Bridging Text Mining and Quantum Simulations for the Design of 2D Monochalcogenide Materials

The formulation of a structured framework dedicated to
the systematic
review of the literature, identification of potential compounds, theoretical
quantum chemistry characterization, and assessment of the significance
of stability descriptors is essential for accelerating the discovery
of two-dimensional materials for technological applications. In this
work, we selected the two-dimensional monochalcogenides (MQ), which
have attracted increased interest due to their potential applications
in future devices, as prototype materials for the present investigation.
Our framework started with a natural language processing analysis
of more than 5400 articles, revealing a growing research interest
in the two-dimensional MQ compounds, especially in electronics, energy,
and fundamental studies. This led to the selection of 27 diverse MQ
compounds across 13 distinct two-dimensional structural phases for
density functional theory calculations, resulting in an extensive
database of physicochemical properties. We evaluated formation enthalpies,
uncovering clear stability trends (e.g., stability declines with heavier
chalcogens, dynamic robustness of GeSe), and evaluated equilibrium
lattice parameters, noting predictable expansions and significant
anisotropies. Bader charge analysis offered insights into charge-transfer
and ionicity trends. Our electronic structure analysis identified
band gaps (direct or indirect, depending on the M element group and
the weight of the chalcogen) and optical absorption anisotropies significantly
influenced by crystal symmetry and spin orbit coupling. Importantly,
band alignment calculations classified all possible heterojunctions
as type I, II, or III, underscoring their extensive potential for
future optoelectronic device development. Additionally, we incorporated
machine learning, using a random forest approach, along with our density
functional theory calculations to accurately forecast trends in energetic
properties. This analysis identified the electronic charge as a highly
significant stability descriptor among the 34 descriptors, underscoring
its importance for future machine learning endeavors. Hence, this
study offers a streamlined framework for the characterization and
discovery of promising two-dimensional materials, highlighting the
synergy between data-driven analysis and quantum-based simulations.