- Email: [email protected]
TiO2 thin film deposition from solution using microwave heating
Thin Solid Films 365 (2000) 12±18 www.elsevier.com/locate/tsf
TiO2 thin ®lm deposition from solution using microwave heating Elena Vigil a, Lahcen Saadoun b, Jose A. AylloÂn c,*, Xavier DomeÁnech c, Inti Zumeta a, Rafael RodrõÂguez-Clemente b a Instituto de Materiales y Reactivos, Universidad de la Habana, Cuba Institut de CieÁncia de Materials de Barcelona (CSIC) Campus de la UAB, 08193 Cerdanyola, Spain c Departament de QuõÂmica, Universitat AutoÁnoma de Barcelona, Edi®ci Cn, Campus de la UAB, 08193 Cerdanyola, Spain b
Received 23 January 1999; received in revised form 24 September 1999; accepted 3 November 1999
Abstract A new method has been devised for the deposition of TiO2 thin ®lms on conducting glass using a microwave reactor. The substrates are immersed in a diluted homogeneous aqueous solution which was prepared by mixing equal volumes of a ¯uorine-complexed titanium(IV) solution (Ti 3:4 £ 1022 M) and 6:8 £ 1022 M boric acid solution. Low microwave power and short deposition time have been used. The TiO2 layers obtained are well-adhered, homogenous, with good specularity and colored by interference of re¯ected light. Their thickness is in the range of 100±500 nm. SEM experiments denote that ®lms are formed by small crystallites having linear dimensions under 100 nm. Crystal dimensions depend on microwave power and deposition time. The layers show a high degree of crystallinity and the observed crystal phase is anatase. Microwave heating has proved to be an ef®cient and inexpensive method for solution growth of TiO2 ®lms; it should also be of importance for other materials layers grown from solution. q 2000 Elsevier Science S.A. All rights reserved. Keywords: Titanium oxide; Deposition process
1. Introduction Several reports on TiO2 thin ®lms obtained by different methods have recently appeared in the literature [1±6] due to the importance of these ®lms in different applications. Compact thin ®lms of TiO2 on conducting glass are used in new types of solar cells: liquid and solid dye-sensitized photoelectrochemical solar cells [7,8], as well as, solar cells with an organic or inorganic extremely thin absorber [9,10]. These thin ®lms are also of interest for the photo-oxidation of water [1], photocatalysis [11], electrochromic devices [12] and other uses [13]. Due to the need of cost competitive devices in these application areas, simple and inexpensive techniques are required for ®lm deposition. Usually ®lms have been obtained using suspensions or pastes containing nanocrystalline TiO2, which are applied using some simple method like doctors blade [14] or screenprinting [15]. In Refs. [1,8] it is argued that a compact layer next to the conducting glass diminishes the dark current and avoids short circuit with the conducting glass. Regarding deposition methods, Yoshimura [16] argues in favor of soft-solution processing for advanced inorganic * Corresponding author. Tel.: 1 34-3-581-1927; fax: 1 34-3-581-2920. E-mail address: [email protected] (J.A. AylloÂn)
materials for fabrication in aqueous solutions without ®ring, sintering or melting and using low temperatures. This author suggests hydrothermal and/or electrochemical methods because they are low temperature methods for in situ fabrication of crystalline thin ®lms [16]. This means quality improvement, lower costs and environmental friendly processing. On the other hand, microwave dielectric heating is rapidly becoming an established procedure in synthetic chemistry [17]. In this paper, a new soft-solution processing method using microwaves is used for depositing TiO2 ®lms with relative ease and low temperature. Characterization of the ®lms reveals that they are formed by crystals with linear dimensions under 100 nm which form a compact layer. The crystal phase is anatase according to X-ray diffraction and optical absorbance. 2. Film preparation The precursor aqueous solution was freshly prepared by mixing equal volumes of a ¯uorine-complexed titanium(IV) solution (Ti 3:4 £ 1022 M) and 6:8 £ 1022 M boric acid solutions. The ®rst mentioned solution was prepared from titanium tetraisopropoxide (TIP), HF and NH4F. TIP (10.0 ml, 0.034 mol) was dissolved in 50 ml of ethanol in which
0040-6090/00/$ - see front matter q 2000 Elsevier Science S.A. All rights reserved. PII: S00 40-6090(99)0094 6-3
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
aqueous HF (6.0 ml 40% aqueous solution, 0.136 mol) was added. Afterwards a solution of NH4F (2.52 g, 0.068 mol) in water (100 ml) was also added. The clear solution so obtained was heated in an open vessel until the volume was reduced to c.a. 30 ml in order to evaporate alcohols. The residue was diluted to 1.0 l with water. Films were grown on 10 V/A ITO conducting glass, kindly supplied by Flabeg (Pilkington Group). Before the microwave process, substrates were cleaned with organic solvents, sonnicated in 30% HNO3 during 1 min and rinsed with water. The electric resistance of the conducting glass corresponding to a ®xed distance of 5 mm before and after this cleaning procedure was measured. Changes in the resistance values were within the experimental uncertainty (,2 V); therefore, the ITO layer thickness was not affected. A Maxidigest MX 350 (Prolabo) microwave furnace operating at 2.45 GHz and a maximum power of 300 W was used. The conducting glass was cut into 15 £ 20 mm plates and they were hung amidst 80 ml of the growing solution. Such a rather relatively large amount of solution was used in order to obtain an appreciable amount of precipitates in the solution which were also analyzed [18]. The substrate immersed in the solution was sonnicated for 3 min before microwave heating. Different microwave power and deposition time were tested. At least three samples were grown with each of the selected regimes. The values of heating time and power, which de®ne each regime, are shown in Table 1. After microwaving the substrates were taken out of the solution and rinsed. Temperature was measured immediately after the processes corresponding to each of the given regimes. Except for 20±120 and 20±30 regimes, the temperatures were less than 1008C. For the last two listed in Table 1, the temperature was below 908C. Actual ®lm growth temperatures could be larger due to the localized heating of the substrates originated from microwaves interacting with the ITO conducting layer on the glass [17]. 3. Film characterization and properties 3.1. Physical appearance and thickness of the ®lms After ®lm growth, samples with deposition time equal or above 30 min had a whitish appearance, due to the presence of TiO2 crystals on the surface. These crystals were loosely
