Creation of High-Refractive-Index Amorphous Titanium Oxide Thin

Jul 23, 2012 - High-refractive-index films have been widely used for diverse optical ..... Notation: ξ, the correlation length estimated by fitting t...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/Langmuir

Creation of High-Refractive-Index Amorphous Titanium Oxide Thin Films from Low-Fractal-Dimension Polymeric Precursors Synthesized by a Sol−Gel Technique with a Hydrazine Monohydrochloride Catalyst Wataru Shimizu,† Satoshi Nakamura,† Takaaki Sato,‡ and Yasushi Murakami*,† †

Division of Chemistry and Materials, Faculty of Textile Science and Technology, and ‡International Young Researchers Empowerment Center, Shinshu University, 3-15-1 Tokida, Ueda, Nagano 386-8567, Japan S Supporting Information *

ABSTRACT: Amorphous titanium dioxide (TiO2) thin films exhibiting high refractive indices (n ≈ 2.1) and high transparency were fabricated by spin-coating titanium oxide liquid precursors having a weakly branched polymeric structure. The precursor solution was prepared from titanium tetra-n-butoxide (TTBO) via the catalytic sol−gel process with hydrazine monohydrochloride used as a salt catalyst, which serves as a conjugate acid−base pair catalyst. Our unique catalytic sol−gel technique accelerated the overall polycondensation reaction of partially hydrolyzed alkoxides, which facilitated the formation of liner polymer-like titanium oxide aggregates having a low fractal dimension of ca. 5/3, known as a characteristic of the so-called “expanded polymer chain”. Such linear polymeric features are essential to the production of highly dense amorphous TiO2 thin films; mutual interpenetration of the linear polymeric aggregates avoided the creation of void space that is often generated by the densification of high-fractaldimension (particle-like) aggregates produced in a conventional sol−gel process. The mesh size of the titanium oxide polymers can be tuned either by water concentration or the reaction time, and the smaller mesh size in the liquid precursor led to a higher n value of the solid thin film, thanks to its higher local electron density. The reaction that required no addition of organic ligand to stabilize titanium alkoxides was advantageous to overcoming issues from organic residues such as coloration. The dense amorphous film structure suppressed light scattering loss owing to its extremely smooth surface and the absence of inhomogeneous grains or particles. Furthermore, the fabrication can be accomplished at a low heating temperature of 2.0).12−17 However, evaporation sources or a target were to be placed a short distance (100 °C), which inevitably cause damage to the substrate.11,12,14,16 This fact discourages the use of plastic substrates despite their advantages (e,g., flexibility, workability, and light weight). For example, commonly used plastic substrates such as poly(methyl methacrylate) (PMMA) have such low heat resistance that the films need to be fabricated at temperatures below 80 °C. Attempts to protect the substrate against heat radiation by increasing the target-to-substrate distance resulted in a reduced film density.13,14 Thus, it has so far been difficult to increase the refractive index of amorphous TiO2 thin films without thermal damage to the substrates. Received: September 14, 2011 Revised: July 22, 2012 Published: July 23, 2012 12245

dx.doi.org/10.1021/la3015139 | Langmuir 2012, 28, 12245−12255

Langmuir

Article

effects of an organic residue). The coating solution was prepared via a sol−gel process using a hydrazine monohydrochloride catalyst from titanium-n-butoxide without any inhibitors. Smallangle X-ray scattering (SAXS) was used to confirm the formation of titanium oxide polymers in the precursor solutions. A relationship between the liquid polymer structure and the optical properties of the resulting solid film was revealed by timeresolved SAXS (Tr-SAXS) experiments and spectroscopic ellipsometry. We also demonstrate the successful production of a high-refractive-index amorphous TiO2 film (n > 2.0) on a lowtemperature plastic {poly(methyl methacrylate, i.e., PMMA)} substrate. The resulting films were characterized by means of transmission electron microscopy (TEM), atomic force microscopy (AFM), and UV−vis spectroscopy.

