Study of Grain Boundary: From Crystallization Engineering to Machine Learning

Abstract

Grain boundaries play a vital role in determining the structural, functional, mechanical, and electrical properties of semiconductor materials. Recent studies have yielded great advances in understanding and modulating the grain boundaries via semiconductor crystallization engineering and machine learning. In this article, we first provide a review of the miscellaneous methods and approaches that effectively control the nucleation formation, semiconductor crystallization, and grain boundary of organic semiconductors. Using the benchmark small molecular semiconductor 6,13-bis(triisopropylsilylethynyl) pentacene (TIPS pentacene) as a representative example, the crystallization engineering methods include polymer additive mixing, solvent annealing, gas injection, and substrate temperature control. By studying the grain-width-dependent charge transport, we propose a grain boundary model as a fundamental basis to theoretically understand the intrinsic relation between grain boundary engineering and charge carrier mobility. Furthermore, we discuss the various machine learning algorithms and models used to analyze grain boundaries for the various important traits and properties, such as grain boundary crystallography, energy, mobility, and dislocation density. This work highlights the unique advantages of both crystallization engineering and machine learning methods, demonstrates new insights into discovering the presence of grain boundaries and understanding new properties of materials, and sheds light on the great potential of material application in various fields, such as organic electronics.