Biotemplated syntheses of anisotropic nanoparticles

1. Introduction

Here, in the ‘fourth age of sand’ (Adams 1998), we find ourselves potentially teetering on the edge of a technological precipice. With the realization in 2008 of transistor densities of almost 2 billion per microchip, the most pessimistic observers have concluded that we are only between 10 and 20 years away from the breakdown of Moore's Law (Avila et al. 2007; Harville & Yashchin 2007; Jenkins 2007; Karlgaard 2007; Sheppard 2008). Gloomy predictions that the failure of this law will result in global economic and scientific stagnation are perhaps a little overblown, yet the prospect of no longer being able to keep our collective foot on the technological accelerator is enough to stimulate a great deal of discussion devoted to finding an answer to the impending breakdown of this law (Soref 2005; Close & Wong 2007; Jurvetson 2007; Renkema 2007; Wang et al. 2007). One consequence of this in particular and the burgeoning interest in nanomaterials in general is a drive towards the synthesis of materials with anisotropic morphologies, in order that they may be directly incorporated into devices. The key feature of the nanowire morphology is the confinement of properties such as electron conduction in all dimensions bar one. This allows the nanowire to be used as a direct interconnect between active components in a device (Wang 2003; Shankar & Raychaudhuri 2005; Parthangal et al. 2006; Yoon et al. 2007), with conduction constrained along the interconnect axis. The range of applications of nanowires is growing rapidly as a result, largely due to this ballistic electron transport. The in-built inhibition of electron conduction along non-long axis directions provides an important self-governance to the material, as the addition of ameliorating functions to direct the performance of nanomaterials can be extremely difficult. An excellent review of nanowire fabrication (Shankar & Raychaudhuri 2005) describes in great detail many of the materials that have been synthesized and used as nanowires. Broadly, these materials have been chosen as they exhibit metallic or semi-conducting properties, either for use as simple conductors (Wharam et al. 1988) or for more complex emergent behaviour such as UV lasing (Huang et al. 2001) or photoluminescence (Gudiksen et al. 2002; Chen et al. 2008). Despite concerns over toxicity inherent in nanoscale materials, nanowires of magnetic oxides are even being used in a biomedical role as drug delivery facilitators and targeted ameliorative devices (Bauer et al. 2004). Application of a magnetic field and/or the loading of the desired location in vivo with magnetic particles effectively directs the migration of the anisotropic magnetic nanoparticles to the appropriate location.

With an ever broadening range of uses, the controlled synthesis of anisotropic materials has never been more critical. As Shankar noted, however, there are a number of significant ‘bottlenecks’ and challenges involved in the controlled synthesis of nanowires, due largely to two key factors that work against the creation of anisotropic single crystals: the difficulty of inducing and subsequently retaining anisotropy and also the inhibition of competing side reactions during the synthesis, which will prevent nanowires of the appropriate stoichiometry from forming. The synthetic protocol must therefore be carefully formulated with these two factors in mind. In practice, there are a number of synthetic stratagems available for the production of nanowires. There is a wealth of literature on this subject (e.g. Dong & Li 2002; Shankar & Raychaudhuri 2005; Wang & Li 2006; Wang et al. 2008), which covers these methods in great detail, so it is not the intention of this review to go over this again in great depth. A brief discussion of these strategies is pertinent, however, as many biomediated routes still rely on existing models of anisotropic crystal growth. The syntheses can be grouped into a few well-defined categories, all designed with the purpose of inducing symmetry breaking in the growth of the inorganic phase. First, growth of nanowires can be achieved in some systems by chemical vapour deposition onto a suitable substrate. Nanowires of many materials including ZnS, CdS, ZnSe, Si and Ge (Wei et al. 2008) have been made by this method; although with limited control over side reactions, this method is not suitable for many functional materials. Second, growth of nanowires can be effected by the simple expedient of conducting the vapour–liquid–solid, solution–liquid–solid, vapour–solid growth (Wagner & Ellis 1964; Kolasinski 2006) within the pores of a membrane or substrate (Knaack et al. 2004; Zhao et al. 2005). With this simple physical constraint, nanowires of inorganic materials, particularly oxides such as Ga₂O₃, In₂O₃, TiO₂, V₂O₅, WO₃, ZnO, ZrO₂ and silicon, are easily attainable (Ozin 2005). The use of a ‘heterogeneous’ template (i.e. one which acts as a physical boundary to the reaction environment) has its limitations too though, in that, with little or no active control by the template over the crystallochemical environment within the pore structures, the synthesis of materials with more complex stoichiometries can be extremely difficult. Lastly, nanowire growth can be obtained by the use of a template that forms a homogeneous mixture with the precursor material. It is this last category that has the potential to extend the range of materials synthesized in anisotropic morphology the furthest, as the template can play a more ‘active’ role in limiting possible side reactions during the synthesis. A more intimate interaction between the precursor materials and the templating phase ensures a localized homogeneous reaction environment, preventing the contamination of the nanoparticulate proto-product.

Biotemplated and biomediated control of crystallization is a relatively recent area of investigation, yet one which has already had far-reaching implications for the way in which a complex form is generated in inorganic systems. Experiments in ‘biomimetics’ (Walsh & Mann 1995; Mann 2000) yield complex and often strikingly ‘life-like’ inorganic materials, by following closely (but not exactly) the protocols used in the natural analogue. The fact that biomineralization and therefore biotemplating succeeds so spectacularly is largely due to the complementary interaction between oppositely charged entities. Formation of nuclei in the first instance is a critical determining factor in the growth of an inorganic material, a process which can be slow and lead naturally to ‘classical’ crystal morphologies (cubes, rhombs, prisms, etc.). Crystallization in the presence of organic matter serves as a ready-made charge complementary substrate for the formation of nuclei, thereby allowing crystallization to occur more favourably and in greater number. This naturally leads to smaller crystallites that are ideal for the precise mineralization of bio-organic matter.

In the biotemplated synthesis of anisotropic nanoparticles, a further layer of complexity exists, in that, once anisotropic growth is underway, it is essential to consider how to retain the morphology in the final product. One study used cellulose to form core–shell and hollow nanoparticles of titania in a layer-by-layer method using partially digested cellulose as a template (Nelson & Deng 2006). The aim of the work was to directly replicate the underlying structure of the cellulose and produce highly anisotropic nanoparticles of TiO₂. In order to achieve this, the cellulose was subjected to a sulphuric acid digestion to break down the macroscopic fibres and leave nanofibres. In the layer-by-layer method, poly(diallyldimethylammonium chloride) (PDADMAC) was added to a dilute solution, which contained digested nanowhiskers, followed by titanium (IV) bis(ammonium lactato)dihydroxide (TALH). After calcination, hollow titania nanoparticles were formed (figure 1).

• Download figure
• Open in new tab
• Download PowerPoint

It is obvious from figure 1, however, that the original intention of faithful replication of the template was not achieved. They found that, for high aspect ratio sub-micrometre templates, the eventual nanoparticle shape grew further away from the shape of the template with the deposition of each layer of the inorganic material. This had the effect of altering the aspect ratio of the coated particles dramatically from that expected with an even coating, as the number of bilayers increased (figure 2).

• Download figure
• Open in new tab
• Download PowerPoint

From these data, they postulated that the polyelectrolytes displayed preferential adsorption on the tips of the whiskers. This led to the coating layer becoming disproportionately thicker on the sides than on the tips, with the consequent loss of anisotropy of the particles. When undergoing growth, all particles will try to minimize their surface free energy by reducing their aspect ratio. This means that, for anything other than the thinnest of coatings, preservation of the template shape will not be possible unless the template is spherical. This is a key finding and one that has repercussions for all potential biotemplates. It would appear that, when looking to retain anisotropy in the final material, replication of a biotemplate with the underlying fibrous morphology must be at length scales larger than the nanoscopic if thicker, more robust coatings are desired.