An alternative deposition method may be a wet process involving spin or dip coating of the solution prepared via the sol−gel process.19−30 The resulting films are formed at room temperature under atmospheric pressure. Such mild and moderate conditions should be useful when we target the use of plastic substrates. Polymeric precursors are synthesized from hydrolysis and polymerization reactions of titanium alkoxides. The packing density of the TiO2 films depends on the primary conformation of the precursor polymers in the coating solution.31 The weakly branched polymeric precursors are expected to provide a denser film, and the highly branched precursors generally leading to globular particle-like aggregates may create void spaces in the product film because of their apertures. Because of the high reactivity of titanium alkoxides, it has so far been difficult to achieve such low-fractal-dimension polymeric structure of titanium oxides, with a sol−gel reaction of titanium alkoxides readily producing particle-like aggregates and immediate precipitation.7 A number of techniques aimed at producing low-fractal-dimension titanium oxide polymeric precursors were attempted on the basis of a sol−gel technique. Rantala et al. fabricated an amorphous TiO2 film with n = 1.81 (at λ = 632.8 nm) using methacrylic acid as an organic ligand at 85 °C.26 However, the use of organic ligands prevented the TiO2 film from being densified because of the organic residue, and the film fabricated from these precursors contained many voids filled with air. As a result, the refractive index of the film did not attain a sufficiently high value. Louis et al. applied a sol−gel technique to the production of high-refractive-index TiO2 thin films using TiCl4 as a starting material, which provided organic-ligandfree TiO2 films.30 In their system, protons prevented titanium oxo-hydroxide precipitation. The film exhibited n ≈ 1.93 even at 60 °C. Cabane and co-workers produced low-fractal-dimension titanium oxide polymers from titanium-n-butoxide (TTBO) with a high proton concentration, where protons act as an inhibitor of the condensation reaction.32,33 An oxygen atom of Ti−OH is protonated, forming Ti−OH2+ species that repel each other.34 The system can produce titanium oxide aggregates having a weakly branched polymer-like structure rather than a globular particle-like structure.33 However, because of the long time (>month) for low-fractal-dimension polymer growth, df ≈ 2, which is sufficient for the formation of the films, the acidcatalyzed sol−gel synthesis would not be very practical for film production. To realize the efficient fabrication of low-fractal-dimension titanium oxide polymers, one needs to introduce a renovated reaction-controlled sol−gel process that can enhance the polycondensation rate of titanium alkoxides to condense partially hydrolyzed reactive species rapidly while suppressing the number of hydrolyzed sites. To overcome all of these issues, we have recently developed a catalytic sol−gel process with a hydrazine monohydrochloride catalyst and titanium alkoxides. The present catalytic sol−gel process enables the rapid production (within several hours) of a transparent sol or gel with no organic ligand. Note that we do not claim that our system is fully composed of inorganic compounds because the TTBO ingredient involves organic residues. An organic-ligand-free reaction here means that we did not use any additional organic ligand to stabilize the titanium alkoxides. In this work, we have attempted to fabricate an amorphous, dense TiO2 film with a high refractive index from titanium oxide precursor solutions below 60 °C, aimed at overcoming all of the above-mentioned issues (e.g., creation of void space or inhomogeneous grains, high heating temperature, and side

2. EXPERIMENTAL SECTION 2.1. Sample Preparation.35 Materials. Tetra-n-buthoxide (TTBO, Ti(OC4H9)4 > 97%) was purchased from Kanto Chemical, Japan. 2-Propanol (C3H7OH > 99.5%) was purchased from Wako Pure Chemical Industries, Japan. Hydrazine monohydrochloride (N2H4 HCl > 98%) was purchased from Tokyo Chemical Industry, Japan. All reagents were used without further purification. Synthesis. To prepare the film coating solution, TTBO, 2-propanol, distilled water, and the hydrazine monohydrochloride catalyst were mixed together. We tested different molar ratios in the precursor solutions by treating the water content as a tunable parameter while fixing [N2H4 HCl]/[Ti(OC4H9)4] to 0.04. [Ti(OC4H9)4] was adjusted to 0.5 mol L−1 for all precursors. Table 1 summarizes the sample

