A concept for synthesis planning in solid-state chemistry.

There is a widely-held belief that the preparation of new solid-state compounds based on rational design is not possible. Herein, we present a concept that points the way towards a rational design of syntheses in solid-state chemistry. The foundation of our approach is the representation of the whole material world, that is, the known and not-yet-known compounds, on an energy landscape, which gives information about the free energies of these compounds. From this it follows that all chemical compounds capable of existence are present on this landscape. Thus the chemical synthesis always corresponds to the discovery of compounds, not their creation. Consequently, the first step in planning a synthesis can and must be to identify a synthesizable compound. Up to now, materials capable of existence are discovered in the course of an experimental exploration of the energy landscape; however, an a priori identification of a synthesis goal requires an exploration using theoretical methods. In contrast to those computational approaches currently employed for structure determination for fixed composition and already known unit cells, our aims clash with such restrictions and full global optimizations have to be performed on the landscape. Although for reasons of computational feasibility the accuracy of the energy calculations is not yet as high as one would wish, our approach proves to be surprisingly robust. One always finds the already known compounds of a given chemical system, and, in addition, further plausible structure candidates are discovered. The second step of a rational planning of syntheses is the design of feasible synthesis routes. Modeling such routes requires highly accurate computations for realistic thermodynamic conditions, however this is usually beyond our current capabilities. Thus, we have not seriously pursued such a deductive approach; instead we have attempted, to reproduce the "computational annealing" employed during our structure predictions in the experiment. Educts, generated by vapor deposition methods, that are disperse on an atomic level are found to react with surprisingly low activation energies to give highly crystallized products. However, even this technique does not yet provide the possibility to selectively synthesize a specific solid compound. For this final step, modeling and experimental control of nucleation processes will be the key ingredient. Only when viewed superficially, our goal of a "rational design" of solid-state syntheses and the "high-throughput" syntheses are in contradiction. But an exhaustive exploration of the unimaginably large combinatorial diversity of chemistry remains beyond our capabilities, even with an exceedingly high throughput. The future of solid-state synthesis will be found in a union of these two conceptual approaches.