13
stuck, mainly on the side of the glass without ITO layer, and strongly scattered light, causing the whitish appearance. They were easily removed mechanically and a colored transparent ®lm remained very well adhered. For characterization samples were thoroughly cleaned and sonnicated in deionized water. The ®lm was visible with the naked eye because of the color change observed by re¯ection. The ITO layers used were bluish when observed by light re¯ection. TiO2 layers on ITO showed different colors (see Table 1). These colored ®lms were very well adhered. Also the layer homogeneity was visible due to the interference phenomena. Layers were homogeneous in the central area. In some borders small interference rainbows formed indicating different thickness. It should be pointed out that the substrate cleaning is very determinant. A Tenkor Instrument was used for per®lometry measurements of the samples. In order to obtain a step, samples were covered partially with a Te¯on ribbon. Only gradual slopes were produced which together with the substrate original surface made the thickness measurement dif®cult and introduced larger experimental uncertainties. Samples produced with regimes 20±30 and 10±30 were the thicker ones (between 300 and 500 nm). The thinner layers were the yellow ones (see Table 1), with thickness under 300 nm. The samples grown during 2 h using 60 W microwave power (20±120 regime) do not seem to ®t the general scheme because they were as thin as those of regimes 10± 20 and 65±3. They are thinner than the ones grown with the same power and less time. On the other hand, they had more loosely stuck TiO2 after deposition than the rest. Thickness determination using SEM (Hitachi S-570, 10± 30 KV) was possible on one sample obtained with regime 20±120. Fig. 1 shows a thickness around 0.15 mm. 3.2. Surface topography The ®lm homogeneity was more closely revealed by the color intensity produced by interference when using optical microscopy. The same colors reported in Table 1 were observed for each layer under the microscope when using a re¯ection microscope. In some samples, even though, the layer could show only one color when observed with the naked eye, differences could be found under the microscope due to the presence of small not fully covered areas. Speci-
Table 1 Analyzed samples Regime a
Microwave power (W)
Heating time (min)
Re¯ection-interference color
Thickness range (nm)
20±120 20±30 10±30 10±20 65±3
60 60 30 20 195
120 30 30 20 3
Yellow Pinkish or green Orange yellow or pinkish Yellow Yellow
,300 300±500 200±400 ,300 ,300
a
Sample regime were named according to the percentage of maximum microwave power available (300 W) and microwaving time.
14
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
bution of the freshly obtained precipitates was analyzed using a Malvern Zetasizer 3 for some of the regimes (see Fig. 4). Larger agglomerates are formed in the solution compared with the corresponding size of the crystals on the layers. 3.3. X-ray diffraction
Fig. 1. SEM photograph showing a cross-sectional view of a TiO2 ®lm grown with regime 20±120. The TiO2 ®lm interface with the conducting layer is indicated with an arrow.
®cally, in ®lms grown with regime 20±120 scratches were visible. Films from different regimes were gold-covered and examined with scanning electron microscopy (SEM) to investigate their microcrystalline structure and surface characteristics. In Fig. 2a,b, ®lms grown during the same time but with different microwave power are compared. It is observed that samples 20±30 showed a better surface coverage than samples 10±30. Nonetheless, this was not the case for samples 20±120, where scratches were visible even with the optical microscope. The in¯uence of the used regime on crystal size is illustrated in the micrographs of Fig. 3. SEM images in this ®gure show crystals or grains with linear dimensions smaller than 100 nm. It is noteworthy how they are closely packed together. It can be observed that samples with smaller thickness show a higher nanocrystalline character, that is, they are formed by smaller grains. To complement the crystal size analysis, the mass distri-
Cu Ka radiation (l 0:154056 nm) was used for all Xray diffraction experiments performed with Rigaku Rota¯ex RU-200 B and Siemens D-3400 apparatus. X-ray diffraction of a 20±30 ®lm is shown in Fig. 5b together with diffraction corresponding to the ITO substrate (Fig. 5a). Comparison of these ®gures reveals the appearance of peaks at 25.3 and 37.88, which correspond to crystal planes of the anatase structure (101) and (004), respectively. The second one is very close to a peak of the ITO substrate. Taking into account the relative intensities, Ihkl, of the two peaks for anatase (I101 =I004 100=20), diffraction from planes (004) is relatively stronger than that from planes (101). These lines were also observed by common X-ray diffraction spectra corresponding to regime 10±30. For thinner layers corresponding to regimes 10±20, 65±3 and 20±120 low angle Xray diffraction was necessary. Low-angle X-ray diffraction spectrum is shown in Fig. 5c corresponding to a sample 10± 30. Lines are observed corresponding to the anatase structure at values of 2u equal to 25.3, 37.8, 48.1, and 53.9 which correspond to planes (101), (004), (200) and (105), respectively. Also in this low angle diffraction experiment, the line intensity relations are not those that correspond to complete random order of crystals. It should also be observed in Fig. 5c that peaks are very well-de®ned with a small half-width, usually attainable only after heat treatment of the TiO2 deposits (2).
