Journal of Colloid and Interface Science 316 (2007) 160–167www.elsevier.com/locate/jcis

Sol–gel synthesis and morphological control of nanocrystalline TiO2 viaurea treatment

Li-Heng Kao, Tzu-Chien Hsu ∗, Hong-Yang Lu

Institute of Materials Science and Engineering, National Sun Yat-Sen University, Kaohsiung 80424, Taiwan

Received 5 January 2007; accepted 18 July 2007

Available online 1 August 2007

Abstract

Nanocrystalline TiO2 rods and hollow tubes with an engraved pattern on the surface have been prepared by a novel anionic template-assistedsol–gel synthesis via urea treatment and under hydrothermal condition. X-ray diffractometry (XRD) results indicate that these nanocrystallinesconsist predominantly of anatase TiO2, with minor amounts of rutile and brookite. Scanning and transmission electron microscopy (SEM andTEM) analyses reveal these rods and hollow tubes may result from the aggregates of nanorods of ∼10 nm in diameter. The crystallographic facetingfound from TEM further reveals the polymorphic nature of the nanocrystalline TiO2 thus prepared. A “reverse micelle” formation mechanismtaking into account the hydrothermal temperature, the pH effect of the sol–gel system, the isoelectric point, the formation of micelles, and theelectrostatic interaction between the anionic surfactant and the growing TiO2 particulates is proposed to illustrate the competition between thephysical micelle assembly of the ionic surfactants and the chemical hydrolysis and condensation reactions of the Ti precursors.© 2007 Elsevier Inc. All rights reserved.

Keywords: Nanocrystalline TiO2; Anionic template-assisted; Sol–gel; Urea; Hydrothermal; Reverse micelle; Isoelectric point; pH effect

1. Introduction

Recently, nanocrystalline titania (TiO2) has attracted muchattention because of its potential applications in environmen-tal purification [1,2], catalysis and photocatalysis [3–5], gassensors [6], dielectric ceramics, pigments, and high efficiencydye-sensitized solar cells [7–10]. These nanocrystalline TiO2

have been prepared with various forms covering from nanopar-ticles [11,12], thin films [13,14], nanotubes [15] to mesoporousstructure [16] and others [17–20]. In particular, nanoscale TiO2

tubes and wires have attracted considerable attention becausethey have large surface area and high photocatalytic activity[20,21]. Nanotubular TiO2 materials with diameters ∼10 nmhave received intensive attention due to its potential applica-tions in the fields which are traditionally dominated by carbonnanotubes. On the other hand, hollow fibers of TiO2 with largerouter diameter of 150–600 nm and aspect ratio of ∼30 preparedwith an amphiphilic supramolecular organogelator as the struc-

* Corresponding author. Fax: +886 7 5250179.E-mail address: tjhsu@facmail.nsysu.edu.tw (T.-C. Hsu).

0021-9797/$ – see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2007.07.062

ture directing agent has been reported. By self-assembling intoa fibrous structure, this agent acted as a template later in the sol–gel polymerization of the titanium precursors to become hol-low tubular titania [22]. On the other hand, use of membranesin template synthesis such as the porous alumina membranesand the nanoporous track-etch polymeric membranes in thepresence of certain additives has also received intensive atten-tion [23]. In this approach, the pores within these nanoporousmembranes play the same role as the templated surfactants incontrolling nanostructures of the desired materials.

Various crystalline phases have been identified for thesereported nanocrystalline titania. Among the major five poly-morphs of titania, the rutile phase is the most stable one at alltemperatures under ambient pressure; the anatase phase is meta-stable; the brookite phase is the least stable one; whereas TiO2-II and TiO2-III can be derived from the anatase or brookitephase under pressure. TiO2-B is the monoclinic form of tita-nium dioxide. This mineral can be found in weathering rimson tektite and perovskite and appears as lamellae in anatasefrom hydrothermal veins; its density is much lower than thatof the other three polymorphs. Polymorphic transformations of

L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167 161

anatase to rutile and of brookite to rutile do not take place re-versibly [24].

