Machine-Learning-Guided Prediction Models of Critical Temperature of Cuprates

Abstract

Cuprates have been at the center of long debate regarding their superconducting mechanism; therefore, predicting the critical temperatures of cuprates remains elusive. Herein, using machine learning and first-principles calculations, we predict the maximum superconducting transition temperature (<i>T</i>_c,max) of hole-doped cuprates and suggest the functional form for <i>T</i>_c,max with the root-mean-square-error of 3.705 K and <i>R</i>² of 0.969. We have found that the Bader charge of apical oxygen, the bond strength between apical atoms, and the number of superconducting layers are essential to estimate <i>T</i>_c,max. Furthermore, we predict the <i>T</i>_c,max of hypothetical cuprates generated by replacing apical cations with other elements. Among the hypothetical structures, the cuprates with Ga show the highest predicted <i>T</i>_c,max values, which are 71, 117, and 131 K for one, two, and three CuO₂ layers, respectively. These findings suggest that machine learning could guide the design of new high-<i>T</i>_c superconductors in the future.