Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802, USA.
J Chem Phys. 2017 Oct 21;147(15):152711. doi: 10.1063/1.4995425.
We use first-principles density-functional theory to characterize the binding sites and diffusion mechanisms for a Ga adatom on the GaAs(001)β2(2 × 4) surface. Diffusion in this system is a complex process involving eleven unique binding sites and sixteen different hops between neighboring binding sites. Among the binding sites, we can identify four different superbasins such that the motion between binding sites within a superbasin is much faster than hops exiting the superbasin. To describe diffusion, we use a recently developed local superbasin kinetic Monte Carlo (LSKMC) method, which accelerates a conventional kinetic Monte Carlo (KMC) simulation by describing the superbasins as absorbing Markov chains. We find that LSKMC is up to 4300 times faster than KMC for the conditions probed in this study. We characterize the distribution of exit times from the superbasins and find that these are sometimes, but not always, exponential and we characterize the conditions under which the superbasin exit-time distribution should be exponential. We demonstrate that LSKMC simulations assuming an exponential superbasin exit-time distribution yield the same diffusion coefficients as conventional KMC.
我们使用第一性原理密度泛函理论来描述 Ga 原子在 GaAs(001)β2(2×4)表面上的吸附位置和扩散机制。在这个体系中，扩散是一个复杂的过程，涉及十一个独特的吸附位置和十六种不同的相邻吸附位置之间的跃迁。在这些吸附位置中，我们可以识别出四个不同的超区，使得超区内的吸附位置之间的运动比离开超区的跃迁快得多。为了描述扩散，我们使用了最近开发的局部超区动力学蒙特卡罗（LSKMC）方法，该方法通过将超区描述为吸收马尔可夫链，加速了传统的动力学蒙特卡罗（KMC）模拟。我们发现，对于本研究中探测到的条件，LSKMC 比 KMC 快 4300 倍。我们描述了从超区退出的时间分布，并发现这些时间分布有时是指数分布，但并不总是指数分布，我们还描述了超区退出时间分布应该是指数分布的条件。我们证明了假设超区退出时间分布为指数分布的 LSKMC 模拟可以得到与传统 KMC 相同的扩散系数。