Notwithstanding these difficulties in synthesis, many classes of biopolymers have been used as morphological and/or functional chemical templates in the creation of anisotropic inorganic materials, and it is the intention of this review to highlight the work that has been done to date and hopefully signpost potential areas of future work.

2. Biotemplates

(a) Saccharides

Simple monosaccharides have been used in biotemplating, largely in their role as ready sources of carbon. One interesting development was the use of sucrose in the synthesis of silicon carbide materials as whiskers and nanotubes, via a synthetic route using mesoporous silica as a sacrificial solid template (Yang et al. 2004). The pore structure of a mesoporous silica was infiltrated by a sucrose solution and the whole was subjected to calcination. SiC materials were obtained via carbothermal reduction of the mesoporous silica as the sucrose carbonized. By varying the temperature and length of carbothermal reduction, it was found that the morphology of the silicon carbide could be modified. In addition to the bulk SiC material, a high proportion of nanotubes and whiskers were able to be obtained, with a high degree of crystallographic control (figure 3). It was determined that the optimum conditions for whiskers to grow in the (111) direction were when the calcination proceeded at high temperatures and extended times, typically in the range of 1250–1300°C for 14 hours.

• Download figure
• Open in new tab
• Download PowerPoint

Whisker diameters were found to be in the region of 50–90 nm, with lengths well over 20 μm. In terms of the surface area, it was found that SiC materials with whiskers had a surface area of 120–145 m² g⁻¹ with a pore volume of 0.42 cm³ g⁻¹. Increasing the carbothermal reduction period to 20 hours, at the same calcination temperatures, resulted in the formation of solid SiC nanotubes of diameter 60–100 nm and approximately 10 μm in length. The compaction of the material resulted in a higher surface area of up to 190 m² g⁻¹.

Selenium nanotubes were synthesized in a one-step hydrothermal method, using Na₂SeO₃ as the selenium source and glucose as the reducing agent (Chen & Gao 2006). Synthesis of selenium nanotubes usually produces polycrystalline or amorphous materials, but by using glucose in a hydrothermal synthesis at 180°C for 10 hours, reaction with Na₂SeO₃ produced a grey precipitate with a cotton-like fibrous appearance. Differences in the contrast of nanostructures on the transmission electron microscopy (TEM) images indicated that the materials were primarily hollow tubular structures with some solid nanorods present. The nanotubes were between 100 and 400 nm in diameter, with a 20 and 40 nm wall thickness. The length of the nanostructures was variable, with some nanotubes in the tens of micrometres in length. High-resolution TEM (HRTEM) images showed pronounced lattice fringes with a calculated interplanar spacing of 0.49 nm, close to the (001) lattice spacing of trigonal selenium (t-Se). The facile synthesis of t-Se using glucose is particularly useful as t-Se has been shown to be an important elemental semiconductor, owing to its high photoconductivity and large piezoelectric and thermoelectric effects. Mechanistically, it is likely that the glucose is acting as a weak reducing agent for Na₂SeO₃ under the hydrothermal conditions. This allows for a dissolution–reprecipitation leading to a spontaneous growth of nanostructures. With elongated reaction times, highly reactive Se reprecipitates and crystallizes on the surface of the t-Se nanoparticles leading to extended nanotube lengths.

(b) Cellulose

Cellulose is formed exclusively from 1→4-glycosidic-linked β-glucose subunits, which give rise to a straight-chained polymer. With extensive hydrogen bonding between the molecular chains, the formation of rod-like microfibrils imbues cellulose with considerable stability and strength. With the inherent structural complexity exhibited by cellulose, its use as a template is extensive. Many studies have made use of the fibrous nature of cellulose in order to impart a fibre-like morphology to the inorganic phase. For example, cellulose has been used to create calcium carbonate (CaCO₃) aggregates, which replicate the underlying cellulose template (Zheng et al. 2007). By sonicating cotton fibres while sequentially adding calcium and carbonate ions, they were able to mineralize the fibres with CaCO₃. Subsequent calcination of the composite material produces hollow fibres of CaCO₃ (figure 4). This method was able to be applied equally well to create CaCO₃ replicas of woven cellulose in the form of cotton cloth.

• Download figure
• Open in new tab
• Download PowerPoint

In a similar manner, cellulose has been used to form nanotubes of electrically conductive indium tin oxide (ITO) (Aoki et al. 2006). In this work, indium methoxyethoxide (In(OCH₂CH₂OMe)₃) and tetraisopropoxytin–isopropanol adduct (Sn(OⁱPr)₄.ⁱPrOH) were dissolved in methoxyethanol and isopropanol–methanol (1 : 1, v/v). Combining the two solutions gave the precursor material, which was then heated to 50°C. By passing this hot solution over a piece of filter paper in a vacuum filtration assembly, individual cellulose fibres in the filter paper were coated with the ITO precursor. After calcination at 450°C, the resultant ITO replica of the cellulose fibres was formed as a layer of 90–220 μm in thickness (figure 5).

• Download figure
• Open in new tab
• Download PowerPoint

The electrical conductivity of the templated fibres was confirmed, with the highest conductivity arising from an ITO sheet with an In/Sn ratio of 93.5/6.5 at 0.53 S cm⁻¹. This value is higher than that observed in similarly nanostructured ITO films prepared by other templated routes.

Members of this group have also looked at the coating of cellulose fibres in order to impart an anisotropic morphology to an organic material. In their work, oxidative polymerization of pyrrole monomer using copper chloride in 2-propanol was carried out in the presence of cellulose, again very simply by filtering the solution through a piece of commercially available filter paper (Huang et al. 2005). This method is of importance as, recently, there has been considerable interest in the production of conductive textiles (Swallow & Thompson 2001; Kushal & Shraddha 2007; Rattfalt et al. 2007). The successful coating of fibres with conductive polymers can be hard to achieve, however, as most polymers will tend to exhibit colloidal behaviour when interacting with substrates, leading to agglomeration and sedimentation. By finely controlling the deposition time, however, they found no such agglomeration when polymerizing pyrrole on cellulose fibres (figure 6).

• Download figure
• Open in new tab
• Download PowerPoint

From the SEM images, it is clear that the composite sheet is composed of continuous interconnected nanofibre networks that retain the original morphology of the cellulose fibres. On closer inspection, they found that the electron beam-induced rupture of the PPy layer revealed a cable-like core–shell structure. TEM imaging (figure 7) clearly showed a flat and homogeneous polypyrrole layer around each individual cellulose fibre.

• Download figure
• Open in new tab
• Download PowerPoint

By changing the deposition time, it was found that the thickness of the polypyrrole layer was able to be controlled quite precisely between 20 and 25 nm. Although no measure of the conductivities of these composites was made, this simple biotemplating technique remains an extremely efficient method of producing conjugated polymeric materials with highly anisotropic morphologies.

