Efficient hybrid machine learning model for inverse design of porous boron nitride with high thermal conductivity

IF 2.4 4区物理与天体物理 Q3 PHYSICS, CONDENSED MATTER

Solid State Communications Pub Date : 2025-06-17 DOI:10.1016/j.ssc.2025.116051

Salman Khajeh , Arian Mayelifartash , Ahmad Ranjbar , Mohammad Ali Abdol , Sadegh Sadeghzadeh , Gianaurelio Cuniberti , Thomas D. Kühne , Hadi Kaveh

引用次数: 0

Abstract

Two-dimensional porous boron nitride (BN) is widely used in electronics, catalysis, and membranes. Its thermal conductivity is influenced by hole density and distribution. This study uses molecular dynamics simulations with a hybrid machine learning approach to predict BN's thermal conductivity. The hybrid model, combining convolutional neural networks and multilayer perceptrons, achieves high accuracy (RMSE = 0.01, R² = 0.99). A novel inverse design method links hole distribution to enhanced conductivity, optimizing designs with just 1024 samples. These findings demonstrate the power of machine learning in advancing physical insights and solving complex design challenges efficiently.

查看原文本刊更多论文

高导热多孔氮化硼反设计的高效混合机器学习模型

二维多孔氮化硼（BN）广泛应用于电子、催化、膜等领域。其导热系数受孔密度和孔分布的影响。本研究使用分子动力学模拟和混合机器学习方法来预测BN的导热性。该混合模型将卷积神经网络与多层感知器相结合，达到了较高的准确率（RMSE = 0.01, R2 = 0.99）。一种新的反设计方法将孔分布与增强导电性联系起来，只需1024个样本即可优化设计。这些发现证明了机器学习在提高物理洞察力和有效解决复杂设计挑战方面的强大力量。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Solid State Communications 物理-物理：凝聚态物理

CiteScore

3.40

自引率

4.80%

发文量

287

审稿时长

51 days

期刊介绍： Solid State Communications is an international medium for the publication of short communications and original research articles on significant developments in condensed matter science, giving scientists immediate access to important, recently completed work. The journal publishes original experimental and theoretical research on the physical and chemical properties of solids and other condensed systems and also on their preparation. The submission of manuscripts reporting research on the basic physics of materials science and devices, as well as of state-of-the-art microstructures and nanostructures, is encouraged. A coherent quantitative treatment emphasizing new physics is expected rather than a simple accumulation of experimental data. Consistent with these aims, the short communications should be kept concise and short, usually not longer than six printed pages. The number of figures and tables should also be kept to a minimum. Solid State Communications now also welcomes original research articles without length restrictions. The Fast-Track section of Solid State Communications is the venue for very rapid publication of short communications on significant developments in condensed matter science. The goal is to offer the broad condensed matter community quick and immediate access to publish recently completed papers in research areas that are rapidly evolving and in which there are developments with great potential impact.