A

wantca©

K

1

rbepMwsrhf

pitee3t

0d

Journal of Materials Processing Technology 192–193 (2007) 561–566

Hydrothermal synthesis and characterization of mesoporousSnO2/SnO2–SiO2 on neutral template

Jie Zhu a,∗, B.Y. Tay b,1, Jan Ma a,2

a School of Materials Science & Engineering, NanYang Technological University, Blk N4.1, NanYang Avenue, Singapore 639798, Singaporeb Singapore Institute of Manufacturing Technology (SIMTech, A*STAR), 71 NanYang Drive, Singapore 638075, Singapore

bstract

Mesoporous SnO2/SnO2–SiO2 with disordered pores and high thermal stability up to 600 ◦C were fabricated using hydrothermal treatmentith/without impurity oxide (SiO2) doping. Neutral surfactant (tetradecylamine, TDA) was selected as the structure directing agent based onsol–gel-derived process. The results showed that with careful control of the synthesis condition, mesoporous SnO2–SiO2 with small SnO2

˚ 2 3

anocrystallites (∼25 A), large specific surface area (211–340 m /g), and pore volume (0.23–0.3 cm /g) could be obtained after the removal ofemplate at temperatures of 400–600 ◦C. The effects of synthesis parameters such as aging time, hydrothermal treatment, Si/Sn molar ratio andalcination temperature were characterized by thermogravimetric analysis, wide/small-angle X-ray scattering (SAXS), HRTEM, and BET gas (N2)dsorption/desorption measurements. 2007 Elsevier B.V. All rights reserved.

a; Hy

ssDrstmf[lmuoe4swo

eywords: Mesoporous; Thermal stability; Surfactant template; Tin oxide/silic

. Introduction

Since the first report on well-ordered siliceous M41S mate-ials in 1992 [1,2], enormous amount of investigations haveeen devoted to the development of mesoporous materials withxpanded chemical compositions, e.g. metal oxides, sulfides andhosphates. Compared with the original silica/aluminosilicate41S materials, those non-siliceous mesostructured materialsere expected to give more functional applications in cataly-

is, photo-optoelectronic, and electrochemical devices. Severaleviews on the progress of non-siliceous mesoporous materialsave been published, dealing with the synthesis methodologies,ormation mechanisms, and applications [3–7].

A wide range of mesoporous metal oxides, sulfides andhosphates have been synthesized based on organic templat-ng technique. However, most of these materials suffer fromhe poor thermal stability against the removal of templates atlevated temperatures [4]. Tin (IV) oxide is a typical transpar-

nt n-type semiconducting material with wide band gap about.6 eV [8]. Due to its unique electrical and optical properties,in oxide has been widely used in many advanced applications

∗ Corresponding author. Tel.: +65 6790 4590; fax: +65 6790 9081.E-mail address: x030037@ntu.edu.sg (J. Zhu).

1 Tel.: +65 6793 3545; fax: +65 6792 2779.2 Tel.: +65 6790 4590; fax: +65 6790 9081.

rao∼wlta

924-0136/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.jmatprotec.2007.04.071

drothermal treatment; Crystal growth

uch as, chemical sensors [9,10], electrodes for dye-sensitizedolar cells and lithium ion battery [11–13], catalysts [14], etc.espite different working mechanisms in these devices, mate-

ial properties such as large specific surface area, pore volume,mall and uniformly distributed crystallites (<4 nm), and highhermal stability (400–600 ◦C) are usually preferred. Among

any synthesis techniques, sol–gel route is an attractive methodor fabrication of porous tin oxide with various morphologies15–17]. Materials composition can be easily modified in thisow-energy chemical pathway. The first attempt to synthesizeesoporous tin oxide was reported by Ulagappan and Rao

sing anionic surfactant [18]. The SnO2–surfactant compositebtained possess a mesophase with d-spacing of 3.2 nm. How-ver, the mesoporous structure was destroyed after calcination at27 ◦C, mainly due to the tin oxide crystal growth. Continuoustudy by applying various surfactants and process modificationere reported by several other groups [19–26]. With the usef neutral template, improved textural mesoporosity and envi-onmentally benign recovery of the expensive surfactant can bechieved [27]. Up to now, the maximum specific surface areaf mesoporous tin oxide reported using neutral template was112 m2/g after the removal of surfactant at 400 ◦C [25]. Here,

e report on the synthesis of mesoporous tin oxide with very

arge surface area and pore volume templated by neutral surfac-ant. Thermal stability enhancement by hydrothermal treatmentnd impurity oxide doping were investigated separately.