As indicated earlier, obtaining a perfect register between the biopolymer template and the mineral phase is not necessarily a simple matter. One possible solution to improving the homogeneity of cellulose coating is to functionalize the biomolecule in order to increase the specificity of interaction between the inorganic phase and the substrate. Cellulose is amenable to functionalization as the hydroxyl groups on the molecule can be reacted, giving a range of polymers with different properties. Esterification of cellulose is one simple option, which can create film-forming and fibre-forming derivatives such as nitrocellulose and cellulose acetate. Functionalization of cellulose in this way was employed in the synthesis of silica nanotubes (Zollfrank et al. 2007). This work was based on the observation that some synthetic and biological polymers that contain secondary and tertiary amino groups within the chain, or primary amino functions on the side groups, are capable of catalysing the formation of silica spheres with diameters ranging from 50 to 400 nm. The proposed mechanism involves the influence of N-methylpropylamino groups, which provide steric fixation and catalyse the condensation of silica precursors. The group therefore introduced the oligopropylamino side chain responsible for the control of biosilica deposition in diatoms into the cellulose biopolymer. They observe that the cellulose forms a rigid backbone owing to intramolecular hydrogen bridges between the secondary hydroxyl group at the C3 position of an anhydroglucose unit (AGU) and the oxygen of the pyrane ring in the next AGU. This means that the free rotation around the glycosidic bond is hindered, thereby ‘locking-in’ the anisotropic morphology. In addition, intermolecular network formation of the cellulose chains is suppressed owing to the functionalization with oligopropylamine. Because these oligopropylcellulose-functionalized molecules are soluble in water, the silica formation is able to be performed under mild conditions with water-soluble silica precursors, such as tetrakis(2-hydroxyethyl) orthosilicate (THEOS), or by using the more common tetraethyl orthosilicate (TEOS) in its pre-hydrolysed form. In the functionalization experiment, dipropylenetriamine (DPTA) was reacted with the cellulose tosylate. Silica nanotubes were then synthesized from DPTA cellulose solutions by incubating the solution with the silica precursors.

Calcination at 1000°C resulted in the removal of the cellulosic template leaving hollow silica nanotubes with diameters between 10 and 30 nm, inner core diameters of approximately 3 nm and lengths of up to 500 nm. The group postulated that, in addition to DPTA cellulose molecules acting as molecular templates for the formation of silica nanotubes, the oligopropylamine derivatives act as a catalyst in the polycondensation of the silica precursors during the sol–gel synthesis, ensuring a coherent coverage of the biopolymer.

(c) Dextran

Another biopolymer that is beginning to find use as a template is dextran (Walsh et al. 2003, 2004, 2007; Gonzalez-McQuire et al. 2005). Dextran is a complex, branched polysaccharide composed of a large number of glucose units in a wide range of lengths so that the molecular weight can vary between 1000 and several hundred thousand daltons.

The backbone of the dextran molecule consists of α1→6-glycosidic linkages, with branches forming primarily from α1→3 links. Depending on the source, the dextran can also be branched from α1→2 and α1→4 linkages. The structural complexity of dextran, in particular the multi-branched nature of the molecule, has made it an interesting molecule to use as a biotemplate. By calcining dextran with silver nitrate, it has been determined that the composite system could be used to produce tangles of silver nanowires (Kong et al. 2006). By holding the composite paste at 180°C for 12 hours, reduction by the aldehyde groups on the dextran produced discrete silver nanoparticles, the sintering of which was prevented by the slowly oxidizing biopolymer. They propose that selective binding of dextran to certain crystallographic faces of the silver nanoparticles, coupled with Ostwald ripening, produces outgrowth of some silver nanowires (figure 8).

• Download figure
• Open in new tab
• Download PowerPoint

SEM images show that the nanowires are highly uniform, with lengths between 20 and 80 μm and widths of approximately 270 nm. Electron diffraction confirmed that these nanowires were cubic single crystals of silver. The group noted that, when a higher concentration of dextran was used, more nanoparticles and fewer nanowires were produced. In these cases, there will be a greater number of Ag nuclei generated from the redox reaction, while the higher viscosity prevents nanowire outgrowth, largely as a result of the biotemplate ‘cocooning’ effect.

(d) Starch

Starch is a ubiquitous, complex glucose polymer, a mixture of amylose and amylopectin, in a 1 : 4 ratio. Starch possesses an inherent molecular anisotropy in that it is composed of long polysaccharide chains. This motif can be easily transferred to other materials introduced into starch solutions. Nanowires of polypyrrole are one such example (Shi et al. 2007). Polypyrrole is a simple polymer constructed from pyrrole heterocycles that shows particularly high conductivity (McNeill et al. 1963; Diaz et al. 1981). In the nanowire study, it was found that, in a starch solution, the addition of pyrrole monomers resulted in the adsorption of the pyrrole onto the starch via hydrogen bonding. The adsorbed monomers were then able to be polymerized, the polymerization following the chains of starch molecules to form PPy nanowires. The group found that, in addition, the PPy nanowires could be electrochemically generated on various electrodes including tin-doped indium oxide (ITO) film, stainless steel, titanium, gold and graphite using starch as a template for growth. SEM and TEM images of the PPy materials showed a uniform wire-like PPy nanostructure with an average diameter of approximately 100 nm, with lengths varying from hundreds of nanometres to several micrometres. Formation of PPy in the absence of starch resulted in the uncontrolled polymerization and growth of PPy aggregates rather than nanowires, either with or without a conductive substrate.

Another conductive material, tellurium, has been made in the nanowire form by the use of a starch template. Tellurium is useful as a semiconductor, with a narrow band gap of approximately 0.35 eV and a pronounced piezoelectric effect. Tellurium nanowires have been synthesized previously, using hydrazine as a reducing agent and a combination of surfactants that rely on or produce harmful reagents and by-products (Mayers & Xia 2002; Mo et al. 2002). It was found that, by using starch as a reducing sugar, the compound H₂TeO₄.2H₂O could be reduced under hydrothermal conditions to form tellurium nanowires in a high yield (Lu et al. 2005). As well as being more environmentally friendly, the starch acted to sequester the tellurium salt, thereby ensuring a homogeneous distribution of the desired phase throughout the composite precursor material. A typical synthesis consisted of a simple admixture of H₂TeO₄.2H₂O in a starch solution, followed by 15 hours in an autoclave at 160°C. TEM images show the presence of long nanowires of widths approximately 25 nm and lengths of several tens of micrometres (figure 9).

• Download figure
• Open in new tab
• Download PowerPoint

The preferred direction of growth of the nanowires was determined from HRTEM as being the [001] direction. The mechanism of formation is likely to be the direct replication of the underlying anisotropic molecular chain motif of the starch, as seen in the PPy study.

(e) Alginate

Alginate is a polysaccharide derived largely from the family of brown seaweeds, which includes the kelp Laminaria. Chemically, the polymer is an unbranched, anionic polysaccharide, consisting of blocks of polyguluronate (–G–)_n, polymannuronate (–M–)_n and alternating (–G–M–)_n residues, the proportions of which depend on the species of seaweed from which the alginate is extracted. This seemingly simple structure gives rise to a morphological complexity, which perfectly marries form to function. Within a particular species of seaweed, the amount and position of the poly-G and poly-M blocks can vary, depending on the desired properties that the alginate must confer to the organism. As it is the poly-guluronate blocks that strongly sequester divalent cations such as Ba²⁺ and Ca²⁺, regions of the seaweed that need to be resistant to tidal forces, for example the holdfast and stalk, will be rich in poly-G residues, whereas the fronds that require flexibility to move with the tide will be richer in poly-M residues. The ability of poly-G blocks to bind cations strongly is a feature of the molecular shape conferred on these blocks by the chemical nature of the residue linkages. When two poly-G regions on separate alginate molecules align, a pocket is created in which mechanically strengthening cationic cross-linking can occur readily. As the poly-G regions are regularly spaced in any given alginate, this leads to a regular array of cross-linking junction zones, dubbed the ‘egg-box’ model of ion binding (Grant et al. 1973). It should be immediately obvious that this useful feature can be exploited for the controlled growth of other cationic species.

