FEATURE ARTICLE A general template-based method for the preparation of nanomaterials John C. Hulteen and Charles R. Martin*† Department of Chemistry, Colorado State University, Fort Collins, CO 80523, USA This article reviews a general template-based approach for the preparation of nanomaterials. The method involves the synthesis of a desired material within the pores of a nanoporous membrane.We have termed this approach ‘template synthesis’ because the pores within these nanoporous membranes act as templates for the synthesis of nanostructures of the desired material.Because the pores within these membranes are cylindrical and of uniform diameter, monodisperse nanocylinders of the desired material are obtained. Depending on the chemistry of the pore wall and material, these nanocylinders may be either hollow (a tubule) or solid (a fibril or nanowire). This template process will be shown to be a very general approach in the fabrication of nanotubes and fibrils composed of a variety of materials including polymers, metals, semiconductors, carbons, and other materials.While there has been a significant amount of research in the Introduction area of template synthesis of conductive polymer nanostruc- Many methods for the fabrication of nanoparticles have been tures, this has been reviewed elsewhere recently.32,33 developed, ranging from lithographic techniques to chemical methods.1,2 This research group has been exploring a fabri- Membranes used cation method termed template synthesis for the preparation of a variety of micro- and nano-materials.3–30 This process Most of the work in template synthesis, to date, has entailed involves synthesizing a desired material within the pores of a the use of two types of nanoporous membranes, ‘track-etch’ porous membrane.Because the membranes that are used have polymeric membranes and porous alumina membranes. cylindrical pores of uniform diameter, a nanocylinder of the However, there are a variety of other templates that could desired material is obtained in each pore.Depending on the be utilized. properties of the material and the chemistry of the pore wall, this nanocylinder may be solid (a nanofibril) or hollow (a Track-etch nanotubule). A number of companies (such as Nucleopore and Poretics) sell There are a variety of interesting and useful characteristics microporous and nanoporous polymeric filtration membranes associated with template synthesis.Probably the most useful that have been prepared by the track-etch method.34 This feature of this method is that it is extremely general with method entails bombarding a non-porous sheet of the desired regard to the types of materials that can be prepared.For material (standard thickness range from 6 to 20 mm) with example, we have used this method to prepare both nanotub- nuclear fission fragments to create damage tracks in the ules and nanofibrils composed of conductive polymers,4–13 material, and then chemically etching these tracks into pores. metals,14–25 semiconductors,26,27 carbon28–30 and other mate- The resulting membranes contain randomly distributed cylin- rials. Next, both tubular and fibrillar nanostructures with very drical pores of uniform diameter, Fig. 1(A,B). The commercial small diameters can be prepared. For example, conductive membranes are available with pore diameters as small as polymer nanowires with diameters as small as 3 nm have been 10 nm (pore density ca. 109 pores cm-2). These commercial prepared using this method.31 Also, because the membranes membranes are prepared from polycarbonate or polyester; employed contain cylindrical pores of uniform diameter,mono- however, a number of other materials are also amenable to disperse nanocylinders of the desired material, whose dimen- the track-etch process.34 sions can be carefully controlled, are obtained.Finally, these Owing to the random nature of the pore-production process, tubular or fibrillar nanostructures can be assembled into a the angle of the pores with respect to the surface normal can variety of architectures. The nanostructures can remain inside be as large as 34°.35 Therefore, depending on the specific pore the pores of the template membrane or they can be freed from diameter and pore density of the track-etched membrane, a the template membrane and collected as an ensemble of free number of pores may actually intersect within the membrane.nanoparticles. Alternatively, if the nanostructure-containing This is a problem when theoretically modelling the optical membrane is attached to a surface and the membrane is properties of template-synthesized nanometals, a topic of great removed, an ensemble of micro- or nano-structures that pro- interest to our group.18–20 For example, theory predicts a trude from the surface like the bristles of a brush can be specific wavelength maximum in the absorption band of iso- obtained.lated metal nanoparticles.18–20 However, physical contact The intent of this article is to provide an overview of the between the metal nanoparticles synthesized within the pores template method.We will start with a brief description of can shift this absorption maximum by 200 nm or more.36 the types of membranes used for template synthesis. Next, the dierent types of chemistries that have been used to prepare Porous alumina template-synthesized nanostructures will be reviewed. Finally, we will discuss fundamental properties and applications of Porous alumina membranes are prepared via the anodization template-synthesized metal and semiconductor nanostructures.of aluminium metal in an acidic solution.37 These membranes contain cylindrical pores of uniform diameter arranged in a hexagonal array, Fig. 1(C,D). However, unlike the track-etch † E-mail: crmartin@lamar.colostate.edu J.Mater. Chem., 1997, 7(7), 1075–1087 1075Other nanoporous materials Tonucci et al. have recently described a nanochannel array glass with pore diameters as small as 33 nm and pore densities as high as 3×1010 pores cm-2.39 Beck et al. have prepared a new class of mesoporous zeolites with large pore diameters.40 Douglas et al. have shown that the nanoscopic pores in a protein derived from a bacterium can be used to transfer an image of these pores to an underlying substrate.41 Clark and Ghadiri have prepared arrays of polypeptide tubules.42 Finally, both Ozin1 and Schollhorn43 have discussed a wide variety of nanoporous solids that could be used as template materials.Template synthetic strategies The limits to which materials can be used in template synthesis are defined by the chemistry required to synthesize the material.Nearly any material can in principle be synthesized within these nanoporous membranes, provided a suitable chemical pathway can be developed. Typical concerns that need to be addressed when developing new template synthetic methods include the following: (1) will the precursor solutions used to prepare the material ‘wet’ the pore (i.e., hydrophobic/hydrophilic considerations); (2) will the deposition reaction proceed too fast resulting in pore blockage at the membrane surface before tubule/fibre growth can occur within the pores; (3) will the host membrane be stable (i.e., thermally and chemically) with respect to the reaction conditions? The following is a general outline of five representative chemical strategies that have been used in our laboratory to conduct template synthesis within the alumina and polymeric template membranes.Electrochemical deposition Electrochemical deposition of a material within the pores is accomplished by coating one face of the membrane with a metal film (usually via either ion sputtering or thermal evaporation) and using this metal film as a cathode for electroplating. 