Table 1. Molar Ratio of Water and 2-Propanol to TTBO in the Precursor Solutions and the Reaction Time in the Sol−Gel Process precursor solution

water

2-propanol

reaction time (h)

W2 W225 W25

2.00 2.25 2.50

21.07 21.02 20.95

6 1 0.5

preparation. The solution was mixed slowly at 0 °C to avoid the generation and precipitation of TiO2 particles. Without hydrazine monohydrochloride in this preparation, fine white particles were immediately generated and precipitated as soon as the solutions were mixed. The mixed solution was then stirred at 25 °C for several hours. Silicon (100) (25 × 25 × 1 mm3) and PMMA (50 × 50 × 1 mm3) substrates were coated with the solution by spin coating (3500 rpm) at 25 °C for 60 s. The coated films were dried in a closed container below 10% relative humidity at 25 °C for 1 h, which was followed by a heating process at 40−150 °C for 1 h. 2.2. Characterizations. SAXS experiments were carried out on precursor solutions in which a SAXSess camera (Anton Paar, Austria) was employed. A Göbel mirror and a block collimator provided a focused monochromatic X-ray beam of Cu Kα radiation. Twodimensional scattering patterns recorded by an image-plate (IP) detector (Cyclone, Perkin-Elmer) were integrated into one-dimensional scattering intensities, I(q), where q is the magnitude of the scattering wave vector. All I(q) data were corrected for background scattering from a capillary and solvents, and the absolute intensity calibration was made using water as a secondary standard.36 A model-independent collimation correction (desmearing) procedure was made via an indirect Fourier transformation (IFT) routine. All fitting and fractal analyses were performed on the collimation-corrected SAXS intensities. The refractive index, extinction coefficient, and thickness of the films were measured with a UVISEL M200 (Horiba Jobin Yvon, Japan) spectroscopic ellipsometer by using DeltaPsi 2 simulation software (Horiba Jobin Yvon). The film thickness was also determined from a cross-sectional image of the films using an S-5000 (FE-SEM, Hitachi, Japan) fieldemission scanning electron microscope. A transmission electron 12246

dx.doi.org/10.1021/la3015139 | Langmuir 2012, 28, 12245−12255

Langmuir

Article

Figure 1. SAXS intensities, I(q), on an absolute scale (A) and a modified Kratky plot, q5/3I(q) versus q, (B) for the precursor solutions prepared at different water molar ratios: W2 (red), W225 (blue), and W25 (green). The solid lines shown in panel A represent optimized fit curves based on the Ornstein−Zernike (OZ) equation. In the right panels, photographs of the titanium oxide liquid precursors derived from TTBO with (C) and without (D) the hydrazine monohydrochloride catalyst are shown. The molar ratio of water and IPA to TTBO was identical in C and D. microscope (TEM) image of the amorphous TiO2 thin film was obtained by means of a JEM-2010 transmission electron microscope (JEOL, Japan) with an acceleration voltage of 200 kV. Finely ground powders of the peeled thin film were dispersed in methanol by ultrasonication. One drop of the suspension was put on a 150 mesh carboncoated copper grid and then dried at 60 °C in air. The crystallinity of the dried and heated precursor powders was estimated using an X-ray diffractometer (XRD, Rigaku, RINT2500HF) that was operated at 40 kV and 40 mA with Cu Kα radiation (λ = 0.1542 nm) in the 2θ range between 20 and 70°. The surface morphology was observed using an atomic force microscope (AFM, Seiko Instruments, SPA400), and the surface root-mean-square roughness values were calculated from AFM images. The film transparency was measured with a spectrophotometer (Hitachi, U-4100). Visible light transmission was calculated as the average transmission in the wavelength range from 400 to 780 nm.

