Investigation and understanding of the mechanical properties of MXene by high-throughput computations and interpretable machine learning

Abstract

2D transition metal carbides, nitrides, and carbonitrides (MXenes) have become prominent in energy storage, catalysis, environmental and nanoelectronics applications. In addition to their outstanding merits of electrical conductivity and electrochemically activity, the mechanical properties play crucial roles in all these applications but are usually ignored by researchers. Only quite a few publications have studied their mechanical properties and these studies still only represent a limited range of predictions compared to the large family of MXenes. Utilizing high-throughput computations and data-driven methods, this work studies the tensile stiffness (E2D) and strength (σs) of 157 types of MXenes that could be potentially fabricated. We find MXenes with the tensile stiffness ranging from 81.71 to 561.4 N/m, while 42 structures show higher stiffness than the well-known graphene and transition metal dichalcogenide (TMDs) monolayers. The E2D strongly depends on the thickness, bond strength and surface terminations. Surprisingly, we find that the surface terminations can significantly improve the E2D, and the improvement can reach nearly 100%. By using the recently developed interpretable machine learning method, we obtain the analytical formula of E2D with efficient and physically interpretable descriptors that can predict the stiffness accurately. As for the σs, it significantly scales with E2D (σs≈E2D/β, β=10.449), and can be also improved by surface terminations as well as E2D. In addition, we also find a few of MXene semiconductors, although most of MXenes are metals. Moreover, the mechanical properties of these MXenes are better than the widely used monolayer semiconductor TMDs. These MXenes semiconductors with a moderate band gap may open a new avenue for fabricating next-generation 2D nanoscale electronics, such as all-in-one monolayer devices.