Fig. 2. SEM photographs of TiO2 ®lms grown during the same time (30 min) but with different microwave power: (a) 60 W microwave power (regime 20±30) and (b) 30 W microwave power (regime 10±30).
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
15
Fig. 3. SEM photographs of TiO2 ®lms grown under different conditions: (a) 60 W microwave power during 120 min (regime 20±120), (b) 60 W microwave power during 30 min (regime 20±30), (c) 30 W microwave power during 30 min (regime 10±30) and (d) 30 W microwave power during 20 min (regime 10±20).
3.4. Optical transmittance Transmittance of TiO2 ®lms on conducting glass were measured with a Varian Cary 05 spectrophotometer using the conducting glass substrate as the blank. Spectra corresponding to the thicker layers from regime 20±30 and 10±30 are shown in Fig. 6a. Spectra have been processed in order to obtain the energy band-gap from the expression corresponding to indirect gap semiconductors
a hnhn / hn 2 Eg 2 Taking into consideration the specularity of the ®lms, it is assumed that specular re¯ectivity can be disregarded as compared to absorbance in the high absorption range. In this case, the absorbance is directly proportional to the absorption coef®cient. From the plot in Fig. 6b, a value of
the energy gap of 3.25 eV is obtained. This value is well over the reported value for rutile (3.0 eV) and corresponds satisfactorily with the reported values for anatase. 4. Discussion From the experimental results obtained the ®lm growth from solution using microwaves has proved to be a simple and inexpensive method. Small times of deposition can be used, which are not possible with conventional heating. As explained before, 80 ml of solution were used in order to study also the precipitates in the solution [18]; but much smaller amount of solution is required for growing the thickness obtained. In that case, power and/or time can be reduced. As compared to our previous experiments for
16
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
Fig. 4. Mass distribution per aggregate size of the freshly obtained precipitates for some of the used regimes.
growing TiO2 ®lms [6], layer growth is more ef®cient with the present precursor solution. In [6] several microwave processes with fresh solution were necessary, one on top of the other, to obtain a layer thickness in the range obtained in the present case with a single microwave heating process. It is expected that several successive processes could also be used with the present precursor solution to obtain thicker ®lms. If we compare thickness obtained in regimes 10±30 with
Fig. 5. XRD spectra corresponding to: (a) ITO glass substrate, (b) TiO2 ®lm on the ITO glass substrate obtained using 60 W microwave power during 30 min (regime 20±30). (c) Low-angle X-ray diffraction pattern corresponding to a TiO2 ®lm obtained using 30 W microwave power during 30 min (regime 10±30). The arrows indicate the peaks corresponding to TiO2 anatase.
Fig. 6. (a) Transmittance of the TiO2 ®lms on conducting glass for sample obtained with regimes 20±30 and 10±30 (conducting glass substrate was used as blank) and (b) spectral dependence of the absorbance showing energy gap value for these same samples.
20±30 and 10±30 with 10±20, the TiO2 ®lms seem to grow more as the microwave power increases and as the time increases. There is an optimum time for each microwave power after which the layers do no become any thicker. This is clear not only because of the thickness of the ®lms (Table 1) and ®lm coverage (Fig. 2a,b), but also because 120 min proved to be a very long time for 60 W microwave power (20% of maximum power). It is not completely clear why these ®lms are even thinner than those grown with the same power during 30 min. It is considered that in an early growth stage the ®lms are formed from TiO2 that nucleates on the surface; which is a different mechanism than that resulting from TiO2 precipitated in the solution and `sticking' to the surface. Apparently, the ®rst produces well-adhered layers. We also believe that nucleation on the surface might proceed at a faster pace than in the solution at the beginning of microwave heating helped by `localized heating' of the ITO glass due to its free electrons [17]. That is, microwaves may interact with free electrons and these may pass the acquired energy to lattice vibrations. The importance of such effect requires further research, in spite of the fact that, well-adhered TiO2 layers do not grow on the side of the glass without conducting layer. Once the crystals precipitated in the solution acquire a large concentration, their collisions on the surface should impair layer growth; on the other hand, `localized heating' must be less effective
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
Fig. 7. SEM photograph of TiO2 powder precipitated in the solution during a process in which 60 W microwave power during 30 min were used (regime 20±30).