A Kratky plot, q2I(q) versus q, is an instructive expression often used to highlight the fractal nature of an ideal or Gaussian coil polymer (df ≈ 2). Instead, we used a modified Kratky plot, q5/3I(q) versus q, as shown in Figure 1B. The water contents modified the fractal nature of the structure as df decreased with decreasing water content, W25 ≈ W225 > W2. W25 and W225 gave a high-q plateau (q5/3I(q) ∝ q0) in this expression, demonstrating q−5/3 behavior in q ≳ 1 nm−1. The deduced df ≈ 5/3 = 1.67 for W25 and W225 coincides well with df of an expanded chain (a polymer chain with excluded volume in a good solvent). W2 (i.e., the liquid precursor with a restricted amount of water) showed a slightly but definitely gentler high-q slope of q−4.7/3 corresponding to df ≈ 1.54, which sensitively shows up as a positive slope in the modified Kratky plot. Decreased water content required a longer reaction time for linear polymeric titanium oxide growth, and the lower reactivity of the titanium alkoxides led to lower df of the polymeric aggregates. For a further quantitative description of the structure of the polymeric titanium oxide precursors, we scrutinized the shape of the scattering curves in which we thoroughly tested three different models (e.g., the Debye,43 the Ornstein−Zernike (OZ),44 and the Fisher−Burford (FB)45 expressions for the titanium oxide liquid precursor systems). The Debye expression for ideal or Gaussian polymer chains takes into account the connectivity and flexibility of the polymer chain and is given by

3. RESULTS AND DISCUSSION 3.1. Structure of the Precursor Solutions as Obtained by SAXS. SAXS has been used to investigate the conformation and the fractal nature of polymer-like and globular-particlelike aggregates produced in a sol−gel process33,37,38 To monitor the microstructure of titanium oxide aggregates in the liquid precursors, we investigated their static structures by means of SAXS.39 Figure 1 shows SAXS experiments on the titanium oxide precursor solutions (W2, W225, and W25) prepared in a varied molar ratio of water. It is established that SAXS efficiently distinguishes linear polymers from globular particles owing to their fairly different scattering behavior (Figure S1, Supporting Information). Figure S1 compares the typical scattering behavior of a polymer in solution and a colloidal particle dispersion using simulated scattering functions; see the Supporting Information. The branching of polymer chains may impose a somewhat particlelike nature on the polymer chains. For instance, starbranched polymers,40 dendrimers,41 nanoparticles comprising tightly cross-linked liner chain precursors,42 and sol−gel-derived high-fractal-dimension silica aggregates38 give a local maximum in a Kratky plot, as do globular particles. We found that all of the titanium oxide precursor solutions were to be assigned to a liner polymeric structure in terms of their scattering patterns. Power law behavior was observed in the intermediate- to high-q regime, indicating the emergence of the fractal structure (i.e., self-similarity of the internal structure independent of the chosen length scale). The fractal dimension, df, directly appears in the slope of I(q), as represented by the following equation: I(q) ∝ q−d f

I(q) = I(0)Pideal(x) = 2I(0)

exp( −x) + x − 1 x2

(2)

2 2

where x is equal to Rg q , with Rg being the radius of gyration reflecting the overall geometry of the polymer chains. In a more general and model-independent manner, Rg can be evaluated by a Guinier plot based on the Guinier equation: ⎡ R 2q2 ⎤ g ⎥ I(q) = I(0)exp⎢ − ⎢⎣ 3 ⎥⎦

(3)

The OZ equation, corresponding to an exponentially decaying correlation function, is given by I(q) =

I(0) 1 + ξ 2q2

(4)

where ξ is the correlation length reflecting the length of the internal density fluctuation of the polymer chain and I(0) is the asymptotic zero-q intensity. The shape of the OZ function is

(1) 12247

dx.doi.org/10.1021/la3015139 | Langmuir 2012, 28, 12245−12255

Langmuir

Article

Figure 2. SAXS intensity, I(q), and a Kratky plot, q2I(q) versus q, of the W225 titanium oxide precursor solution with the optimized fit curves based on Debye (blue), Ornstein−Zernike (OZ) (green), and Fisher−Burford (FB) (red) equations. In the inset, an enlarged view of I(q) in the low-q regime (qξ < 1) is displayed.

