发表机构
Yale University（耶鲁大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文以Lennard-Jones体系为基准，结合分子动力学、跳跃式正向通量采样和经典成核理论，量化了势能截断对成核速率的影响，发现截断半径变化可致速率改变约十个数量级，并提出了基于CNT的外推框架和最优截断选择标准。
AI 中文摘要
成核速率对热力学驱动力呈指数级敏感，因此可能强烈依赖于长程分子间相互作用的处理方式。本文以Lennard-Jones（LJ）体系为基准，结合分子动力学（MD）模拟、跳跃式正向通量采样（jFFS）和自由能计算，量化了势能截断对熔化热力学、均质晶体成核动力学及计算成本的影响。在截断半径范围$2.5\sigma\le r_c\le 6\sigma$内，零压下的熔化温度变化约11%，而成核速率变化约十个数量级。通过引入经典成核理论（CNT），我们表明这种显著的动力学敏感性主要源于截断引起的液固两相化学势差的变化。基于这一观察，我们发展了一个基于CNT的框架，用于将有限截断速率外推至全势极限，并估计在其他截断半径和温度下预期的速率偏差。这些发现也为截断选择提供了系统依据：最优截断应在将偏离全势速率的偏差控制在可接受范围内的同时最小化计算成本。在$kT/\epsilon=0.5$时，根据这些标准，$r_c=4\sigma$提供了合理的折中。我们进一步证明，传统的均质尾校正不能作为可靠替代方案，因为它们无法一致地解释成核过程中存在的液体、晶体和界面环境。我们的发现强调，在非均匀环境中成核及其他界面相变的模拟中，需要将截断方案的指定和验证作为力场开发的一个组成部分。
英文摘要
Nucleation rates are exponentially sensitive to the thermodynamic driving force and can therefore depend strongly on the treatment of long-range intermolecular interactions. Here, using the Lennard--Jones (LJ) system as a benchmark, we combine molecular dynamics (MD) simulations, jumpy forward-flux sampling (jFFS), and free-energy calculations to quantify the effect of potential truncation on melting thermodynamics, homogeneous crystal nucleation kinetics, and computational cost. Within the cutoff-radius range $2.5σ\le r_c\le 6σ$, the melting temperature at zero pressure varies by approximately 11%, while the nucleation rate changes by approximately ten orders of magnitude. By invoking classical nucleation theory (CNT), we show that this pronounced kinetic sensitivity originates primarily from cutoff-induced changes in the chemical potential difference between the liquid and crystalline phases. Building on this observation, we develop a CNT-based framework for extrapolating finite-cutoff rates to the full-potential limit and for estimating the expected rate deviations at other cutoff radii and temperatures. These findings also provide a systematic basis for cutoff selection: the optimal cutoff should minimize computational cost while keeping the deviation from the full-potential rate within acceptable bounds. At $kT/ε=0.5$, $r_c=4σ$ provides a reasonable compromise according to these criteria. We further demonstrate that conventional homogeneous tail corrections do not offer a reliable alternative, as they cannot consistently account for the liquid, crystalline, and interfacial environments present during nucleation. Our findings highlight the need to specify and validate the truncation scheme as an integral component of force-field development in simulations of nucleation and other interfacial phase transitions within inhomogeneous environments.
Comments14 pages, 10 figures, 1 table