5 ssing

2

2

taistrw

2

wmtS0DdtunortmtTirtatid

laactb

ttt3r

wea

2

lT2wSwt

st0tdf1wtbd

2

gXucr(it0s

3

3

atcwatttthe weight losses of three as-prepared samples continued till∼400 ◦C, which indicates that TDA template was successfullyincorporated into the composite materials. Calcination temper-ature of 400 ◦C is sufficient to remove the organic template.

62 J. Zhu et al. / Journal of Materials Proce

. Experimental

.1. Raw materials

Tetraethylorthosilicate (C8H20O4Si, ≥99.0%) was obtained from Fluka. Tinetrachloride (SnCl4, 99.0%), tetradecylamine (C14H31N, 98.5%), and 28 wt.%mmonia solution (H2N–OH, 99.99%) were obtained from Sigma–Aldrich.so-Propanol (AR grade) was used as co-solvent and homogenizing agent forurfactant solution preparation. Deionized water was always used throughouthe synthesis wherever necessary. Silver nitrate (AgNO3) was applied to testesidual chlorine ions in waste solution. All the chemicals were used as receivedithout further analysis and purification.

.2. Synthesis—part I

Using tetradecylamine as template, the synthesis of mesoporous tin oxideith additional hydrothermal treatment is described as follows. Different fromost of the previously reported works, SnO2 precursor sol solution and surfac-

ant (tetradecylamine, TDA) solution were prepared separately before mixing.nCl4 precursor solution of 0.1 M concentration was prepared by mixing.02 mol (5.21 g) of SnCl4 (99%) into 200 ml deionized water with stirring.iluted ammonium hydroxide solution of 14 wt.% concentration was addedropwise into SnCl4 solution to cause condensation and the pH was adjustedo 9.8 at ambient temperature. White precipitate appeared in the SnCl4 solutionpon the addition of ammonium hydroxide, and the sol solution became gelati-ous at pH ∼4, as the IEP (reported to be 3.9) was approached. Further additionf ammonium hydroxide reduced the viscosity of SnCl4 sol with the solutionemaining white, suggesting the formation of relatively large oligomers. To keephe surfactant/Sn molar ratio at 0.3, surfactant aqueous mixture was prepared by

ixing 1.28 g of TDA (C14H31N, 98.5%) into 160 ml deionized water at roomemperature. iso-Propanol (Analytical reagent grade) of 65 ml was added intoDA/water mixture to improve the solubility of TDA. SnCl4 solution was added

nto TDA mixture slowly with stirring and the solution mixture was reacted atoom temperature for 2 h. Aging temperature was maintained at 75 ◦C for allhe samples prepared with aging time at 24, 48 and 75 h without stirring. Afterging, the white precipitate were filtered and washed with distilled water for 10imes to remove excess surfactant and Cl− ions until no precipitate was foundn the waste solution when AgNO3 was tested. The filtered white powders wereried in an oven at ∼60 ◦C with airflow for overnight before calcination.

Certain amounts of the as-prepared wet samples were loaded into a Teflon-ined tube inside a hydrothermal treatment bomb. 20 ml deionized water wasdded as HT aqueous medium. The as-prepared samples with various periods ofging time were hydrothermally treated at 120 ◦C for 24 h. The heating rate andooling rate were selected to be 5 and 10 ◦C/min separately. After HT treatmenthe samples were filtered and washed with distilled water and dried in an ovenefore calcination.

After drying the samples were calcined at 400 ◦C for 3 h in airflow to removehe surfactant templates. The whole process took place by 3 steps. In the first stephe temperature was raised to 120 ◦C and was held for 30 min. In the second stephe temperature was raised to the desired calcination temperature and held forh. Finally the samples were cooled down to ambient temperature. The heating

ate and cooling rate used for all samples are 5 and 10 ◦C/min.For simplicity, the samples were labeled hereafter using format of “TmHn”,

here “m” represents the aging time, and “n” represents the HT temperature,.g. T24H120 represents the sample with aging time of 24 h and HT temperaturet 120 ◦C.

