Soap-like molecules serve as a scaffold for remarkably well-ordered, porous germanium skeletons. The nanometre-sized features of these semiconductor frameworks confer unique optical and electronic properties.
As semiconductor materials are shrunk to the nanoscale, their physical properties begin to alter: colours change, melting points decrease, electron energy bands turn into discrete levels, and reactive surface areas become proportionately larger as particle size decreases1. This applies not only to discrete semiconductor particles, but also for extended ‘mesoporous’ framework structures of nanometre scale. In this issue, two papers2,3 detail approaches to the synthesis of periodic porous structures made of germanium. The optical properties of these materials are shown to depend on their dimensions, composition and the presence of other molecules attached to their surface.
Porosity can be an extremely useful characteristic in a material, with the attendant large surface area benefiting applications in which molecules interact with a surface, such as sensors and catalysts. Silicon with nanoscale holes has been widely studied since its luminescent properties were discovered4. But porous forms of a fellow member of group 14 of the periodic table — named ekasilicon by Mendeleev, but rechristened germanium by its discoverer, Clemens Winkler, in honour of his home country — have been investigated much less. Germanium's interest lies in its semiconducting nature and use in transistors, and also in its applications as a component of fibre-optic cables, as the focusing element in infrared-sensitive night-vision systems and as a polymerization catalyst in the manufacture of soft-drinks bottles. Size-dependent properties can also be achieved with germanium at relatively large feature sizes, and, in a mesoporous form, unusual electronic and optical properties are to be expected.
But few procedures exist to synthesize porous germanium. Etching and vapour-deposition techniques have been used to form germanium films with somewhat random pore structures and feature sizes5,6. Now, Armatas and Kanatzidis (page 1122)2 and Tolbert and colleagues (Sun et al., page 1126)3 describe the use of a versatile technique known as surfactant templating to synthesize mesoporous germanium structures with cubic and hexagonal geometry, respectively.
Surfactant templating involves the self-assembly of inorganic building-blocks through their electrostatic interactions with organic ‘amphiphilic’ surfactant molecules that, like those in soap, have both hydrophilic and hydrophobic properties7. According to how these interactions vary, different geometries are obtained as the inorganic components link to form a framework. Forms of silica with well-ordered cubic or hexagonal channel systems, or with disordered worm-like tunnels or layered structures, have already been synthesized using this technique7,8. In these mesostructured materials, inorganic regions alternate with organic template regions at typical distances of several nanometres. To obtain a mesoporous solid with accessible channels and large surface areas, the organic template must be removed, often by combustion, without the inorganic framework collapsing. If this procedure works, it results in a structure that, although not necessarily crystalline on the atomic scale (mesoporous silica, for example, has amorphous walls), often shows regular order on the scale of the templated pore size.
The successful synthesis of mesostructured germanium must address several challenges. Chief among these is finding a soluble germanium precursor that self-assembles in the presence of an amphiphilic surfactant and that can be linked into a continuous framework. Self-assembly is itself promoted by the choice of a solvent that forms a hydrogen-bonded network but does not react with the precur-sor. The current papers2,3 use precursors that, although different, are both related to the compounds formed when highly electropositive metals combine with elements from the middle of the periodic table9. These compounds were pioneered by Eduard Zintl in the 1920s and 1930s. That of Armatas and Kanatzidis2 has the chemical formula Mg2Ge and contains Ge4− units in a crystalline lattice, whereas that of Tolbert and colleagues3, K2Ge9, consists of more complex (Ge92−)n polymer chains derived from Ge94− cluster ions.
To form an ordered, mesostructured germanium, Armatas and Kanatzidis add GeCl4 to their reaction mixture to provide Ge4+ bridges that connect the Ge4− units of the precursor. Tolbert and colleagues, on the other hand, link up their germanium chains by mild oxidation, taking care to avoid germanium oxide phases that would result from over-oxidation. The end result is, in the first case, cubic channel structures2 and, in the second, two-dimensional, hexagonal channel structures3.
As in the case of mesoporous silica, the walls of both materials are amorphous on an atomic scale, but periodic on the mesoscale of the frameworks themselves. As anticipated, the materials exhibit size-dependent effects that result from so-called quantum confinement, a phenomenon that affects the allowed energy values of the material's electrons at smaller scales. In particular, compared with spectra in bulk germanium, substantial shifts towards shorter, ‘bluer’ wavelengths are observed in the optical absorption or luminescence spectra of 1-nm-thick germanium walls.
A final challenge was the removal of the surfactant without destroying the mesostructure or converting germanium to its oxide. In the system derived from Mg2Ge, this proved to be very difficult: Armatas and Kanatzidis state that “attempts to remove the surfactant from mesostructured germanium by thermal decomposition led to contraction of the inorganic framework with consequent loss of pore ordering”2. Interesting properties are, however, observed in this system even without surfactant removal. Its optical bandgap — a crucial parameter in determining the conduction properties of a material — could, for instance, be smoothly varied through controlled partial oxidation and the generation of sub-oxide GeOx species in the framework.
In the (Ge92−)-based system, mild oxidation followed by an ion-exchange step generates cross-linkages in the (Ge92−)n that allow surfactant removal to take place. This results in a large surface area of 500 m2 per gram of material, and the opened structure permits incorporation of guest molecules that shift the energies of electrons at the semiconductor's surface and so can modify its conductivity.
These papers2,3 both represent significant advances in the structuring of germanium frameworks. In a sense, the papers are complementary: they show the versatility of Zintl-compound chemistry in combination with surfactant templating and provide different avenues towards designed, ordered semiconductor geometries. Tolbert and colleagues3 illustrate a way to produce true mesoporosity; Armatas and Kanatzidis2 show that a cubic structure with continuous walls and continuous channels can be achieved, which is often a more desirable architecture to facilitate the access of guests to the whole pore system.
This work will no doubt spawn new mesoporous semiconductor systems of relevance in sensing, detection and communications. Opportunities exist for tuning the properties of these systems with mixed compositions, including the mixed Ge/Si systems3 and Zintl clusters involving other group-14 elements, or by using other linking molecules to con-nect Zintl compounds using Armatas and Kanatzidis's method. With the precedents of structural and compositional variety set by mesoporous oxides, the stage is set for mesoporous semiconductors with properties that can be engineered.