17–22,44,45 This method has been used to prepare a variety of metal nanowires including copper, platinum, gold, silver, and nickel in both track-etch and alumina templates. Typical gold nanowires are shown in Fig. 2(A). The lengths of these nanowires can be controlled by varying the amount of metal deposited. By depositing a small amount of metal, short wires Fig. 1 Electron micrographs of polycarbonate (A and B) and alumina can be obtained; alternatively, by depositing large quantities (C and D) template membranes. For each type of membrane an image of metal, long needle-like wires can be prepared.18–20 This of a larger pore membrane is presented (A and C) so that the characteristics of the pores can be clearly seen.Images of a membrane ability to control the length or aspect ratio (length to diameter) with extremely small pores are also presented (B and D). A, SEM of of the metal nanowires is especially important in our optical the surface of a polycarbonate membrane with 1 mm diameter pores. investigations because the optical properties of nanometals are B, TEM of a graphite replica of the surface of a polycarbonate dependent on the aspect ratio.18–20,24 membrane with 30 nm diameter pores.The pores appear ‘ragged’ Hollow metal tubules can also be prepared via this method, owing to the artifact of using a graphite replica. C and D, TEMs of Fig. 2(B).17,22 To obtain tubules, one must typically chemically microtomed sections of alumina membranes with 70 nm (C) and 10 nm (D) diameter pores.derivatize the pore walls so that the electrodeposited metal preferentially deposits on the pore wall; that is, a molecular anchor must be applied. For example, gold tubules have been prepared by attaching a cyanosilane to the walls of the alumina membranes, the pores in these membranes have little or no tilt template membrane prior to metal depositions.17,22,46 Owing with respect to the surface normal resulting in an isolating, to the large number of commercially available silanes, this non-connecting pore structure.Although such membranes are method can provide a general route for tailoring the pore walls sold commercially (Whatman), a very limited number of pore in the alumina membranes. diameters are available. We have, however, prepared mem- Electrochemical deposition can also be used to synthesize branes of this type with a broad range of pore diameters.18,20 conductive polymers [such as polypyrrole, polyaniline, or We have made membranes with pore diameters as large as poly(3-methylthiophene)] within the pores of these template 200 nm and as small as 5 nm, and we believe that even smaller membranes.10,13 When these polymers are synthesized within pores can be prepared. Pore densities as high as 1011 the pores of track-etched polycarbonate membranes, the polypores cm-2 can be achieved,38 and typical membrane thickness mer preferentially nucleates and grows on the pore walls, can range from 10 to 100 mm.The higher pore density is resulting in polymeric tubules at short polymerization times, important if one wanted to mass-produce a nanomaterial by Fig. 2(C).By controlling the polymerization time, we can the template method. Membranes with high pore density would produce thin-walled tubules, thick-walled tubules or solid allow a greater number of nanostructures to be produced per fibrils. The reason why the polymer preferentially nucleates and unit area of template membrane. 1076 J. Mater. Chem., 1997, 7(7), 1075–1087Fig. 2 Electron micrographs of tubules and fibrils. A, TEM of a microtomed section of an alumina template membrane showing Au nanofibrils that are 70 nm in diameter within the pores. B, SEM of an array of Au microtubules. C, TEM of three polypyrrole nanotubules. The outer diameter is ca. 90 nm; the inner diameter is ca. 20–30 nm.grows on the pore walls is straightforward.8 Although the membrane is immersed into a Au plating bath containing AuI and a reducing agent, which results in Au plating on the monomers are soluble,the polycationic forms of these polymers are completely insoluble. Hence, there is a solvophobic compo- membrane faces and pore walls. The key feature of the electroless deposition process is that nent to the interaction between the polymer and the pore wall.There is also an electrostatic component because the polymers metal deposition in the pores starts at the pore wall. Therefore, after short deposition times, a hollow metal tubule (Fig. 3) is are cationic and there are anionic sites on the pore walls.8 obtained within each pore while long deposition times result in solid metal nanowires.Unlike the electrochemical deposition Electroless deposition method where the length of the metal nanowire can be Electroless metal deposition involves the use of a chemical controlled at will, electroless deposition yields structures that reducing agent to plate a metal from solution onto a surface.47 run the complete thickness of the template membrane.This method diers from electrochemical deposition in that However, the inside diameter of the tubules formed via electro- the surface to be coated need not be electronically conductive. less deposition can be controlled at will by varying the metal We have developed methods by which gold and other metals deposition time.15,16 Of course the outside diameter is deter- can be plated from solution onto the surfaces of both the mined by the diameter of the pores in the template membrane.plastic and alumina membranes.15 This method involves applying a sensitizer (typically Sn2+) to the membrane surfaces Chemical polymerization (pore walls and faces). The sensitizer binds to the surfaces via complexation with surface amine, carbonyl, and hydroxy Chemical template synthesis of a polymer can be accomplished by simply immersing the membrane into a solution containing groups.This sensitized membrane is then activated by exposure to Ag+ resulting in the formation of discrete nanoscopic Ag the desired monomer and a polymerization reagent. This process has been used to synthesize a variety of conductive particles on the membrane’s surfaces.Finally, the Ag-coated J. Mater. Chem., 1997, 7(7), 1075–1087 1077Fig. 3 TEM showing a microtomed section of a Au nanotubulecontaining membrane.The Au tubules are the black rings. The elliptical appearance is caused by the microtoming process. Pore diameter 50 nm; plating time 10 min. polymers within the pores of various template membranes. 