Sonwane C G, Li Q
National Institute of Standards and Technology, 836/221, 100 Bureau Drive, Gaithersburg, MD 20899, USA.
J Phys Chem B. 2005 Mar 31;109(12):5691-9. doi: 10.1021/jp044775w.
In the present manuscript, we have presented the simulation of nanoporous aluminum oxide using a molecular-dynamics approach with recently developed dynamic charge transfer potential using serial/parallel programming techniques (Streitz and Mintmire Phys. Rev. B 1994, 50, 11996). The structures resembling recently invented ordered nanoporous crystalline material, MCM-41/SBA-15 (Kresge et al. Nature 1992, 359, 710), and inverted porous solids (hollow nanospheres) with up to 10 000 atoms were fabricated and studied in the present work. These materials have been used for separation of gases and catalysis. On several occasions including the design of the reactor, the knowledge of surface diffusion is necessary. In the present work, a new method for estimating surface transport of gases based on a hybrid Monte Carlo method with unbiased random walk of tracer atom on the pore surface has been introduced. The nonoverlapping packings used in the present work were fabricated using an algorithm of very slowly settling rigid spheres from a dilute suspension into a randomly packed bed. The algorithm was modified to obtain unimodal, homogeneous Gaussian and segregated bimodal porous solids. The porosity of these solids was varied by densification using an arbitrary function or by coarsening from a highly densified pellet. The surface tortuosity for the densified solids indicated an inverted bell shape curve consistent with the fact that at very high porosities there is a reduction in the connectivity while at low porosities the pores become inaccessible or dead-end. The first passage time distribution approach was found to be more efficient in terms of computation time (fewer tracer atoms needed for the linearity of Einstein's plot). Results by hybrid discrete-continuum simulations were close to the discrete simulations for a boundary layer thickness of 5lambda.
在本手稿中，我们使用分子动力学方法，并结合最近开发的动态电荷转移势，利用串行/并行编程技术（Streitz和Mintmire，《物理评论B》，1994年，第50卷，第11996页），对纳米多孔氧化铝进行了模拟。在本工作中，制备并研究了与最近发明的有序纳米多孔晶体材料MCM - 41/SBA - 15（Kresge等人，《自然》，1992年，第359卷，第710页）相似的结构以及包含多达10000个原子的倒置多孔固体（空心纳米球）。这些材料已被用于气体分离和催化。在包括反应器设计在内的多种情况下，表面扩散知识是必要的。在本工作中，引入了一种基于混合蒙特卡罗方法的新方法，用于估计气体在孔隙表面的输运，该方法采用示踪原子在孔隙表面进行无偏随机游走。本工作中使用的非重叠堆积是通过一种算法制备的，该算法将稀悬浮液中的刚性球体非常缓慢地沉降到随机填充床中。对该算法进行了修改，以获得单峰、均匀高斯分布和分离双峰的多孔固体。这些固体的孔隙率可通过使用任意函数进行致密化或从高度致密的颗粒进行粗化来改变。致密化固体的表面曲折度呈现出倒钟形曲线，这与以下事实一致：在非常高的孔隙率下，连通性会降低，而在低孔隙率下，孔隙会变得难以接近或成为死端。就计算时间而言（爱因斯坦图的线性所需的示踪原子较少），首次通过时间分布方法被发现更有效。混合离散 - 连续模拟的结果与边界层厚度为5λ的离散模拟结果接近。