By introducing metal cations to a solution of alginate, there will be preferential uptake of the metallic species into the poly-G regions, leading to controlled nucleation and growth of nanoparticles when subsequent processing steps are taken. This stratagem was adopted in the synthesis of single-crystal nanowires of an yttrium–barium–copper-oxide (YBCO) superconductor (Schnepp et al. 2008). Morphological control of superconductor growth is of prime importance as it has previously been shown that a significant reduction in superconducting properties occurs with an increase in either the polycrystallinity of the sample or in the non-alignment of grain boundaries (Klie et al. 2005). Traditionally produced materials consist of large and irregularly shaped crystals, and therefore the overall performance of the superconductor is limited. Attempts had been made to synthesize YBCO as nanowires, but complexities associated with control over stoichiometry, avoidance of phase separation and suitable growth catalysts led many to believe that, for such a complex material, nanowire growth was simply impossible (Ozin 2005). In the work by Schnepp et al., use was made of the alginate egg-box binding model to produce ordered arrays of barium cations, which were able to subsequently act as preferential sites of nucleation for barium carbonate nanoparticles. The cocooning effect of the alginate biopolymer prevented the coalescence of the barium nanoparticles, leading to uniform, homogeneously dispersed nanoparticles throughout the matrix. On calcination, once the correct temperature was reached, growth of superconductor needles occurred radially from partially embedded barium carbonate nanoparticles exposed on the surface of the amorphous material. SEM revealed that the material comprised a complex microstructure of fibrous aggregates, up to several micrometres in thickness (figure 10a). These fibres had smooth surfaces and a circular cross section of 50–80 nm in diameter (figure 10b). On further investigation, TEM imaging of fractured fibres indicated that the fibres themselves comprised bundles of straight-sided nanowires, typically 10 nm in diameter (figure 10c). Electron diffraction showed that these fibres were uniaxially aligned, with a preferred direction of growth (figure 10d). Superconducting quantum interference device (SQUID) magnetometry showed that these nanowires had a critical temperature of 77 K, which is only slightly lower than the 85–89 K typical for Y124.

• Download figure
• Open in new tab
• Download PowerPoint

(f) Schizophyllan

Schizophyllan (SPG) is a natural polysaccharide present in the fungus Schizophyllum commune. Molecularly, SPG can be considered to consist of three β-(1-3) glucoses and one β-(1-6) glucose side chain linked at every third main-chain glucose. Chains of SPG are therefore able to self-interact producing a hydrophobic cavity. This useful morphological feature has been used to create a silica material that comprised nanofibres (Numata et al. 2005). In this work, they used the molecular structure as a direct morphological template in which to grow fibres of silica, making use of the hydrophobic cavity of the one-dimensionally aligned hydrocolloid as a constrained reaction volume. The group were interested in this approach as they had previously used fibres of an organo-gel to produce fibrous silica, although they had experienced difficulties in removing the organic template without subsequent loss of inorganic structural integrity. Calcination to remove the organic part of the system would invariably result in structural collapse as outgassing occurred. By synthesizing the silica inside the organic template, calcination would leave the inorganic part intact. The synthesis consisted simply of a solution of SPG in dimethylsulphoxide (DMSO) that was mixed with TEOS and a large excess of water to give an H₂O/DMSO ratio of 95/5. After ageing for a week at room temperature, the group dialysed the system, which they found induced precipitation of the silica. TEM results were striking in that the SPG produced highly anisotropic fibres of silica of approximately 15 nm in diameter.

(g) Chitosan

Chitosan (2-amino-2-deoxy-β(1→4)-d-glucan) is formed by the deacetylation of chitin, a naturally occurring polysaccharide found in the cell walls of fungi and the exoskeletons of insects and crustaceans. It is an ideal biopolymer for biotemplating research, as it has the ability to preferentially sequester transition and post-transition metal ions from aqueous solutions (Ogawa et al. 1993) and is able to withstand temperatures of up to 200°C without undergoing any significant alteration of molecular configuration (Bengisu & Yilmaz 2002). This stability arises from the formation of chain-like, helical bundles on crystallization. Charge stabilization of the inorganic–biopolymer composite structure is achieved by the positively charged amino groups on the chitosan associating with anionic species in the synthesis.

Wang et al. looked at the effect of complexing the chitosan with silk fibroin prior to the incorporation of hydroxyapatite (HAp), in an effort to improve the crystallinity of the composite material (Wang & Li 2007). Silk fibroin was used as it is a simple protein (17 amino acid residues) that is readily available from silk moth cocoons and has the ability to cross-link with chitosan, forming porous networks in three dimensions. In this way, the crystallization volume was altered, with the intention of altering the morphology of HAp. The synthesis of chitosan/silk fibroin/HAp materials was achieved by adding an acidic solution of chitosan containing H₃PO₄ to a suspension of silk fibroin in Ca(OH)₂. The pH of the synthesis mixture initially was then adjusted to 9.0 by the addition of ammonium hydroxide, in order to ensure that the HAp phase of calcium phosphate was formed preferentially. This was confirmed by both XRD and FR-IR, which showed that a monophasic HAp material was produced. TEM images of the crystals showed that, for both control and chitosan/silk fibroin/HAp materials, there was no change in the morphology, both sets having the characteristic HAp needle-like morphology. The group found that the crystals were between 20 and 50 nm in length and 10 nm in width. Although no morphological change was observed, the biotemplating appeared to induce a degree of aggregation of the needles, forming discrete bundles.

In a similar manner to the work with alginate, nanowires of high-temperature superconductors have also been synthesized using chitosan. It was shown that it was possible to control the crystal morphology of the high-temperature superconductor YBa₂Cu₄O₈ (Y124) using the ability of chitosan to sequester metal ions to provide sites of preferred nucleation and growth and to prevent the sintering of particles during calcination (Hall 2006; figure 11).

• Download figure
• Open in new tab
• Download PowerPoint

SQUID magnetometry showed the typical temperature-dependent magnetic behaviour that was characteristic of the Y124 phase, with a T_c of 89 K. The morphology of Y124 crystals prepared from chitosan gels was dependent on the heating rate used in the calcination stage. Nanowires formed specifically at a heating rate of 1°C min⁻¹, due to outgrowth of the embedded nanoparticles forming nanowires from 350°C onwards. This temperature corresponds to the formation of metal oxide species, and the results suggest that sintering of the nucleated nanoparticles is prevented by the extended presence of the carbonized, amorphous biopolymer matrix to producing small, nanoparticulate centres for subsequent Y124 outgrowth.

(h) Collagen and silk

We have already seen how polysaccharides are able through the formation of glycosidic links to produce morphological complexity in both themselves and the inorganic materials that they template. Variations in the monosaccharide building blocks and the way these are conjoined through the glycosidic bond provide a rich variation on a structural theme, which organisms (including the materials chemist) can use as templates in the creation of complex inorganic phases. The other main structural macromolecular assemblies found in nature are the proteins. As the glycosidic bond is to polysaccharides, so the peptide bond is to proteins. In nature, proteins with complex form involved in the creation and maintenance of structural features such as cell walls and ion transport are classified broadly into two types according to their quaternary structure. These proteins tend to form either fibrous or globular morphologies and thereby take quite distinct roles in vivo.

Fibrous proteins are found only in the animal kingdom and can be characterized by the relatively restricted palette of amino acids that form the protein. They tend to feature large, hydrophobic R side groups, which induce self-aggregation and impart a high degree of water insolubility to the protein, making it particularly useful as an invariant structural component in the aqueous environment of animal cells. Forming rod or wire-like morphologies, the fibrous proteins constitute the connective tissues in animals, such as collagen, keratin and elastin. With a defined morphology hard-wired into the fibrous proteins and a physico-chemical preference for hydrophobic environments, it is not surprising that fibrous proteins have been used as biotemplates for the mineralization of anisotropic inorganic phases. One of the most widely investigated, owing to its critical role in bone formation in the animal kingdom, is collagen. The mineralization of bone is directed largely by the layout of the fibrils of collagen, hydroxyapatite crystallizing along these structures, producing uniaxially aligned crystals that are oriented along the fibrils.