6,9,48,49 As with electrochemical polymerization, the polymer preferentially nucleates and grows on the pore walls, resulting in tubules at short deposition times and fibres at long times.Conventional (electronically insulating) plastics can also be Fig. 4 SEM images of carbon tubules (A) and fibrils (B) with an outer chemically synthesized within the pores of these template diameter of 200 nm prepared in an alumina template membrane; membranes.For example, polyacrylonitrile tubules can be membrane was removed for imaging purposes prepared by immersing an alumina template membrane into a solution containing acrylonitrile and a polymerization Chemical vapour deposition initiator.28,29 The inside diameter of the resulting polyacrylon- A major hurdle in applying chemical vapour deposition (CVD) itrile (PAN) tubules is varied by controlling the time the techniques to template synthesis has been that deposition rates membrane remains in the polymerization bath.These PAN are often too fast. As a result, the surface of the pores becomes tubules have been further processed to create conducting blocked before the chemical vapour can traverse the length of graphitic carbon tubules and fibrils in alumina membranes, the pore.We have, however, developed two template-based Fig. 4.28,29 This is accomplished by heating the PAN tubules– CVD syntheses that circumvent this problem. The first entails alumina membrane composite to 700 °C under an argon flow the CVD of carbon within porous alumina membranes which or under vacuum. has been achieved by our group30 and others.50 This involves placing an alumina membrane in a high-temperature furnace Sol–gel deposition (ca. 700 °C) and passing a gas such as ethene or propene Sol–gel chemistry typically involves hydrolysis of a solution of through the membrane. Thermal decomposition of the gas a precursor molecule to obtain first a suspension of colloidal occurs throughout the pores, resulting in the deposition of particles (the sol) and then a gel composed of aggregated sol carbon films along the length of the pore walls (i.e., carbon particles.The gel is then thermally treated to yield the desired tubules are obtained within the pores). The thickness of the product. We have recently conducted various sol–gel syntheses walls of the carbon tubes is again dependent on total reaction within the pores of the alumina membranes to create both time and precursor pressure.tubules and fibres of a variety of inorganic semiconducting The second CVD technique utilizes a template-synthesized materials including TiO2, ZnO and WO3.26 First, an alumina structure as a substrate for CVD deposition.51 For example, template membrane is immersed into a sol for a given period we have used a CVD method to coat an ensemble of gold of time, and the sol deposits on the pore walls.After thermal nanotubules with concentric TiS2 outer nanotubules. The first treatment, either a tubule or fibril of the desired semiconductor step of this process requires the electroless plating of Au tubules is formed within the pores, Fig. 5. As with other template or fibrils into the pores of a template membrane. The Au synthesis techniques, longer immersion times yield fibres while surface layer is removed from one face of the plated membrane, brief immersion times produce tubules. and the membrane is dissolved away. The resulting structure The formation of tubules after short immersion times indi- is an ensemble of Au tubules or fibrils protruding from the cates that the sol particles adsorb to the alumina membrane’s remaining Au surface layer like the bristles of a brush, Fig. 6(A). pore walls. This is expected because the pore walls are nega- This structure is exposed to the precursor gases used to carry tively charged while the sol particles used to date26 are out the CVD synthesis of TiS2. As indicated in Fig. 6(B) the positively charged (a similar situation to what was described Au tubules become coated with outer TiS2 tubules. for conductive polymers). It has also been found that the rate of gelation is faster within the pore than in bulk solution.26 Composite nanostructures This is most likely due to the enhancement in the local concentration of the sol particles due to adsorption on the We have shown that a large number of dierent chemical techniques can be used to prepare tubules or fibrils that are pore walls. 1078 J. Mater. Chem., 1997, 7(7), 1075–1087Fig. 5 SEM images of TiO2 tubules and fibrils prepared in an alumina membrane with 200 nm diameter pores. The sol was maintained at 15 °C, and the immersion time varied from 5 to 60 s. A, Immersion time=5 s; remnants of the TiO2 surface layer can be seen in this image.B, Immersion time=25 s. C, Immersion time=60 s. alumina membrane via the sol–gel process discussed above, Fig. 7(A). After thermal treatment of the TiO2 tubules, conductive polypyrrole nanowires were grown using the chemical polymerization method inside the semiconductor tubules, Fig. 7(B). TiO2 is a promising material for photoelectrochemical energy production, and it has been shown that high surface area TiO2 has a higher photoeciency.52 Therefore, these TiO2–polypyrrole nanocomposites should be excellent photocatalysts because these template-synthesized structures have very high surface area.One problem in using high surface area TiO2 as a photocatalyst is the low electrical conductivity of the material.However, this tubular nanocomposite structure should solve this problem because each TiO2 tubule has its own current-collecting electrode inside. Another method for the construction of a two-component Fig. 6 SEM images of an ensemble of Au tubules before (A) and after (B) CV deposition of the outer TiS2 tubules. The tubules are protruding from the substrate Au surface layer.composed of a single material. However, one can imagine a host of applications where composite tubular nanostructures would be necessary. Examples might include concentric nanocapacitor or nanobattery tubules. We have recently developed chemical strategies for preparing such concentric tubular nanostructures.51 These composites have very high interfacial surface areas between concentric layers of materials.High interfacial areas are obtained because the interfaces are parallel to the long axis of the composite tubular nanostructure. The fabrication of a semiconductor–conductor tubular nanocomposite will introduce this concept of sequential tubular synthesis.51 This composite was prepared in a 60 mm thick Fig. 7 SEM images of TiO2 nanotubules prepared by sol–gel methods alumina template membrane with 200 nm diameter pores.before (A) and after (B) filling with the polypyrrole nanowires. Outer diameter of tubular composite is 200 nm. First, TiO2 tubules are synthesized within the pores of the J. Mater. Chem., 1997, 7(7), 1075–1087 1079concentric composite has already been described, Fig. 6.51 Gold the capacitors are connected in parallel from the surfaces of the template membrane. This will require that all of the tubules are electrolessly synthesized within the template membrane pores.The membrane is dissolved away, and a thin film electronically conductive outer tubules are electronically insulated from the conductive inner nanowires. of TiS2 is synthesized on the surface of the Au tubules via CVD.TiS2 is a Li+-intercalation material for Li-based Finally, self-assembly chemistry54 can also be used as a synthetic step to prepare tubular composites. For example, Au rechargeable batteries. We have recently shown that templatesynthesized Li+-intercalating materials can provide higher tubules were synthesized within the 1 mm pores of a polycarbonate template membrane via the electroless deposition discharge capacities than conventional electrodes made from the same material.53 As with the photoconductor materials, method.The inside diameter of these tubules was ca. 500 nm, and the length of the tubules was 1.0 mm. The Au tubule- many Li+-intercalation materials have