because of the existence of the grown TiO2 layer. Further, it must be considered that ®lms suffer erosion from the precipitated TiO2 in the solution during the 90-min. difference between experiments with 20±120 regime and those with 20±30 regime. Certainly, as already mentioned scratches are visible on the surface. The highest concentration and the largest agglomerates of precipitated TiO2 in the solution are obtained for this 20±120 regime (see Fig. 4). Further experiments are required to fully explain this. On the other hand, as was pointed out before, crystals are smaller the thinner the layer (See Fig. 3 and Table 1). This indicates, and can be observed in some areas of the SEM photos in Fig. 3, that as growth proceeds new deposits cover more than one crystal and larger crystals form from smaller ones; the length of frontiers diminishing for the new grown material. Agglomeration of smaller, primary TiO2 particles to form larger ones was also observed by (2) and reference cited therein. We believe that both SEM photographs in Fig. 3a (regime 20±120) and in Fig. 3d (regime 10±20), show smaller nanocrystals produced during an early growth stage of the layer, in agreement with previous considerations. It is characteristic of the ®lms obtained that crystals are compactly stacked. If Fig. 3 is compared with Fig. 1b of Ref. [1] where TiO2 ®lms are also grown on conducting glass, a different growth habit is observed. We believe this might be due to an enhanced effect of the ITO conducting layer produced by microwaves since SEM images of crystals precipitated in the solution showed a different crystal habit more similar to Fig. 1b in Ref. [1]. In Fig. 7 a SEM
17
photograph of crystals precipitated in the solution in a 20± 30 regime is shown. Flower like crystals are observed, as well as, in Ref. [1] where TiO2 layers are also deposited on conducting glass. Closer and more ordered packing is observed for the microwave-grown layers on ITO (see Fig. 3). Therefore, the reported microwave method could be of interest for obtaining the thin, compact TiO2 layer used in dye-sensitized solar cells to avoid direct contact of the liquid or solid electrolyte with the conducting layer [8]. The ordered packing could be due to preferential orientation of TiO2 nanocrystallites. In Fig. 5c, the relations of peak intensities corresponding to TiO2 anatase planes do not correspond with the known relations for complete random order. This indicates a preferential growth. The narrow diffraction peaks observed in the low angle diffraction pattern of Fig. 5c con®rm the high degree of crystallinity in spite of small crystal dimensions. Even without any sintering, they show higher crystallinity than other reports [4,6]. The anatase phase present is highly convenient for some applications of these layers [20,21]. Simplicity, low temperature, fast growth, good adherence, closed-packing, good crystallinity without sintering, anatase phase and nanostructured-character are some of the advantages found with the performed deposition experiments. This new method using microwave heating could provide faster and better results for thin ®lms grown from solution than obtained when using other procedures as chemical bath deposition (CBD) [19] or soft-solution methods [16]. Full explanation of thin ®lm deposition mechanisms and of the exact role of microwaves deserve special attention and need further research. 5. Conclusion TiO2 ®lms deposited using the reported new technique show good adherence and specularity. The layers are nanostructured and closely packed for some of the different regime employed. X-ray diffraction experiments show narrow peaks revealing a good crystallinity usually obtained only after sintering when using other deposition from solution methods. Microwave heating required short times, low temperatures and is relatively inexpensive. The characteristics of the ®lms, as well as, the ease of the layer deposition, indicate that this is a perspective technique for TiO2 ®lms for different applications. It should also be possible to use it for obtaining other materials ®lms grown from solution. References [1] L. Kavan, M. GraÈtzel, Electrochim. Acta 40 (1995) 643. [2] S.D. Burnside, V. Shklover, C. BarbeÂ, et al., Chem. Mater. 10 (1998) 2419. [3] P. Sawunyama, A. Yasumori, K. Okada, Mater. Res. Bull. 33 (1998) 795. [4] S. Deki, Y. Aoi, J. Mater. Res. 13 (1998) 883.
18
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
[5] S. Yin, Y. Inoue, S. Uchida, Y. Fujishiro, T. Sato, Inoue, J. Mater. Res. 13 (1998) 844. [6] E. Vigil, L. Saadoun, R. RodrõÂguez-Clemente, J.A. AylloÂn, X. DomeÁnech, J. Mater. Sci. Lett. 18 (1999) 1067. [7] B. O'Regan, M. GraÈtzel, Nature 353 (1991) 737. [8] U. Bach, D. Lupo, P. Comte, J.E. Moser, F. WeissoÈrtel, J. Salbeck, H. Spreitzer, M. GraÈtzel, Nature 395 (1998) 583. [9] K. Tennakone, G.R.R.A. Kumara, A.R. Kumarasinghe, K.G.U. Wijayantha, P.M. Sirimane, Semicond. Sci. Technol. 10 (1995) 1689. [10] S. Siebentritt, K. Ernst, C.-H. Fischer, R. Konenkamp, M.C. LuxSteiner, in: H.A. Ossenbrink, P. Helm, H. Ehmann (Eds.), Proc. 14th Eur. Photovoltaic Sol. Energy Conf. Exhibit., Barcelona, H.S. Stephens & Associates, Bedford, UK, 1997, pp. 1823±1827. [11] R.L. Pozzo, M.A. Baltanas, A.E. Cassano, Catalysis Today 39 (1997) 219. [12] T.-S. Kang, D. Kim, K.-J. Kim, J. Electrochem. Soc. 145 (1998) 1982.
[13] Y. Paz, Z. Luo, L. Rabenberg, A. Heller, J. Mater. Res. 10 (1995) 2842. [14] A. Kay, M. GraÈtzel, Sol. Energy Mater. Sol. Cells 44 (1996) 99. [15] C.J. BarbeÂ, F. Arendse, P. Comte, et al., J. Am. Ceram. Soc. 80 (1997) 3157. [16] M. Yoshimura, J. Mater. Res. 13 (1998) 796. [17] C. Gabriel, S. Gabriel, E.H. Grant, B.S.J. Halstead, D.M.P. Mingos, Chem. Soc. Rev. 27 (1998) 213. [18] L. Saadoun, J.A. AylloÂn, E. Vigil, R. RodrõÂguez-Clemente, X. Domenech, unpublished. [19] P.K. Nair, M.T.S. Nair, V.M. Garcia, et al., Sol. Energy Mater. Sol. Cells 52 (1998) 313. [20] L. Kavan, M. Gratzel, J. Rathousky, A. Zukal, J. Electrochemical Soc. 143 (1996) 394. [21] L. Kavan, K. Kratochvilova, M. Gratzel, J. Electroanal. Chem. 394 (1995) 93.
