Computational Discovery of New Zeolite-Like Materials

We present a database of computationally predicted zeolite-like materials. The materials were identified by a Monte Carlo search of Si atom positions as the number of unique atoms, density, space group, and unit cell of the crystalline material was systematically explored. Over 2.7M unique structures were identified, with roughly 10% within the +30 kJ/mol Si energetic band above α-quartz in which the known zeolites lie. Predicted structures within this band have geometric and topological characteristics similar to that of the known zeolites. Known zeolites are shown to lie on the low-density edge of the distribution of predicted structures. Dielectric constants and X-ray powder diffraction patterns are calculated. Strategies for chemical synthesis of these materials are discussed, a low-density subset of the materials is identified as particularly interesting, and the complementarity of these materials to high-throughput methods is discussed. These structures have been deposited in two publicly available databases.

本研究构建了一个经计算预测的类沸石材料（zeolite-like materials）数据库。该数据集通过蒙特卡洛（Monte Carlo）搜索硅原子位置构建，系统探索了晶体材料的独特原子数目、密度、空间群及晶胞参数。研究共筛选出超过270万个独特结构，其中约10%位于α-石英上方30 kJ/mol的硅基能量带内，已知沸石即处于该能量带中。该能量带内的预测结构，其几何与拓扑特征与已知沸石高度相似。研究表明，已知沸石恰好位于预测结构分布的低密度边缘。本研究已计算得到这些材料的介电常数（dielectric constants）与X射线粉末衍射（X-ray powder diffraction）图谱。此外，本研究讨论了此类材料的化学合成策略，甄别出其中的低密度子集为极具研究价值的对象，并阐述了该类材料与高通量（high-throughput）实验方法的互补性。上述结构已被存入两个公开可访问的数据库中。