low electrical conductivities. However, the current-collecting Au electrode inside containing membrane was then immersed in a solution of hexadecyl thiol causing the thiol to self-assemble onto the each TiS2 tubule should again solve this problem. We have shown that the TiS2–polypyrrole composite nanostructures inner surfaces of the Au tubules.The template membrane was dissolved away and the freed tubules were collected by reversibly intercalate and deintercalate Li+, and we are currently investigating the charge–discharge kinetics and capacit- filtration. When these gold–thiol tubules were placed in water, they ies of these tubular composite battery electrode materials.An alternative set of chemistries was used to fabricate floated at the air/water interface owing to the presence of the hydrophobic thiol on the inside of the tubule.In contrast, a conductor–insulator–conductor composite consisting of carbon–polyacrylonitrile–gold concentric tubules, Fig. 8.51 tubules that were not treated with the thiol filled with water and sank.51 Because self-assembly provides a general way to Initially, polyacrylonitrile (PAN) tubules were chemically polymerized within the pores of an alumina membrane followed apply a large number of dierent chemical functionalities to the inner (and outer) surfaces of such tubules, composite by thermal carbonization resulting in conductive carbon tubules, Fig. 8(B). The PAN polymerization step was then tubules with diverse inner and outer chemistries should be possible. repeated creating insulating PAN tubules within the carbon tubules, Fig. 8(C). A gold film was then sputtered onto one These have been just afew examplesof the types of composite structures that can be fabricated with template synthesis.face of the membrane. Using this film, Au nanowires were electroplated within the inner PAN tubules resulting in the Composites composed of a variety of dierent conducting, insulating, semiconducting, photoconducting and electroactive desired concentric tubular C–PAN–Au composite structures, Fig. 8(D). We are currently using this synthetic strategy and materials have been prepared. The limits as to how many dierent components each composite can contain is limited others to prepare ensembles of nanocapacitors, where all of Fig. 8 A, SEM image of the surface of the alumina template membrane. B, The carbon tubules obtained after dissolution of the template membrane; C, as per B but after polymerization of a PAN tubule within each carbon tubule.D, After electrodeposition of a Au nanowire within each PAN tubule. As noted, the carbon–PAN–Au composites were prepared by performing the appropriate chemistries in sequence leaving the alumina membrane intact; however, it is easier to image these extremely small structures by dissolving the membrane. 1080 J.Mater. Chem., 1997, 7(7), 1075–1087only by the initial diameter of the template pore and the rate Structural characterization of material deposition. The diameter of the electroplated gold nanoparticles is equivalent to the pore diameter of the alumina template membrane. Thus, Au nanoparticles with dierent diameters can be fabri- Optical properties of gold nanoparticles cated in alumina membranes containing dierent pore diameters.The aspect ratio is controlled by changing the amount We18–20 and others38,55 have been investigating the properties of Au electrochemically deposited into the pores. However, we of nanometals prepared within the pores of alumina have found that it is not possible to quantitatively predict the membranes.Through confinement of metals to a nano-sized aspect ratio of the Au nanoparticles because the plating current dimension, a variety of changes occur in the optical,18–20,55 eciency varies from membrane to membrane.24 Hence, it is electronic56 and magnetic38,57 properties. The first demon- not possible to calculate the aspect ratio of the Au nanoparticle stration of template synthesis for the creation of nanometal obtained from the known quantity of Au deposited and the fibrils was by Possin in 1970.58 Earlier work in which nanomet- pore diameter and density.Therefore, transmission electron als were used to colour alumina is also of interest.59 Nanometal- microscopy (TEM) analysis of the Au nanoparticles synthe- containing membranes of this type have also been used as sized in each membrane is necessary to determine the lengths selective solar absorbers.60 Finally, magnetic metals have been (and aspect ratios) of the nanoparticles.24 A TEM image of a deposited within the pores of such membranes to make vertical transverse section of a Au nanoparticle–alumina composite is magnetic recording media.61 shown in Fig. 2(A). When dierent amounts of Au are elec- This research group18–20 and others55 have been primarily trodeposited within the pores of the template membrane, we interested in the fundamental optical properties of nano- can produce Au nanoparticle shapes that are prolate, spheroid cylinders of Au imbedded into alumina membranes. The or oblate.24 colours of colloidal suspensions of Au can range from red to purple to blue depending on the diameter of the particles,62 Optical characterization and we have been able to demonstrate analogous colours for Au particles electroplated into the alumina template mem- The dierences in the shapes of the Au nanoparticles result in brane.18–20 These colours result from shape-induced changes changes in the optical absorption properties of the comin the plasmon resonance band of the Au nanoparticle which posite.18–20,24 Such changes are clearly visible as a membrane’s corresponds to the wavelength of light that induces the largest colour can vary from a bright red to deep blue to turquoise electric field on the nanoparticles.depending on the particle shape.18–20,24 The alumina membranes are optically transparent, so the colours are predominantly due to the Au nanoparticles.It should also be noted Fabrication that the parallel orientation of the pores in the alumina membrane confines the Au particles to a single dimensional The gold nanoparticles are prepared using the electrodepos- alignment. Correspondingly, there is no ambiguity in particle ition method discussed above, Fig. 9.18–20 First, Ag is deposited orientation which is a necessary feature for theoretical model- onto one face of an alumina template membrane to provide a ling of the absorption spectrum. conductive film for electrodeposition, Fig. 9(A). The membrane Fig. 10 shows the experimental absorption spectra for a is placed, Ag film side down, on a glass plate and covered with variety of Au nanoparticle–alumina composites.The Au par- a Ag plating solution. Then, short Ag ‘plugs’ or ‘posts’ are ticle aspect ratio (length/diameter) varies from 7.7 to 0.38, and electrochemically grown into the pores, Fig. 9(B). These Ag nanoposts are used as foundations onto which the Au nanoparticles are electrochemically grown, Fig. 9(C). Finally, the Ag foundations are removed with a nitric acid wash resulting in an array of Au nanoparticles imbedded within the pores of the alumina membrane, Fig. 9(D). Fig. 9 Fabrication procedure for Au nanoparticle–alumina composite. A, Ag is sputtered on one side of the host alumina membrane. Fig. 10 Experimental absorption spectra for the Au nanoparticle con- B, Membrane is placed sputtered side down onto a glass plate, and a Ag