Novel Approach for Clustering Zeolite Crystal Structures.

Informatics approaches play an increasingly important role in the design of new materials. In this work we apply unsupervised statistical learning for identifying four framework-type attractors of zeolite crystals in which several of the zeolite framework types are grouped together. Zeolites belonging to these super-classes manifest important topological, chemical and physical similarities. The zeolites form clusters located around four core framework types: LTA, FAU, MFI and the combination of EDI, HEU, LTL and LAU. Clustering is performed in a 9-dimensional space of attributes that reflect topological, chemical and physical properties for each individual zeolite crystalline structure. The implemented machine learning approach relies on hierarchical top-down clustering approach and the expectation maximization method. The model is trained and tested on ten partially independent data sets from the FIZ/NIST Inorganic Crystal Structure Database.

Colloidal aggregation with mobile impurities.

The structure and aggregation kinetics of diffusion-limited cluster-cluster three-dimensional monomeric aggregates and gels is investigated as a function of the molar fraction of two types of impurities. In one case the impurities are allowed to aggregate among themselves whereas in the other the impurities are mobile monomers that remain as such during the whole aggregation process. Computer simulations are performed on a simple cubic lattice for which the functionality of the aggregating particles is effectively 6. The first type of impurity shows a decrease in the fractal dimension when compared to that of a one component system at the same concentration. As a consequence of this decrease, the gelation concentration is lowered. At higher concentrations a gelling to nongelling transition was observed. In the nongelling regime the colloidal aggregates are kept apart by the impurity clusters, developing a local ordering. For the monomeric type of impurity, at large impurity molar fractions, a nonstructured nongelling phase appears at high enough concentration, in which the colloidal aggregates are kept apart by the sea of mobile impurities that inhibits the formation of a gel. Smaller molar fractions of mobile monomeric impurities strongly affect both the fractal dimension and the kinetics of the aggregating colloid.

Cluster-cluster aggregation in binary mixtures

The structure and aggregation kinetics of three-dimensional clusters composed of two different monomeric species at three concentrations are thoroughly investigated by means of extensive, large-scale computer simulations. The aggregating monomers have all the same size and occupy the cells of a cubic lattice. Two bonding schemes are considered: (a) the binary diffusion-limited cluster-cluster aggregation (BDLCA) in which only the monomers of different species stick together, and (b) the invading binary diffusion-limited cluster-cluster aggregation (IBDLCA) in which additionally monomers of one of the two species are allowed to bond. In the two schemes, the mixed aggregates display self-similarity with a fractal dimension d(f) that depends on the relative molar fraction of the two species and on concentration. At a given concentration, when this molar fraction is small, d(f) approaches a value close to the reaction-limited cluster-cluster aggregation of one-component systems, and when the molar fraction is 0.5, d(f) becomes close to the value of the diffusion-limited cluster-cluster aggregation model. The crossover between these two regimes is due to a time-decreasing reaction probability between colliding particles, particularly at small molar fractions. Several dynamical quantities are studied as a function of time. The number of clusters and the weight-average cluster size display a power-law behavior only at small concentrations. The dynamical exponents are obtained for molar fractions above 0.3 but not at or below 0.2, indicating the presence of a critical transition between a gelling to a nongelling system. The cluster-size distribution function presents scaling for molar fractions larger than 0.2.