Computational design and phase prediction of transition metal dichalcogenides using machine learning techniques

Two-dimensional transition metal dichalcogenides (TMDs) offer unprecedented opportunities for fundamental science and technology. These materials most commonly adopt either a trigonal prismatic phase (1H) or an octahedral symmetry phase (1T) and display diverse electronic and structural properties. Research efforts to synthesize new TMDs have led to the recent discovery of Janus TMDs, with two different chalcogens on the two faces of a monolayer. Their synthesis increases the combinatorial possibilities for synthesizing TMDs. In this work, we used high throughput calculations in combination with machine learning methods to accelerate the hunt for newer TMDs for various applications. We explored different physiochemical factors (descriptors) that govern not only the stability of TMDs but also their preferred phases. The latter has been a long-standing problem, and we show that one needs to go beyond descriptors considered within Pauling’s ratio rules.