Author's personal copy

Available online at www.sciencedirect.com

Thin Solid Films 516 (2008) 957 – 966 www.elsevier.com/locate/tsf

Natural and persistent superhydrophilicity of SiO2/TiO2 and TiO2/SiO2 bi-layer films S. Permpoon a,b , M. Houmard a,b , D. Riassetto b , L. Rapenne b , G. Berthomé a , B. Baroux a , J.C. Joud a , M. Langlet b,⁎ a

Laboratoire de Thermodynamique et de Physico-Chimie Métallurgique, ENSEEG-INPG, BP 75, Domaine Universitaire, 38402 Saint Martin d'Hères, France b Laboratoire de Matériaux et de Génie Physique, ENSPG-INPG-MINATEC, 3 parvis Louis Néel, BP 257, 38016 Grenoble Cedex 1, France Received 11 July 2006; received in revised form 19 April 2007; accepted 4 June 2007 Available online 13 June 2007

Abstract Sol–gel SiO2/TiO2 and TiO2/SiO2 bi-layer films have been deposited from a polymeric SiO2 solution and either a polymeric TiO2 mother solution (MS) or a derived TiO2 crystalline suspension (CS). The chemical and structural properties of MS and CS bi-layer films heat-treated at 500 °C have been investigated by Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, and transmission electron microscospy. Water contact angle measurements show that MS SiO2/TiO2 and CS TiO2/SiO2 bi-layer films exhibit a natural superhydrophilicity, but cannot maintain a zero contact angle for a long time over film aging. In contrast, CS SiO2/TiO2 bi-layer films exhibit a natural, persistent, and regenerable superhydrophilicity without the need of UV light. Superhydrophilic properties of bi-layer films are discussed with respect to the nature of the TiO2 single-layer component and arrangement of the bi-layer structure, i.e. TiO2 underlayer or overlayer. © 2007 Elsevier B.V. All rights reserved. Keywords: Superhydrophilicity; Sol–gel thin films; SiO2–TiO2 system; Self-cleaning applications

1. Introduction It is well known that the photo-induced hydrophilicity of titanium oxide, preferentially in its anatase polymorphic form, confers a self-cleaning functionality to TiO2 surfaces. This behaviour ensues from surface oxygen vacancies (O2⁎), which are created through an oxydo-reduction of TiO2 (Ti4+ + e− → Ti3+ and 2O2− + 2h+ → O2⁎) induced by photo-generated electron (e−)/hole (p+) pairs. Surface oxygen vacancies can then be saturated by OH groups, through a molecular or dissociative adsorption of atmospheric water, which yields a superhydrophilic surface, i.e. a surface showing a water contact angle of zero [1,2]. However, the photo-induced superhydrophilicity does not persist in time in the absence of UV radiation, which limits its field of application because in real conditions surfaces are not permanently exposed to UV. A large number of articles have reported that the addition of SiO2 into TiO2 films enhances the photo-induced super⁎ Corresponding author. Tel.: +33 4 56 52 93 22; fax: +33 4 56 52 93 01. E-mail address: [email protected] (M. Langlet). 0040-6090/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2007.06.005

hydrophilicity, which can be maintained for a certain time in the absence of UV radiation [3–8]. The effects of SiO2 addition on the photo-induced hydrophilicity of TiO2 films have been studied [5,9,10]. Published works suggest that the improved hydrophilicity of SiO2–TiO2 composite films ensues from an enhanced acidity of Si–O–Ti bonds at the SiO2–TiO2 interfaces, which would induce a greater amount of hydroxyl groups at the film surface. Several models have been proposed to describe acidity of SiO2–TiO2 composites, which attribute acidity to the charge imbalance developed along Si–O–Ti heterolinkages owing to the difference in the coordination geometry of Si4+ and Ti4+ cations [11–14]. Lewis and/or Bronsted acid sites are thus formed. Some authors have considered interface interactions in SiO2–TiO2 bi-layer films with respect to the local structures, either on SiO2 or TiO2 overlayers [12,15–17]. Through X-ray photoelectron spectroscopy (XPS) experiments, Sanz et al. evidenced the formation of cross-linking Ti–O–Si bonds at the TiO2–SiO2 planar interface, which led to significant core level shifts and changes in the electronic structure of a SiO2 outer surface [15]. Gao and Wachs reported a Raman study on TiO2– SiO2 bi-layer films that confirmed the formation of Ti–O–Si

Author's personal copy

958

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

bonds at the planar interface, which were shown to be similar to those evidenced for SiO2–TiO2 composite films [12]. Other authors also compared the interactions at SiO2–TiO2 planar interfaces for SiO2 and TiO2 overlayers [16,17]. They proposed a similar description for both interfaces, which were formed by the cross-linking of Ti–O–Si bonds, as determined by angular XPS [16] and Electron Energy Loss Spectroscopy (EELS) [17]. All these articles did not report on the hydrophilic properties of bi-layer films. More recently, a sustained superhydrophilicity has been reported when a TiO2 film is thoroughly covered by a SiO2 overlayer [18,19]. Guan et al. mentioned an enhanced amount of OH groups at the SiO2 outer surface and so-formed Si–OH surface bonds were thought to be more stable than Ti–OH surface bonds, which increased in-time persistence of the superhydrophilicity [18]. However, Hattori et al. indicated that the rate of photo-wettability leading to superhydrophilicity decreased with increasing the thickness of a SiO2 overlayer [19]. Tada et al. also indicated that a SiOx coated TiO2 film showed improved photocatalytic activity only in the case of a very thin SiOx monolayer [20]. This enhanced activity was attributed to an increase in the electrostatic attraction of adsorbents at the outer surface, which ensued from a SiO2–TiO2 interfacial charge transfer. In a previous article, we have indicated that sol–gel derived SiO2–TiO2 composite films could show a natural and persistent superhydrophilicity, i.e. a water contact angle of zero without UV radiation [21]. This property had never been reported before for such films. According to previous literature, we tentatively attributed this natural superhydrophilicity to an enhanced acidity at SiO2–TiO2 granular interfaces. We were then encouraged to test whether this unusual property could exist in SiO2–TiO2 bilayer films. In this work, sol–gel derived bi-layer films deposited from two kinds of TiO2 sols have been investigated. The film hydrophilic properties and their in-time persistence are studied and discussed with respect to the nature of TiO2 films and arrangement of the bi-layer structures, i.e. SiO2 or TiO2 overlayer. 2. Experimental details