It is generally claimed that the sol–gel synthesis of ceramicoxides offers advantages such as high purity, good homogene-ity, low processing temperature, and also the possibility of mak-ing new nanocrystalline solids outside the range of normal glassformation [25,26]. Factors affecting the ultimate properties ofceramic materials in a typical sol–gel process may include thetype of solvent, the reactivity of metal precursor, pH of the re-action medium, and reaction temperature, among many factors[27,28]. By controlling these material or processing parameters,different surface chemistry and microstructure of ceramic ox-ides can be obtained. However, the precipitates derived by sol–gel synthesis are typically amorphous; this requires a high tem-perature heat treatment to induce the crystallization during thepost-gel stage. On the other hand, hydrothermal synthesis hasbeen carried out at a relatively low temperature (<250 ◦C) toproduce sufficiently crystalline ceramic solids as compared tocalcination [26,29,30]. In the hydrothermal synthesis, morphol-ogy, composition, structure, grain size, and crystalline phasecan be controlled by changing the hydrothermal parameterssuch as reaction temperature and pressure, pH values, sol com-position, type of the solvent and additive, and the aging time[11,31,32]. Among these processing parameters, the hydrother-mal temperature is regarded as the most crucial one since itcontrols the steam pressure of this closed aqueous system.

It has been reported that the addition of liquid ammonia inthe sol–gel synthesis of some inorganic oxides may result inprecipitation. Yada et al. [33] obtained a hexagonal structurealumina by a homogeneous precipitation method using urea,from which the ammonia was generated at an elevated tem-perature. Banerjee et al. [34] employed urea as the hydrolyz-ing agent to control the hydrolysis rate in the preparation of ahexagonal mesoporous nickel oxide. To our best understand-ing, the urea approach has not been applied to the preparationof TiO2 nanotubes. However, a different surfactant-mediatedtemplate laurylamine hydrochloride was adopted by Peng etal. [35], which resulted in a TiO2 tubules with mesostructuralwalls, the outer diameter and the wall thickness of the tita-nia microtubules being 2–8 and 0.2–2 µm. We report in thisstudy a novel anionic template-assisted sol–gel approach viaurea treatment and under hydrothermal condition for synthe-sizing the nanocrystalline TiO2 rods and hollow tubes, con-sisting of aggregates of nanorods of ∼10 nm in diameter. Thenanocrystalline TiO2 thus prepared appears to have an engravedsurface morphology and consists mainly of the anatase and mi-nor amount of rutile and brookite. It is also demonstrated thatmicrostructure and phase content of the nanocrystalline TiO2can be tailor-made by properly manipulating those parame-ters during the combined sol–gel and hydrothermal process-ing.

2. Experimental

Titanium tetrachloride (TiCl4; 99.9%, Acros) was used asthe titanium source; sodium dodecyl sulfate (SDS, CH3(CH2)11OSO3Na; 99%, Aldrich) and urea (H2NCONH2; 98%, Acros)

were used as received. Urea was employed as a hydrolysis agentto control the hydrolysis rate, it was also used in this work tocontrol the pH value of sol–gel system by a hydrolysis reaction(Eq. (1)) at a temperature greater than 80 ◦C, from which theammonia could be generated [34]:

(NH2)2CO + 3H2O → 2NH+4 + 2OH−+ CO2. (1)

A specific amount of SDS was first dissolved in deionized(DI) water and mixed with urea. TiCl4 was then added dropwiseto this highly viscous aqueous solution with rigorous stirring.Caution on the quick release of HCl should be taken when TiCl4was added, due primarily to its extreme sensitivity to the mois-ture. TiCl4, SDS, urea, and DI water were mixed in a molarratio of 1:2:30:60. The mixture was stirred at 40 ◦C for 1 h toyield a transparent solution. The solution was then heated un-til the pH had increased to 2.2; it was immediately transferredinto a sealed Teflon-coated autoclave and was further heated atan elevated hydrothermal temperature for 48 h. The pH valueof the final white slurry thus prepared was 5.6 (90 ◦C), 8.52(120 ◦C), 9.12 (150 ◦C), and 10.6 (180 ◦C). Since the autoclaveis a closed system under high temperature and high pressure, itis not feasible to dynamically measure the pH values in situ dur-ing the sol–gel process. Instead of estimating the dynamics ofthe pH during hydrothermal treatment, an alternative approachwas adopted. The sol–gel reaction was terminated after 24 h andthe pH value was measured. It was found for all the four sam-ples the pH values at 24 h had reached a value as high as that ofat 48 h, indicating that effect of ammonia on the microstructurewas mostly profound in the first half period of the hydrothermaltreatment.