The affinity of collagen for inorganic materials has been exploited in the templated growth of silica (Eglin et al. 2005). The group observed by birefringence studies that a solution of collagen in acetic acid naturally formed chiral nematic liquid crystals, comprising the collagen helical structures. Increasing the pH by the addition of ammonia induced the helices to self-assemble into a gel consisting of fibrillar structures. The group then exposed these gels to the vapour of a volatile alkoxysilane in order that hydrolysis and condensation of silica would occur upon and within the fibrillar gel matrix. They found that this was simply achieved by the exposure of open vials of gels to the vapours of tetramethoxysilane in a closed system for 6 days. Confirmation of the retention of the collagen liquid crystalline structure by polarized light microscopy showed that the birefringence and the long-range order was maintained in the silica–collagen hybrid material once condensation was complete. Analysis of the composite gels by TEM revealed a high degree of replication of the collagen structure, with silica forming along the collagen fibrils.

As a final step, the group was able to calcine the hybrid materials and confirm retention of the fibrillar structure in the silica phase (figure 12). The composite materials were first calcined at 120°C for 2 days in order to increase the densification of the silica, followed by heating at 600°C for 2 hours to remove the collagen template. Once calcined, BET of the silica material revealed that the average pore size was 2.2±0.2 nm, which was slightly larger than the typical collagen helix diameter. The group also investigated the removal of the collagen template by more benign methods: digestion of the collagen by either 1 M HCl or the enzyme collagenase, both of which produced similar results to calcination. The group concluded that the fine pore size in the final material is suggestive of the silica condensing around single helices of collagen.

• Download figure
• Open in new tab
• Download PowerPoint

Similarly, formation of silica fibres around collagen was demonstrated, although without the formation of a liquid crystal gel by raising pH (Ono et al. 1999). In this method, a buffered solution of collagen was left to spontaneously form fibres over a period of one month. In doing so, large fibrous structures were formed, although no description was given as to any spontaneous ordering in solution. As with the work by Eglin et al., the collagen fibres were then exposed to a volatile siloxane; in this case by mixing the collagen solution with a solution of TEOS. Hydrolysis and condensation of the silica was then able to occur during quiescent ageing of the mixed solution, in order that mechanical disruption of the fibre structures was minimized. Once the silica condensation was complete, the group analysed the results by TEM. This revealed that individual fibres had been coated evenly with silica, producing 50–100 nm thick structures that contained a central core of collagen. Interestingly, the fibres had a serrated edge, with a period of 60–80 nm. This corresponds well to the groove in the collagen fibre coils, which was calculated to be 67 nm. Calcination of the composite material resulted in structurally stable, hollow silica tubes without much structural collapse.

Aside from collagen, the other fibrous protein most intensively investigated as a biotemplate is silk. Used by man since time immemorial, it is a particularly useful material as it is incredibly strong for its size, with some spider dragline silks reportedly having a higher tensile strength than a commensurate steel wire (Kenney et al. 2002). It is composed of a small number of amino acids, among which glycine constitutes almost 50 per cent of the protein; the small size of the glycine residue leading in part to the efficient packing and thereby enhanced structural stability of the silk protein fibre through the formation of extensive hydrogen bonding. Mao et al. investigated the use of spider dragline silk in the growth of anisotropic crystals of HAp with preferred crystallographic orientation (Cao & Mao 2007). It was found that, on incubating strands of silk in a supersaturated HAp solution, elongated crystals grew along the silk fibres. Electron diffraction of these crystals showed that all of the crystals were HAp and that the (002) reflection, which the group normally observed (in control experiments) to form a ring, indicative of a random polycrystalline arrangement, was instead in the form of a very short arc, indicating that the HAp crystals were nucleated with preferred crystallographic orientation. Analysis of the disposition of these crystals on the spider silk showed that the crystals were oriented with their c-axes preferentially oriented at an angle of 72.9° with respect to the long axis of the silks. The group rationalized the mechanism of preferred orientation as being due to crystal plane matching, stereochemistry matching and the electrostatic interaction between the silk and the HAp phase. Experiments on the crystallization of other inorganic phases on the silk fibres, namely Au and CdS, showed an affinity for nucleation, but with no preferred orientation on growth. Invoking an epitaxial matching mechanism, the group rationalized the preferred orientation of HAp crystals as being due to the close match between the a–b plane of HAp and the b–c plane of silk protein nanocrystals (figure 13).

• Download figure
• Open in new tab
• Download PowerPoint

The group found that the a–b plane of HAp and the b–c plane of the silk are almost structurally identical, with only a small degree of mismatch of approximately 7.5 per cent. Electrostatic attraction of Ca²⁺ ions to the oriented silk β-sheets results in strong binding and the creation of sites of preferred nucleation and subsequent growth of HAp. Once Ca²⁺ is bound to the silk surface, ions will be electrostatically attracted to it, leading to the eventual formation of oriented HAp. The group finally observed that, as spider silk is totally biocompatible, the opportunity exists for their composite material to act as a bone implant and tissue engineering scaffold.

Li et al. (2005) examined the effect of electrospinning in the controlled mineralization of HAp using silk. In this study, the group grew HAp on fibres spun from a mixture of silkworm (Bombyx mori) silk and polyethylene oxide (PEO). The PEO solution contained poly(l-aspartate (poly-Asp)), which the group introduced as an analogue to the associate proteins found in natural bone. The solution of silk/PEO/poly-Asp was subjected to electrospinning at an applied voltage of 12 kV to generate an electric field strength of 0.6 kV cm⁻¹. Once spun, the fibres were immersed in 100 per cent methanol in order to induce transformation to the fibrous β-sheet structure. As before, a simple soaking of the silk in a solution of buffered CaCl₂ (200 mM), followed by transfer to a Na₂HPO₄ solution (120 mM), produced pronounced mineralization of the fibres. At low poly-Asp concentrations, the group found that HAp mineralization occurred, albeit with a random orientation of the crystallites. Once the poly-Asp concentration rose above 200 mg g⁻¹ of silk, however, oriented growth along the fibres of the silk template was observed.

In addition to its role as a director of crystallization, the fibrous nature of silk has been exploited as a sacrificial morphological template. Volatile butoxides of titanium and zirconium have been used in order to selectively coat fibres of silk (He & Kunitake 2004). After soaking silk fibres in a solution of the butoxide, the fibres were removed and the volatile phase was allowed to hydrolyse and condense in the moisture of the air. This cycle of soaking and exposure to air was repeated 10 times. Calcination of the composite material at 450°C for 4 hours resulted in the removal of the silk template and a consolidation of the titania phase. SEM investigations of the titania revealed a porous structure of filaments approximately 5 μm wide. This compares with the original width of the silk fibre of 9 μm, indicating an appreciable degree of shrinkage on mineralization. The titania coating was not even though, giving rise to a hierarchical pore structure comprising large pores of approximately 100 nm in diameter and smaller pores of approximately 10 nm in diameter. The group noted that the porous structure of the silk may be aiding the production of porosity in the titania in addition to the intra-crystallite pores formed by titania crystals growing apart during the condensation step. Zirconia-templated silk fibres resembled the titania fibres, albeit with a poorly crystalline coating as zirconia is amorphous at the calcination temperatures used in their study. For both titania and zirconia, the group demonstrated their use as nanoreactors for the controlled growth of gold nanoparticles, producing monodisperse nanoparticles of gold which were sub-10 nm in diameter.

