Accurate and efficient prediction of the stability and structure–stability relationship is important to discover materials; however, it requires tremendous efforts via traditional trial-and-error schemes. Here, we presented a small-data set machine learning (ML) method to accelerate the discovery of promising ternary transition metal boride (MAB) candidates. Based on data sets obtained by ab initio calculations, we developed three robust neural networks to predict the decomposition energy (ΔH_d) and assess the thermodynamic stability of 212-typed MABs (M₂AB₂). The quantitative relation between ΔH_d and stability was unraveled by several composition-and-structure descriptors. Three hexagonal M₂AB₂, i.e., Nb₂PB₂, Nb₂AsB₂, and Zr₂SB₂, were discovered to be stable with negative ΔH_d, and 75 metastable MABs were identified with ΔH_d less than 70 meV/atom. Finally, the dynamical stability and mechanical properties of MABs were investigated by ab initio calculations, whose results further verified the reliability of our ML models. This work provided a ML approach on small data sets to accelerate the discovery of compounds and expanded the MAB phase family to V_A and VI_A groups.

中文翻译：

---

通过小数据集上的机器学习加速过渡金属硼化物的发现

准确有效地预测稳定性和结构-稳定性关系对于发现材料非常重要；然而，这需要通过传统的试错方案付出巨大的努力。在这里，我们提出了一种小数据集机器学习（ML）方法来加速发现有前途的三元过渡金属硼化物（MAB）候选物。基于从头计算获得的数据集，我们开发了三个稳健的神经网络来预测分解能 (Δ H _d ) 并评估 212 型 MAB (M ₂ AB ₂ ) 的热力学稳定性。Δ H _d和稳定性之间的定量关系通过几种成分和结构描述符阐明。三六角M ₂AB ₂，即Nb ₂ PB ₂、Nb ₂ AsB ₂和Zr ₂ SB ₂被发现是稳定的，具有负ΔH d _，并且鉴定出75个亚稳态MAB具有小于70meV/原子的ΔH _{d 。}最后，通过从头计算研究了 MAB 的动态稳定性和机械性能，其结果进一步验证了我们的机器学习模型的可靠性。这项工作提供了一种针对小数据集的机器学习方法，以加速化合物的发现，并将 MAB 相系列扩展到 V _A和 VI _A组。