TiO2 thin ®lm deposition from solution using microwave heating Elena Vigil a, Lahcen Saadoun b, Jose A. AylloÂn c,*, Xavier DomeÁnech c, Inti Zumeta a, Rafael RodrõÂguez-Clemente b a Instituto de Materiales y Reactivos, Universidad de la Habana, Cuba Institut de CieÁncia de Materials de Barcelona (CSIC) Campus de la UAB, 08193 Cerdanyola, Spain c Departament de QuõÂmica, Universitat AutoÁnoma de Barcelona, Edi®ci Cn, Campus de la UAB, 08193 Cerdanyola, Spain b
Received 23 January 1999; received in revised form 24 September 1999; accepted 3 November 1999
Abstract A new method has been devised for the deposition of TiO2 thin ®lms on conducting glass using a microwave reactor. The substrates are immersed in a diluted homogeneous aqueous solution which was prepared by mixing equal volumes of a ¯uorine-complexed titanium(IV) solution (Ti 3:4 £ 1022 M) and 6:8 £ 1022 M boric acid solution. Low microwave power and short deposition time have been used. The TiO2 layers obtained are well-adhered, homogenous, with good specularity and colored by interference of re¯ected light. Their thickness is in the range of 100±500 nm. SEM experiments denote that ®lms are formed by small crystallites having linear dimensions under 100 nm. Crystal dimensions depend on microwave power and deposition time. The layers show a high degree of crystallinity and the observed crystal phase is anatase. Microwave heating has proved to be an ef®cient and inexpensive method for solution growth of TiO2 ®lms; it should also be of importance for other materials layers grown from solution. q 2000 Elsevier Science S.A. All rights reserved. Keywords: Titanium oxide; Deposition process
1. Introduction Several reports on TiO2 thin ®lms obtained by different methods have recently appeared in the literature [1±6] due to the importance of these ®lms in different applications. Compact thin ®lms of TiO2 on conducting glass are used in new types of solar cells: liquid and solid dye-sensitized photoelectrochemical solar cells [7,8], as well as, solar cells with an organic or inorganic extremely thin absorber [9,10]. These thin ®lms are also of interest for the photo-oxidation of water [1], photocatalysis [11], electrochromic devices [12] and other uses [13]. Due to the need of cost competitive devices in these application areas, simple and inexpensive techniques are required for ®lm deposition. Usually ®lms have been obtained using suspensions or pastes containing nanocrystalline TiO2, which are applied using some simple method like doctors blade [14] or screenprinting [15]. In Refs. [1,8] it is argued that a compact layer next to the conducting glass diminishes the dark current and avoids short circuit with the conducting glass. Regarding deposition methods, Yoshimura [16] argues in favor of soft-solution processing for advanced inorganic * Corresponding author. Tel.: 1 34-3-581-1927; fax: 1 34-3-581-2920. E-mail address: [email protected] (J.A. AylloÂn)
materials for fabrication in aqueous solutions without ®ring, sintering or melting and using low temperatures. This author suggests hydrothermal and/or electrochemical methods because they are low temperature methods for in situ fabrication of crystalline thin ®lms [16]. This means quality improvement, lower costs and environmental friendly processing. On the other hand, microwave dielectric heating is rapidly becoming an established procedure in synthetic chemistry [17]. In this paper, a new soft-solution processing method using microwaves is used for depositing TiO2 ®lms with relative ease and low temperature. Characterization of the ®lms reveals that they are formed by crystals with linear dimensions under 100 nm which form a compact layer. The crystal phase is anatase according to X-ray diffraction and optical absorbance. 2. Film preparation The precursor aqueous solution was freshly prepared by mixing equal volumes of a ¯uorine-complexed titanium(IV) solution (Ti 3:4 £ 1022 M) and 6:8 £ 1022 M boric acid solutions. The ®rst mentioned solution was prepared from titanium tetraisopropoxide (TIP), HF and NH4F. TIP (10.0 ml, 0.034 mol) was dissolved in 50 ml of ethanol in which
0040-6090/00/$ - see front matter q 2000 Elsevier Science S.A. All rights reserved. PII: S00 40-6090(99)0094 6-3
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
aqueous HF (6.0 ml 40% aqueous solution, 0.136 mol) was added. Afterwards a solution of NH4F (2.52 g, 0.068 mol) in water (100 ml) was also added. The clear solution so obtained was heated in an open vessel until the volume was reduced to c.a. 30 ml in order to evaporate alcohols. The residue was diluted to 1.0 l with water. Films were grown on 10 V/A ITO conducting glass, kindly supplied by Flabeg (Pilkington Group). Before the microwave process, substrates were cleaned with organic solvents, sonnicated in 30% HNO3 during 1 min and rinsed with water. The electric resistance of the conducting glass corresponding to a ®xed distance of 5 mm before and after this cleaning procedure was measured. Changes in the resistance values were within the experimental uncertainty (,2 V); therefore, the ITO layer thickness was not affected. A Maxidigest MX 350 (Prolabo) microwave furnace operating at 2.45 GHz and a maximum power of 300 W was used. The conducting glass was cut into 15 £ 20 mm plates and they were hung amidst 80 ml of the growing solution. Such a rather relatively large amount of solution was used in order to obtain an appreciable amount of precipitates in the solution which were also analyzed [18]. The substrate immersed in the solution was sonnicated for 3 min before microwave heating. Different microwave power and deposition time were tested. At least three samples were grown with each of the selected regimes. The values of heating time and power, which de®ne each regime, are shown in Table 1. After microwaving the substrates were taken out of the solution and rinsed. Temperature was measured immediately after the processes corresponding to each of the given regimes. Except for 20±120 and 20±30 regimes, the temperatures were less than 1008C. For the last two listed in Table 1, the temperature was below 908C. Actual ®lm growth temperatures could be larger due to the localized heating of the substrates originated from microwaves interacting with the ITO conducting layer on the glass [17]. 3. Film characterization and properties 3.1. Physical appearance and thickness of the ®lms After ®lm growth, samples with deposition time equal or above 30 min had a whitish appearance, due to the presence of TiO2 crystals on the surface. These crystals were loosely
