Discovery of materials with extreme work functions by high-throughput density functional theory and machine learning

The work function is the key surface property that determines how much energy is required for an electron to escape the surface of a material. This property is crucial for thermionic energy conversion, band alignment in heterostructures, and electron emission devices. Data-driven predictions of bulk material properties have been widely explored and work functions of elemental crystals have been studied thoroughly. However, the work functions of more complex compounds have not been investigated with a data-driven framework yet. Here, we present a high-throughput workflow using density functional theory (DFT) to calculate the work function of 29,270 surfaces (23,603 slab calculations) that we created from 2,492 bulk materials, including up to ternary compounds. Based on this database we develop a physics-based approach to featurize surfaces and use supervised machine learning to predict the work function. Our (random forest) model achieves a mean absolute test error of 0.19 eV, which is more than 4 times better than the baseline and comparable to the accuracy of DFT. This surrogate model enables rapid predictions of the work function ($\sim 10^5$ faster than DFT) across a vast chemical space and facilitates the discovery of material surfaces with extreme work functions for energy conversion and electronic applications.