0.13. The solution was aged at room temperature for 2 days before deposition. The second method relied on the preparation of a crystalline suspension (CS) of TiO2 nano-crystallites in absolute ethanol [24]. This suspension was prepared from the mother solution using a multistep procedure. The mother solution was firstly diluted in an excess of deionized water (H2O/TIPT molar ratio of 90) and then autoclaved at 130 °C for 6 h. Autoclaving yielded the crystallization of TiO2 particles in the liquid phase. An exchange procedure was then performed in order to remove water from the sol and to form a crystalline suspension in absolute ethanol. The final TiO2 concentration in ethanol was 0.24 M. For more data, the whole procedure has been described in a previous paper [24]. The final sol was composed of TiO2 nano-crystallites of about 6 nm in diameter. Previous works showed that SiO2 as well as MS and CS TiO2 sols were very stable, which indicated that no gelation took place in polymeric SiO2 and TiO2 (MS) sols, while no significant crystal aggregation took place in CS TiO2 sols. Consequently, all these sols could be used for several weeks in reproducible film deposition conditions. Single- and bi-layer films were deposited at room temperature on (100) silicon wafers by spin-coating (300 μL of each sol, spin speed of 3000 rpm). Prior to deposition, the substrates were ultrasonically cleaned with ethanol for 3 min, then rinsed with distilled water, and dried with air spray. Single-layer SiO2 and TiO2 films were first deposited and heat-treated at 500 °C for 2 h. Some single-layer films were studied as-prepared. Bi-layer films were produced through the subsequent deposition of a complementary SiO2 or TiO2 single-layer component followed by additional heat-treatment at 500 °C for 2 h. For comparison, intermediary and final heat-treatments at 110 °C were also punctually tested. Heat-treatments were performed in air and the samples were directly introduced in the pre-heated oven. After heat-treatment, the films were cooled to room temperature under ambient condition. Hereafter, structures composed of a TiO2 (SiO2) layer deposited on a SiO2 (TiO2) layer are denoted as TiO2/SiO2 (SiO2/TiO2) bi-layer films. In this study, the thickness of SiO2 and TiO2 components were fixed at around 200 and 40 nm, respectively, as shown by ellipsometric measurements performed on single-layer films. Some deviations from these data will be discussed in Section 3.4.

2.1. Sol and film preparations 2.2. Characterizations Sol–gel derived films were deposited from silica and titania sols. A SiO2 polymeric sol was prepared by diluting tetraethoxysilane (TEOS) in absolute ethanol, deionized water, and hydrochloric acid (HCl), according to a previously published procedure [22]. A concentrated sol was first prepared with a TEOS concentration of 2.35 M, a H2O/TEOS molar ratio of 2.2, and a pH of 3.5. This solution was aged at 60 °C for 2 days. Then, it was diluted in additional absolute ethanol to get a final TEOS concentration of 1.5 M. TiO2 films were deposited from two kinds of TiO2 sols which were prepared using two different sol–gel routes. A classical method yielded a polymeric mother solution (MS), which was prepared by mixing tetraisopropyl orthotitanate (TIPT) with deionized water, hydrochloric acid, and absolute ethanol as a solvent [23]. TIPT concentration in the solution was 0.4 M, and the TIPT/H2O/HCl molar composition was 1/0.82/

The films were characterized by Fourier transform infrared (FTIR) transmission spectroscopy in the range of 4000– 250 cm−1 with a resolution of 4 cm−1 using a Bio-Rad FTS165 spectrometer. Spectra corresponding to 300 scans were recorded in room atmosphere after purging the measurement chamber with dry air for 15 min. The spectra were analyzed after subtraction of the bare substrate spectrum. Surface analysis was performed by XPS using a XR3E2 apparatus from Vacuum Generator employing an Mg Kα source (1253.6 eV). The X-ray source was operated at 15 kV for a current of 20 mA. Before collecting data, the samples were put in equilibrium for 24 h in an ultra high vacuum chamber (10−10 mbar). Photoelectrons were collected by a hemispherical analyzer at 30° take-off angle. All spectra were calibrated with C1s peak at 284.7 eV.

Author's personal copy

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

Fig. 1. Water contact angle variations vs aging time in the absence of UV radiation for TiO2 (a), SiO2 (b), TiO2/SiO2 (c), and SiO2/TiO2 films (d) heattreated at 500 °C. TiO2-derived single- and bi-layer films were deposited from a MS sol.

Transmission electron microscope (TEM) studies were performed on film cross-sections. For these studies, crosssectional samples were first thinned down to 5–10 μm by tripod polishing. Then, thinning was completed by argon ion-milling to obtain a very thin preparation compatible with electron transmission. In order to avoid irradiation damage during ion thinning, a PIPS-GATAN ion-milling apparatus was used with low angle (6°) and low voltage (2.5–3 kV). High resolution TEM (HRTEM) and energy filtered TEM (EFTEM) studies were performed with JEOL-2010 LaB6 and JEOL-2010 FEF instruments, respectively, both operating at 200 keV. EFTEM images were extracted from EELS spectra using a ‘threewindow’ method that allowed extrapolating the EELS spectrum background, for background correction, and filtering a small energy window of 10 eV around the Si–L2,3 peak at 108.5 eV. This method yielded EFTEM images that allowed a mapping of silicon through the film cross-section. Surface hydrophilicity of the films was quantified from water contact angle measurements. Experiments were performed at 20 °C in an environmental chamber using a KRUSS G 10 goniometer connected with a video camera. Several water droplets of 0.5 μL volume were spread on the samples and water contact angles were measured at different points of the thin film surface for statistical purpose. The effects of film aging on the hydrophilic properties were analyzed using a same statistical procedure. During aging, the films were stored in a dark place, in such a way that aging effects only traduced natural hydrophilic properties of the films, i.e. did not traduce any photo-induced effects. In this work, no exposure to UV was performed.