13
stuck, mainly on the side of the glass without ITO layer, and strongly scattered light, causing the whitish appearance. They were easily removed mechanically and a colored transparent ®lm remained very well adhered. For characterization samples were thoroughly cleaned and sonnicated in deionized water. The ®lm was visible with the naked eye because of the color change observed by re¯ection. The ITO layers used were bluish when observed by light re¯ection. TiO2 layers on ITO showed different colors (see Table 1). These colored ®lms were very well adhered. Also the layer homogeneity was visible due to the interference phenomena. Layers were homogeneous in the central area. In some borders small interference rainbows formed indicating different thickness. It should be pointed out that the substrate cleaning is very determinant. A Tenkor Instrument was used for per®lometry measurements of the samples. In order to obtain a step, samples were covered partially with a Te¯on ribbon. Only gradual slopes were produced which together with the substrate original surface made the thickness measurement dif®cult and introduced larger experimental uncertainties. Samples produced with regimes 20±30 and 10±30 were the thicker ones (between 300 and 500 nm). The thinner layers were the yellow ones (see Table 1), with thickness under 300 nm. The samples grown during 2 h using 60 W microwave power (20±120 regime) do not seem to ®t the general scheme because they were as thin as those of regimes 10± 20 and 65±3. They are thinner than the ones grown with the same power and less time. On the other hand, they had more loosely stuck TiO2 after deposition than the rest. Thickness determination using SEM (Hitachi S-570, 10± 30 KV) was possible on one sample obtained with regime 20±120. Fig. 1 shows a thickness around 0.15 mm. 3.2. Surface topography The ®lm homogeneity was more closely revealed by the color intensity produced by interference when using optical microscopy. The same colors reported in Table 1 were observed for each layer under the microscope when using a re¯ection microscope. In some samples, even though, the layer could show only one color when observed with the naked eye, differences could be found under the microscope due to the presence of small not fully covered areas. Speci-
Table 1 Analyzed samples Regime a
Microwave power (W)
Heating time (min)
Re¯ection-interference color
Thickness range (nm)
20±120 20±30 10±30 10±20 65±3
60 60 30 20 195
120 30 30 20 3
Yellow Pinkish or green Orange yellow or pinkish Yellow Yellow
,300 300±500 200±400 ,300 ,300
a
Sample regime were named according to the percentage of maximum microwave power available (300 W) and microwaving time.
14
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
bution of the freshly obtained precipitates was analyzed using a Malvern Zetasizer 3 for some of the regimes (see Fig. 4). Larger agglomerates are formed in the solution compared with the corresponding size of the crystals on the layers. 3.3. X-ray diffraction
Fig. 1. SEM photograph showing a cross-sectional view of a TiO2 ®lm grown with regime 20±120. The TiO2 ®lm interface with the conducting layer is indicated with an arrow.
®cally, in ®lms grown with regime 20±120 scratches were visible. Films from different regimes were gold-covered and examined with scanning electron microscopy (SEM) to investigate their microcrystalline structure and surface characteristics. In Fig. 2a,b, ®lms grown during the same time but with different microwave power are compared. It is observed that samples 20±30 showed a better surface coverage than samples 10±30. Nonetheless, this was not the case for samples 20±120, where scratches were visible even with the optical microscope. The in¯uence of the used regime on crystal size is illustrated in the micrographs of Fig. 3. SEM images in this ®gure show crystals or grains with linear dimensions smaller than 100 nm. It is noteworthy how they are closely packed together. It can be observed that samples with smaller thickness show a higher nanocrystalline character, that is, they are formed by smaller grains. To complement the crystal size analysis, the mass distri-
Cu Ka radiation (l 0:154056 nm) was used for all Xray diffraction experiments performed with Rigaku Rota¯ex RU-200 B and Siemens D-3400 apparatus. X-ray diffraction of a 20±30 ®lm is shown in Fig. 5b together with diffraction corresponding to the ITO substrate (Fig. 5a). Comparison of these ®gures reveals the appearance of peaks at 25.3 and 37.88, which correspond to crystal planes of the anatase structure (101) and (004), respectively. The second one is very close to a peak of the ITO substrate. Taking into account the relative intensities, Ihkl, of the two peaks for anatase (I101 =I004 100=20), diffraction from planes (004) is relatively stronger than that from planes (101). These lines were also observed by common X-ray diffraction spectra corresponding to regime 10±30. For thinner layers corresponding to regimes 10±20, 65±3 and 20±120 low angle Xray diffraction was necessary. Low-angle X-ray diffraction spectrum is shown in Fig. 5c corresponding to a sample 10± 30. Lines are observed corresponding to the anatase structure at values of 2u equal to 25.3, 37.8, 48.1, and 53.9 which correspond to planes (101), (004), (200) and (105), respectively. Also in this low angle diffraction experiment, the line intensity relations are not those that correspond to complete random order of crystals. It should also be observed in Fig. 5c that peaks are very well-de®ned with a small half-width, usually attainable only after heat treatment of the TiO2 deposits (2).