not the same as that of the Debye function, Pideal(q), but rather resembles it. ξ can be interpreted as the mesh size of the transient polymer network.46,47 The Fisher−Burford (FB) equation45 I(q) =

Table 2. Microscopic Geometries of the Liquid Precursors Prepared at Different Water Concentrations and the Optical Properties and Thicknesses of the Resulting Amorphous TiO2 Thin Films Fabricated on a Si Substrate with Low-Temperature Heat Treatment at 80 °Ca

I(0) [1 + (2R g,FB 2q2 /3df )]d f /2

precursor solution

(5)

W2 W225 W25

accounts for lower fractal dimensions (df < 2), where Rg,FB is the (apparent) radius of gyration in its expression. When df = 2, this coincides with the OZ expression. As a typical example, a comparison of FB, Debye, and OZ fits the W225 titanium oxide precursor solution and is given in Figure 2. As highlighted in the Kratky plot (Figure 2B), the FB expression with df ≈ 5/3 gave a far better description of the experimental I(q) than did the Debye and OZ expressions in the fractal regime, where I(q) ∝ q−5/3. We found a systematic deviation of the FB curve from the experimental I(q) in the low-q regime (Figure 2A). Such a deviation is certainly not negligible in the Guinier regime (qRg < 1). Consequently, Rg,FB obtained with the FB fit is not consistent with the actual Rg determined with a Guinier plot. Furthermore, the FB equation does not involve the correlation length, ξ, explicitly. One possible assumption may be to rely on the relation Rg,FB = (3)1/2ξ FB. However, such an assumption is valid only for a very dilute system and again needs to refer to uncertain Rg,FB. As shown in the inset of Figure 2A, the low-q portion of the experimental I(q) can be well fitted either by the Debye or OZ equation, with the Debye giving a slightly better fit than the OZ. Rg obtained with the Debye fit coincides well with that evaluated with a Guinier plot. Because ξ is practically evaluated from the low-q portion of I(q) where qξ < 1 (q < ∼0.3−0.4 nm−1 for the present systems) by monitoring the crossover point from Guinier to fractal scattering, we believe that despite the marked high-q deviation a fitting analysis based on the OZ equation was almost sufficient to evaluate ξ. The mesh size for W2, ξ = 2.1 nm, was found to be the smallest. Increasing the water content increased ξ, with values of 2.7 and 2.9 nm obtained for W225 and W25, respectively. Accordingly, Rg for W2, W225, and W25 was an increasing function of the water content and was estimated to be 3.4, 4.1, and 4.5 nm, respectively (Table 2). The crossover from expanded coil-like scattering [I(q) ∝ q−5/3] to rodlike scattering [I(q) ∝ q−1] can be clearly seen for W225 and W25 at q ≈ 3.5 nm−1, giving a

ξ Rg (nm) (nm) 2.1 2.7 2.9

3.4 4.1 4.5

film thickness (nm)

n550

n633

k380

91 117 118

2.141 1.991 1.926

2.110 1.968 1.905

3.10 × 10−3 3.10 × 10−3 2.10 × 10−3

a

The thickness and optical constants were determined by an analysis of ellipsometric spectra. Notation: ξ, the correlation length estimated by fitting the SAXS curves based on the Ornstein−Zernike equation (eq 4); Rg, the radius of gyration determined with Guinier plot; n550, refractive index at λ = 550 nm; n633, refractive index at λ = 633 nm; k380, extinction coefficient at λ = 380 nm.