foundation is deposited electrochemically.C, Au is electrochemi- taining membranes. The spectrum with the highest absorbance maximum is for the membrane containing the aspect ratio (length/diameter) cally deposited onto the Ag foundations. D, Ag is removed with nitric acid. of 7.7, followed by 2.7, 1.3, 0.77, 0.54, 0.46 and 0.38 respectively. J. Mater. Chem., 1997, 7(7), 1075–1087 1081the diameter of each Au particle is constant at ca. 52 nm. The higher values of the background or double-layer charging currents. In the case of the NEE, the polycarbonate is stretch- reduction in absorption intensity with decreasing aspect ratio is expected due to the decrease in the metal volume fraction oriented during fabrication to improve mechanical properties. Upon heating above the glass transition temperature of the composites.The shift in the absorption maximum from 518 nm (aspect ratio 7.7) to 738 nm (aspect ratio 0.38) is (ca. 150 °C), the membrane relaxes, shrinks, and seals the junction between the Au nanowires and the polymer predicted from simulated spectra obtained using a dynamic Maxwell–Garnett theory.24 membrane.14,15 This article shows that the template method can be used to fabricate Au nanoparticles with various diameters and aspect Current response of the NEE ratios.Shifts in the absorption maximum and changes in the Two dierent electrochemical response limiting cases can be absorption intensity of the Au nanoparticle–alumina com- observed at an NEE, the ‘total overlap’ and ‘radial’ response.14 posites have been studied as a function of both particle Which limiting case is achieved depends strongly upon the diameter and aspect ratio.Current work involves determining distance between the electrode elements and the timescale (e.g., the eects of heating the Au nanoparticle–alumina composite. scan rate) of the electrochemical experiment. When the elec- Changes in the Au nanoparticles aspect ratios, optical proper- trode elements are in close proximity and the scan rate is ties and crystal structure have been observed.relatively low, the diusion layers at each electrode element overlap, Fig. 12(A). This overlap results in a single diusion Nanoelectrode ensembles layer that covers the total geometric area of the NEE. Linear diusion occurs to the entire NEE surface, and conventional One very exciting application of template synthesis is in the peak-shaped voltammograms are obtained. Also, the total area of electrochemistry.Nanoelectrodes oer opportunities to faradaic current is equivalent to that obtained at an electrode perform electrochemistry in highly resistive media63,64 and to of equivalent geometric area whose entire surface area is gold.investigate the kinetics of redox processes that are too fast If the electrode elements are located far apart and the to measure at conventional macroscopic electrodes.65–68 (By timescale of the experiment is relatively fast, the diusion macroscopic electrodes we mean disk-shaped electrodes with layers at each electrode act independently resulting in a radial diameters of the order of 1 mm.) We have used the template diusion field at each individual electrode element, Fig. 12(B). method to prepare ensembles of Au nanodisk electrodes where The voltammogram in this case has a sigmoidal shape, and the diameters of the Au disks are as small as 10 nm. the predicted total faradaic current is equivalent to the sum of the current generated at each individual electrode element Fabrication within the NEE.Fig. 13 shows a series of SEM images of NEEs with varying Using the electroless Au deposition procedure, Au nanowires are synthesized within the pores of a polycarbonate track-etch average distances between the electrode elements.14 The NEEs were fabricated from polymer template membranes with dier- membrane. In addition, both faces of the membrane become coated with thin Au films.If one of these surface Au films is ent pore densities but similar pore diameters. Fig. 14 presents the faradaic response of an electroactive species [trimethylami- removed, the disk-shaped ends of the Au nanowires traversing the membrane are exposed. These nanodisks can be used as nomethylferrocene (TMAFc+)] at each of these NEEs.14 The NEE with the highest electrode element density [Fig. 13(A), active elements in an ensemble of nanoelectrodes. Fig. 11 shows a schematic of such a nanoelectrode ensemble (NEE).14 Fig. 14(A)] shows a peak-shaped voltammogram indicative of the total overlap response. In contrast, the NEE with the Electrical contact is made to the remaining surface layer which acts as a common current collector for all the nanoelectrode lowest electrode element density [Fig. 13(D), Fig. 14(D)] shows the expected sigmoidal voltammogram. The other elements. A consistent problem associated with micro- and nano- two NEEs have an intermediate nanoelectrode density [Fig. 13(B,C), Fig. 14(B,C)] and show an intermediate electrodes is achieving an ecient seal between the conductive element and the host material.If a good seal is not achieved, response. We can demonstrate quantitatively that the NEEs in solution can creep into this junction resulting in significantly Fig. 13(A,D) are operating in the total overlap and radial Fig. 11 Schematic of an edge view of a nanoelectrode ensemble. The nanometal fibrils running through the pores of the template membrane are shown. The lower ends of the fibrils define nanodisks which are Fig. 12 Schematic of a side view of NEEs and the corresponding the electrodes. The opposite (upper) ends of the nanofibrils are connected to a common metal film which is used to make electrical diusion fields for the total overlap (A) and radial (B) limiting electrochemical response contact to the nanodisks. 1082 J.Mater. Chem., 1997, 7(7), 1075–1087Fig. 13 SEM images of the surfaces of NEEs showing the disk-like electrode elements prepared from membranes with varying pore densities. Average distance between pores are: A, 0.25 mm; B, 1.1 mm; C, 3.5 mm; D, 17.5 mm. Diameters of electrode elements are 100 nm (A,D) and 200 nm (B,C). response modes by comparing experimental and simulated voltammograns.Such a comparison is shown in Fig. 15. The simulated voltammogram in Fig. 15(A) is based on the reversible total overlap limiting case, and the experimental voltammogram is the same as Fig. 14(A).14 The quantitative Fig. 14 Cyclic voltammograms (50 mV s-1) for 50 mm TMAFc+ in Fig. 15 Simulated and experimental voltammograms for NEEs prepared from Fig. 13A and D.Scan rate and solution same as Fig. 14. 