Testing Predictions from Density Functional Theory at Finite Temperatures: $\beta_2$-Like Ground States in Co-Pt

TL;DR

This study critically evaluates the accuracy of DFT-based methods in predicting phase stability in Co-Pt alloys at finite temperatures, revealing significant discrepancies with experimental data and limitations of current approaches.

Contribution

The paper demonstrates that standard DFT predictions of phase stability in Co-Pt alloys are inaccurate at finite temperatures and shows that hybrid functionals like HSE06 do not resolve these issues.

Findings

DFT predictions do not match experimental phase diagrams.

Hybrid functional HSE06 fails to correct stability predictions.

DFT has fundamental limitations for late 3d-5d binary alloys.

Abstract

We perform a critical assessment of the accuracy of DFT-based methods in predicting stable phases within the Co-Pt binary alloy. Statistical mechanical analysis applied to zero kelvin DFT predictions yields finite-temperature results that can be directly compared with experimental measurements. The predicted temperature-composition phase diagram is qualitatively incompatible with experimental observations, indicating that the predicted stability of long-period superstructures as ground states in the Co-Pt binary is incorrect. We also show that recently suggested methods to better align DFT and experiment via the hybrid functional HSE06 are unable to resolve the discrepancies in this system. Our results indicate a need for better verification of DFT based phase stability predictions, and highlight fundamental flaws in the ability of DFT to treat late 3$d$-5$d$ binary alloys.