(i) Lipids

Lipids are long-chain molecules characterized by their ability to be solubilized in fats (triglycerides) and include a wide range of molecules such as sterols, vitamins A, D and E and phospholipids. Owing to their wide compositional variations, the lipids have found many diverse functions within organisms, most notably as energy storage media and in the maintenance of structural stability in cell membranes. It is in this latter role that the lipids have inspired their use as complex biotemplates for mineralization. It is the phospholipids that are of particular interest, as their molecular structure contains both hydrophilic polar and hydrophobic non-polar regions, thereby allowing the molecule to exhibit surfactant-like behaviour.

One of the most common morphologies adopted by lipids is the lipid tubule, as the amphiphilic behaviour of lipid bilayers leads naturally to the formation of curved surfaces in order to minimize unfavourable lipid–solvent interactions. By curling up in this manner, the lipid tubule is a ready-made hollow cylindrical template of nanoscale dimensions. Many common inorganic phases have been templated by lipid tubules, the first of which was silica, reported in the early 1950s (Nemetschek & Hofmann 1953). Subsequently, researchers have varied the dimensions of the tubes by assembling multi-layered lipid structures, or by subtle alteration of the synthetic protocol in order to produce a wide range of products (Archibald & Mann 1993; Patil et al. 2003). A good example of the most usual synthetic approach is given by Ji et al. (2007). In their work, a lipid was synthesized to consist of tri-proline and glutamic acid dialkyl amide groups. When added to water at a concentration of 5 mg ml⁻¹, and subjected to ultrasonication, the lipid formed tubules. The addition of TEOS to the lipid solution at a concentration of 50 mg ml⁻¹ resulted in the formation of a gel after a week of quiescent storage (Ji et al. 2007). TEM of the tubules (stained with phosphotungstate) prior to TEOS addition revealed that they were several micrometres in length and approximately 200 nm wide. After gelation, TEM images of the composite material revealed that the gel consisted of high yields of hollow nanotubes with a very tight size distribution. The nanotubes were all between 5 and 10 μm in length, with a 200 nm diameter and 20 nm wall thickness, indicating that the silica produced a particularly fine replication of the lipid. At this point, calcination of the composite material to remove the lipid would result in a marked structural collapse of the silica replica. In their work, however, the group lyophilized the aqueous gel under vacuum (1 Pa) for 48 hours prior to calcination. By forming a xerogel in this manner, the structure of the lipid could be preserved, as the destruction of the morphology by outgassing during calcination is minimized. After calcination, the wall thickness of the nanotubes was determined to be approximately 8 nm, indicating a degree of shrinkage, although, overall, the anisotropic morphology was retained.

A more complex form of siliceous material was investigated by Patil et al. (2003) in the formation of an organoclay around a chiral lipid template. Organoclays are of particular interest, as their enhanced functionality allows for a greater range of applications (e.g. sequestration and drug release) than for simpler, non-functionalized clays alone. The group used a chiral phospholipid, 1,2-bis(10,12-tricosadiyonyl)-sn-glycero-3-phosphatidylcholine (DC_8,9PC), as the templating phase for the creation of complex organoclay structures, owing to the ability of DC_8,9PC to form helical ribbons and fibres by self-assembly. A dispersion of the self-assembled lipid structures in an ethanol–water solution was able to be mineralized by the simple addition of 3-(2-aminoethyl-3-aminopropyl)trimethoxysilane (EDTMS) and magnesium chloride, producing high-fidelity replicas of the lipid consisting of Mg-phyllo(organo)silicate clay materials. The high quality of the replication was confirmed by TEM, which showed helical ribbons or tubules of an electron-dense material of 0.55–0.8 μm in diameter, with lengths greater than several micrometres. XRD showed that the inorganic phase was indeed the desired phyllo(organo)silicate clay and that also peaks due to the lipid lamellar mesophase could be seen, confirming the retention of the complex-ordered structure in the replica material. From a close examination of the TEM images, the inner and outer walls of the tubules were able to be identified, now decorated with the organoclay. This revealed that the wall thickness of the material was 35 nm, which corresponded to a six-lipid-bilayer-wide assembly. When the group increased the immersion time of the lipid in EDTMS, thicker structures up to 60 nm were observed. This suggested that clay clusters were interacting with the zwitterionic phosphocholine headgroups of the lipid on both the inner and outer walls of the tubules. Changing the order of the reactant addition produced similar assemblages of mineralized lipid tubules, which indicated a general electrostatic mechanism of mineralization, rather than a more active role of EDTMS as a facilitator of nucleation and growth of tubule assemblies.

In a similar manner, DC_8,9PC was used as a lipid template on which to grow aluminium phases (Chappell & Yager 1992). Simple addition of aluminium chloride to an ethanolic solution of DC_8,9PC produced an even coating of an aluminium-containing phase. The group was unable to obtain an electron diffraction pattern for this material, suggesting that it was a poorly or non-crystalline phase of an aluminium hydroxide-like material. However, they found that these aluminium-coated tubules were an eminently suitable substrate on which to subsequently grow crystalline aluminium phases. Redispersion of the aluminium-coated tubules in ethanol, followed by the addition of sodium bicarbonate, produced in some instances a crystalline solid, which replicated the tubule structure completely. Again, the group found that electron diffraction and this time XRD produced inconclusive results, although a tentative assignment of the crystalline phase as a sodium hydroxyaluminium carbonate was made.

The same DC_8,9PC lipid was used in the creation of copper nanospirals (Price et al. 2003). In their work, use was made of the seams created by ribbon formation on the lipid, rather than the body of the lipid itself.

It had been noted in previous studies on the interaction between DC_8,9PC lipids and nanoparticulate metals that the colloidal particles would adsorb preferentially on the seams of the tubule. As Price et al. (2003) found, however, as the interaction between the lipid and the metal occurs through van der Waals and hydrogen bonding, there is invariably enough of the metal adsorbing to the lipid walls to result in complete coverage of the lipid. The group was able to restrict metallization to the lipid seams by the introduction of a polyelectrolyte that enhanced the specificity of the metal templating phase for the seam areas. By sequentially adsorbing oppositely charged polyelectrolyte layers of sodium poly(styrenesulphonate) (PSS) and poly(ethyleneimine) (PEI) onto the lipid tubule seams, the binding of Pd(II) nanoparticles was able to be achieved preferentially at these sites owing to the strong electrostatic affinity of the negatively charged Pd(II) nanoparticles for the terminal positively charged PEI layer of the multilayer polyelectrolyte/lipid assembly. Areas of Pd were then used as catalytic sites for the preferential deposition of copper onto the composite assembly. SEM and TEM images of the copper-coated composite showed that after prolonged (2 hours) exposure to the copper deposition, some spiral structures are formed, which mimic the underlying spiral structure of the lipid seams. The success of replication of the lipid structure was determined by the measurement of the pitch angle of the copper nanospirals. The group found that a pitch angle of 48°±9° in the copper nanostructures correlated well with that of the starting lipid (42°±6°).