959

measured on SiO2 and TiO2 single-layer films. Films illustrated in Figs. 1a–d and 2a–d were heat-treated at 500 °C, while the bilayer film illustrated in Fig. 2e was heat-treated at 110 °C. We first pay attention to films heated at 500 °C. It is observed that freshly prepared MS and CS TiO2 single-layer films (Figs. 1a and 2a) present hydrophilic properties comparable to those of a SiO2 film (Figs. 1b and 2b). These films exhibit a natural contact angle of around 15–20°. After 8 weeks aging in the dark, the water contact angle of both MS and CS TiO2 films has increased by about 35– 40°. In contrast, the SiO2 film exhibits a slow contact angle increase over the same aging period, which illustrates the natural hydrophilicity of a silica surface and the greater stability of Si– OH surface bonds. TiO2/SiO2 bi-layer films (TiO2 overlayer) exhibit an initial water contact angle of 6° and 0° for MS (Fig. 1c) and CS (Fig. 2c) films, respectively. These values, which are much weaker than those measured on TiO2 and SiO2 single-layer films, indicate a natural superhydrophilicity that can be related to the existence of planar TiO2–SiO2 interfaces. As explained in Introduction, such interfaces are probably characterized by Si– O–Ti heterolinkages. However, the natural superhydrophilicity of TiO2/SiO2 bi-layer films does not persist in time, i.e. the contact angle of MS and CS TiO2/SiO2 films increases by about 35–40° over an aging period of 8 weeks. Both MS (Fig. 1d) and CS (Fig. 2d) SiO2/TiO2 bi-layer films (SiO2 overlayer) initially show a natural superhydrophilicity with a water contact angle of zero. In contrast to TiO2/SiO2 films, this property can maintain for a certain aging period. Natural superhydrophilicity of the SiO2/ TiO2 (MS) film persists for a period of 2 weeks. After that, the contact angle is observed to slowly increase with time, reaching a value of 13° after aging for 8 weeks. Persistence of the natural superhydrophilicity is much longer for the SiO2/TiO2 (CS) film, since no contact angle increase is observed over an aging period of 8 weeks. Here again, these properties can be related to the existence of a SiO2–TiO2 interface. However, our data indicate that this interface promotes a better property when the bi-layer is constituted of a SiO2 overlayer rather than a TiO2 one. Such natural and persistent superhydrophilic properties of SiO2/TiO2 bi-layer films have never been reported before. These unusual properties can first be associated to the greater natural hydrophilicity of a SiO2 surface. However, two arguments

3. Results and discussions 3.1. Hydrophilicity of single- and bi-layer films The hydrophilic properties of SiO2/TiO2 and TiO2/SiO2 bilayer films were studied with respect to aging time. Figs. 1 and 2 depict in-time variations over aging of the natural water contact angle, i.e. water contact angle measured without UV irradiation, for bi-layer films deposited from MS and CS TiO2 sols, respectively. Contact angle variations are also compared to those

Fig. 2. Water contact angle variations vs aging time in the absence of UV radiation for TiO2 (a), SiO2 (b), TiO2/SiO2 (c), and SiO2/TiO2 films (d) heattreated at 500 °C, and for a SiO2/TiO2 film heat-treated at 110 °C (e). TiO2derived single- and bi-layer films were deposited from a CS sol.

Author's personal copy

960

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

indicate that superhydrophilic properties of SiO2/TiO2 bi-layer films do not entirely rely on the intrinsic properties of the SiO2 overlayer. Firstly, a pure SiO2 film does not exhibit any superhydrophilicity. Secondly, for a same structure, a SiO2/TiO2 (MS) film cannot maintain its natural superhydrophilicity for a long time, while a SiO2/TiO2 (CS) film exhibits a much more persistent superhydrophilicity. This difference not only confirms that superhydrophilic properties are related to interface effects but also shows that nature of the TiO2 underlayer influences these effects. Let us recall that CS TiO2 films are deposited from a suspension of nano-crystallites. As-deposited CS films are thus well crystallized, while a heat-treatment at around 500 °C is necessary to reach a similar crystallization degree in MS films. Previous studies showed that physico-structural properties of CS TiO2 films did not vary significantly with heat-treatment temperature up to 500 °C [25]. Thus, a heat-treatment was performed at 110 °C in order to test the wettability properties of a SiO2/TiO2 (CS) bi-layer film processed at low temperature. Fig. 2e shows that this film exhibits a natural water contact of 65°, which slowly increases over aging. It appears therefore that, contrary to a same SiO2/TiO2 film heat-treated at 500 °C, this film does not present any natural hydrophilicity. As explained above, such a discrepancy cannot be related to any differences in physico-structural properties of TiO2 components heat-treated at 110 or 500 °C. It is therefore possible that the formation of a SiO2–TiO2 interface able to promote enhanced surface hydrophilicity requires a sufficiently high temperature heat-treatment. Such a treatment would in turn influence the formation of Si–O–Ti heterolinkages at the planar interface. Furthermore, Fig. 2b and e show that, for a same SiO2 outer surface, a SiO2/TiO2 film heat-treated at 110 °C is much less hydrophilic than a SiO2 single-layer film heat-treated at 500 °C. Alkoxy groups arising from the silicon precursor are still present in SiO2 films heat-treated at 110 °C, while they are completely decomposed after heat-treatment at 500 °C [22]. It can therefore be inferred that the presence of alkoxy groups strongly affect the natural hydrophilicity of a SiO2 surface. Thus, in the following parts we will focus on films heat-treated at 500 °C. To summarize, wettability measurements suggest that the occurrence and persistence of natural superhydrophilicity rely on two separate aspects. On the one hand, occurrence of a natural superhydrophilicity seems to be governed by the presence and nature (MS or CS) of a SiO2–TiO2 interface. On the other hand, it is interesting to note that all single- and bi-layer films with an outer TiO2 layer lose their hydrophilic properties with a comparable rate, i.e. the contact angle increases by about 35– 40° over an aging period of 8 weeks, while for single- and bi-layer films with an outer SiO2 layer, the contact angle increases by 15° or less over the same aging period. It suggests that the rate of hydrophilicity loss follows two kinds of distinct behaviours depending on the outer surface nature, irrespective of the presence or not of an interface. Best properties of a SiO2/TiO2 CS film would therefore arise from a combination of interface and outer layer effects. However, even such a film was observed to lose its superhydrophilicity after a prolonged aging period. A contact angle value of 5° was measured after 12 weeks aging. This film was then submitted to aspersion with cold (20 °C) deionized water for 1 min. The contact angle of 0° measured after aspersion