.3. Synthesis—part II

Neutral surfactant templated tin oxide powders with TEOS (tetraethy-orthosilicate) as crystal growth suppressing agent was prepared as follows.in (IV) chloride sol solution of 0.1 M concentration was prepared by mixing

.605 g of SnCl4 (99%) with 100 ml deionized water. The pH of the sol solutionas adjusted to 7–8 by ammonia (28 wt.%) to cause condensation. To keep thei:Sn molar ratio at 0.25 and 0.06, tetraethylorthosilicate (TEOS) with differenteights was added slowly to the tin chloride sol solution during which the reac-

ions of SnCl4/TEOS mixture take place for 20 min at room temperature with

Technology 192–193 (2007) 561–566

tirring. The final pH of the mixture was adjusted to 10.8 by ammonia solu-ion. Semi-transparent tetradecylamine (TDA) solution was prepared by mixing.64 g of TDA (98.5%) with 80 ml deionized water and 30 ml iso-propanol, sohat TDA/Sn molar ratio was kept as 0.3:1. The TDA solution was slowly intro-uced to SnCl4/TEOS mixture; the aqueous mixture was sent for aging at 75 ◦Cor 24 h without stirring. Hydrothermal treatment was applied to the mixture at20 ◦C for 24 h. The white precipitates were washed and filtered with distilledater repeatedly until no precipitation was observed when AgNO3 was added to

he filtered waste. The white aggregated powders were dried at 60 ◦C overnightefore calcination. The calcination was conducted in a similar way as previouslyescribed from 400 to 600 ◦C.

.4. Characterization

Weight variation upon the thermal treatment was characterized by thermo-ravimetry (TGA HR 2590, V5.4A) at a heating rate of 5 ◦C/min. Wide-angle-ray powder diffraction was conducted on a Shimadzu 6000 diffractometersing Cu K� radiation of wavelength 1.54056 A. Meso-scale structures wereharacterized by small-angle X-ray scattering (Bruker Nanostar) using Cu K�

adiation with wavelength of 1.54184 A. Transmission electron microscopyTEM) images were obtained from Jeol JEM 2010, the samples were dispersedn ethanol under ultrasonic and transferred to copper grid. Nitrogen adsorp-ion/desorption isotherms were measured at 77K for relative pressures from.05 to 0.995 using ASAP2000 adsorption apparatus from Micromeritics. Theamples were degassed under vacuum at 90 ◦C overnight before analysis.

. Results and discussions

.1. Hydrothermal treatment

TG analyses of hydrothermally treated as-prepared T24, T48nd T75 samples are shown in Fig. 1. The total weight losses ofhe three samples range from 30 to 40 wt.%, and the differencesame from the surface adsorbed surfactants. Below ∼120 ◦C, theeight loss can be mainly attributed to the removal of surface

dsorbed moisture. The major weight reduction started from 150o 400 ◦C, which was related to the decomposition of organicemplates and surface dehydroxylation. Complete decomposi-ion of formal TDA usually occurs below 200 ◦C. However,

Fig. 1. TG analysis of as-prepared (A) T24, (B) T48 and (C) T75 samples.

J. Zhu et al. / Journal of Materials Processing Technology 192–193 (2007) 561–566 563

Fig. 2. XRD patterns of as-prepared (A) T24, (B) T24H120, and their calcinedsamples at 400 ◦C in (C) and (D), respectively.

Fs

mscpfo

Fs

Table 1Average crystal sizes of various samples with/without hydrothermal treatment

Samples Ave. Crystal Size (nm)

As-prepared Calcined at 400 ◦C

T24 1.9 6.3T24H120 3.2 5.3T48 2 5.2T48H120 2.3 4.7TT

easa

sitcWid2sissacWtsit

ig. 3. XRD patterns of as-prepared (A) T48, (B) T48H120, and their calcinedamples at 400 ◦C in (C) and (D), respectively.

XRD results of the samples with/without hydrothermal treat-ent of different aging periods are shown in Figs. 2–4. For all the

amples, rutile SnO2 crystalline phase were confirmed by three

haracteristic peaks, diffracted from (1 1 0), (1 0 1) and (2 1 1)lanes separately. All the as-prepared samples possess broadull-width-at-half-maximum (FWHMs), indicating the presencef nanocrystallites. Upon calcination, the peaks became sharp-

ig. 4. XRD patterns of as-prepared (A) T75, (B) T75H120, and their calcinedamples at 400 ◦C in (C) and (D), respectively.

wacrc

afbtritpwtTaats

75 2 4.175H120 2.5 3.8

ned due to the crystal growth. Using Deby–Scherrer equationnd assuming that the peak broadening was mainly due to theize effect, the average crystal size of each sample was calculatednd listed in Table 1.