5 mMNaNO3 forNEEs prepared from the membranes shown in Fig. 13 J. Mater. Chem., 1997, 7(7), 1075–1087 1083agreement between the simulated and experimental voltammo- elements and a geometric area equivalent to that of the macroelectrode at various low concentrations of TMAFc+. grams confirms that the NEEs at this pore density and scan rate are in the total overlap electrochemical response.It is While the voltammograms essentially look identical to those obtained at the macroelectrode, the concentrations of the important to point out that there are no adjustable parameters in this simulation. electroactive species at the NEE are three orders of magnitude lower than those for the macroelectrode. The detection limit The simulated voltammogram in Fig. 15(B) assumes a single 100 nm diameter disk electrode, but the total current is multi- at the macroelectrode was determined to be ca. 2 mM while the detection limit at the NEE was ca. 2 nM.15 plied by the number of electrodes within the geometric area of the NEE. The experimental voltammogram is equivalent to Template synthesis has been shown to provide a simple means of creating ensembles of nanoelectrode ensembles.These Fig. 14(D). The quantitative agreement between the simulated and experimental voltammograms proves that the radial elec- NEEs can achieve electroanalytical detection limits that are three orders of magnitude lower than detection limits obtained trochemical response has been achieved at this NEE. Again, there are no adjustable parameters in this simulation. at conventional macroelectrodes.We are currently investigating fabrication processes that allow the use of NEEs in nonaqueous solvents. Detection limits A possible application of these NEEs is the ultra trace detection Metal nanotube membranes of electroactive species. We have recently shown that NEEs with 10 nm diameter disks operating in the total overlap mode We close our discussion of metal nanostructures with an show electroanalytical detection limits that are three orders of interesting new type of membrane consisting of Au nanotubes magnitude lower than detection limits obtained at macroscopic that span the complete thickness of the membrane.We have Au disk electrodes of comparable geometric area.15 This occurs previously mentioned that by controlling the electroless Au because in the total overlap mode, the total faradaic signal deposition time, the inside diameters of these tubes can be generated at the NEE is equivalent to that obtained at the controlled at will.We recently asked the question, can tubes conventional macroelectrode of equivalent geometric area. with inside diameters that approach the sizes of molecules be However, the background double-layer charging current is prepared, and if so, what applications might exist for such significantly less because these currents are proportional only nanotubule containing membranes? to the active Au area.The ratio of active area to geometric area for a 10 nm NEE is approximately 0.001.15 As a result, Fabrication the background current is reduced by three orders of magnitude, and detection limits can be improved by three orders of Typical templates used to prepare the metal nanotubule membranes were 6 mm thick polycarbonate membranes with 50 nm magnitude.An example of this enhancement in detection limits at an pore diameters and 6×108 pores cm-2. Gold was electrolessly plated onto the walls of the pores yielding a Au nanotube NEE is shown in Fig. 16.15 Fig. 16(A) shows voltammograms at a conventional Au macroelectrode at various low within each pore. Variation in the plating time has been shown to produce Au tubules with internal diameters ranging from concentrations of TMAFc+. As expected, the faradaic signal eventually vanishes into the double-layer charging currents as 34 to 1.4 nm.16 The diameter of these Au tubules was determined from measurements of gas (He) flux across the mem- the concentration of TMAFc+ decreases. Fig. 16(B) shows voltammograms at a NEE with 10 nm diameter electrode brane.16 Because the electroless processplates on the membrane surface as well as within the pores, electrical contact with the surfaceallows electrical control of the potentialinside the pores.Ion-selective membranes The ion transport properties of these Au nanotubule-containing membranes were studied using a U-tube concentration cell where the membrane separates two diering aqueous solutions, Fig. 17.16 In an initial experiment, diering concentrations of KCl were placed on each side of the membrane, and reference electrodes were inserted into each solution to measure the membrane potential (Em).When the diameters of the Au nanotubules approached 2 nm or less, the membranes dis- Fig. 16 Cyclic voltammograms at 100 mV s-1 in aqueous TMAFc+ at (A) a gold macrodisk electrode in 50 mM NaNO3 [TMAFc+]=0.5 (a), 7.8 (b), 15.6 (c), 31.2 mM (d ); (B) a 10 nm NEE in 1 mM NaNO3 Fig. 17 Schematic of a U-tube concentration cell [TMAFc+]=0.5 (a), 7.8 (b), 15.6 (c), 31.2 nM (d). 1084 J. Mater. Chem., 1997, 7(7), 1075–1087played near ideal cation-permselective behaviour, i.e., these cations in both fundamental and applied electrochemistry. Because the Au tubules have dimensions of the order of membranes transport cations but reject anions.16 This behaviour occurs because Cl- adsorbs strongly to Au, and as a molecular sizes and are quite monodisperse, we have been exploring the possibility of separating molecules based upon result, the Au tubules have an excess of negative charge (Cl-) on their inner surfaces.This causes anions to be excluded from dierences in their physical dimensions. the pores. Ion permselectivity can also be controlled by directly chang- Semiconductor nanotubules and nanofibres ing the potential applied to the Au nanotubules.For this work, it was essential to use an anion that does not adsorb to Au Electrochemical methods have been used previously as a means of depositing semiconductor materials into the pores of a because we wanted to control the charge in the Au tubes and not have it predetermined due to excess charge from counter- template membrane.27 However, this section will discuss the properties of semiconductor tubules and fibrils synthesized by ion adsorption.Because F- does not adsorb to Au, KF was chosen as the electrolyte. The U-tube assembly was used again, a much more versatile deposition method, sol–gel chemistry.26 but this time the membrane was connected to the working electrode lead of a potentiostat.The potential applied to the Structural characterization Au nanotube membrane varied from -0.5 to +0.5 V vs. Upon the confinement of a semiconductor to nanoscopic Ag/AgCl. The membrane was placed between solutions of dimensions, the first two questions that arise are: can we see 10 mM and 1 mM KF, and Em values were measured at each evidence for quantum confinement, and what is the crystal applied potential. structure of the material? TiO2 fibrils have been synthesized The dashed lines at the top and bottom of Fig. 18 are the within the pores of both 200 nm and 22 nm pore diameter Em values that would be achieved if the nanotubule membrane alumina membranes.26 The sol–gel fabrication of TiO2 fibrils showed ideal cation and ideal anion permselectivity, respect- within the pores of alumina membranes was described earlier.ively. At negative applied potentials, the nanotubule membrane An absorption spectrum of the template alumina membrane shows ideal cation permselectivity, whereas at positive applied containing these fibres showed an abrupt increase in potentials the membrane shows ideal anion permselectivity.absorbance at an approximate wavelength of 389 nm. This This selectivity occurs because at negative applied potentials, corresponds to the bandgap of bulk TiO2.70 This suggests that an excess negative charge is present on the walls of the Au the diameter of these fibrils is too large to see evidence for tubes. This results in the exclusion of anions from the tubes. quantum confinement in the absorption spectrum.We are At positive applied potentials, the opposite situation occurs: capable of preparing alumina template membranes with pore cations are excluded and anions are transported. diameters