面向低功耗光存储应用的超快激光诱导有序相变的第一性原理研究

Non-volatile phase-change memory (PCM) controlled by ultrafast laser is a key candidate for optical memory/computing towards big-data applications. Conventional phase transitions induced by laser are often governed by thermal melting which costs a large amount of energy. Recent experiments find that ultrafast laser can also induce phase transitions from crystalline phase to crystalline phase, which is named the “order-to-order phase transition” (OPT) in this project. The absence of high-temperature melting makes the OPT a promising proposal for low-power-consumption memory. However, there still exists some problems in the discovered OPT: ①the final phase is unstable which do not favor the memory applications; ②the physical mechanism is unclear which hinders the exploration of memory based on ultrafast laser induced OPT. Based on first-principles calculations, this project proposes an innovative idea of directional searching to solve these problems. In this idea, the OPT with a stable final state is decomposed into two physical processes, namely, ultrafast laser induced directional motions of atoms and subsequent overcoming of barriers to form a stable phase. Then the two processes will be used as criterions to filter and design the material system and excitation condition for OPT, which can accelerate the progress of memory technology based on OPT. The main contents of this project focus on the dynamic mechanisms and control factors of the two physical processes, which provide theoretical guidance for the exploration of OPT. The detailed contents contains: analyses on the physical mechanisms of ultrafast laser induced directional motions of atoms in a series of phase change materials; understanding of the microscopic dynamics of laser induced OPT; summarization of the material system and excitation condition for laser induced OPT. Finally, the project will try to propose a theoretical proposal for memory devices based on ultrafast laser induced OPT.

超快激光激励的非易失性相变存储技术是面向大数据应用的光存储、光计算中的关键候选技术。传统光控相变的物理机制以热熔化为主，最近发现超快激光可诱导从晶态到晶态的有序相变，为低功耗存储提供了新思路，但其主要问题是①现有体系的末态相不稳定，未能满足存储应用的要求；②超快激光诱导有序相变的机理不清楚，导致光控有序相变的探索缺少方向性。本项目立足于第一性原理方法解决以上问题，提出一种定向搜索的创新思路，其主要内涵是，将具有稳定末态相的有序相变分解为光激发诱导原子定向运动和原子移动跨越势垒进入新稳态两个物理过程，以这两个过程作为材料体系和激发条件的筛选与设计依据，有助于快速锁定末态相稳定的有序相变体系。具体研究着重于以上两个过程的动力学机制和控制因素：分析各类相变存储材料中光激发驱动原子运动的原理和有序相变的动力学机制，为材料体系和激发条件的选择与设计提供理论依据，并尝试提出光控有序相变的设计方案。

以超快激光驱动非易失性材料相变为基础的存储技术具有速度快、非接触远程控制等优点，是面向未来大数据和人工智能应用的光存储、光计算等先进应用中的关键候选技术。然而，传统光控相变的物理机制以热效应为主，功耗高、速度受限于形核-生长过程。而超快激光诱导从晶态到晶态的有序相变为低功耗存储提供了新思路，其主要问题是理论机制不够清楚，仍然缺乏鲁棒的可非易失性转变的材料体系。本项目力图揭示光控有序相变的理论机制并以此为基础提出具体的光控非易失性有序结构转变理论方案。理论机制方面，首先基于密度泛函理论的第一性原理计算方法从电子、原子尺度研究光激发驱动结构相变的驱动力和规律，揭示出光激发的轨道选择性和对称性依赖可用于控制原子的定向运动，并证明静态的激发态势能面计算可用于判断相变趋势。进一步地，采用含时密度泛函理论分子动力学方法研究非易失性结构转变的动力学机制，揭示出驱动超快有序相变的关键是原子的相干运动，而保证结构转变的非易失性则依靠电子、原子的退相干作用，这些过程中最核心的物理因素是激发态的电声耦合作用。以揭示出的理论机制为基础，项目总结出实现光控非易失性有序结构转变的潜在体系应为类似具有派尔斯扭曲/姜-泰勒扭曲的材料，例如铁电材料和电荷密度波材料。以具有铁电性的典型相变存储材料GeTe为例，项目证明了低功耗光控超快非易失性结构转变是可行的。另外，项目进一步探索了面向光电存储应用的不同相变/铁电材料中的结构转变机制，除已发表的成果外，仍有部分工作仍在进行中，预计未来两年内陆续发表。本项目的科学意义在于为利用超快激光精细控制材料物性提供了理论参考。特别地，本项目聚焦的有序相变由超快激光的非热效应驱动，不经历传统的熔化或形核-生长过程，其速度可达亚皮秒尺度，是面向未来光存储/计算应用的极具潜力的新方案。