A deft colloidal templating process allows simple-cubic crystals to be formed from more readily available complex precursors. It's a promising way to produce the regular crystals much in demand for photonics.
Colloidal crystals — arrangements of generally spherical particles between around 10 and 100 nanometres in size — typically assume a dense face-centred-cubic (f.c.c.) order, with particles at each vertex and in the middle of each side of the crystal's constituent cubic cells. This kind of packing is tight, filling about 74% of the available space; but the small gaps left can be infiltrated by other, smaller particles to form a replica of the original f.c.c. template after this has been removed.
Writing in Angewandte Chemie International Edition, Li et al.1 show how such a structure can be made to reassemble when its template is removed thermally (calcinated), to form a less densely packed, simple-cubic structure (Fig. 1). Their work represents a new way to synthesize crystals with structures that deviate from dense sphere packings2, and that until now were hard to make. Such crystals are much in demand, for instance as photonic crystals that act on light as semiconductors do on electrons.
Binary nanoparticle systems of various compositions have already been assembled into different packing motifs3,4. But to create non-close-packed structures of larger particles (in the 100-nanometre range), either the forces in the system must be maintained in a delicate balance5, or prestructured surfaces are required to act as a template6.
Li and colleagues' approach is different, and involves first infiltrating the pores of a close-packed colloidal crystal of poly(methylmethacrylate) (PMMA) spheres with a cocktail of chemicals in solution — titanium isopropoxide, triethylphosphate, the surfactant Brij 56 and acetylacetone. On calcination, this precursor is converted to a binary titanium dioxide–phosphorus pentoxide (TiO2–P2O5) structure that assumes cuboid and spheroidal shapes in the two differently sized gaps found in the f.c.c. PMMA crystal. If the particles remain fully connected, an 'inverse opal structure', essentially a negative replica of the PMMA crystal, is obtained when the template is calcinated. If, on the other hand, these connections are lost, the particles become randomly oriented. The authors' result represents a 'third way' between these two extremes.
This happy medium is made possible because the larger and energetically favoured cuboid structures grow, through a process known as Ostwald ripening, at the expense of the smaller spheres, which eventually vanish. Adjacent sheets of the f.c.c. structure are then assumed to collapse on each other, leading to the observed simple-cubic packing. Simple cubic might in fact be the closest — and thus energetically favoured — packing possible for the new arrangement: whereas for spheres, the space-filling factor of 74% for f.c.c. or hexagonal packing is the best that can be achieved, ideal cubes pack closest in tetragonal layers with a space filling of 100%. As Li and colleagues' cuboids are somewhere between a perfect cube and a sphere, it is difficult to predict theoretically what their closest packing would be.
Meanwhile, the interesting question is how general the pathway described by the authors is and what applications the new materials might have. The authors have observed similar ordered regions in a zirconium dioxide–phosphorus pentoxide (ZrO2–P2O5) system, and speculate that the presence of a phosphate-rich phase is crucial to the formation of these regions. Such a phase has a low melting point, and could provide the liquid component necessary for the growth of the cuboid particles at the expense of the rounded ones. The principle involved is quite general, and should be applicable to other systems, with appropriate changes to concentrations, temperatures and times of treatment.
Whether a wider range of structures can be obtained is more difficult to predict. Clearly, a parent structure with differently sized voids is needed. For the formation of a simple-cubic packing, the fact that holes are on top of particles, and vice versa, in adjacent layers of the parent f.c.c. structure, allowing the replica layers to collapse into each other on calcination, seems to be important. Cubes from a redispersed simple-cubic colloidal crystal do not reassemble into an ordered structure. Preordering thus has a crucial role. With cuboids deviating from the perfect cube shape, as in Li and colleagues' work, there are structures with higher packing density than the simple cubic packing. An example is a stacking with every second layer shifted, resulting in a 'body-centred' tetragonal (stretched-cube) structure (Fig. 2).
Cuboids that have different surface curvatures could also induce variations in the surface properties and effective cohesion forces of the structure, allowing position-sensitive interactions between nanoparticles that show promise for the design of other structures. If, for instance, Li and colleagues' process1 could be driven to result in tetrahedral building-blocks, these could form other interesting structures, including some with fewer symmetry elements. This symmetry lowering enables remarkable changes to occur in the photonic bandgap properties that are so essential to the use of these materials as photonic crystals for guiding the propagation of light7. The potential of colloidal crystals for photonic applications increases markedly if their features are around the same size as the wavelength of visible light.
The standard approach to generating photonic crystals, rather like the standard techniques for creating patterned circuits in the semiconductor industry, uses lithography, albeit in this case for the generation of three-dimensional structures. Compared with that approach, nanoparticle self-assembly as previously implemented has two great drawbacks: it produces too many defects, and it is essentially restricted to f.c.c. lattices. Li and colleagues' method shows a way to avoid the second drawback. If nanoparticle self-assembly is to have photonic applications, attention must now turn to ridding the simple-cubic packing of old defects inherited from the parent template, and of new ones introduced during the replication process.
About this article
Chemical Physics Letters (2009)