原子单位
材料科学
等离子体
带隙
平滑的
原子扩散
光电子学
化学
结晶学
物理
量子力学
计算机科学
计算机视觉
作者
Yongjie Zhang,Jin Tang,Shaoxiang Liang,J. Zhao,Mengyuan Hua,Chun Zhang,Hui Deng
标识
DOI:10.1016/j.ijmachtools.2024.104119
摘要
β-Ga2O3, known as a next-generation wide-bandgap transparent semiconducting oxide (TSO), has considerable application potential in ultra-high-power and high-temperature devices. However, fabricating a smooth β-Ga2O3 substrate is challenging owing to its strong mechanical strength and chemical stability. In this study, an atomic-scale smoothing method named plasma-enabled atomic-scale reconstruction (PEAR) is proposed. We find that three reconstruction modes, namely, 2D-island, step-flow, and step-bunching, can be identified with the increase in the input power; only the step-flow mode can result in the formation of an atomically smooth β-Ga2O3 surface (Sa = 0.098 nm). Various surface and subsurface characterizations indicate that the smooth β-Ga2O3 surface shows excellent surface integrity, high crystalline quality, and remarkable photoelectric properties. The atomic-scale density functional theory-based calculations show that the diffusion energy barrier of a Ga atom is only 0.46 eV, thereby supporting the atomic mass migration induced by high-energy plasma irradiation in the experiment. Nanoscale molecular dynamics simulations reveal that O atoms firstly migrate to crystallization sites, followed by Ga atoms with a lower migration rate; reconstruction mainly proceeds along the <010> direction and then expands along the <100> and <001> directions. The millimeter-scale numerical simulations based on the finite element method demonstrate that the coupling of the thermal and flow fields of plasma is the impetus for PEAR of β-Ga2O3. Furthermore, the smoothing generality of PEAR is demonstrated by extending it to other common TSOs (α-Al2O3, ZnO, and MgO). As a typical plasma-based atomic-scale smoothing method, PEAR is expected to enrich the theoretical and technological knowledge on atomic-scale manufacturing.
科研通智能强力驱动
Strongly Powered by AbleSci AI