The average crystal sizes of as-prepared T24, T48 and T75amples are all about 2 nm, suggesting that crystal growth wasnsensitive to aging time at 75 ◦C. However, the effect of agingime on crystal thermal stability was reflected by differentrystal sizes after calcination and hydrothermal treatment.ith calcination at 400 ◦C for 3 h, there was a large increase

n crystal size for each sample. And the final crystal sizeecreases from ∼6 to 4 nm as the aging time increases from4 to 75 h. Further, there was also a reduction in crystal size ofamples after hydrothermal treatment when the aging time wasncreased. This result indicates that although the initial crystalizes were almost same for each sample, the concentration ofurface hydroxyl groups was lower for the sample with longerging time. It was well known that –OH groups was the mainause of crystal growth for those sol–gel-derived products [28].ith additional hydrothermal treatment and after calcination,

he crystal size was further suppressed as compared to thoseamples without HT treatment. Therefore, HT was able tomprove the thermal stability of SnO2 crystallites by eliminatinghe surface hydroxyl groups during aqueous environment. Itas also noticed that the effect of HT tended to reduce as the

ging time was increased. In our experiment, the optimumondition was obtained at aging time of 75 h with HT at 120 ◦C,eflected by a minimum crystal size difference of 0.3 nm afteralcination.

For comparison, small-angle X-ray scattering patterns of T24nd T75 samples are shown in Figs. 5 and 6, respectively. Theormation of as-prepared mesophase composite was confirmedy the presence of single diffracted peak at low 2θ region. Uponhe removal of organic template, the single diffracted peak wasetained but shifted to lower 2θ angle. The increase in peakntensity was related to the increase of electron density con-rast between air and inorganics. The single and broad reflectedeak at low 2θ revealed the formation of mesoporous structureith limited range of ordering. After the removal of surfac-

ants, enlargement in d-spacing was observed for both T24 and75 samples, which was caused mainly by the crystal growth

nd coalescence. The peak positions for T75 calcined samplesre larger than those of T24 samples; this is in agreement withhe XRD results that the crystal sizes of the corresponding T24amples are larger.

564 J. Zhu et al. / Journal of Materials Processing Technology 192–193 (2007) 561–566

Fig. 5. SAXS patterns of as-prepared (A) T24, (B) T24H120, and their calcinedsamples in (C) and (D), respectively.

Fs

(ar

Fr

ccfcsp

gBItafeo0A

ig. 6. SAXS patterns of as-prepared (A) T75, (B) T75H120, and their calcinedamples in (C) and (D), respectively.

As a typical example, the transmission electron microscopyTEM) images of T75 samples are shown in Fig. 7. Theverage crystal size from TEM images agrees well with theesults form XRD characterization. The formation of mesophase

rcsd

Fig. 7. TEM images of (a) as-prepared T75 sample and (b) T75

ig. 8. (a) N2 gas adsorption isotherm of calcined T75H120 sample and (b)espective pore size distribution.

omposite was clearly revealed in Fig. 7a due to the phaseontrast. Upon the removal of templates, the mesopores wereormed as the interparticles voids. The average lattice spacingalculated in Fig. 7b is 0.34 nm which corresponds to the d-pacing value (0.335 nm) of 1 1 0 planes calculated from XRDattern.

The porous features of T samples were characterized by N2as adsorption method. The adsorption/desorption isotherm andJH pore size distribution curve of are shown in Fig. 8. The type

V isotherm with hysteresis loop shown in Fig. 8a is characteris-ic of mesoporous materials. The initial large adsorption volumet low relative pressure region indicates the large specific sur-ace area. The presence of hysteresis loop is a result of capillaryffect when N2 gas is adsorbed/desorbed. The large incrementf adsorbed volume in the relative pressure range from 0.2 to.7 was caused by the liquid N2 filling inside the mesopores.nother increase above relative pressure of 0.9 was probably

elated to the N2 filling of the voids between secondary parti-le agglomerations. The pore size distribution curve in Fig. 8bhows that the calcined T75H120 sample possesses uniformlyistributed mesopores with the size centered at ∼4–5 nm.