showed that this film could easily recover its natural superhydrophilicity, which was again observed to persist over an additional aging period. This behaviour is meaningful since it indicates that the superhydrophilicity of SiO2/TiO2 CS films can be maintained for a long period of time through a simple periodic water rinsing. Such a property allows thus envisaging a long-term self-cleaning functionality without the presence of UV radiation. In contrast, this behaviour could not be observed on a SiO2/TiO2 MS film. After a same aging period of 12 weeks, this film exhibited a contact angle of 16°. Water aspersion yielded a contact angle of 5°, but this angle was observed to increase again immediately after aspersion, reaching a value of 16° after further aging for 1 week. These data do not mean that, in the latter case, superhydrophilicity cannot be regenerated in more adapted conditions, but they confirm that the TiO2 underlayer nature strongly influences the superhydrophilic properties of SiO2/TiO2 bi-layer films, best properties being reached from a crystalline suspension approach. 3.2. XPS characterization XPS is a suitable method traditionally used to study surface chemical properties, which can provide insights in surface mechanisms yielding a natural superhydrophilicity. In this work, XPS was used for investigating the surface chemical state of freshly deposited single- and bi-layer films, i.e. the films were studied within the first 24 h following deposition. As mentioned in the Introduction, the hydrophilic properties of SiO2 and TiO2 surfaces rely on the presence of surface OH groups. Thus, we first paid attention to these species, which were studied from a deconvolution of the O1s peak located at around 533 and 530 eV for SiO2 and TiO2 outer surfaces, respectively. The O1s peak could be decomposed in two components, i.e. O–H and Si–O (or Ti–O) components, using 10% Lorentzian/Gaussian functions (see an illustration in insert of Fig. 3). The O–H component essentially traduces the presence of surface hydroxyl groups. Other possible C–O and H2O components were not accounted for owing to their very weak intensity. Fig. 3 shows the relative intensity of the OH component, normalized with respect to the total O1s peak intensity, for single- and bi-layer films heat-

Fig. 3. Variations of the relative intensity of the O–H component, as deduced from deconvolution of the XPS O1s peak, for various films deposited from MS ( ) and CS (□) sols and subsequently heat-treated at 500 °C. Insert shows an example of deconvolution of the O1s peak for a SiO2 film.

▪

Author's personal copy

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

▪

Fig. 4. Variations of the C1s peak relative intensity for various films deposited from MS ( ) and CS (□) sols and subsequently heat-treated at 500 °C.

treated at 500 °C. The surface hydroxyl amount is similarly high (35% relative intensity of the O–H component) for the SiO2 single-layer and SiO2/TiO2 bi-layer films, irrespective of the MS or CS TiO2 component nature, while it is similarly much weaker (12% relative intensity) for the TiO2 single-layer and MS or CS TiO2/SiO2 bi-layer films. On the one hand, the greater amount of surface OH groups illustrates the fact that a silica surface is naturally more hydrophilic than a titania one. On the other hand, data of Fig. 3 do not depict any differences in the amount of surface OH groups for a SiO2 (TiO2) single-layer film and SiO2/ TiO2 (TiO2/SiO2) bi-layer films, which might be correlated with differences in their hydrophilic properties, as illustrated in Figs. 1 and 2. It is known that carbon contamination present at the outer surface can drastically alter the film hydrophilicity. Thus, we also paid attention to the C1s XPS peak (284.7 eV). Intensity of this peak, normalized with respect to the total intensity of O1s and Ti2p (or Si2p) peaks, is presented in Fig. 4 for the same films as those illustrated in Fig. 3. Since previous studies have shown that alkoxy groups, which might contribute to the C1s peak, are not present in SiO2 and TiO2 films heated at 500 °C [21,22], it is inferred that this peak essentially reflects the amount of carbon contamination at the outer surfaces. Fig. 4 indicates that a TiO2 single-layer and TiO2/SiO2 bi-layer films (MS or CS) suffer a similarly high contamination, while this contamination is much weaker and comparable for a SiO2 single-layer and SiO2/TiO2 bi-layer films (MS or CS). Thus, a clear correlation can be observed between data illustrated in Figs. 3 and 4, i.e. compared to carbon contamination, the amount of hydroxyl groups exactly follows reverse trends with respect to film nature. On the one hand, this correlation traduces that carbon contamination is a limiting factor for hydrophilic properties and/or more hydrophilic surface are less prone to be contaminated than less hydrophilic ones. On the other hand, it confirms that the amount of surface OH groups and the level of carbon contamination depicted by XPS are not correlated at all with the water contact angles measured on fresh films. Let us recall that the films were put in equilibrium for 24 h in ultra high vacuum before collecting XPS data. The vacuum atmosphere can influence the surface chemical properties by promoting partial carbon and water desorption, which in turn relies on the respective surface reactivity and affinity of each

961

sample toward carbon and water. Thus, XPS data do not necessarily correspond to the actual surface properties and cannot strictly be correlated to hydrophilic properties of the films, which are measured in ambient atmosphere. It is therefore inferred that XPS data more presumably traduce the ability of the surface to retain OH groups rather than the actual OH amount. Accordingly, a comparison between Figs. 1 2 and 3 shows that there is a close correlation between the rate of hydrophilicity loss, which is related to a lack of retention of surface OH groups, and the OH amount depicted by XPS data. All films with a weak and similar surface OH amount (singleand bi-layers with an outer TiO2 layer) lose their hydrophilic properties with a fast and comparable rate, while films with a greater and similar surface OH amount (single- and bi-layers with an outer SiO2 layer) do not show any contact angle variation or follow a very slow rate of contact angle increase. In summary, data illustrated in Figs. 1–4 seem to confirm that the rate of hydrophilicity loss is governed by the nature of the overlayer, i.e. the superhydrophilicity persistence relies on the presence of a SiO2 overlayer, irrespective of the presence of an interface. This SiO2 overlayer influence can be related to the weaker sensitivity of silica surfaces to carbon contamination, compared to titania surfaces, and to the greater stability of Si– OH surface groups compared to Ti–OH surface groups. 3.3. FTIR characterization According to previous arguments, it is believed that XPS data do not traduce actual surface chemical states of our films. For this reason, we performed complementary studies using FTIR spectroscopy. Since FTIR spectra are collected in ambient atmosphere, they should be less influenced by characterization conditions than XPS data. However, atmospheric water can adsorb at the film surface during spectrum acquisition, which can bother the final interpretation of OH absorption bands. Thus, in order to minimize water adsorption, the measurement chamber was systematically purged with dry air for 15 min before collecting IR data. FTIR studies were performed on a

Fig. 5. FTIR spectra in the low wavenumber region for TiO2 (a), SiO2 (b), TiO2/ SiO2 (c), and SiO2/TiO2 films (d). TiO2-derived films were deposited from a CS sol. All films were heat-treated at 500 °C. Spectra were collected before (dotted lines) and after (continuous lines) aging in the dark for 1 month. Insert shows a magnified view of the 1000–750 cm− 1 region for a freshly deposited SiO2/TiO2 film.