approaching 5 nm or smaller. Correspondingly, we For any combination of metal and electrolyte, there is a are currently attempting to prepare fibrils small enough to potential called the potential of zero charge (p.z.c.) where there provide evidence for quantum confinement.is no excess charge on the metal. At this potential the nano- Electron diraction has been employed to determine the tubule membranes should show neither cation nor anion crystal structure of the template-synthesized TiO2 fibrils.26 permselectivity, and Em should approach 0 mV.Em for the Fig. 19(A) shows a TEM image of 15 nm diameter TiO2 tubule-containing membrane does go from the ideal cation nanofibres with the membrane dissolved away. The small fibres permselective value, through zero, to the ideal anion permselec- are arranged in bundles which can contain anywhere from 2 tive value. Furthermore, the potential at which Em approaches to 10 or more fibres.Fig. 19(B) shows the indexed electron zero is close to the reported p.z.c. (-4 mV).69 diraction pattern obtained from the centre of the fibril bundle We have demonstrated that these Au nanotubule-containing on the left side of the main feature in Fig. 19(A). The orien- membranes can be cation permselective, anion permselective, tations of the images are the same, i.e., the c* axis in Fig. 19(B) or non-selective depending on the potential applied to the is parallel to the fibril bundle axis in Fig. 19(A). These data membrane. These membranes can be as permselective as the show that the fibrils are highly crystalline anatase-phase TiO2, commercially available Nafion polymer and should have appli- with the c* axis of the anatase oriented along the long axis of the fibril.Small fibril bundles throughout the sample display the same crystalline orientation; i.e., the reciprocal lattice direction [110] is almost always parallel to the electron beam, and the c* axis is along the fibril axis. We have concluded that these fibrils crystallize as long, prismatic crystals with the rare, Fig. 18 Variation of Em with potential applied to the membrane [1 mM KF on the low concentration (l) side, and 10 mM KF on the high concentration (h) side of the membrane; tubule radius ca. 1.1 nm]. The potential of the membrane was controlled with a potentiostat vs. a Fig. 19 A, TEM image of a bundle of 15 nm diameter TiO2 fibrils. Ag/AgCl reference electrode immersed in the side-h solution. Em was measured with the membrane under potentiostatic control. B, Corresponding electron diraction pattern.J. Mater. Chem., 1997, 7(7), 1075–1087 1085and metastable, anatase mineralogical orientation [001] with The decomposition data can be used to determine quantitatively the rate of photodecomposition. If a pseudo-first-order {100}.71 rate law with respect to the salicylic acid concentration is plotted versus reaction time, rate constants for the decompo- Photocatalysis sition of salicylic acid can be determined, Fig. 20(B).73,75–77 A standard application of TiO2 has been as a photocatalyst The slope of these lines provides the decomposition rate for the decomposition of organic molecules.72–76 This is a constant. The thin film catalyst has a rate constant of surface reaction that is thought to involve absorption of a UV 0.003 min-1 while the fibrillar catalyst shows an increased rate photon by TiO2 to produce an electron–hole pair which reacts constant of 0.03 min-1.This order of magnitude increase in with water to yield hydroxyl and superoxide radicals. These reaction rate is much smaller than the 315 times enhancement radicals can then oxidize the organic molecule. Template- predicted.This is not surprising because the thin film TiO2 synthesized TiO2 structures should increase the TiO2 surface undoubtedly has some degree of surface roughness resulting area and correspondingly increase the decomposition reaction in higher surface areas and higher decomposition rates than rates. For example, TiO2 fibrils can be synthesized within the predicted.Also, scanning electron microscopy (SEM) analysis pores of a 60 mm thick alumina membrane with 200 nm of the fibrillar TiO2, Fig. 5(C), shows that the fibres ‘lean’ diameter pores.26 The TiO2 fibril-containing membrane is against each other, possibly shading large portions of the attached to an epoxy surface, and the membrane is dissolved surface from the sunlight resulting in lower decomposition away.The calculated surface area of the immobilized fibrils rates than predicted. is 315 cm2 of TiO2 surface area per cm2 of planar geometric Single-crystal TiO2 fibrils can be fabricated via template area. This suggests that, in principle, an enhancement of 315 synthesis and sol–gel chemistry. Also, owing to the increased in the catalytic rate of organic decomposition on template- surface area of the TiO2 fibril array, the decomposition rate of synthesized TiO2 fibres is possible versus a thin film TiO2 an organic molecule increases.However, this prototype fibrillar catalyst. Through the use of tubular structures and/or template catalyst is not optimal. We are currently working on processes membranes with smaller diameter pores (with correspondingly to optimize the fibril arrays by varying the fibril diameter and higher pore densities and surface area) even larger increases in aspect ratio and the distance between the fibrils.We are also the rate would be predicted. exploring additional applications of these TiO2 nanofibres We have studied the decomposition of salicylic acid over including electrochemistry, battery research, photoelectro- time on an array of immobilized TiO2 fibres, Fig. 5(C), with chemistry and enzyme immobilization. exposure to sunlight, Fig. 20(A).26 The upper curve follows the concentration of salicylic acid for a solution containing no TiO2 catalyst, and no significant decomposition is observed. Conclusions The small increase in salicylic acid concentration has been The template method has become a very simple yet powerful ascribed to the evaporation of water during the exposure to process for the synthesis of nanomaterials. This article has sunlight.The middle curve follows salicylic acid decomposition described a host of chemistries that are now available for the on a thin film of TiO2, and the bottom curve shows a marked template synthesis of a wide variety of nanomaterials including increase in decomposition of salicylic acid for the template- metals, polymers, carbon, and semiconductors. Applications synthesized TiO2 fibres.have ranged from fundamental optical studies to ultra trace molecular detection to high surface area catalysis. What does the future hold for template synthesis? From a fundamental viewpoint, our group is interested in fabricating nanostructures with significantly smaller diameters in order to explore further the eects of size on the properties of materials. We are also developing new chemistries so that tubules and fibrils composed of an even larger variety of materials are available.New applications for template-synthesized nanomaterials are also being developed. We are exploring applications in photocatalysis, chemical analysis, bioencapsulation, biosensors, bioreactors, molecular separations, and electronic and electrooptical devices.Finally, it is clear that if practical applications are to be realized, methods for mass producing template-synthesized nanostructures will be required. This work would not have been possible without the eorts of a number of hardworking and highly motivated graduate students and postdocs.They include Vinod P. Menon, Zhihua Cai, Junting Lei, Wenbin Liang, Ranjani V. Parthasarathy, Charles J. Brumlik, Gabor L. Hornyak, Leon S. Van Dyke, Colby Foss, Matsuhiko Nishizawa, Reginald M. Penner, Charles J. Patrissi, Veronica M. Cepak, Brinda B. Lakshmi, Guangli Che and Kshama B. Jirage. Financial support from the National Science Foundation, the Oce of Naval Research and the Department of Energy is also gratefully acknowledged.We also wish to thank the Colorado State University Electron Microscopy Center. Fig. 20 A, Photodecomposition of salicylic acid in sunlight. Data for References no photocatalyst, the thin film TiO2 photocatalyst, and the fibrillar (200 nm) TiO2 photocatalyst are shown.B, First-order kinetics of the 1 G. A. Ozin, Adv. Mater., 1992, 4, 612. 