Anisotropic assemblages of more functional inorganic materials have also been realized using lipids. An early realization of this was by Mann et al., who used lipid tubules as a template for the formation of magnetic and non-magnetic iron oxides (Archibald & Mann 1993). Similarly, Shimizu et al. reported on the templated growth of vanadium oxide, titania and tantalum oxide using a self-assembled lipid tubule system, which contained secondary ammonium sites as positively charged headgroups (Ji & Shimizu 2005). Their lipid assembly produced tubular structures with outer diameters of 80 nm and quite thin walls, corresponding to only one bilayer in thickness. Addition of an ethanolic solution of titanium isopropoxide Ti(OPr)₄ to an aqueous dispersion of the lipid did not produce templated mineralization, merely bulk condensation of titania that rapidly threw a sediment. Ti(OPr)₄ will rapidly undergo hydrolysis and self-condensation when in contact with water, and so the presence of large amounts in which the lipids are dispersed understandably resulted in uncontrolled titania formation. The group solved this problem by carrying out the mineralization with vitrified lipid nanotubes in ice. They suggest that at 0°C a slow permeation of the alcohol throughout the ice will carry the titania throughout the structure. Encountering an unfrozen film of water around each tubule would then induce condensation of the titania only in those regions in intimate contact with the tubule, thereby effectively constraining mineralization to the tubule surfaces. After lyophilization of the vitrified composite material, SEM revealed that there were a large number of intact tubular structures. Calcination of these structures at 500°C resulted in pure titania hollow tubes. The group was also able to apply their methodology to the synthesis of tantalum and vanadium oxides, producing hollow tubes in each case.

An active means of controlling inorganic growth by use of a lipid layer was demonstrated by the use of a composite organic template consisting of anionic DNA and cationic lipid membranes (Liang et al. 2003). These were found to spontaneously self-assemble into a multi-lamellar structure. The lipid component comprised binary mixtures of the neutral lipid dioleoyl-phosphatidylcholine and the cationic lipid dioleoyltrimethylammonium propane, which were mixed with an aqueous DNA solution to form self-assembled complexes. Chains of DNA were found to arrange themselves parallel across the lipid membrane structure, forming a one-dimensional lattice. By using lipids to induce periodicity in the DNA molecules, the group were able to organize Cd²⁺ ions within the interhelical pores of the DNA chains, thereby providing sites of preferential reaction. On treatment with H₂S, the Cd²⁺ formed CdS nanorods, which were all aligned with the (002) planes parallel to the negatively charged sugar-phosphate DNA backbone, indicating the fine control over the crystallographic properties afforded by the preferential growth along the DNA chains (figure 14).

• Download figure
• Open in new tab
• Download PowerPoint

Confirmation of the nanoparticulate nature of the CdS nanorods was afforded by both TEM and wide-angle X-ray scattering, which showed broadened peaks, indicative of nanoscale crystallite size. The group noted that the pronounced tilt of the nanorods of 60° matched the orientation of the DNA sugar-phosphate backbone, which is tilted by approximately 60° with respect to the helix axis in B-form DNA when projected onto a two-dimensional plane. The negatively charged ridge on the DNA molecule in this orientation is therefore responsible for the nucleation of cations of Cd²⁺, leading to CdS growth with a pronounced tilt.

(j) DNA

In addition to the work done on DNA and lipids in co-assembly, the DNA molecule itself has been used directly as a template to form anisotropic structures. The amount of work that has been done on the templating of the DNA molecule is vast and it is beyond the scope of this review to cover them all in detail, so the reader is directed to some recent reviews on this subject for an in-depth treatment (Niemeyer 2000; Sharma et al. 2006; Tanaka & Shionoya 2006; Sotiropoulou et al. 2008). However, some of the most recent work on the templating of DNA is presented here, to give a flavour of the current state of the field.

As would be expected, one of the most common inorganic phases, silica, has been successfully grown around DNA to produce nanowires. Numata et al. (2004) were able to achieve this, but only by solving a couple of critical organic–inorganic compatibility issues. First, in neutral or alkaline media, both DNA and the ubiquitous silica precursor TEOS are anionic. This is a significant problem, as charge complementarity is the major driving force in biotemplated mineralization. Second, TEOS is insoluble in water, while DNA is virtually insoluble in any solvent other than water. The group solved both of these issues by replacing the sodium counterions associated with the DNA molecule with a guanidinium/ammonium complex group, thereby rendering the surface of the DNA cationic and imbuing the molecule with a certain degree of amphiphilicity. This allowed the DNA to be solubilized in a chloroform:methanol mixture, into which TEOS was incorporated. Sol–gel hydrolysis and condensation of the TEOS produced rod-like silica structures of up to 1 mm in length. The group confirmed that DNA was acting as a template via UV–vis spectrometry of the product, which showed a broadened spectrum, but which contained the basic spectral features of the cationic DNA solution, indicating that the DNA was still present encased within the silica matrix. TEM images of the calcined silica nanorods proved that they were hollow, with an inner core diameter of 10 nm, very close to the diameter of the cationic DNA (7 nm). Calcination of the composite structures removed the DNA successfully, with little or no structural collapse of the tubular structure.

Another alteration of the DNA molecule prior to mineralization was demonstrated by using palladium nanoparticles as catalytic sites of nucleation and growth of cobalt (Gu & Haynie 2008). By complexing Pd(II) to the DNA molecule, the group were able to reduce the metal in situ to Pd(0) with dimethylamine borane to produce Pd nanoparticles that decorated the surface of the DNA. On incubation in a CoCl₂/boric acid/sodium citrate bath, cobalt nanoparticles were able to be grown preferentially on the palladium nanoparticles. After incubation, the group found that the DNA molecules had been completely metallized, resulting in cobalt nanowires between 40 and 60 nm in width and several micrometres in length. In control experiments (without decoration of the DNA by palladium), the group was unable to produce cobalt metallization of the DNA molecules; the solution remaining pink throughout, indicative of uncomplexed hydrated cobalt species.

The sol–gel approach was employed in a rather simpler method to achieve the coating of DNA by titania. By fixing DNA molecules on a flat substrate, Fujikawa et al. (2005) were able to deposit drops of ethanolic titanium n-butoxide (Ti(O_nBu)₄), which effectively formed a titania covering over the entire surface. Once dry, the composite material was calcined to leave titania nanowires with diameters between 10 and 180 nm.

DNA has also been used to infiltrate the gallery spaces of certain clays. In doing so, the clay provides an ‘armour-plating’ to the DNA molecule, rendering it much more resistive to thermal and chemical denaturation. This was achieved by the addition of DNA to an aminopropyl-functionalized magnesium phyllosilicate (AMP) clay suspended in water (Patil et al. 2007). Prior to addition, the clay is exfoliated by sonication, resulting in nanosheets and clusters of clay layers, thereby facilitating the adsorption of DNA directly into the gallery spaces. Once incorporated, the composite material formed lamellar structures and highly elongated nanowires and nanofilaments that were flexible and continuous over several micrometres in length. The group determined that these filaments comprised individual nanowires between 15 and 30 nm in width. On drying, the clay materials were found to still contain DNA by FTIR spectroscopy (which showed characteristic absorbances for both DNA and AMP clay) and by XRD, which showed that the AMP gallery spacings had increased to 3.6 nm from 1.6 nm. The fact that DNA was so effectively trapped within the clay gallery spaces was demonstrated by recording the melting temperature profiles of free and intercalated DNA molecules at 260 nm via a UV–vis spectrophotometer. In each instance, the group obtained sigmoidal melting curves, indicative of denaturation of the DNA double helix, but the temperature at which melting occurred differed significantly. In free DNA, the melting temperature was determined to be 63°C, whereas in intercalated DNA it was 83°C. The group postulated that the significant thermal resistivity afforded to the DNA by the AMP clay was a result of multipoint electrostatic interactions between pendent aminopropyl groups covalently attached to the clay and counter-charged entrapped DNA molecules.

(k) Viruses

At its very simplest, a virus consists of genetic material, either RNA or DNA, surrounded by a protein coat. The interest for the biotemplator comes in the structures adopted by the protein capsid. There are many variations, but the main structure types can be summarized as helical/filamentous and spherical/polyhedral. Helical capsids consist of a single type of protein molecule that self-interacts and forms a cylindrical assembly around the genetic material via electrostatic interactions. In this manner, the helical viruses are able to form highly anisotropic assemblies, sometimes several micrometres in length. This cooperative assembly between the protein and the genetic material creates a tubular structure that, if the genetic material is subsequently removed, can be used as a constrained reaction environment. This is in addition to the opportunities for mineralization afforded by the outer surface of the protein coat.