The measured autogenous pressures inside the autoclave un-der different hydrothermal temperatures were 0.92 bar (90 ◦C),2.1 bar (120 ◦C), 5.13 bar (150 ◦C), and 10.97 bar (180 ◦C),which are very close to the values estimated from ASMESteam Tables (0.70 bar (90 ◦C), 1.99 bar (120 ◦C), 4.76 bar(150 ◦C), and 10.03 bar (180 ◦C)) [36]. The product was re-peatedly washed with DI water to remove the organic moieties,filtered with a 0.45 µm filter paper, and then dried in air. Itwas then calcined at 800 ◦C (at a heating rate of 4 ◦C/min) for4 h in air to further remove the residual organic moieties. Thecalcined product was then re-washed with DI water to removewater-soluble impurities such as sodium sulfate formed duringcalcination.

Crystalline phases were determined by the X-ray diffractom-etry (XRD, Siemens D-5000, Karshrule, Germany) with CuKα

radiation and Ni filter operating at 40 kV/30 mA. Microstruc-ture was analyzed by the field-emission scanning electron mi-croscopy (FE-SEM, JEOL™ 6330, Tokyo, Japan) operating at20 kV. Sample for SEM was first mounted onto a carbon adhe-sive pad which was attached to an aluminum stub, it was thenair-dried and gold-coated (Pelco SC-6 sputter-coater). Samplesultrasonicated and filtered on holey carbon grids were exam-ined by the transmission electron microscopy (TEM, JEOL™AEM 3010, Tokyo, Japan) operating at 200 kV.

162 L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167

3. Results and discussion

3.1. Phase identification

The XRD patterns of the as-prepared nanocrystalline TiO2samples synthesized under different hydrothermal conditionsshown in Fig. 1 reveal diffraction peaks of (101), (004), (200),and (211), which are characteristic of the anatase phase. Therelative broad peaks suggest low crystallinity among the foursamples measured; the crystallinity increases with increasinghydrothermal temperatures. The result is quite different fromthe traditional (or typical) sol–gel process in which only amor-phous phase can be obtained from the precipitates derived bysol–gel process before calcinations; and further higher temper-ature heat treatment is normally required to induce crystalliza-tion. Thus, the hydrothermal treatment may be regarded as analternative to calcination for promoting the crystallization [11].A minor diffraction peak assigned as the rutile (110) found insample SDS-90 can also discernibly be identified from Fig. 1.

XRD patterns of calcined nanocrystalline TiO2 prepared un-der different hydrothermal conditions are shown in Fig. 2. Dif-fraction peaks of anatase (tetragonal, I41/amd (No. 141)), ru-tile (tetragonal, P 4/mnm (No. 136)) and brookite (orthorhom-bic, Pcab (No. 61)) (corresponding to JCPDS No. 21-1272,21-1276 and 29-1360, respectively) are identified unambigu-ously. The fact that no discernible peak was identified in thelow range of 2θ = 1–10◦ has ruled out the existence of theamorphous mesoporous structure [37,38]. This is contradictoryto what was expected in our initial design of experiments andis also inconsistent with those reported in the literature [34,39].An amorphous mesoporous structure would usually be formedin the sol–gel-derived ceramic oxides prepared by the assistanceof ionic template without any hydrothermal treatment. It can bepostulated that this hydrothermal treatment in this study has a

profound effect on the ultimate microstructure of the nanocrys-talline TiO2.

Summarized in Table 1 are the phase contents determinedfrom the integrated XRD peak intensities of anatase (101), ru-tile (110), and brookite (121) by a numerical deconvolutionmethod [40]; also listed are the crystallite sizes calculated byScherrer equation. It is obvious that the crystallite sizes un-der various hydrothermal conditions change considerably; butmost are within the range of 30–50 nm, except for SDS-150(∼13 nm). Table 1 also indicates that the meta-stable anatase isthe dominant phase for all the samples; other minor phases in-clude the more stable rutile (for the low-temperature samples)and the least stable brookite (for the high-temperature samples).Thus, increasing hydrothermal temperature seems to encour-age a phase transformation from rutile to brookite, while phasecontent of anatase remains roughly the same. The results areconsistent with the reported findings [25]; it was argued thatthe dissolution of an acidified titania sol at low temperatureswas slow, resulting in a slow crystallization. This crystallizationprocess would be governed by the thermodynamics, not the ki-netics; therefore, the most stable rutile phase should be formedat low hydrothermal temperatures. On the other hand, the meta-stable anatase or brookite should be the favorable phases underhigher hydrothermal temperatures, due kinetically to a fasterdissolution and a more rapid precipitation.