Fig. 2. SEM photographs of TiO2 ®lms grown during the same time (30 min) but with different microwave power: (a) 60 W microwave power (regime 20±30) and (b) 30 W microwave power (regime 10±30).
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
15
Fig. 3. SEM photographs of TiO2 ®lms grown under different conditions: (a) 60 W microwave power during 120 min (regime 20±120), (b) 60 W microwave power during 30 min (regime 20±30), (c) 30 W microwave power during 30 min (regime 10±30) and (d) 30 W microwave power during 20 min (regime 10±20).
3.4. Optical transmittance Transmittance of TiO2 ®lms on conducting glass were measured with a Varian Cary 05 spectrophotometer using the conducting glass substrate as the blank. Spectra corresponding to the thicker layers from regime 20±30 and 10±30 are shown in Fig. 6a. Spectra have been processed in order to obtain the energy band-gap from the expression corresponding to indirect gap semiconductors
a hnhn / hn 2 Eg 2 Taking into consideration the specularity of the ®lms, it is assumed that specular re¯ectivity can be disregarded as compared to absorbance in the high absorption range. In this case, the absorbance is directly proportional to the absorption coef®cient. From the plot in Fig. 6b, a value of
the energy gap of 3.25 eV is obtained. This value is well over the reported value for rutile (3.0 eV) and corresponds satisfactorily with the reported values for anatase. 4. Discussion From the experimental results obtained the ®lm growth from solution using microwaves has proved to be a simple and inexpensive method. Small times of deposition can be used, which are not possible with conventional heating. As explained before, 80 ml of solution were used in order to study also the precipitates in the solution [18]; but much smaller amount of solution is required for growing the thickness obtained. In that case, power and/or time can be reduced. As compared to our previous experiments for
16
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
Fig. 4. Mass distribution per aggregate size of the freshly obtained precipitates for some of the used regimes.
growing TiO2 ®lms [6], layer growth is more ef®cient with the present precursor solution. In [6] several microwave processes with fresh solution were necessary, one on top of the other, to obtain a layer thickness in the range obtained in the present case with a single microwave heating process. It is expected that several successive processes could also be used with the present precursor solution to obtain thicker ®lms. If we compare thickness obtained in regimes 10±30 with
Fig. 5. XRD spectra corresponding to: (a) ITO glass substrate, (b) TiO2 ®lm on the ITO glass substrate obtained using 60 W microwave power during 30 min (regime 20±30). (c) Low-angle X-ray diffraction pattern corresponding to a TiO2 ®lm obtained using 30 W microwave power during 30 min (regime 10±30). The arrows indicate the peaks corresponding to TiO2 anatase.
Fig. 6. (a) Transmittance of the TiO2 ®lms on conducting glass for sample obtained with regimes 20±30 and 10±30 (conducting glass substrate was used as blank) and (b) spectral dependence of the absorbance showing energy gap value for these same samples.
20±30 and 10±30 with 10±20, the TiO2 ®lms seem to grow more as the microwave power increases and as the time increases. There is an optimum time for each microwave power after which the layers do no become any thicker. This is clear not only because of the thickness of the ®lms (Table 1) and ®lm coverage (Fig. 2a,b), but also because 120 min proved to be a very long time for 60 W microwave power (20% of maximum power). It is not completely clear why these ®lms are even thinner than those grown with the same power during 30 min. It is considered that in an early growth stage the ®lms are formed from TiO2 that nucleates on the surface; which is a different mechanism than that resulting from TiO2 precipitated in the solution and `sticking' to the surface. Apparently, the ®rst produces well-adhered layers. We also believe that nucleation on the surface might proceed at a faster pace than in the solution at the beginning of microwave heating helped by `localized heating' of the ITO glass due to its free electrons [17]. That is, microwaves may interact with free electrons and these may pass the acquired energy to lattice vibrations. The importance of such effect requires further research, in spite of the fact that, well-adhered TiO2 layers do not grow on the side of the glass without conducting layer. Once the crystals precipitated in the solution acquire a large concentration, their collisions on the surface should impair layer growth; on the other hand, `localized heating' must be less effective
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
Fig. 7. SEM photograph of TiO2 powder precipitated in the solution during a process in which 60 W microwave power during 30 min were used (regime 20±30).