persistence length of lp ≈ 1.8 nm, whereas such a transition is somewhat vague for W2 in the available q range. All of the above-mentioned findings demonstrate that titanium oxide polymers having an expanded polymer-chain-like structure were successfully synthesized using the hydrazine monohydrochloride catalyst in a reasonably short time (within several hours). Kinetic models for polymer growth (i.e., diffusion-limited aggregation (DLA) and reaction-limited aggregation (RLA)) have been discussed in the sol−gel process of silicon alkoxides.31,48 Under strongly acidic conditions of pH < 2, the silica precursor polymer growth was governed by RLA, forming weakly branched structures (df ≈ 2). Likewise, low-fractaldimension titanium oxide polymers were produced from titanium alkoxides under strongly acidic conditions.33 As shown in Figure 1C,D, the titanium oxide precursor solutions prepared via the hydrazine monohydrochloride-assisted catalytic sol−gel process were transparent without any precipitation (Figure 1C) whereas whitish precipitates were immediately generated in solution without the catalyst (Figure 1D) because of the high reactivity of titanium alkoxides. In this respect, we can say that lowering the reactivity enables the production of lowfractal-dimension titanium oxide polymeric precursors, which may correspond to a switch from DLA to RLA growth. 3.2. Role of a Hydrazine Monohydrochloride Catalyst in the Rapid Synthesis of the Low-Fractal-Dimension Polymeric Titanium Oxide Aggregates. The major strategy in previously reported synthesis techniques aimed at producing 12248

dx.doi.org/10.1021/la3015139 | Langmuir 2012, 28, 12245−12255

Langmuir

Article

A hydrazinium ion, [H2N−NH3]+, releases a proton acting as an acid whereas a hydrazine, [H2N−NH2], accepts a proton, serving as a base.51 If such an equilibrium state exists in the reaction system, then hydrazine promotes the deprotonation (eq 7) as a base catalyst and a hydrazinium ion promotes dehydroxylation (eq 8) or dealkoxylation (eq 9) as an acid catalyst. Thereby, the overall dehydrative polycondensation reaction of the partially hydrolyzed titanium alkoxides is efficiently accelerated by hydrazine monohydrochloride.51,52 3.3. Optical Properties of the Amorphous TiO2 Thin Films Fabricated on a Silicon Substrate. Silicon(100) substrates were coated with precursor solutions W2, W225, and W25, dominated by the weakly branched titanium oxide polymers, by spin coating, and then heat treatment was applied to the dried spin-coated films at varied temperatures (40 ≤ T/°C ≤ 150). To investigate the optical properties (i.e., refractive index, n, and extinction coefficient, k), of the resulting thin films, we conducted a spectroscopic ellipsometry measurement. Figure 3A,B shows complex optical constants, the n and k curves, of the low-temperature (80 °C) heat-treated TiO2 thin films on a Si substrate as a function of the wavelength of light, λ. Table 2 lists the n values at 550 and 633 nm, the k values at 380 nm, and the film thickness. A noteworthy fact is that, as targeted, all of the thin films exhibited relatively high n values (>1.9 at 633 nm) despite such a mild heating condition, which is obviously advantageous for widening a range of applications of the TiO2 optical coating. The high n values obtained also imply that dense amorphous films were produced, owing to the mutual interpenetration of the expanded chainlike titanium oxide polymers (df ≈ 5/3) during the drying and curing processes while avoiding void space creation. Figure 3C shows the relation between the mesh size (correlation length), ξ, of the extended chainlike titanium oxide polymers in the liquid precursors and the refractive indices, n, of the resulting TiO2 thin film. Importantly, as shown in Figure 3C, we found that the n values of the thin films increased almost linearly with decreasing water content in the precursor solutions, demonstrating that the titanium oxide polymer having a smaller ξ was transcribed into finer and denser solid structures in the thin film (Table 2). This view is well supported by our recent study on nanoporous silica production via the catalytic sol−gel process;38 silica powders having exclusively micropores (diameter 5 nm) because of their excluded volume effect. The k values were found to be almost identical for all of the films at 380 nm, giving 390 nm. We infer that the small k values were accomplished thanks to the absence of organic residues in our ligand-free synthesis strategy because an organic-ligand-coordinated, stabilized titanium alkoxide generally imposes coloration on the resulting solid products because of its visible-light absorption. In addition, the thin films containing no such organic ligands are expected to be denser and therefore to be equipped with higher refractive indices. 3.4. Structures of Amorphous TiO2 Thin Films. TEM images of the TiO2 thin films fabricated from the weakly branched titanium oxide polymer solutions (W2, W225, and W25) with heat treatment at 80 °C are presented in Figure 4. In an optical thin film containing particles dispersed in a matrix, an optical transmission loss may arise from Mie scattering if the refractive index difference between a matrix and particles is large enough and when the correlation length of the density fluctuation of the aggregates of the dispersed particles becomes comparable to the wavelength of light. In the TEM images, no obvious voids or particulates were evident at this magnification. The texture of the films was not porous, showing no traces of agglomerated particles, but rather appeared dense and monolithic. If crystalline particles are dispersed in an amorphous thin film, they often act as scattering centers for light and become the source of birefringence. In the inset of Figure 4A, we present

