提升和理论算法的发展,理论设计方法可以在短时间内探索成百上千种材料,相比于采用试错法的实验方式,具有开发周期短且成本低等特点,逐渐成为新材料研发的关键步骤.通过将物理原则与高通量计算、智能优化算法和机器学习等理论设计策略相结合,可以准确高效地探索性能优异的光电半导体材料.本文概述了光电半导体材料的设计策略和研究进展.首先,介绍基于第一性原理计算光电性质的方法,并分析相关物理性质的成因及其在光电半导体设计方面的意义;然后,结合具体的研究成果对高通量材料筛选、物理原则导向的材料设计、基于智能算法的材料搜索和基于机器学习方法的材料发现等不同的光电半导体设计策略进行概述,为该领域的理论设计方向提供指导原则;最后,对光电半导体设计方面的工作进行总结,并对该领域未来的发展进行展望.

The rapid development of the semiconductor industry has motivated researchers'passion for accelerating the discovery of advanced optoelectronic semiconductors.Optoelectronic semiconductors have played a significant role in energy and electronic applications in recent decades and can convert light to electric energy and inverse conversion.As relevant theories and methodologies have developed,we can feasibly predict the optoelectronic properties of hundreds of materials by first-principles calculations within short timeframes at a practical level of accuracy.A fundamental understanding of the structure-property relationship of optoelectronic semiconductors is essential to fabricate novel materials and highperformance devices.Ideal optoelectronic semiconductors require tunable band gaps,high absorption coefficient,broad absorption spectrum,high charge carrier mobility,and long charge diffusion length,which enable a broad range of photovoltaic and optoelectronic applications.These optoelectronic properties can be explored comprehensively and efficiently by theoretical design methods.Furthermore,compared with the experimental processes of trial-and-error,the theoretical design method has the advantages of a short development cycle and low cost.Therefore,the theoretical design method gradually becomes critical in researching and developing new materials.Combining physical principles with theoretical design strategies such as high-throughput screening,intelligent optimization algorithms,and machine learning,optoelectronic semiconductor materials can be accurately and efficiently explored.This review summarizes the optoelectronic semiconductor materials'design strategies and some typical design examples.In the first section,we introduce the methods of calculating several representative optoelectronic properties based on first principles calculations and analyze the causes of relevant physical properties and their significance in the application of optoelectronic semiconductors.This chapter not only introduces the calculation method and physical meaning of these optoelectronic properties in general cases but also explains the matters needing attention in some exceptional cases,which can be used as a reference for the relevant work in the calculation of these properties.Then in the second section,we introduce the guiding principles of four different optoelectronic semiconductor designing strategies,including high-throughput materials screening,physical principles instructed materials design,materials prediction based on intelligent algorithms,and materials discovery based on machine learning.The high-throughput screening strategy is based on first-principles calculations of a group of structures with similar characteristics.During the design process,candidate structures will be accurately and efficiently screened based on the multi-level screening funnel designed by expected targets.Physical principles instructed materials design strategy aims to discover the target structures inversely and further analyze their underlying physical principles from some phenomena.The strategy of materials prediction based on the intelligent algorithm explores the potential energy surface through global optimization algorithms such as genetic or evolutionary algorithms and predicts ideal stable structures according to the component information.The strategy of materials discovery based on machine learning can extract valuable information through artificial intelligence algorithms from massive data,reorganize existing knowledge and discover hidden relationships between different materials.It is practical to solve the short board of DFT calculations when dealing with complex chemical space and large systems with the machine learning method.To illustrate these theoretical design modalities,we provide various examples of representative optoelectronic materials,such as perovskites,metal oxides,wide bandgap semiconductors,and organic-inorganic semiconductors.Summaries and prospects for the computational design of optoelectronic semiconductors are further emphasized at the end of the review.