SVC_and_SISSO_files

Three new antimonides were discovered in the ternary K-Cd-Sb systems by a multifaceted approach that involves (i) employing a machine learning algorithm to pinpoint the compositional regions with low formation energy, (ii) rapid experimental compositional screening aided by hydride route; (iii) determination of optimal synthesis temperature from in-situ high-temperature powder X-ray diffraction data. In addition, we develop machine learning approaches to categorize and predict compositional regions in which a specific structural motif (layered, framework-, or cage-like) is more likely to form for the K-containing ternary intermetallic. These predictive models have been successfully tested on the recently discovered compounds, unlocking the potential of further optimization of targeted synthesis of the compounds of interest (with the specific structure details) from the favorable compositional regions (with low formation energy) using rapid compositional screening via hydride route本研究通过多维度研究策略，在三元钾-镉-锑（K-Cd-Sb）体系中发现了三种新型锑化物（antimonide），该策略包含以下三个环节：(i) 采用机器学习算法（machine learning algorithm）精准定位低形成能（formation energy）的成分区间；(ii) 借助氢化物法（hydride route）辅助开展快速实验成分筛选；(iii) 基于原位高温粉末X射线衍射（powder X-ray diffraction）数据确定最优合成温度。 此外，本研究开发了机器学习方法，用于分类并预测含钾三元金属间化合物中更易形成特定结构基元（structural motif，层状、骨架型或笼状）的成分区间。 上述预测模型已在本次研究中新近发现的化合物上完成验证，为后续通过氢化物法开展快速成分筛选、从低形成能的有利成分区间优化合成具备特定结构细节的目标化合物，解锁了应用潜力。