Author's personal copy

962

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

Fig. 6. FTIR spectra in the hydroxyl region for TiO2 (a), SiO2 (b), TiO2/SiO2 (c), and SiO2/TiO2 films (d). TiO2-derived films were deposited from a CS sol. All films were heat-treated at 500 °C. Spectra were collected before (dotted lines) and after (continuous lines) aging in the dark for 1 month.

SiO2 single-layer film and CS-derived single- and bi-layer films heat-treated at 500 °C, before and after aging in the dark for one month. As shown in Fig. 5, in the low wavenumber region, i.e. between 1200 and 250 cm−1, single- and bi-layer film spectra exhibit the typical absorption bands of Si–O–Si bonds in silica (1075, 800, and 440 cm−1) and Ti–O–Ti bonds in anatase (425 and 260 cm−1). A detailed assignment of these bands has been made in our previous article [21]. Fig. 5 shows that silica bands slightly increase in intensity over aging, this feature being clearly observed for the most intense band at 1075 cm−1, while no variation of anatase bands can be observed. These features will be discussed below. A weak band is also observed at around 920 cm−1 for the SiO2 film and the bi-layer films (see an illustration in insert of Fig. 5 for a freshly deposited SiO2/TiO2 film). This band is not detected for a TiO2 single-layer film, and can be attributed to a very small amount of Si–OH groups in the SiO2 component and/or Si–O–Ti heterolinkages at the SiO2– TiO2 interface of bi-layer films [26,27]. However, owing to its weak intensity, no reliable conclusion could be drawn about the exact assignment of this band and no significant variation could be evidenced with aging. IR absorption spectra in the hydroxyl region (3800– 3200 cm−1) are illustrated in Fig. 6. Following observations

can be done for freshly deposited films. In the case of a TiO2 film (Fig. 6a), no signal is detected showing not only that the film is free of any detectable Ti–OH or water species within its thickness, but also that the amount of surface OH groups, which might account for the moderate hydrophilicity of this film (Fig. 2a), is below the sensitivity threshold of our spectrometer. The IR spectra of SiO2-derived single- and bi-layer films show a broad absorption band between 3800 and 3200 cm−1. For silica samples, it is well known that this band arises from the overlapping of stretching vibrations corresponding to different kinds of hydroxyl groups [14,26]. Basically, this band is composed of three regions: i/ a large component with an absorption maximum centered around 3500 cm−1, which is commonly associated to absorbed free water and to Si–OH groups (silanols) linked to molecular water through hydrogen bonds, the latter giving rise to a component located at higher wavenumbers than the former, ii/ an intermediary region around 3650 cm−1, where pairs of silanols mutually linked through hydrogen bonds are depicted, and iii/ a high wavenumber peak or shoulder at around 3740 cm−1, which corresponds to isolated surface silanols. Literature indicates that similar hydroxyl bands are observed for titania samples, bands corresponding to free or linked Ti– OH groups being slightly shifted toward lower wavenumbers compared to silica [14,28]. The OH band depicted for a SiO2 single-layer film (Fig. 6b) and a TiO2/SiO2 bi-layer film (Fig. 6c) shows a local maximum at around 3500 cm−1 and a pronounced high wavenumber shoulder at around 3740 cm−1, which indicate the presence of water and Si–OH groups. These spectra are very similar in their intensity and shape, which suggests that both films contain a comparable amount of water and OH groups. This observation contrasts with XPS data illustrated in Fig. 3 for same films, which depicted a greater amount of surface OH groups for the SiO2 single-layer. Besides, it is not possible to correlate infrared OH bands with hydrophilicity differences depicted for these films in Fig. 2b and c. However, it should be mentioned that, since spectra are collected in transmission, OH bands evidenced by FTIR spectroscopy do not necessarily correspond to species present at the film surface. It cannot be excluded that OH species are present in the thickness of a SiO2 film. In particular, the presence of

Fig. 7. Cross-section TEM images of SiO2/TiO2 (CS) (a), SiO2/TiO2 (MS) (b), and TiO2/SiO2 (CS) (c) bi-layer films heat-treated at 500 °C.

Author's personal copy

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

963

Fig. 8. Cross-section HRTEM images showing the SiO2–TiO2 interfacial region of SiO2/TiO2 (CS) (a), SiO2/TiO2 (MS) (b), and TiO2/SiO2 (CS) (c) bi-layer films heattreated at 500 °C.

silanol species would ensue from the incomplete polycondensation of hydrolyzed alkoxy groups in the silica network. The similarity between spectra of SiO2 and TiO2/SiO2 films might therefore express that the signal of eventual surface OH species is very weak and OH bands essentially depict residual water and/or Si–OH groups present in the film thickness, which are comparable for both SiO2 film components. Compared to the spectra of SiO2 and TiO2/SiO2 films, that of a SiO2/TiO2 bilayer film (Fig. 6d) exhibits a more intense OH band, this difference being particularly marked in the water region around 3500 cm−1. Since the SiO2 single-layer film and the outer SiO2 layer of the SiO2/TiO2 bi-layer film underwent the same heattreatment at 500 °C for 2 h, it is inferred that the content in water and Si–OH groups within thickness is comparable for both films. It can therefore be concluded that differences depicted in the spectra of SiO2 and SiO2/TiO2 films illustrate a much greater amount of surface OH species for the latter, which can in turn be related to its natural superhydrophilicity, as illustrated in Fig. 2d. These conclusions are supported by FTIR characterizations performed on the same films after 1 month aging in the dark.

The TiO2 film spectrum (Fig. 6a) does not reveal any evolution over aging, i.e. no OH species are detected in the aged film spectrum. In contrast, spectra of the SiO2 (Fig. 6b) and TiO2/ SiO2 films (Fig. 6c) illustrate significant modifications. The silanol-related high wavenumber shoulder is less pronounced after aging, while the free water component appears to be comparatively more intense. Such changes possibly traduce a dehydroxylation mechanism according to the condensation reaction: Si–OH + Si–OH → Si–O–Si + H2O. This condensation reaction can in turn explain a slight increase of the Si–O–Si band intensities observed in the spectra of aged SiO2 and TiO2/ SiO2 films (Fig. 5b and c). However, as previously mentioned, it is difficult for these films to infer on the respective contribution of surface species and species present within the thickness. The condensation reaction was also evidenced for the aged SiO2/ TiO2 bi-layer film (Fig. 5d). Besides, Fig. 6d indicates a noticeable intensity decrease of the OH band after aging this film for 1 month. This decrease is appreciated in the whole OH (water and silanol) region and the intensity decrease in the silanol region is more marked than for aged SiO2 and TiO2/SiO2 films. However, this greater decrease does not necessarily imply

Fig. 9. HRTEM (a) and EFTEM images (b) of a same CS SiO2/TiO2 cross-sectional area. The CS SiO2/TiO2 film was heat-treated at 500 °C.

