1D van der Waals Molecular Wires: Exfoliation Energies, Electronic Properties and Machine Learning Discovery

Two-dimensional (2D) materials derived from van der Waals (vdW)-bonded layered crystals have been the subject of considerable research focus, but analogous one-dimensional (1D) materials have received less attention while also exfoliable and useful for optical or electronic applications. Using density-functional-theory-based methods, we find binding energies of several 1D families to be within typical ranges possible for 2D materials. We compute the electronic properties of a variety of insulating, semiconducting, and metallic wires and find differences that could enable the identification of and distinction between 1D, 2D, and 3D forms during mechanical exfoliation onto a substrate. 1D wires from chemical families of the forms PdBr2, SbSeI, and GePdS3 are likely distinguishable from bulk materials via photoluminescence. Machine learning methods are used to predict undiscovered 1D materials, targeting conductive and magnetic compositions. Models trained on different data subsets recover similar predicted unsynthesized materials, illustrating robustness. We focus on classes of materials tailored for potential experimental synthesizability, specifically for chalcogen, halogen, and pnictogen-containing compounds.