Scholarlink

<p>Au-Cu bimetallic clusters are attractive for catalysis and electronic applications owing to their synergistic effects and tunable structures. We propose an improved Elite Niching Adaptive Differential Evolution (ENADE) algorithm for their structural optimization. Building on conventional differential evolution, ENADE introduces an elite pool to improve convergence, a multi-population parallel strategy to strengthen global search, combined sphere-cutting crossover and atomic-exchange operators, and a restart mechanism coupled with niching to prevent premature convergence. Using the Gupta potential, the algorithm was validated on Au-Cu sequences (<i>N</i> = 13, 19, 23, 38), benchmark systems with well-characterized global minima. ENADE reached a 70% success rate for the <i>N</i> = 38 global minimum, and across 20 independent <i>N</i> = 55 runs the best-energy standard deviation was only 0.002&#xa0;eV, significantly outperforming basin-hopping, genetic-algorithm, and conventional DE baselines. The located structures serve as a consistency check rather than new structural claims: they reproduce the known Cu-core/Au-surface segregation driven by size mismatch and the expected truncated-octahedral <i>N</i> = 38 motif, with alloy clusters more stable than pure-metal counterparts. The main contribution is therefore the ENADE framework and its robustness on chemically ordered bimetallic landscapes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Global structural optimization and stability study of Au-Cu bimetallic clusters based on Elite Niching Adaptive Differential Evolution

  • 期刊论文

  • 发表日期：

    2026年7月7日

查看全文
  • Yuxiang Tang,
  • Yuxin Zhang,
  • Hua Shi,
  • Yuheng Chen,
  • Xiaomin Wu

摘要

Au-Cu bimetallic clusters are attractive for catalysis and electronic applications owing to their synergistic effects and tunable structures. We propose an improved Elite Niching Adaptive Differential Evolution (ENADE) algorithm for their structural optimization. Building on conventional differential evolution, ENADE introduces an elite pool to improve convergence, a multi-population parallel strategy to strengthen global search, combined sphere-cutting crossover and atomic-exchange operators, and a restart mechanism coupled with niching to prevent premature convergence. Using the Gupta potential, the algorithm was validated on Au-Cu sequences (N = 13, 19, 23, 38), benchmark systems with well-characterized global minima. ENADE reached a 70% success rate for the N = 38 global minimum, and across 20 independent N = 55 runs the best-energy standard deviation was only 0.002 eV, significantly outperforming basin-hopping, genetic-algorithm, and conventional DE baselines. The located structures serve as a consistency check rather than new structural claims: they reproduce the known Cu-core/Au-surface segregation driven by size mismatch and the expected truncated-octahedral N = 38 motif, with alloy clusters more stable than pure-metal counterparts. The main contribution is therefore the ENADE framework and its robustness on chemically ordered bimetallic landscapes.