3.2. Morphological characterization

Under a lower magnification, FE-SEM image for SDS-90shown in Fig. 3a reveals a long fibrous structure with an as-pect ratio ∼30, a value much higher than those reported datawhich are typically less than 10 [18,19]; an engraved patternon the surface of SDS-90 sample is also discernible in Fig. 3b,not found in the other samples prepared at higher hydrothermal

Fig. 1. XRD patterns of nanocrystalline TiO2 rods and hollow tubes prepared at four hydrothermal temperatures indicated; samples measured before calcination.Patterns were compared to the JCPDS data (F: rutile); “s” and “u” denote peaks corresponding to SDS and urea, respectively.

L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167 163

Fig. 2. XRD patterns of nanocrystalline TiO2 rods and hollow tubes prepared at four hydrothermal temperatures indicated; samples measured after calcination.Patterns were compared to the JCPDS data (F: rutile; ": brookite).

Table 1Summary of phase contents and crystallite sizes

Material Anatase Rutile Brookite

Crystallite size (nm)a Content (%)b Crystallite size (nm)a Content (%)b Crystallite size (nm)a Content (%)b

SDS-90 44.0 88.7 46.6 11.3 – –SDS-120 47.0 87.8 42.3 3.4 38.5 8.7SDS-150 13.4 74.4 – – 12.6 25.6SDS-180 49.9 88.5 – – 32.0 11.5

a Calculated by Scherrer equation.b Calculated using the equation in Ref. [38].

Fig. 3. FE-SEM images of nanocrystalline TiO2 rods and hollow tubes (a) and the engraved surface (b) for SDS-90.

164 L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167

Fig. 4. FE-SEM images of nanocrystalline TiO2 rods, hollow tubes, and platelets prepared at different hydrothermal temperatures: (a) SDS-90, (b) SDS-120,(c) SDS-150, and (d) SDS-180.

temperatures. Note that SDS-90 maintained a pH value underthe isoelectric point (IEP) of TiO2 (∼5.8) during its sol–gelsynthesis. A closer examination on SDS-90 is given in Fig. 4a,from which a well defined rod-like or hollow-tube-like mor-phology can be identified with the rod diameter of 50–300 nmand the hollow tube diameter of 200–600 nm (all have an aspectratio ∼30); note that some nanorods of ∼10 nm and with thesame aspect ratio in diameter can also be discernibly identified.The morphology found here is very similar to those reportedTiO2 hollow fibers which adopted an amphiphilic compoundand without hydrothermal treatment [22,40,41]. In our case, ananionic surfactant SDS was adopted as the liquid crystal tem-plate; then the base-catalyzed sol–gel synthesis was followedunder hydrothermal conditions. Due to the high temperatureand high pressure in the autoclave under the hydrothermal con-dition, a completely different reaction mechanism and forma-tion sequence of liquid crystal template and sol–gel synthesismay be rationalized. The steam is generated under high temper-atures to produce a hydrostatic pressure which in turn imposes

a profound effect on the ultimate microstructure of the ceramicoxides thus prepared. This autogenous hydrostatic pressure canbe as high as 11 bars under a hydrothermal temperature of180 ◦C (exact values can be found in Section 2). Therefore, itcan be said that in this aqueous system in the sealed autoclave,the temperature has much higher impact on the sol–gel reactionrate, the morphology, as well as the reaction mechanism.

When the hydrothermal temperature is increased, it is ob-served from Fig. 4 that the rods and hollow tubes are graduallytransformed into a tabular or platelet structure, possibly dueto the autogenous pressure generated under hydrothermal con-ditions. Note that the pH value for SDS-120, SDS-150, andSDS-180 during their sol–gel processing was above the IEPof TiO2. These tabular or platelet structures remain almost thesame physical size as the rods and tubes (same aspect ratio).The straight and parallel stripes (referred to Figs. 4b and 4c) onthe surface of the rods and hollow tubes may result from theaggregates of the smaller and finer nanorods of ∼10 nm in di-ameter [31,32].

L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167 165

Fig. 5. TEM images of nanocrystalline TiO2 hollow tubes and platelets prepared at different hydrothermal temperatures: (a) SDS-90, (b) SDS-120, (c) SDS-150,and (d) SDS-180.