Author's personal copy

964

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

that the aged SiO2/TiO2 film loses its hydrophilicity faster than the other films, but more presumably traduces that this film initially possesses a much greater amount of surface OH groups, whose variations with aging can thus be appreciated. Accordingly, the OH band of the SiO2/TiO2 film aged for 1 month is still more intense than that depicted for aged SiO2 and TiO2/ SiO2 films. This observation probably traduces a greater persistence of surface OH groups that can in turn be related to the persistent superhydrophilicity of this film, as depicted in Fig. 2d. 3.4. Structural characterization of SiO2–TiO2 interfaces Previous XPS and FTIR studies provided complementary information on surface chemical states that could be correlated with the hydrophilic properties of bi-layer films. XPS suggests that a SiO2 outer surface promotes a greater in-time stability of surface OH species, which governs the persistence of hydrophilic properties. FTIR indicates that, compared to a SiO2 single-layer, a CS TiO2 underlayer induces a greater amount of OH species at the surface of a SiO2 overlayer, which yields enhanced superhydrophilic properties of CS SiO2/TiO2 bi-layer films. In this section, we present structural studies that were performed by TEM to understand how the TiO2 underlayer nature can influence outer SiO2 surface properties. Fig. 7 shows cross-section images of CS SiO2/TiO2, MS SiO2/TiO2, and CS TiO2/SiO2 bi-layer films heat-treated at 500 °C. Cross-section of the TiO2 underlayer or overlayer is clearly observed for the three bi-layer films. A statistical image analysis showed that, for MS SiO2/TiO2 (Fig. 7b) and CS TiO2/SiO2 (Fig. 7c) samples, the TiO2 component thickness was about 40 nm, which confirmed ellipsometric measurements performed on single-layer TiO2 films deposited on silicon. In contrast, the TiO2 thickness of around 60 nm deduced from TEM images for the CS SiO2/TiO2 film was about 30% greater (Fig. 7a). Besides, cross-section images of both former films indicate a marked SiO2–TiO2 interface, while this interface is more diffuse for the latter film. When considering brightness contrasts with the SiO2 film component, cross-section of the TiO2 film component appears significantly darker for both former bilayer films than for the latter. These observations are confirmed by HRTEM images presented in Fig. 8, which show magnified areas of SiO2–TiO2 interfacial regions for the three bi-layer films. Provided that the overall ion-thinned sample thickness is comparable, which is assessed in Figs. 7 and 8 by a similar brightness of the SiO2 component for the three bi-layer films, brightness contrasts of HRTEM images are very sensitive to atomic weighs of the elements (Z-contrast). Since the molecular weigh of TiO2 is greater than that of SiO2, the TiO2 film crosssections logically appear much darker than the SiO2 ones. However, brightness contrasts evidenced in Figs. 7 and 8 would indicate that the TiO2 component has a weaker weigh density, i.e. is more porous, in the case of a CS SiO2/TiO2 film (Figs. 7a and 8a) than for MS SiO2/TiO2 (Figs. 7b and 8b) and CS TiO2/SiO2 films (Figs. 7c and 8c). Furthermore, bright regions observed in TEM images across the TiO2 film thicknesses, might account for internal porosity and are much more accentuated for the CS SiO2/ TiO2 film.

In a previous paper, we have indicated that CS and MS TiO2 films undergo a thermally activated densification process during heat-treatment at 500 °C and CS TiO2 films heat-treated at 500 °C exhibit a significantly greater pore size (several nanometers) than MS TiO2 films heat-treated at a same temperature (a few tenths of nanometers) [29]. These data would explain contrast differences between a CS SiO2/TiO2 bi-layer (Figs. 7a and 8a) with a more porous TiO2 component and a MS SiO2/ TiO2 bi-layer (Figs. 7b and 8b) with a denser TiO2 component. However, they cannot account for differences in CS SiO2/TiO2 (Figs. 7a and 8a) and CS TiO2/SiO2 (Figs. 7c and 8c) bi-layer films, since the TiO2 component of these bi-layers is a priori comparable, i.e. it has been deposited from a same sol and heattreated at a same temperature. We suppose, therefore, that such differences might be induced by arrangement of the bi-layer structures, i.e. whether the SiO2 film is deposited below or above the CS TiO2 film. In the latter case, since the CS TiO2 underlayer has rather large intergranular pores, it is likely that, during liquid film deposition of the overlayer, the SiO2 sol can impregnate pores of the TiO2 film through a certain thickness. During subsequent sol–gel reaction and heat-treatment at 500 °C, the impregnated sol would yield a SiO2 component located within pores of the TiO2 underlayer. It would give rise to a SiO2–TiO2 composite underlayer whose HRTEM signature (Figs. 7a, 8a) can differ from that of a CS overlayer (Figs. 7c, 8c) owing to differences in weigh densities of SiO2–TiO2 composite and pure TiO2 films. However, it is important noting that brightness contrasts in TEM images do not only account for weigh density variations or internal porosity [30,31]. Ion thinning performed during sample preparation can locally alter the sample and induce contrasts due to local thickness variations. Beside, elastic interactions (diffraction) can also affect brightness contrasts in TEM images. In order to evidence a possible liquid impregnation and to estimate the impregnation depth, we performed EFTEM imaging on a CS SiO2/TiO2 cross-sectional sample. This imaging method is based on the extraction and mapping of a particular high energy signature detected in EELS spectra of samples under TEM observation. When electrons inelastically diffuse through a thin sample, they lose energy that excites different energetic transitions. In the high energy part (typically several tens of eV) of the derived spectra, the so-called core-loss region, each core-loss edge of the spectrum is characteristic for a specific chemical element. EFTEM images are very few sensitive to local thickness variations or elastic interactions, which can affect brightness contrasts in TEM images. Consequently, contrasts in EFTEM only account for chemical information thus allowing an elemental mapping of the sample [30,31]. In this study, EELS spectra were filtered around the Si–L2,3 peak. HRTEM and EFTEM images of a same CS SiO2/TiO2 film cross-sectional area are shown in Fig. 9a and b, respectively. Both images show quite similar brightness contrasts, which indicates that the HRTEM image of Fig. 9a traduces, at least partially, local weigh density variations. In the EFTEM image of Fig. 9b, presence of silicon is depicted by bright regions, which can be appreciated by a grey appearance for SiO2 and a white appearance for the Si substrate, while TiO2 regions appear in black. This image unambiguously shows that