because of the existence of the grown TiO2 layer. Further, it must be considered that ®lms suffer erosion from the precipitated TiO2 in the solution during the 90-min. difference between experiments with 20±120 regime and those with 20±30 regime. Certainly, as already mentioned scratches are visible on the surface. The highest concentration and the largest agglomerates of precipitated TiO2 in the solution are obtained for this 20±120 regime (see Fig. 4). Further experiments are required to fully explain this. On the other hand, as was pointed out before, crystals are smaller the thinner the layer (See Fig. 3 and Table 1). This indicates, and can be observed in some areas of the SEM photos in Fig. 3, that as growth proceeds new deposits cover more than one crystal and larger crystals form from smaller ones; the length of frontiers diminishing for the new grown material. Agglomeration of smaller, primary TiO2 particles to form larger ones was also observed by (2) and reference cited therein. We believe that both SEM photographs in Fig. 3a (regime 20±120) and in Fig. 3d (regime 10±20), show smaller nanocrystals produced during an early growth stage of the layer, in agreement with previous considerations. It is characteristic of the ®lms obtained that crystals are compactly stacked. If Fig. 3 is compared with Fig. 1b of Ref. [1] where TiO2 ®lms are also grown on conducting glass, a different growth habit is observed. We believe this might be due to an enhanced effect of the ITO conducting layer produced by microwaves since SEM images of crystals precipitated in the solution showed a different crystal habit more similar to Fig. 1b in Ref. [1]. In Fig. 7 a SEM
17
photograph of crystals precipitated in the solution in a 20± 30 regime is shown. Flower like crystals are observed, as well as, in Ref. [1] where TiO2 layers are also deposited on conducting glass. Closer and more ordered packing is observed for the microwave-grown layers on ITO (see Fig. 3). Therefore, the reported microwave method could be of interest for obtaining the thin, compact TiO2 layer used in dye-sensitized solar cells to avoid direct contact of the liquid or solid electrolyte with the conducting layer [8]. The ordered packing could be due to preferential orientation of TiO2 nanocrystallites. In Fig. 5c, the relations of peak intensities corresponding to TiO2 anatase planes do not correspond with the known relations for complete random order. This indicates a preferential growth. The narrow diffraction peaks observed in the low angle diffraction pattern of Fig. 5c con®rm the high degree of crystallinity in spite of small crystal dimensions. Even without any sintering, they show higher crystallinity than other reports [4,6]. The anatase phase present is highly convenient for some applications of these layers [20,21]. Simplicity, low temperature, fast growth, good adherence, closed-packing, good crystallinity without sintering, anatase phase and nanostructured-character are some of the advantages found with the performed deposition experiments. This new method using microwave heating could provide faster and better results for thin ®lms grown from solution than obtained when using other procedures as chemical bath deposition (CBD) [19] or soft-solution methods [16]. Full explanation of thin ®lm deposition mechanisms and of the exact role of microwaves deserve special attention and need further research. 5. Conclusion TiO2 ®lms deposited using the reported new technique show good adherence and specularity. The layers are nanostructured and closely packed for some of the different regime employed. X-ray diffraction experiments show narrow peaks revealing a good crystallinity usually obtained only after sintering when using other deposition from solution methods. Microwave heating required short times, low temperatures and is relatively inexpensive. The characteristics of the ®lms, as well as, the ease of the layer deposition, indicate that this is a perspective technique for TiO2 ®lms for different applications. It should also be possible to use it for obtaining other materials ®lms grown from solution. References [1] L. Kavan, M. GraÈtzel, Electrochim. Acta 40 (1995) 643. [2] S.D. Burnside, V. Shklover, C. BarbeÂ, et al., Chem. Mater. 10 (1998) 2419. [3] P. Sawunyama, A. Yasumori, K. Okada, Mater. Res. Bull. 33 (1998) 795. [4] S. Deki, Y. Aoi, J. Mater. Res. 13 (1998) 883.
18
E. Vigil et al. / Thin Solid Films 365 (2000) 12±18
[5] S. Yin, Y. Inoue, S. Uchida, Y. Fujishiro, T. Sato, Inoue, J. Mater. Res. 13 (1998) 844. [6] E. Vigil, L. Saadoun, R. RodrõÂguez-Clemente, J.A. AylloÂn, X. DomeÁnech, J. Mater. Sci. Lett. 18 (1999) 1067. [7] B. O'Regan, M. GraÈtzel, Nature 353 (1991) 737. [8] U. Bach, D. Lupo, P. Comte, J.E. Moser, F. WeissoÈrtel, J. Salbeck, H. Spreitzer, M. GraÈtzel, Nature 395 (1998) 583. [9] K. Tennakone, G.R.R.A. Kumara, A.R. Kumarasinghe, K.G.U. Wijayantha, P.M. Sirimane, Semicond. Sci. Technol. 10 (1995) 1689. [10] S. Siebentritt, K. Ernst, C.-H. Fischer, R. Konenkamp, M.C. LuxSteiner, in: H.A. Ossenbrink, P. Helm, H. Ehmann (Eds.), Proc. 14th Eur. Photovoltaic Sol. Energy Conf. Exhibit., Barcelona, H.S. Stephens & Associates, Bedford, UK, 1997, pp. 1823±1827. [11] R.L. Pozzo, M.A. Baltanas, A.E. Cassano, Catalysis Today 39 (1997) 219. [12] T.-S. Kang, D. Kim, K.-J. Kim, J. Electrochem. Soc. 145 (1998) 1982.
[13] Y. Paz, Z. Luo, L. Rabenberg, A. Heller, J. Mater. Res. 10 (1995) 2842. [14] A. Kay, M. GraÈtzel, Sol. Energy Mater. Sol. Cells 44 (1996) 99. [15] C.J. BarbeÂ, F. Arendse, P. Comte, et al., J. Am. Ceram. Soc. 80 (1997) 3157. [16] M. Yoshimura, J. Mater. Res. 13 (1998) 796. [17] C. Gabriel, S. Gabriel, E.H. Grant, B.S.J. Halstead, D.M.P. Mingos, Chem. Soc. Rev. 27 (1998) 213. [18] L. Saadoun, J.A. AylloÂn, E. Vigil, R. RodrõÂguez-Clemente, X. Domenech, unpublished. [19] P.K. Nair, M.T.S. Nair, V.M. Garcia, et al., Sol. Energy Mater. Sol. Cells 52 (1998) 313. [20] L. Kavan, M. Gratzel, J. Rathousky, A. Zukal, J. Electrochemical Soc. 143 (1996) 394. [21] L. Kavan, K. Kratochvilova, M. Gratzel, J. Electroanal. Chem. 394 (1995) 93.