Shown in Fig. 5a is a TEM image of hollow tube with anouter diameter of ∼600 nm and an inner diameter of ∼200 nmfor SDS-90, in addition to the rods found from SEM (Fig. 4a).The yield of the TiO2 nanotubes and nanorods synthesized fromtheir precursors is estimated from stoichiometric calculation tobe 81%; meanwhile the ratio between the hollow tubes to therods is estimated from the SEM image to be about 1/5. Theclear parallel and straight stripes appeared on the surface of therod, tube, and platelet in Figs. 5a–5c are in fact the crystallo-graphic faceting that occurs due to anisotropic (solid-to-vapor)surface energy. This faceting exists because atoms are packedin different density along crystal planes, e.g., fcc close-packedplane on {111} along 〈110〉. Likewise for TiO2, regardless ofits polymorphs (the anatase is the dominant phase from XRDresults in this study), would have different atomic packing den-sity, and so faceting appears. Crystals with different structure

usually appear faceted. On the other hand, these stripes may bethe aggregates of nanorods as suggested from SEM observa-tions.

With increasing hydrothermal temperatures, the rod or thetube gradually transform into a tabular or platelet structure(Figs. 5c, 5d). These TEM images show consistent results asthe SEM in Fig. 3. When proper chemicals are inserted inside,these types of TiO2 rods and hollow tubes may possess potentialapplications due to their unusual catalytic, electric, and opticalproperties.

3.3. Formation mechanism

In sol–gel synthesis using transition metal chlorides as theprecursors, hydrolysis and condensation reactions take placevery rapidly [37,42,43]. In preparing TiO2 using SDS as the an-

166 L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167

Fig. 6. Formation of nanocrystalline TiO2 rods and hallow tubes is favorable at lower temperatures when the pH value is below IEP of TiO2 (path I); while highertemperatures encourage the formation of a tabular structure of TiO2 (path II).

ionic surfactant under basic conditions with urea, the formationof SDS micelles as the templates may have to compete with thehydrolysis and condensation of Ti precursors. At a hydrother-mal temperature of 90 ◦C in which the basic urea has not beenable to completely release its hydroxyl ions, the reaction systemappears to have a pH value below IEP of TiO2, thus allowinga positively charged TiO2 surface due to rapid hydrolysis andcondensation. The negatively charged SDS micelles originallypresent in the system are now forced to break down and areattached to the positive TiO2 surface, forming the so-called “re-versed micelle” (path I in Fig. 6). These SDS-coated TiO2 parti-cles then aggregate into stacked TiO2 nanorods; further calcina-tion at higher temperatures results in the nanocrystalline TiO2rods and hollow tubes. Another possible mechanism of chemi-cally induced self-transformation of amorphous solid particlesto account for the fabrication of hollow inorganic microsphereshas been mentioned [17]. This approach was based on mor-phologically confined processes of Oswald ripening where nosacrificial surfactant template was involved. The profound ef-fect found in this study on the microstructure due to the pres-ence of urea, from which ammonia is chemically produced,can also be regarded as a result of chemically induced phasetransformation. On the other hand, the “reverse micelle” mech-anism was associated with certain anionic surfactant-templatedsol–gel processes such as dodecyl sulfate (SDS) and sodiumbis(2-ethylhexyl) sulfosuccinate; note that no urea was involvedin the morphological control of TiO2 materials in these studiesand that the particle sizes thus prepared are within the approxi-mately same range as in this study [44–46].

It can now be rationalized that the engraved pattern in Fig. 3bmay result from the complete destruction of the anionic organicsurfactants SDS on the surface of the nanocrystalline TiO2 dur-ing calcination. Initially, SDS was designed to self-assembleinto a template upon which sol–gel reaction was expected to

take place subsequently, an approach similar to the process ofthe commercially available MCM-41 silica. Obviously, the ob-served morphology is not consistent with the generally acceptedformation mechanism of the typical liquid crystal template-assisted sol–gel synthesis of ceramic powders. The anionic sur-factants appear to have altered the reaction kinetics of hydroly-sis and condensation of Ti precursors.

On the other hand, at higher hydrothermal temperatures, acomplete urea reaction encourages a basic sol–gel reaction sys-tem; this allows the surface of TiO2 to be negatively chargedbecause the pH of the sol–gel system is now above the IEP ofTiO2. Therefore, the also negatively charged SDS micelles areincompatible to the TiO2 particles and are excluded from thesol–gel processing of TiO2 upon further condensation (path IIin Fig. 6). Without the constraint of the reverse micelles by therejected SDS molecules, formation of rods and hollow tubesbecomes unfavorable and a plate-like morphology with a roughsurface may form after the hydrothermal treatment. Further cal-cination results in the nanocrystalline TiO2 platelets as shownin Figs. 4d and 5d.