Author's personal copy

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

SiO2 is present across the whole TiO2 film thickness, which demonstrates that the SiO2 sol totally impregnated rather large pores of the CS TiO2 underlayer during deposition of the overlayer. It can therefore be concluded that the underlayer of a CS SiO2/TiO2 bi-layer consists of a SiO2–TiO2 composite film located at the SiO2-substrate interface. EFTEM imaging has not been performed for MS SiO2/TiO2 and CS TiO2/SiO2 bi-layer films. However it is likely that, for those two films, liquid impregnation of the underlayer cannot take place in significant extent during deposition of the overlayer. On the one hand, in the case of a MS SiO2/TiO2 film, the MS TiO2 underlayer has very small intergranular pores that can hardly be impregnated. On the other hand, in the case of a CS TiO2/SiO2 film, previous (unpublished) works have shown that our SiO2 underlayer heat-treated at 500 °C is very dense. It is thus inferred that occurrence or not of a liquid impregnation can definitely explain brightness contrasts observed in Figs. 7 and 8, since a SiO2–TiO2 composite film has a weaker molecular weigh than a pure TiO2 film. Besides, it is possible that, in the case of a CS SiO2/TiO2 bi-layer film, liquid impregnation of the SiO2 sol can modify the arrangement of TiO2 crystallites present in the CS TiO2 underlayer, through a mechanism that has not been elucidated yet, which would in turn explain why in this case the TiO2 film component has a much greater thickness than in both other bi-layer films. We previously indicated that Si–O–Ti heterolinkages are likely to be formed at SiO2–TiO2 interfaces. Thus, MS SiO2/TiO2 and CS TiO2/SiO2 bi-layer films would essentially be composed of Si–O–Ti heterolinkages located at a planar SiO2–TiO2 interface, while such heterolinkages would be created at granular interfaces of a SiO2–TiO2 composite film in the case of a CS SiO2/TiO2 bilayer. It is thus inferred that, owing to a larger SiO2–TiO2 contact surface, the amount of Si–O–Ti heterolinkages is much greater in this latter case. 3.5. Superhydrophilic properties of bi-layer films: discussion In a previous work, we have shown that SiO2–TiO2 composite films, constituted of TiO2 crystallites embedded in a SiO2 amorphous matrix, exhibit natural and persistent superhydrophilic properties [21], comparable to those evidenced in the present work for CS SiO2/TiO2 bi-layer films. We have postulated that, according to well established bibliographic data [11– 14], these unusual properties are due to interfacial Si–O–Ti heterolinkages that promote the formation of TiO6−2 or SiO4+4/3 units inducing charge imbalances at SiO2–TiO2 granular interfaces. Deprotonated TiO6−2 and/or protonated SiO4+4/3 units present at the composite film surface can favor adsorption of H3O+ and/or OH− ions, thus inducing enhanced molecular or dissociative water adsorption and leading to natural superhydrophilic properties of composite films. We believe that superhydrophilic properties of bi-layer films might also originate from such depronated or protonated interfacial units, which can be discussed as follows. We have shown that existence and persistence of natural superhydrophilicity in bi-layer films follows the order CS TiO2/ SiO2 b MS SiO2/TiO2 ≪ CS SiO2/TiO2 and, in same operating

965

conditions, superhydrophilicity of the latter films can be regenerated, which is not achieved for both former films. According to previous studies and arguments, both former films are constituted of a SiO2–TiO2 planar interface where Si–O–Ti heterolinkages and related negative or positive charge imbalances can be present. Since the electrical neutrality of the whole bi-layer film must be satisfied, such interfacial charge imbalances are likely to be compensated for by counter charges at the bi-layer film outer surface. So-formed negative or positive surface charges would in turn favor the molecular or dissociative water adsorption, as previously postulated for SiO2–TiO2 composite films, yielding an initial natural superhydrophilicity of CS TiO2/SiO2 and MS SiO2/TiO2 films. Furthermore, a MS SiO2/TiO2 film benefits from the presence of a SiO2 outer layer that promotes a greater in-time stability of adsorbed OH species. This outer layer would explain a certain persistence of superhydrophilicity for the MS SiO2/TiO2 film that is not observed for a CS TiO2/SiO2 film with a TiO2 outer layer. However, in both cases no long-term persistence of superhydrophilicity is observed. It can be related to the limited amount of interfacial negative or positive charges, whose appearance is restrained at a planar interface, which induces a limited amount of negative or positive charges at the outer surface. In contrast, a CS SiO2/TiO2 bi-layer film would not only benefit from the existence of a SiO2 outer layer but also from the SiO2–TiO2 composite nature of the underlayer. For this bilayer, Si–O–Ti heterolinkages are not concentrated at a planar SiO2–TiO2 interface but are distributed at granular SiO2–TiO2 interfaces. Thus, owing to a much greater SiO2–TiO2 contact surface, such a composite layer can promote a noticeably larger amount of intergranular Si–O–Ti heterolinkages, leading to a greater charge imbalance in the interfacial region. According to the electrical neutrality criterion, it is therefore inferred that the amount of negative or positive charges at the outer surface of CS SiO2/TiO2 films should be much more important than for other bi-layer films, which would in turn favor a more efficient water adsorption and, consequently, enhanced natural superhydrophilic properties. Finally, it should be mentioned that morphological features can also influence the wettability of our bi-layer films, in particular the surface roughness. Water contact angles measured on a hydrophilic rough surface are usually smaller than those measured on a same non-rough surface. Thus, the roughness of bilayer films may in principle influence their wettability properties. However, two arguments indicate that, in the present case, roughness presumably plays a secondary role. In a previous article, we indicated that the RMS roughness of MS TiO2 and CS TiO2 films was about 1 and 8 nm, respectively [29]. In this study, we showed that these roughness values were presumably too weak to significantly influence wettability properties of TiO2 films. Furthermore, SiO2 films studied in the present work are extremely smooth, with a RMS roughness of only 0.4 nm. Since existence and persistence of natural superhydrophilicity in bi-layer films follows the order CS TiO2/SiO2 b SiO2/TiO2, it means that the much smoother SiO2/TiO2 bi-layer film exhibits better wettability properties. It seems, therefore, that roughness effects do not predominantly act on the superhydrophilicity of our bi-layer

