Abstract The global demand for data storage and processing is increasing exponentially. To deal with this challenge, massive efforts have been devoted to the development of advanced memory and computing technologies. Chalcogenide phase-change materials (PCMs) are currently at the forefront of this endeavor. In this Review, we focus on the mechanisms of the spontaneous structural relaxation – aging – of amorphous PCMs, which causes the well-known resistance drift issue that significantly reduces the device accuracy needed for phase-change memory and computing applications. We review the recent breakthroughs in uncovering the structural origin, achieved through state-of-the-art experiments and ab initio atomistic simulations. Emphasis will be placed on the evolving atomic-level details during the relaxation of the complex amorphous structure. We also highlight emerging strategies to control aging, inspired by the in-depth structural understanding, from both materials science and device engineering standpoints, that offer effective solutions to reduce the resistance drift. In addition, we discuss an important new paradigm – machine learning – and the potential power it brings in interrogating amorphous PCMs as well as other disordered alloy systems. Finally, we present an outlook to comment on future research opportunities in amorphous PCMs, as well as on their reduced aging tendency in other advanced applications such as non-volatile photonics.

中文翻译：

---

揭示控制相变存储材料电阻漂移的结构起源

摘要 全球对数据存储和处理的需求呈指数级增长。为了应对这一挑战，已投入大量精力开发先进的内存和计算技术。硫属化物相变材料 (PCM) 目​​前处于这一努力的最前沿。在这篇综述中，我们关注非晶 PCM 的自发结构松弛机制——老化，这会导致众所周知的电阻漂移问题，从而显着降低相变存储器和计算应用所需的器件精度。我们回顾了最近在揭示结构起源方面的突破，这些突破是通过最先进的实验和 ab initio 原子模拟实现的。重点将放在复杂非晶结构松弛过程中不断发展的原子级细节上。我们还强调了控制老化的新兴策略，其灵感来自于从材料科学和器件工程的角度深入的结构理解，这些策略提供了减少电阻漂移的有效解决方案。此外，我们讨论了一个重要的新范式——机器学习——以及它在询问非晶 PCM 以及其他无序合金系统方面带来的潜在力量。最后，我们提出了一个展望，以评论非晶 PCM 的未来研究机会，以及它们在其他先进应用（如非易失性光子学）中降低的老化趋势。我们讨论了一个重要的新范式 - 机器学习 - 以及它在询问非晶 PCM 以及其他无序合金系统方面带来的潜在力量。最后，我们提出了一个展望，以评论非晶 PCM 的未来研究机会，以及它们在其他先进应用（如非易失性光子学）中降低的老化趋势。我们讨论了一个重要的新范式 - 机器学习 - 以及它在询问非晶 PCM 以及其他无序合金系统方面带来的潜在力量。最后，我们提出了一个展望，以评论非晶 PCM 的未来研究机会，以及它们在其他先进应用（如非易失性光子学）中降低的老化趋势。