4. Conclusions

A novel approach adopting the anionic template-assistedsol–gel synthesis via urea treatment and under hydrothermalprocessing to synthesize the nanocrystalline TiO2 rods and hol-low tubes has been demonstrated in this study. Similar hol-low tubes with same range of diameters have been reportedbut using a completely different amphiphilic surfactant. In-stead of an amorphous as is reported in most sol–gel-derivedceramic oxides before calcinations, the as-prepared nanocrys-talline TiO2 has an anatase phase, although with a low degreeof crystallinity. The calcined nanocrystalline TiO2 is found tobe mostly anatase, with minor amount of rutile and brookite.

L.-H. Kao et al. / Journal of Colloid and Interface Science 316 (2007) 160–167 167

Low hydrothermal temperatures are found to favor the forma-tion of TiO2 rods and hollow tubes, both may result from theaggregates of TiO2 nanorods. The pH value of the reactingsol–gel system plays a crucial role in determining the chargeon the surface of the reacting TiO2 particles. The pH valueof the sol–gel system under lower hydrothermal temperature isfound to below IEP of TiO2 such that the surface charge of theanatase TiO2 particles becomes positive. This forces the SDSmicelles to form a coating on the TiO2 particles due to elec-trostatic attraction. This “reverse micelle” mechanism results inthe formation of the nanocrystalline TiO2 rods or hollow tubes.Upon calcinations at higher temperatures as the surfactant SDSis incinerated, an engraved TiO2 surface is found. Since higherhydrothermal temperatures encourage the urea reaction to re-lease the hydroxyl ions, the sol–gel system becomes more basicand its pH value is far exceeding the IEP of TiO2. The neg-atively charged SDS micelles are thus rejected from the alsonegatively charged TiO2 particle, resulting in a tabular TiO2structure. These types of TiO2 rods and hollow tubes, whenproper chemicals are inserted inside, may possess potential ap-plications due to their unusual catalytic, electric, and opticalproperties.

Acknowledgments

This work was funded by the National Science Council ofTaiwan through contract NSC-93-2216-E-110-012. One of theauthors (L.H.K.) wishes to thank C.L. Chang for his help withFE-SEM, and Dr. Y.C. Wu of National Taipei University ofTechnology for her help with TEM.

References

[1] W. Choi, M.R. Hoffmann, Environ. Sci. Technol. 31 (1997) 89.[2] Q. Dai, Z. Zhang, N. He, P. Li, C. Yuan, Mater. Sci. Eng. C 8–9 (1999)

417.[3] B. Sun, P.G. Smirniotis, Catal. Today 88 (2003) 49.[4] G. Dagan, M. Tomkiewicz, J. Phys. Chem. 97 (1993) 12651.[5] X. Hong, Z. Wang, W. Cai, F. Lu, J. Zhang, Y. Yang, N. Ma, Y. Liu, Chem.

Mater. 17 (2005) 1548.[6] N. Savage, B. Chwieroth, A. Ginwalla, B.R. Patton, S.A. Akbar, P. Dutta,

Sens. Actuators B 79 (2001) 17.[7] U. Bach, D. Lupo, P. Comte, J.E. Moser, F. Weissortel, J. Salbeck, H. Spre-

itzer, M. Gratzel, Nature 395 (1998) 583.[8] M. Adachi, Y. Murata, J. Takao, J. Jiu, M. Sakamoto, F. Wang, J. Am.

Chem. Soc. 126 (2004) 14943.[9] H. Luo, T. Takata, Y. Lee, J. Zhao, K. Domen, Y. Yan, Chem. Mater. 16

(2004) 846.

[10] S. Nakade, Y. Saito, W. Kubo, T. Kitamura, Y. Wada, S. Yanagida, J. Phys.Chem. B 107 (2003) 8607.

