Effective Descriptor for Screening Single-Molecule Conductance Switches.

The adsorption-type molecular switch exhibits bistable states with an equivalently long lifetime at the organic/inorganic interface, promising reliable switching behavior and superior assembly ability in the electronic circuits at the molecular scale. However, the number of reported adsorption-type molecular switches is currently less than 10, and exploring these molecular switches poses a formidable challenge due to the intricate interplay occurring at the interface. To address this challenge, we have developed a model enabling the identification of diverse molecular switches on metal surfaces based on easily accessible physical characteristics. These characteristics primarily include the metal valency electron concentration, the work function of metal surfaces, and the electronegativity difference of molecules. Using this model, we identified 56 new molecular switches. Employing the gradient descent algorithm and statistical linear discriminant analysis, we constructed an explicit descriptor that establishes a relationship between the interfacial structure and chemical environment and the stability of molecular switches. The model's accuracy was validated through density functional theory calculations, achieving a 90% accuracy for aromatic molecular switches. The conductive switching behaviors were further confirmed by nonequilibrium Green's function transport calculations.