H120 calcined at 400 ◦C. The scale bar represents 10 nm.

J. Zhu et al. / Journal of Materials Processing Technology 192–193 (2007) 561–566 565

Table 2BET results of T samples with various aging time

Samples BET surfacearea (m2/g)

BJH porevolume (cm3/g)

BJH averagepore size (nm)

T24 93 0.197 5.6T24H120 101 0.238 6.1T48 116 0.23 4.9T48120 119 0.26 5.3TT

ulpacstia

hoHfCtrs

3

istlt

Fc

Fig. 10. XRD patterns of SnO2–SiO2 with Si/Sn = 0.06 (A) as-prepared, andcalcined sample at (B) 400 ◦C, (C) 500 ◦C, and (D) 600 ◦C.

Table 3Average crystal sizes of SnO2 samples with SiO2 dopants

Si/Sn molar ratio Average Crystal Size (nm)

As-prepared 400 ◦C 500 ◦C 600 ◦C

00

S2sFctc

io

75 133 0.23 4.375H120 138 0.26 5.0

For comparison, BET surface area, BJH desorption pore vol-me and average pore size of T samples calcined at 400 ◦C areisted in Table 2. It was noticed that BET surface area was almostroportional to the aging time and inversely related to the aver-ge crystal size. This suggests that the surface area is mainlyomposed by the exposed surfaces of crystallites. The thermaltability enhancement by prolonging the aging time and addi-ional hydrothermal treatment was reflected by ∼23% and 11%mprovement in BET surface area and pore volume, respectively,s compared to the maximum values reported [19,25].

These results showed that with increased aging time andydrothermal treatment, thermal stability of mesoporous tinxide can be enhanced by reducing the surface hydroxyl groups.owever, the aging makes the synthesis less effective and the

urther enhancement by hydrothermal treatment is also limited.rystal growth is an intrinsic property of material and effec-

ive crystal growth has to be achieved by impurity doping. Theesults from using SiO2 as SnO2 crystal growth inhibitor will behown in next section.

.2. Doping with SiO2

The influence of SiO2 dopants on the SnO2 crystal growthnhibition was confirmed by the XRD characterizations, as

hown in Figs. 9 and 10. At Si/Sn molar ratio of 0.25, effec-ive crystal growth hindering was clearly observed by the muchess peak sharpening as compared to the previous samples. Allhe diffracted peaks match with rutile SnO2 phase. Amorphous

ig. 9. XRD patterns of SnO2–SiO2 with Si/Sn = 0.25 (A) as-prepared, andalcined sample at (B) 400 ◦C, (C) 500 ◦C, and (D) 600 ◦C.

ntS0c

wtgarc

TBo

S

456

.25 2.2 2.5 2.5 2.7

.6 2.5 3.6 4.2 5.6

iO2 phase is revealed by broad maxima at low 2θ angles from0◦ to 40◦. The FWHMs of sample calcined at even 600 ◦C wastill quite broad, indicating very small crystal size as shown inig. 9D. However, when Si/Sn ratio was reduced to 0.06, lessrystal growth inhibition effect was observed at same calcina-ion temperature. This result directly proves the role of SiO2 asrystal growth inhibitor during calcination.

The calculated average crystal sizes of the samples are listedn Table 3. For the sample with Si/Sn ratio of 0.25, there wasnly about 0.2 nm increment in crystal dimension as the calci-ation temperature increased from 400 to 600 ◦C. This showshat SiO2 impurity dopants are very effective in suppressing thenO2 crystal growth. However, when the ratio was lowered to.06, the effect diminished as reflected by the more pronouncedrystal growth.

The BET adsorption results of calcined SnO2–SiO2 samplesith Si/Sn ratio f 0.25 are given in Table 4. As shown, both

he BET surface area and pore volume have been improved

reatly when SiO2 dopants were used. The maximum surfacerea and pore volume are 340 m2/g and 0.3 cm3/g, respectively,esulted from the sample after calcination at 400 ◦C. As the cal-ination temperature increased to 600 ◦C, there was reduction in

able 4ET adsorption results of calcined SnO2–SiO2 samples with Si/Sn molar ratiof 0.25

amples BET surfacearea (m2/g)

BJH pore volume(cm3/g)

BJH averagepore size (nm)

00 ◦C 340 0.30 3.000 ◦C 262 0.27 3.100 ◦C 211 0.23 3.2

5 ssing

bg

4

fstir(rpcp

R

[[

[

[[[[

[

[[

[

[[

[

[

66 J. Zhu et al. / Journal of Materials Proce

oth surface area and pore volume caused by the thermal crystalrowth.