[11] C.C. Wang, J.Y. Ying, Chem. Mater. 11 (1999) 3113.[12] H. Yu, J. Yu, B. Cheng, J. Mol. Catal. A 253 (2006) 99.[13] N. Negishi, K. Takeuchi, T. Ibusuki, J. Mater. Sci. Lett. 18 (1999) 515.[14] D. Grosso, J. Galo, A.A. Soller-Illia, Adv. Mater. 13 (2001) 1085.[15] J. Yu, H. Yu, Mater. Chem. Phys. 100 (2006) 507.[16] M. Anderson, H. Birkedal, N.R. Franklin, T. Ostomel, S. Boettcher,

A.E.C. Palmqvist, G.D. Stucky, Chem. Mater. 17 (2005) 1409.[17] J. Yu, H. Guo, S.A. Davis, S. Mann, Adv. Funct. Mater. 16 (2006) 2035.[18] P.D. Cozzoli, A. Korrowski, H. Weller, J. Am. Chem. Soc. 125 (2003)

14539.[19] T. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino, K. Niihara, Langmuir 14

(1998) 3160.[20] B.B. Lakshmi, C.J. Patrissi, C.R. Martin, Chem. Mater. 9 (1997) 2544.[21] E.K. Kyriaki, E.V. Xenophon, J. Phys. Chem. 97 (1993) 1184.[22] S. Kobayashi, K. Hanabusa, N. Hamasaki, M. Kimura, H. Shirai, S.

Shinkai, Chem. Mater. 12 (2000) 1523.[23] J.C. Hulteen, C.R. Martin, J. Mater. Chem. 7 (1997) 1075.[24] J.L. Murray, H.A. Wriedt, Phase Diagrams of Binary Titanium Alloys,

ASM International, Metals Park, OH, 1987.[25] B.L. Bischoff, M.A. Anderson, Chem. Mater. 7 (1995) 1772.[26] C.C. Wang, J.Y. Ying, Chem. Mater. 11 (1999) 3113.[27] C.J. Brinker, G.W. Scherer, Sol–Gel Science, Academic Press, San Diego,

1990.[28] L.C. Klein, Sol–Gel Technology for Thin Films, Fibers, Preforms, Elec-

tronics, and Specialty Shapes, Noyes Publications, New Jersey, 1987.[29] M. Wu, J. Long, A. Huang, Y. Luo, Langmuir 15 (1999) 8822.[30] K. Yanagisawa, J. Ovenstone, J. Phys. Chem. B 103 (1999) 7781.[31] Q. Zhang, L. Gao, Langmuir 19 (2003) 967.[32] L. Kavan, M. Kalbac, M. Zukalova, I. Exnar, V. Lorenzen, R. Nesper,

M. Graetzel, Chem. Mater. 16 (2004) 477.[33] M. Yada, M. Machida, T. Kijima, Chem. Commun. (1996) 769.[34] S. Banerjee, A. Santhanam, A. Dhathathreyan, P.M. Rao, Langmuir 19

(2003) 5522.[35] T. Peng, A. Hasegawa, J. Qie, K. Hirao, Chem. Mater. 15 (2003) 2011.[36] J.M. Smith, H.C. Van Ness, Introduction to Chemical Engineering Ther-

modynamics, McGraw–Hill International, New York, 1987.[37] D.M. Antonelli, J.Y. Ying, Angew. Chem. Int. Ed. Engl. 34 (1995) 2014.[38] P. Yang, D. Zhao, D.I. Margolese, B.F. Chmelka, G.D. Stucky, Nature 396

(1998) 152.[39] D.M. Antonelli, Micropor. Mesopor. Mater. 30 (1999) 315.[40] H. Zhang, J.F. Banfield, J. Phys. Chem. 104 (2000) 3481.[41] P.C.A. Alberius, K.L. Frindell, R.C. Hayward, E.J. Kramer, G.D. Stucky,

B.F. Chmelka, Chem. Mater. 14 (2002) 3284.[42] E.L. Crepaldi, G.J.A.A. Soller-Illia, D. Grosso, F. Cagnol, F. Ribot,

C. Sanchez, J. Am. Chem. Soc. 125 (2003) 9770.[43] P. Yang, D. Zhao, D.I. Margolese, B.F. Chmelka, G.D. Stucky, Chem.

Mater. 11 (1999) 2813.[44] S.H. Kang, I. Hirasawa, W.-S. Kim, C.K. Choi, J. Colloid Interface

Sci. 288 (2005) 496.[45] J.D. Hopwood, S. Mann, Chem. Mater. 9 (1997) 1819.[46] M. Li, S. Mann, Adv. Funct. Mater. 12 (2002) 773.