Complexes of the type [M(phen)3](PF6)2 (M = Ni(II), Fe(II), Ru(II) and phen = 1,10-phenanthroline) were found to co-crystallize to form molecular alloys (solid solutions of molecules) with general formula [MA x MB 1–x (phen)3](PF6)2·0.5H2O in which the relative concentrations of the metal complexes in the crystals closely match those in the crystallizing solution. Consequently, the composition of the co-crystals can be accurately predicted and controlled by modulating the relative concentrations of the metal complexes in the crystallizing solution. Although they are chemically and structurally similar, complexes of the type [M(bipy)3](PF6)2 (M = Ni(II), Fe(II), Ru(II) and bipy = 2,2′-bipyridine) display markedly different behavior upon co-crystallization. In this case, the resulting co-crystals of general formula [MA x MB 1–x (bipy)3](PF6)2 have relative concentrations of the constituent complexes that are markedly different from the relative concentrations of the complexes initially present in the crystallizing solution. For example, when the nickel and iron complexes are co-crystallized from a solution containing a 50:50 ratio of each, the result is the formation of some crystals with a higher proportion of iron and others with a higher proportion of nickel. The relative concentrations of the metal complexes in the crystals can vary from those in the crystallizing solutions by as much as 15%. This result was observed for a range of combinations of metal complexes (Ni/Fe, Ni/Ru, and Fe/Ru) and a range of starting concentrations in the crystallizing solutions (90:10 through to 10:90 in 10% increments). To explain this remarkable result, we introduce the concept of “supramolecular selection”, which is a process driven by molecular recognition that leads to the partially selective aggregation of like molecules during crystallization.

Heteroleptic complexes of the type [RuL2L′]­(PF6)2 (L, L′ = combinations of 1,10-phenanthroline (phen) and 2,2′-bipyridine (bipy)) were found to cocrystallize with [Ni­(phen)3]­(PF6)2 to produce cocrystals of [Ni­(phen)3] x ­[RuL2L′]1–x ­(PF6)2. In this report we show that the ability of the complexes to cocrystallize is influenced by the number of common ligands between complexes in solution. Supramolecular selection is a phenomenon caused by molecular recognition through which cocrystals can grow from the same solution but contain different ratios of the molecular components. It was found that systems where L = phen displayed less supramolecular selection than systems where L = bipy. With increasing supramolecular selection, the composition of cocrystals was found to vary significantly from the initial relative concentration in the cocrystallizing solution, and therefore it was increasingly difficult to control the final composition of the resultant cocrystals. Consequently, modulation of concentration-dependent properties such as phase was also found to be less predictable with increasing supramolecular selection. Notwithstanding the complication afforded by the presence of supramolecular selection, our results reaffirm the robustness of the [M­(phen)3]­(PF6)2 structure because it was maintained even when ca. 90% of the complexes in the cocrystals were [Ru­(phen)­(bipy)2]­(PF6)2, which in its pure form is not isomorphous with [M­(phen)3]­(PF6)2. Experiments between complexes without common ligands, i.e., [Ru­(bipy)3]­(PF6)2 cocrystallized with [Ni­(phen)3]­(PF6)2, were found to approach the limit to which molecular recognition processes can be confused into cocrystallizing different molecules to form single cocrystals. For these systems the result was the formation of block-shaped crystals skewered by a needle-shaped crystals.

No abstract