. Conclusion

In summary, mesoporous SnO2/SnO2–SiO2 with large sur-ace area, pore volume and small nanocrystallites have beenuccessfully synthesized using neutral surfactant templatingechnique. The thermal stability of SnO2 framework could bemproved greatly by impurity SiO2 doping with proper molaratio. With Si/Sn ratio of 0.25, the maximum surface area340 m2/g) and pore volume (0.3 cm3/g) were obtained after theemoval of template at 400 ◦C. Mesoporous tin oxide with suchroperties are promising candidates in the applications such asatalysis, chemical sensors, electrodes for electrochemical andhoto-optoelectronic devices.

eferences

[1] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Nature359 (1992) 710–712.

[2] J.S. Beck, C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D.Schmitt, C.T.-W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B.

Higgins, J.L. Schlenker, J. Am. Chem. Soc. 114 (1992) 10834–10843.

[3] A. Sayari, P. Liu, Micropor. Mater. 12 (1997) 149–160.[4] M.E. Davis, Nature 417 (2002) 813–821.[5] C.Z. Yu, B.Z. Tian, D.Y. Zhao, Curr. Opin. Solid State Mater. Sci. 7 (2003)

191–197.

[[

[[

Technology 192–193 (2007) 561–566

[6] A. Stein, Adv. Mater. 15 (2003) 763–781.[7] J.L. Shi, Z.L. Hua, L.X. Zhang, J. Mater. Chem. 14 (2004) 795–806.[8] E.E. Kohnek, J. Phys. Chem. Solids 23 (1962) 1557–1560.[9] J. Watson, K. Ihokura, G.S.V. Coles, Meas. Sci. Technol. 4 (1993) 711–719.10] C.O. Park, S.A. Akbar, J. Mater. Sci. 38 (2003) 4611–4637.11] C. Nasr, P.V. Kamat, S. Hotchandani, J. Phys. Chem. B 102 (1998)

10047–10056.12] T. Stergiopoulos, I.M. Arabatzis, H. Cachet, P. Falaras, J. Photochem.

Photobiol. A 155 (2003) 163–170.13] A. Yu, R. Frech, J. Power Sources 104 (2002) 97–100.14] M.J. Fuller, M.E. Warwick, J. Catal. 29 (1973) 441–450.15] K.L. Chopra, S. Major, D.K. Pandya, Thin Solid Films 102 (1983) 1–42.16] R.S. Hiratsuka, S.H. Pulcinelli, C.V. Santilli, J. Non-Cryst. Solids 121

(1990) 76–83.17] M.J. Hampden-Smith, T.A. Wark, C.J. Brinker, Coord. Chem. Rev. 112

(1992) 81–116.18] N. Ulagappan, C.N.R. Rao, Chem. Commun. 14 (1996) 1685–1686.19] K.G. Severin, T.M. Abdel-Fattah, T.J. Pinnavaia, Chem. Commun. 14

(1998) 1471–1472.20] L.M. Qi, J.M. Ma, H.M. Cheng, Z.G. Zhao, Langmuir 14 (1998)

2579–2581.21] F.L. Chen, M.L. Liu, Chem. Commun. 18 (1999) 1829–1830.22] Y.D. Wang, C.L. Ma, X.H. Wu, X.D. Sun, H.D. Li, Talanta 57 (2002)

875–882.23] T. Hyodo, N. Nishida, Y. Shimizu, M. Egashira, Sens. Actuators B 83

(2002) 209–215.24] N.L. Wu, C.Y. Tung, J. Am. Ceram. Soc. 87 (2004) 1741–1746.

25] S. Zhou, S.M. Lu, Y.X. Ke, J.M. Li, Mater. Lett. 57 (2003) 2679–2681.26] P.D. Yang, D.Y. Zhao, D.I. Margolese, B.F. Chmelka, G.D. Stucky, Nature

396 (1998) 152–154.27] P.T. Tanev, T.J. Pinnavaia, Science 267 (1995) 865–867.28] N.L. Wu, S.Y. Wang, I.A. Rusakova, Science 285 (1999) 1375–1377.