The controlled synthesis of materials with very large pores is an ongoing challenge in the area of materials science. Research into large-pore materials is fueled by their use in catalysis, gas storage, and separation. [1] After intense development of inorganic frameworks in the early 1990s, [2] the subsequent discovery of hybrid porous solids, in which the connection of inorganic moieties is ensured by organic functionalized N-donor or O-donor molecules, paved the way for the rational design of hybrid frameworks. [3][4][5][6][7][8][9][10][11] A new class of materials has emerged at the crossroads of inorganic materials science and coordination chemistry. Among the most illustrative examples of open frameworks are the zinc carboxylate series by Yaghi, OKeeffe, and co-workers, and the transition-metal terephtalates that have remarkable methane-and hydrogen-storage properties, [12] together with, for example, a recent series of 3d transition-metal [13a, 14] and rare-earth [15] -based hybrids that exhibit interesting magnetic properties. [13a] The richness of this area lies in the diversity of topologies (from molecular to 3D) and properties, which is conveyed in the wide choice of metal atoms that are available, combined with a virtually infinite choice of organic counterparts (e.g., carboxylates, phosphonates, polyamines).
In the current search for new and interesting hybrid open frameworks, the predictability of the framework architecture and the control of its dimensionality are essential, even if one is confronted with the underlying issue of polymorphism. [6] The possibility of rational design with these types of solids has rapidly emerged through the use of topological and chemical considerations on existing networks. [7] The concept of rational design is rooted in the fact that topochemically selected reactions govern the construction process of the hybrid framework under hydrothermal conditions. Although metalcontaining secondary building units (SBUs) may not be isolated, their recurrence in a large number of structures suggests that the targeted inorganic subunit already exits in solution before their condensation into the framework structure and thus may be obtained under appropriate synthetic conditions. For example, this aspect is apparent through the iso-reticular synthesis of IRMOFs1-16 [12b] derived from the prototypic MOF-5 structure. [12a] In this context, structural prediction is an important issue. It is crucial to consider how systematic approaches might be computationally developed for producing new hybrid frameworks, with the desire of developing virtual libraries that might be accessible by rational synthesis.
It is worth underscoring here that crystal-structure prediction is now routinely explored in organic chemistry and polymer science, [16,17] in which candidate structures may be predicted by assembling molecular entities through hydrogen-bond intermolecular interactions. In contrast, such developments have only been recently reported for inorganic crystal structures, which are extremely difficult to predict due to their infinite lattices. Pioneered by Newsam et al. in the field of zeolites, [18] crystal-structure prediction [18][19][20][21][22][23][24][25][26][27][28] and rational design [29][30][31] are now at work in the field of open frameworks. Due to their ability to cross hypersurface energy barriers and search for low-energy regions, global optimization techniques are intensively used to predict atomic-scale arrangements of infinite lattices and are able to handle the assembly of atoms, ions, [20,[22][23][24][25] or predefined building units in three dimensions. [21,27,28] We have introduced the concept of building units for the computational prediction of crystal structures with the AASBU method (automated assembly of secondary building units). [21,27,28] This method explores the possible ways of assembling predefined inorganic building units, and focuses on the topology of network-based structures.
Inspired by these recent developments, we present herein the extension of the AASBU method to the realm of hybrid organic-inorganic frameworks and demonstrate its capacity to produce hybrid candidate crystal structures that are built from predefined organic and inorganic counterparts. To our knowledge, no systematic computational strategy has been reported in this field so far.
Indeed, hybrid frameworks offer ideal features for computational developments: although the isolated metal ion, taken alone, lacks directional information, the inorganic unit derived from the metal atom and the organic ligand does (Figure 1). Once the inorganic and organic units are defined, one may assume that there are a limited number of arrangements that are compatible with periodicity and symmetry. With predefined organic and inorganic building units, AASBU simulations are used here to perform their automated assembly in three dimensions, thus exploring the possibilities of connection. The simulations yield a virtual library of candidate hybrid frameworks that are assorted by their space group, cell parameters, and atomic positions.
Our initial efforts were aimed at validating the AASBU approach in the field of hybrid frameworks by simulating existing architectures. In the second instance, we aimed at predicting structures that have not yet been synthesized, both to tackle the issue of polymorphism by limiting the domain of [*] Dr.