a selected-area electron diffraction (SAED) pattern of the W2-based thin film, which confirms the amorphous nature of the TiO2 film. The dense amorphous TiO2 matrix with low void content and no crystalline particles helps us to achieve a high n value, simultaneously suppressing optical scattering owing to its fine textured matrix. We note that the condensation of the weakly branched titanium oxide polymers provided the TiO2 films with ultrafine structural features, which are intrinsic to an optical thin film to avoid Mie scattering and to lower the Rayleigh scattering loss in the visible-light range. Structural differences among the TiO2 thin films generated from the precursors having different transient mesh sizes, ξ, were not evident in the TEM images. The measured differences in the refractive indices (i.e., the higher n value of the film for smaller ξ in the liquid precursor) suggest that a microscopic length scale structure as small as ξ contributed to the enhancement of the optical properties of the film. The film thickness of the W2-derived TiO2 film on the silicon substrate with heating at 60 °C was found to be ca. 110 nm in view of its cross-sectional SEM image (Figure 5A). The AFM image (Figure 5B) and the cross-sectional profile (Figure 5C) of the TiO2 film show that the film is very smooth, with a surface asperity within 1 nm. The root-mean-square surface roughness was estimated to be ca. 0.4 nm. Low roughness values reduce the surface scattering loss and consequently provides excellent optical properties. 3.5. Time Evolution of the Polymeric Liquid Precursors and Optical Properties of the Resulting TiO2 Thin Films. We monitored the time evolution of the W2 liquid precursor 12250

dx.doi.org/10.1021/la3015139 | Langmuir 2012, 28, 12245−12255

Langmuir

Article

Figure 6. Time evolution of the titanium oxide polymeric aggregates in the W2 precursor solution as obtained by SAXS. Variation of SAXS intensity, I(q), on the absolute scale (A) and the radius of gyration Rg (red circles) and the correlation length ξ (blue triangles) (B) of the titanium oxide polymeric aggregates as a function of the reaction time.

Figure 7. Optical properties of the titanium oxide thin films. The refractive indices n of the TiO2 thin films, fabricated on a Si substrate with heat treatment at 80 °C, at λ = 550 nm (○) and λ = 633 nm (●) as a function of reaction time (A) of the mesh size (correlation length), ξ, of the titanium oxide polymeric aggregates in the precursor solutions (B), and of the apparent electron density fluctuation, Δρapp, of the polymeric precursors defined as Δρapp ≡ [I(q → 0)/Rg3]0.5 (C).

