Camino Bruno, Buckeridge John, Chancellor Nicholas, Catlow C Richard A, Ferrari Anna Maria, Warburton Paul A, Sokol Alexey A, Woodley Scott M
Chemistry Department, University College London, 20 Gordon St, London WC1H 0AJ, UK.
School of Engineering and Design, London South Bank University, 103 Borough Rd, London SE10 AA, UK.
Sci Adv. 2025 Jun 6;11(23):eadt7156. doi: 10.1126/sciadv.adt7156.
Alloys, solid solutions, and doped systems are essential in technologies such as energy generation and catalysis, but predicting their properties remains challenging because of compositional disorder. As the concentration of components changes in a binary solid solution [Formula: see text] , the number of possible configurations becomes computationally intractable. Algorithms used in classical optimization methods cannot avoid assessing high-energy states where, for example, simulated annealing is designed to initially spend computational effort. We introduce a scalable, practical, and accurate approach using quantum annealing to efficiently sample low-energy configurations of disordered materials, avoiding the need for excessive high-energy calculations. Our method includes temperature and simulates large unit cells, producing a Boltzmann-like distribution to identify thermodynamically relevant structures. We demonstrate this by predicting bandgap bowing in [Formula: see text] and bulk modulus variations in [Formula: see text] , with results in excellent agreement with experiments.
合金、固溶体和掺杂体系在能源生成和催化等技术中至关重要，但由于成分无序，预测它们的性质仍然具有挑战性。随着二元固溶体[化学式：见正文]中组分浓度的变化，可能的构型数量在计算上变得难以处理。经典优化方法中使用的算法无法避免评估高能态，例如模拟退火最初就被设计用于花费计算精力处理这些高能态。我们引入一种使用量子退火的可扩展、实用且准确的方法，以有效地对无序材料的低能构型进行采样，避免进行过多的高能计算。我们的方法包括温度并模拟大晶胞，产生类似玻尔兹曼分布以识别热力学相关结构。我们通过预测[化学式：见正文]中的带隙弯曲和[化学式：见正文]中的体模量变化来证明这一点，结果与实验结果非常吻合。