One of the most commonly investigated anisotropic viruses, owing to its stability over a wide pH range (3–9), temperatures up to 90°C and organic solvents, is the tobacco mosaic virus (Harrison & Wilson 1999; Culver 2002; McCormick & Palmer 2008). Consisting of approximately 2000 protein subunits, which assemble around an RNA core in a ‘lock washer’ configuration, the protein forms a structure of 18 nm wide and up to 300 nm in length, with an inner core of 4 nm in diameter containing the RNA.

One study looked into the effect of incubating TMV with CuCl₂, a commonly used precursor material in the mineralization of copper compounds (Lee et al. 2006). They found that, at concentrations of CuCl₂ higher than 0.5 mM, the hydrodynamic radius of TMV increased substantially, indicating that undesired spontaneous aggregation had occurred. Furthermore, the group observed that complete mineralization of the TMV capsids was not possible in their system at CuCl₂ concentrations lower than 0.5 mM. They concluded that, in a system with no other additive, it is not possible to achieve dense mineralization of TMV capsids without aggregation. Their studies were complicated by the fact that they found it necessary to replace the buffered solution containing the TMV capsids with water in order to achieve reproducible coatings and to make the analysis of the colloidal interaction between TMV capsids simpler.

No such complications were observed by Shenton et al. (1999) as they were able to keep the TMV capsids in buffered solutions and yet achieve appreciable mineralization of the surface of the virion surfaces, producing anisotropic, polycrystalline assemblages. In their work, they were able to mineralize the surface of TMV capsids with CdS, PbS, iron oxides and silica. In the case of sulphide compounds, incubation of either chlorides or nitrates of cadmium or lead with TMV capsids, followed by diffusion of H₂S gas through the system, produced nanocrystals of CdS or PbS. TEM studies of the CdS mineralized capsids showed that they were coated with a 16 nm thick crust, which lattice imaging revealed were CdS nanoparticles in the zinc blende structure, approximately 5 nm in diameter. The group noted that mineralization appeared to be restricted to the outer surface of the capsid, although given the high electron density of the CdS, a clear picture of the inner surfaces of the virion could not be obtained. For PbS, the particles coating the capsid surface were much larger, approximately 40 nm in diameter, and appeared either prismatic or irregularly shaped. Electron diffraction data of these nanoparticles showed that they were single-domain crystals and possessed the rock salt structure.

Their work was extended by inducing TMV to form into liquid crystalline arrangements, which were subsequently mineralized with silica (Fowler et al. 2001). Making use of the tendency to align in strong magnetic fields, the group placed a buffered aqueous suspension of TMV inside the bore of an NMR machine for 1 hour. This produced a viscous gel that exhibited birefringence, indicative of nematic liquid crystalline ordering. Addition of a mixture of TEOS and aminopropyltriethoxysilane (APTES) to the gel, followed by gentle manipulation of the NMR tube, allowed siliceous species to infiltrate the entire liquid crystal. After quiescent storage for 4 days, the group ensured full condensation of the silica by placing the tube in a bath of water at 45°C for 2 days. Once complete, the gel was removed from the tube and dried under ambient conditions overnight. Finally, the TMV was removed from the system to leave hollow inorganic replicas by calcining the gel to 540°C. TEM of the composite material revealed that the sample comprised fragments of silica/TMV composite with pronounced lattice fringes in the silica of approximately 18 nm in width. The group interpret this to be due to a periodic array of co-aligned, anisotropic TMV particles joined end to end and intercalated within a continuous framework of amorphous silica. On removal of the TMV template, the structure remained largely intact, leaving a mesoporous siliceous body. Porosity was determined to be regular rather than disordered as viewed end-on the silica replicas showed a hexagonal array of 11 nm channels, separated by silica walls of approximately 10 nm in thickness.

Further refinements of the templating conditions have enabled researchers to restrict mineralization to the interior of the TMV virion. By first rendering the surfaces of the TMV conductive with platinum, several studies have shown that subsequent immersion in copper and nickel solutions results in the catalytic electroless deposition of the desired metal in an extremely fine manner (Knez et al. 2003; Balci et al. 2006). Nanowires of 3 nm in diameter and up to 150 nm in length were able to be formed within the interior channel of the TMV virion. These studies were able to restrict the mineralization to the interior of the TMV by using a mutant strain, which did not possess metallophilic amine groups on the surface.

3. Conclusions

The biotemplated synthesis of anisotropic nanomaterials has been successfully demonstrated for many different systems and via several different routes. The one common thread that draws all of this work together is the desire to find efficient synthetic methods in the production of anisotropic nanoparticles; most commonly for simpler materials, but also more recently as a means to achieving this technologically important morphology in more functional ones. Researchers are now discovering that biopolymers are particularly well suited in this role, as we have seen that a biopolymer-mediated synthesis offers some key advantages.

1. Many biopolymers have an extremely strong affinity for metal cations. By chelating metal ions, the biopolymer can act effectively as a ‘cocoon’ for the inorganic phase, thereby ensuring that the inorganic phase will remain (according to reaction conditions) nano-sized.

2. The biopolymer can, if so desired, provide a ready source of carbon during the synthesis, either for use in the formation of carbide phases or as a means of ensuring a localized reducing environment by carbothermal reduction of the inorganic phase.

3. Biopolymers can be obtained from, or can be functionalized with, a range of reactive functional groups such as sulphates, carboxylates, nitrates and sulphides, leading to the incorporation (if so desired) of the functional group in the final product.

4. The majority of biopolymers, particularly the polysaccharides and some proteins, are composed of long molecular chains, which leads naturally to fibrous morphologies on the macroscale. The presence of an anisotropic motif at all length scales in the biopolymer can direct crystal growth naturally in this morphology.

5. Lastly, biopolymers are on the whole extremely cheap and readily accessible. Both cellulose and chitosan are produced in gigaton amounts worldwide, either in the biosphere or as by-products of other industrial processes.

The biotemplated route is one which is conceptually easy to imagine and execute, yet which produces the controlled crystallization of complex inorganic materials, time after time. In 1995, three ‘main thrusts’ in the new field of biomimetics were identified, of which biotemplating plays a significant part (Mann 1995). The main thrusts were:

1. use of natural materials to control crystallization chemically and structurally,

2. use of living organisms to deposit inorganic material, and

3. use of biomineralization concepts to direct syntheses.

This author concludes that the success of the fields of biomimetics and bio-inspired materials chemistry would rely on the interdisciplinary nature of research and, more critically, on the training at an early level of students who would come through the academic system with an open and communal view of science, rather than that of the traditional ‘boxes’ of physical, organic and inorganic chemistry. We are beginning to see the emergence of the first green shoots of this ethos, with young academics now leading research groups that embrace this interdisciplinarity. Nowhere more than in the biotemplated synthesis of anisotropic nanoparticles is this evident; scientists are now routinely invoking principles and practices that run the whole gamut of science, from biology to physics, within the same piece of research. The wide intellectual net that is cast by biotemplating has caught the imaginations of many research groups worldwide and, in doing so, has hopefully managed to, as Mann requested, ‘unlock our imagination from the straightjackets of conventional disciplines’. The necessity for these new materials to find their way into the next generation of devices introduces yet another layer of interdisciplinarity and perhaps will open up new avenues of collaboration between the pure and applied sciences, ensuring that we keep our collective foot firmly down on the technological gas pedal.

Footnotes