Author's personal copy

966

S. Permpoon et al. / Thin Solid Films 516 (2008) 957–966

films, which would essentially rely on the outer surface nature and effects of SiO2–TiO2 interfaces. 4. Conclusion Sol–gel SiO2/TiO2 and TiO2/SiO2 bi-layer films have been deposited from a polymeric SiO2 solution and either a polymeric TiO2 mother solution (MS) or a TiO2 crystalline suspension (CS). The chemical and structural properties of bi-layer films heattreated at 500 °C have been investigated by FTIR, XPS, and TEM characterizations in relation to their hydrophilic properties. Chemical characterizations suggest that an outer SiO2 surface promotes a greater in-time stability of surface OH species, while a CS TiO2 underlayer induces a greater amount of OH species at the outer SiO2 surface than a MS TiO2 underlayer. TEM studies show that a SiO2/TiO2 composite film is present at the SiO2-substrate interface of a CS SiO2/TiO2 bi-layer, while this composite film does not exist for other bi-layers under investigation. These properties can in turn be related to the existence and persistence of a natural superhydrophilicity of bi-layers, which is never observed for SiO2 or TiO2 single-layer films. Water contact angle measurements show that existence and persistence of natural superhydrophilicity in bi-layer films follows the order CS TiO2/SiO2 b MS SiO2/TiO2 ≪ CS SiO2/TiO2 and, in same operating conditions, superhydrophilicity of the latter films can be regenerated, which is not achieved for both former films. Enhanced properties of CS SiO2/TiO2 films are primarily related to the presence of an outer SiO2 layer. They also arise from existence of a SiO2–TiO2 composite layer in the interfacial region. Si–O–Ti heterolinkages developing at granular interfaces in the composite layer are believed to be the origin of important electrical charge imbalances that would promote negative or positive charges at the outer SiO2 surface, thus inducing natural water adsorption and enhanced natural superhydrophilic properties of CS SiO2/TiO2 bi-layer films. Natural, persistent, and regenarable superydrophilic properties of bi-layer films allow envisaging long-term self-cleaning applications in conditions where UV light is not available. Acknowledgement The authors thank D. Lafond from LETI-Grenoble for valuable help in EFTEM experiments. References [1] R. Wang, N. Sakai, A. Fujishima, T. Wanatabe, K. Hashimoto, J. Phys. Chem., B 103 (1999) 2188. [2] N. Sakai, A. Fujishima, T. Watanabe, K. Hashimoto, J. Phys. Chem., B 105 (2001) 3023. [3] M. Machida, K. Norimoto, T. Watanabe, K. Hashimoto, A. Fujishima, J. Mater. Sci. 34 (1999) 2569.

[4] D. Ren, X. Cui, J. Zhen, Q. Zhang, X. Yang, Z. Zhang, L. Ming, J. Sol–gel Sci. Technol. 29 (2004) 131. [5] K. Guan, Surf. Coat. Technol. 191 (2005) 155. [6] H.J. Lee, S.H. Hahn, E.J. Kim, Y.Z. You, J. Mater. Sci. 39 (2004) 3683. [7] M. Maeda, S. Yamasaki, 204th Electrochem. Soc. Meeting, Orlando, USA, October 12–16, 2003, extended abstract 1258. [8] J.C. Yu, J. Yu, W. Ho, J. Zhao, J. Photochem. Photobiol., A Chem. 148 (2002) 331. [9] X. Fu, L.A. Clark, Q. Yang, M.A. Anderson, Environ. Sci. Technol. 30 (1996) 647. [10] K. Guan, B. Lu, Y. Yin, Surf. Coat. Technol. 173 (2003) 219. [11] K. Tanabe, T. Sumiyoshi, K. Shibata, T. Kiyoura, K. Kitagawa, Bull. Chem. Soc. Jap. 47 (1974) 1064. [12] X. Gao, I.E. Wachs, Catal. Today 51 (1999) 233. [13] T. Kataoka, J.A. Dumesic, J. Catal. 112 (1988) 66. [14] Z. Liu, J. Tabora, R.J. Davis, J. Catal. 149 (1994) 117. [15] J.M. Sanz, L. Soriano, P. Prieto, G. Tyuliev, C. Morant, E. Elizalde, Thin Solid Films 332 (1998) 209. [16] B. Gallas, A. Brunet-Bruneau, S. Fisson, G. Vuye, J. Rivory, J. Appl. Phys. 92 (2002) 1922. [17] N. Brun, C. Colliex, J. Rivory, K.Y. Zhang, Microsc. Microanal. Microstruct. 7 (1996) 161. [18] K.S. Guan, H. Xu, B.J. Lu, Trans. Nonferr. Met. Soc. 14 (2004) 251. [19] A. Hattori, T. Kawahara, T. Uemoto, F. Suzuki, H. Tada, S. Ito, J. Colloid Interface Sci. 232 (2000) 410. [20] H. Tada, Y. Kubo, M. Akazawa, S. Ito, Langmuir 14 (1998) 2936. [21] S. Permpoon, G. Berthome, B. Baroux, J.C. Joud, M. Langlet, J. Mater. Sci. 41 (2006) 7650. [22] N. Primeau, C. Vautey, M. Langlet, Thin Solid Films 310 (1997) 47. [23] M. Langlet, M. Burgos, C. Coutier, C. Jimenez, C. Morant, M. Manso, J. Sol–gel Sci. Technol. 22 (2001) 139. [24] M. Langlet, A. Kim, M. Audier, C. Guillard, J.M. Herrmann, J. Mater. Sci. 38 (2003) 3945. [25] M. Fallet, S. Permpoon, J.L. Deschanvres, M. Langlet, J. Mater. Sci. 41 (2006) 2915. [26] G. Orcel, J. Phallippou, L.L. Hench, J. Non-cryst. Solids 88 (1986) 114. [27] R.M. Almeida, J. Sol–gel Sci. Technol. 13 (1998) 51. [28] C.I. Contescu, J.A. Schwarz, in: K.L. Mittal (Ed.), Acid–Base Interactions, vol. 2, 2000, p. 245. [29] M. Langlet, S. Permpoon, D. Riassetto, G. Berthomé, E. Pernot, J.C. Joud, J. Photochem. Photobiol. A 181 (2006) 203. [30] M. Worch, H.J. Engelmann, W. Blum, E. Zschech, Thin Solid Films 405 (2002) 198. [31] R. Pantel, S. Jullian, D. Delille, D. Dutartre, A. Chantre, O. Kermarrec, Y. Campidelli, L.F.T.Z. Kwakman, Micron 34 (2003) 239.