2 Engineering a Small World: From Atomic Manipulation to photodecomposition of salicylic acid with both the thin film and fibrillar TiO2 photocatalyst. Microfabrication, special section of Science, 1991, 254, 1300–1342. 1086 J. Mater. Chem., 1997, 7(7), 1075–10873 C. R. Martin, Chem.Mater., 1996, 8, 1739.S. B. McCullen, J. B. Higgins and J. L. Schlenker, J. Am. Chem. Soc., 1992, 114, 10834. 4 R. V. Parthasarathy and C. R. Martin, J. Polym. Sci., 1996, 62, 875. 41 K. Douglas, G. Devaud and N. A. Clark, Science, 1992, 257, 642. 5 C. R. Martin and R. V. Parthasarathy, Adv. Mater., 1995, 7, 487. 42 T. D. Clark and M. R. Ghadiri, J. Am. Chem. Soc., 1995, 117, 12364. 6 R. Parthasarathy and C. R. Martin, Nature (L ondon), 1994, 369, 43 R. Schollhorn, Chem.Mater., 1996, 8, 1747. 298. 44 S. K. Chakarvarti and J. Vetter, J. Micromech. Microeng., 1993, 7 C. R. Martin, R. Parthasarathy and V. Menon, Synth. Met., 1993, 3, 57. 55–57, 1165. 45 M. J. Tierney and C. R. Martin, J. Phys. Chem., 1989, 93, 2878. 8 C. R. Martin, Adv.Mater., 1991, 3, 457. 46 C.J. Miller, C. A. Widrig, D. H. Charych and M. Majda, J. Phys. 9 Z. Cai, J. Lei, W. Liang, V. Menon and C. R. Martin, Chem.Mater., Chem., 1988, 92, 1928. 1991, 3, 960. 47 Electroless Plating: Fundamentals and Applications, ed. 10 L. S. Van Dyke and C. R. Martin, L angmuir, 1990, 6, 1123. G. O. Mallory and J. B. Hajdu, American Electroplaters and 11 W. Liang and C. R. Martin, J.Am. Chem. Soc., 1990, 112, 9666. Surface Finishers Society, Orlando, FL, 1990, ch. 1, pp. 1–55. 12 Z. Cai and C. R. Martin, J. Am. Chem. Soc., 1989, 111, 4138. 48 J. Lei, Z. Cai and C. R. Martin, Synth. Met., 1992, 46, 53. 13 R. M. Penner and C. R. Martin, J. Electrochem. Soc., 1986, 133, 49 R. V. Parthasarathy and C. R. Martin, Chem.Mater., 1994, 6, 1627. 2206. 50 T. Kyotani, L.Tsai and A. Tomita, Chem.Mater., 1996, 8, 2109. 14 J. C. Hulteen, V. P. Menon and C. R. Martin, J. Chem. Soc., 51 V. M. Cepak, J. C. Hulteen, G. Che, K. B. Jirage, B. B. Lakshmi, Faraday T rans., 1996, 92, 4029. E. R. Fisher and C. R. Martin, Chem. Mater., 1997, 9, 1065. 15 V. P. Menon and C. R. Martin, Anal. Chem., 1995, 67, 1920. 52 B. O’Regan and M. Gra�tzel, Nature (L ondon), 1991, 335, 737. 16 M. Nishizawa, V. P. Menon and C. R. Martin, Science, 1995, 53 M. Nishizawa, K. Muaki, S. Kuwabata and C. R. Martin, 268, 700. J. Electrochem. Soc., in press. 17 C. J. Brumlik, V. P. Menon and C. R. Martin, J.Mater. Res., 1994, 54 G. M. Whitesides and J. P. Mathius, Science, 1991, 254, 1312. 9, 1174. 55 C. K. Preston and M. Moskovits, J. Phys. Chem., 1993, 97, 8495. 18 C.A. Foss Jr., G. L. Hornyak, J. A. Stockert and C. R. Martin, 56 J. T. Masden and N. Giordino, Phys. Rev. B, 1987, 36, 4197. J. Phys. Chem., 1994, 98, 2963. 57 T. M. Whitney, J. S. Jiang, P. C. Searson and C. L. Chien, Science, 19 C. A. Foss Jr., G. L. Hornyak, J. A. Stockert and C. R. Martin, 1993, 261, 1316. Adv. Mater., 1993, 5, 135. 58 G. E. Possin, Rev. Sci. Instrum., 1970, 41, 772. 20 C. A. Foss Jr., G. L. Hornyak, J. A. Stockert and C. R. Martin, 59 T. Asada, Jpn. Pat., 1960, 310401. J. Phys. Chem., 1992, 96, 7497. 60 R. D. Patel, M. G. Takwale, V. K. Nagar and V. G. Bhide, T hin 21 C. J. Brumlik, C. R. Martin and K. Tokuda, Anal. Chem., 1992, Solid Films, 1984, 115, 169. 64, 1201. 61 S. Kawai, in Symposium on Electrochemical T echnology in 22 C. J. Brumlik and C.R. Martin, J. Am. Chem. Soc., 1991, 113, 3174. Electronics, Electrochemical Society, Pennington, NJ, 1987, p. 389. 23 R. M. Penner and C. R. Martin, Anal. Chem., 1987, 59, 2625. 62 J. C. van de Hulst, L ight Scattering by Small Particles, Dover, New 24 G. L. Hornyak and C. R. Martin, J. Phys. Chem., 1997, 101, 1548. York, 1981, pp. 397–400. 25 G. L. Hornyak, C. J. Patrissi and C. R. Martin, T hin Solid Films, 63 S. M. Drew and R. M. Wrightman, J. Electroanal Chem., 1991, submitted. 317, 117. 26 B. B. Lakshmi, P. K. Dorhout and C. R. Martin, Chem. Mater., 64 M. F. Bento, M. J. Medeiros, M. L. Montenegro, C. Beriot and 1997, 9, 857. D. Pletcher, J. Electroanal. Chem., 1993, 345, 273. 27 J. D. Klein, R. D. I. Herrick, D. Palmer, M. J. Sailor, C. J. Brunlik 65 A. Russell, K. Repka, T. Dibble, J. Ghoroghian, J. J. Smith, and C. R. Martin, Chem. Mater., 1993, 5, 902. M. Fleischmann, C. H. Pitt and S. Pons, Anal. Chem., 1986, 58, 28 R. V. Parthasarathy, K. L. N. Phani and C. R. Martin, Adv. Mater., 2961. 1995, 7, 896. 66 A. M. Bond, T. L. E. Henderson, D. R. Mann, W. Thormann and 29 J. C. Hulteen, X. C. Chen and C. R. Martin, in preparation. C. G. Zoski, Anal. Chem., 1988, 60, 1878. 30 G. Che, K. B. Jirage, E. R. Fisher and C. R. Martin, J. Electrochem. 67 N. Oyama, T. Ohsaka, N. Yamamoto, J. Matsui and Soc., submitted. O. J. Hatosaki, J. Electroanal. Chem., 1989, 265, 297. 31 C.-G. Wu and T. Bein, Science, 1994, 264, 1757. 68 Z. J. Karpinski and R. A. Osteryoung, J. Electroanal. Chem., 1993, 32 C. R. Martin, Handbook of Conductive Polymers, in press. 349, 285. 33 C. R. Martin, Acc. Chem. Res., 1995, 28, 61. 69 J. Clavilier and C. N. V. Huong, J. Electroanal. Chem., 1977, 80, 34 R. L. Fleisher, P. B. Price and R. M. Walker, Nuclear T racks in 101. Solids, University of California Press, Berkeley, CA, 1975. 70 B. Enright and D. Fitzmaurice, J. Phys. Chem., 1986, 100, 1027. 35 Poretics Corporation, Livermore, CA, Product Guide, 1995. 71 J. D. Dana (rewritten by C. Polache, H. Berman and C. Frondel), 36 M. Quinten and U. Kreibig, Surf. Sci., 1986, 172, 557. T he System of Mineralogy, John Wiley, NY, 1955. 37 A. Despic and V. P. Parkhutik, in Modern Aspects of 72 A. Fujishima and K. Honda, Nature (L ondon), 1972, 37, 238. Electrochemistry, ed. J. O’M. Bockris, R. E. White and 73 R. W. Matthew, J. Phys. Chem., 1987, 91, 3328. B. E. Conway, Plenum Press, New York, 1989, vol. 20, ch. 6. 74 B. Kraeutler and A. J. Bard, J. Am. Chem. Soc., 1978, 100, 5985. 38 D. AlMawiawi, N. Coombs and M. Moskovits, J. Appl. Phys., 75 D. F. Ollis, C. Hsiao, L. Budiman and C. Lee, J. Catal., 1984, 88, 89. 1991, 70, 4421. 76 K. Okamoto, Y. Yamamoto, H. Tanaka, M. Tanaka and A. Itaya, Bull. Chem. Soc. Jpn., 1985, 58, 2015. 39 R. J. Tonucci, B. L. Justus, A. J. Campillo and C. E. Ford, Science, 77 R. W. Mathews, J. Catal., 1987, 97, 565. 1992, 258, 783. 40 J. S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, K. D. Schmitt, C. T.-W. Chu, D. H. Olson, E. W. Sheppard, Paper 7/00027H; Received 2nd January, 1997 J. Mater. Chem., 1997, 7(7), 1075–1087 1087