Effect of Exchange-Correlation Functionals on Schottky Barriers at Si/Metal Interfaces

TL;DR

This study systematically evaluates computational methods for predicting Schottky barrier heights at Si/metal interfaces, emphasizing the importance of structural and electrostatic consistency, and proposes improved hybrid approaches for accurate, cost-effective predictions.

Contribution

It introduces a physically grounded assessment of exchange-correlation treatments and reference protocols, demonstrating improved accuracy in SBH predictions through mixed hybrid-semilocal methods with strained references.

Findings

Mixed hybrid-semilocal approaches improve SBH accuracy.

Structural and electrostatic consistency are crucial for reliable predictions.

Methodology achieves near-experimental accuracy with balanced computational cost.

Abstract

Accurate prediction of Schottky barrier heights (SBHs) at metal-semiconductor (M-SC) interfaces is essential for understanding and optimizing charge injection in electronic and optoelectronic devices. However, first-principles calculations of SBHs remain challenging due to the combined difficulties of semiconductor bandgap underestimation, metal Fermi level placement, lattice-mismatch, relative geometric alignment and electrostatic potential alignment across heterogeneous interfaces. In this work, we present a systematic and physically grounded assessment of computational strategies for SBH prediction using Si(111)/Metal (Al, Cu, Ag, Au) interfaces as representative test cases. We evaluate multiple exchange-correlation (XC) treatments, in combination with three distinct bulk reference protocols: relaxed bulk, relaxed bulk with spin-orbit coupling, and strained bulk references consistent with the interface geometry. By benchmarking against experimental data, we demonstrate that structural and electrostatic consistency between interface and bulk reference calculations is the dominant factor governing SBH accuracy. We show that mixed hybrid-semilocal approaches combined with strained reference protocols yield uniformly positive and significantly improved SBHs, achieving near-experimental accuracy while maintaining a favorable balance between computational cost and predictive performance. Our results establish a clear and transferable methodology for reliable Schottky barrier prediction and provide practical guidance for large-scale screening and interface engineering.