Accurate prediction on the lattice thermal conductivities of monolayer systems by a high-throughput descriptor. (26th January 2023)

• Record Type: Journal Article
• Authors: Luo, Yufeng; Li, Mengke; Yuan, Hongmei; Cao, Haibin; Liu, Huijun
• Is Part Of: Journal of physics. Volume 56:Number 4(2023)
• Publication Date: 2023-01-26
• Subjects: monolayers -- lattice thermal conductivity -- machine learning -- high-throughput prediction; Physics -- Periodicals; 530
• DOI: 10.1088/1361-6463/aca9db
• Languages: English
• ISSNs: 0022-3727

Abstract

Abstract: Due to the vital importance of heat management in micro- and nano-electronic devices, it is quite necessary to evaluate the lattice thermal conductivity ( κ L ) of two-dimensional (2D) materials. However, the accurate prediction on the κ L has been demonstrated to be a rough task, especially for systems with large unit cell and low symmetry. Here, by using the sure independence screening and sparsifying operator (SISSO) approach, we propose a physically interpretable descriptor to quickly determine the κ L of many potential monolayer systems, which are one of the fast-growing class among numerous 2D materials. It should be noted that the Pearson correlation coefficient between the real and predicted κ L is as high as 0.98, suggesting good reliability of the derived descriptor. Beyond the initial training data, the strong predictive power of our descriptor is further confirmed by good agreement between the predicted κ L and those calculated theoretically or measured experimentally. As such a data-driven descriptor contains only elementary properties of the monolayers, it is very beneficial for high-throughput screening of systems with desired κ L .