(TTBO/H2O/2-propanol/N2H4HCl = 1:2:21:0.04) by means of time-resolved SAXS (Tr-SAXS) for 72 h. Variations of SAXS intensities, I(q), of the W2 liquid precursor are shown in Figure 6A. A pronounced increase in I(q) with the reaction time provides evidence for the growth of the titanium oxide polymers in the precursor solutions. Figure 6B shows the variation of Rg and ξ as a function of the reaction time. The data show that we are able to treat ξ of the titanium oxide liquid precursors as a tunable parameter by controlling the reaction time while preserving the low fractal dimensionality (df ≈ 5/3). The refractive indices, n, of the resulting titanium oxide thin films fabricated from the precursors at various reaction times of 1, 6, 12, 24, 48, and 72 h were evaluated (Figure 7 and Table 3). These films were heat treated at 80 °C. We found that, as shown in Figure 7B, n of the film plotted as a function of the mesh size, ξ, of the corresponding polymeric liquid precursors is clearly a decreasing function of ξ except for the film obtained at very short reaction times (∼1 h), where the concentration of unreacted monomers is likely to be high. This implies that a higher local density in the liquid precursors is a key to increasing the n value of the resulting thin films. Considering the relation I(q) ∝ NΔρ2V2,39 where N and V are the number density and volume of a scattering object, respectively, in the reacting system we can reasonably assume that I(q) ∝ CVΔρ2 because NV is a constant, C. If we define the apparent electron density, Δρapp, of the polymeric aggregates as Δρapp ≡ [I(q → 0)/Rg3]0.5, with reference to I(q) ∝ CVΔρ2, the n values of the thin films are sharply increased with Δρapp as shown in Figure 7C. All of these

Table 3. Variation of Microscopic Geometries of the W2-Based Liquid Precursors with Reaction Time and the Optical Properties and Thickness of the Resulting Amorphous TiO2 Thin Films Fabricated on a Si Substrate with Low-Temperature Heat Treatment at 80 °Ca reaction time, t (h)

ξ (nm)

Rg (nm)

film thickness (nm)

n550

n633

k380

1 6 12 24 48 72

1.4 2.2 2.7 3.2 3.6 4.0

2.2 3.4 4.0 4.6 5.1 5.8

91 92 93 90 92 93

2.131 2.167 2.155 2.110 2.002 2.011

2.103 2.139 2.127 2.084 1.977 1.985

2.96 × 10−3 3.10 × 10−3 4.11 × 10−3 3.31 × 10−3 2.50 × 10−3 3.35 × 10−3

a

The thickness and optical constants were determined by an analysis of the ellipsometric spectra. ξ is the correlation length estimated by fitting the SAXS curves based on the Ornstein−Zernike equation (eq 4). Rg is the radius of gyration determined from a Guinier plot. n550 is the refractive index at λ = 550 nm. n633 is the refractive index at λ = 633 nm. k380 is the extinction coefficient at λ = 380 nm.

data demonstrate that the high refractive index of the amorphous TiO2 films essentially comes from the high local electron density. The n values at 550 and 633 nm, the k values at 380 nm, and the film thickness determined with ellipsometry are summarized in Table 3. Next, following the method reported by Cabane and coworkers,33,34 we synthesized a titanium oxide sol under strongly acidic conditions at 25 °C for 30 days without any catalyst 12251

dx.doi.org/10.1021/la3015139 | Langmuir 2012, 28, 12245−12255

Langmuir

Article

Figure 8. Scattering intensity, I(q), of the titanium oxide liquid precursor solutions synthesized by the acid-catalyzed sol−gel process33 (red) and the hydrazine monohydrochloride catalyst-assisted sol−gel process (black) (A). Photographs of the acid-catalyzed (B) and hydrazine monohydrochloride-catalyzed (C) titanium oxide precursor sols. Refractive indices, n(λ), of the resulting thin films fabricated from the acid-catalyzed sol (red) and the hydrazine monohydrochloridecatalyzed sol (W2) (black) at 80 °C (D). In the inset of panel D is a photograph of the thin film fabricated from the acid-catalyzed sol fabricated on a Si substrate.

fabricated from the W2 precursor at a reaction time of 6 h (e.g., the film thicknesses and n values at λ = 550 and 633 nm and k values at λ = 380 nm for the TiO2 thin films coated on a Si substrate). The n value of the TiO2 film was lower than 1.9 when no heat treatment was applied or the film was heated to