Atmospheric pressure chemical vapour deposition of titanium dioxide coatings on glass

Abstract

Atmospheric pressure chemical vapour deposition of titanium dioxide coatings on glass substrates was achieved by the reaction of TiCl₄ and a co-oxygen source (MeOH, EtOH, ⁱPrOH or H₂O) at 500–650 °C. The coatings show excellent uniformity, surface coverage and adherence. Growth rates were of the order of 0.3 µm min⁻¹ at 500 °C. All films are crystalline and single phase with XRD showing the anatase TiO₂ diffraction pattern; a = 3.78(1), c = 9.51(1) Å. Optically, the films show minimal reflectivity from 300–1600 nm and 50–80% total transmission from 300–800 nm. Contact angles are in the range 20–40° for as-prepared films and 1–10° after 30 min irradiation at 254 nm. All of the films show significant photocatalyic activity as regards the destruction of an overlayer of stearic acid.

---

Introduction

Titanium dioxide has been extensively investigated as a photocatalyst¹ and it has been shown that, in the form of thin films and powders, it will photo-degrade a wide range of organic substances. Semiconducting titania coatings on glass function by absorbing sub-320 nm light that is incident on the surface to promote the formation of an electron and a hole.² These species can either recombine in the bulk or migrate to the surface, where they promote oxidation and reduction reactions. Most commonly, semiconductors such as titanium dioxide are used to promote the photocatalytic destruction of organic material by oxygen at the surface. In this case, the photogenerated holes oxidise any organic species deposited on the surface, whereas the photogenerated electrons help reduce oxygen to water. In such circumstances, the overall reaction is the complete mineralisation of the organic material by oxygen, sensitised by the underlying semiconductor substrate.³ Titania has been shown to photo-degrade bacteria, viruses, herbicides, pesticides and an extensive range of functionalised and unfunctionalised organic molecules.⁴ Commercial applications of photoactive TiO₂ include coatings for bathroom tiles and paving slabs, and as a deodoriser in underground stations.⁵

Thin films of titania have been obtained by chemical vapour deposition (CVD),⁶ physical vapour deposition (PVD),⁷ dip coating⁸ and spin coating.⁹ Titania coatings on glass have been grown by low pressure (LP) CVD using Ti(OⁱPr)₄ and oxygen,¹⁰ and they have also been grown on silica, sapphire and indium phosphide substrates.¹¹ To some extent, the substrate temperature can determine whether anatase or rutile films are grown. Typical growth rates in the LPCVD of TiO₂ were of the order of 0.2 µm h⁻¹.¹² The low pressure CVD of TiCl₄ and oxygen has also been used to form TiO₂ coatings.¹³ Problems with this route include the incorporation of chlorine into the coating. PVD methods employed to form titania coatings include both sputtering and molecular beam methods.¹⁴ Notably, no atmospheric pressure CVD routes to TiO₂ films have yet been reported.

Titanium dioxide has recently been investigated as a semiconducting photocatalyst for solar energy conversion and environmental purification.¹⁵ In addition to its ability to split water and photodegrade organic contaminants, ultraviolet irradiation of titania has been found to induce a patchwork of super-hydrophilicity across the surface.¹⁶ This hydrophilicity is maintained by sunlight, so that contaminants are readily washed away by rainwater. Such surfaces have many practical applications, including the production of self-cleaning and antifogging materials. Titania coatings on glass greatly reduce the need to clean windows, a feature which could find particular uses in motor cars, tall buildings and household glazing.¹⁷ Commercialised titania coatings have been prepared by sol–gel routes.¹⁸ Latterly, spray pyrolysis using titanium(IV)acetate in aqueous suspensions has been patented by PPG.¹⁹ Pilkington glass have commercialised the world’s first self-cleaning coating for window glass, Pilkington Activ^TM.²⁰

Here, we report the atmospheric pressure chemical vapour deposition (APCVD) of titania coatings on glass from the reaction of titanium tetrachloride and various oxygen sources. Notably, growth rates were of the order of 0.3 µm min⁻¹. The titania films are shown to function as effective photocatalysts for the photo-degradation of layers of stearic acid.

Experimental

Nitrogen (99.99%) was obtained from BOC and used as supplied. Coatings were obtained on float-glass with an SiCO coating. The SiCO layer is a barrier coating that contains silicon, carbon and oxygen. It is very effective in stopping diffusion of ions from the glass into the growing CVD film. The float glass used was standard window glass and is essentially a silicate glass that contains significant amounts of other elements, including, sodium, calcium and magnesium. APCVD experiments were conducted on 225 mm × 89 mm × 4 mm pieces of glass using a horizontal-bed cold-wall APCVD reactor. Details of the CVD reactor and its usage conditions have been reported previously.²¹ Gases came directly from a cylinder and were preheated by being passed along 2 m lengths of stainless steel tubing which were coiled and inserted into a tube furnace. The temperatures of all the gas inlet lines were monitored with Pt–Rh thermocouples and Eurotherm heat controllers. Titanium(IV) chloride (99.9%, Aldrich) was used as supplied and placed into a stainless steel bubbler heated to 68 °C on a hot plate, TiCl₄ was introduced into a gas stream by passing hot nitrogen gas through the liquid. Co-oxygen sources (MeOH, EtOH, ⁱPrOH and H₂O) were introduced into a gas stream by passing hot nitrogen through a bubbler heated to 50–80 °C. Streams of TiCl₄ (diluted with nitrogen) and oxygen precursors were mixed by using concentric pipes of 1/4 and 1/2 inches in diameter, the inner pipe being 3 cm the shorter. The concentric pipes were attached directly to the mixing chamber of the coater. Gas flows were adjusted using suitable regulators and flow controllers. The exhaust from the reactor was vented directly into the extraction system of a fume cupboard. All of the apparatus was baked out with nitrogen at 150 °C for 1 h before the runs. Suitable two- and four-way valves (containing VESPEL inserts and rated to 200 °C) allowed the nitrogen lines to be diverted into or away from the bubbler. Deposition experiments were conducted by heating the horizontal bed reactor and the bubbler to the required temperatures before diverting the nitrogen line through the bubbler, and hence to the reactor. Deposition experiments were timed by stop-watch. At the end of the deposition, the bubbler line was closed and only nitrogen was passed over the substrate, which was allowed to cool with the graphite block to ca. 60 °C before it was removed. Coated substrates were handled and stored in air. The large coated glass sample was broken up into ca. 1 cm × 1 cm squares for subsequent XPS, EDAX, SEM, electron probe, Raman and UV studies. Large pieces of glass (ca. 4 cm × 4 cm) were used for measurement of the sheet resistance, X-ray powder diffraction, infrared spectra and contact angles, and for photocatalysis and Scotch tape tests.

The general apparatus used by us for X-ray powder diffraction, Raman, SEM/EDAX, electron microprobe, XPS, UV/vis and reflectance/transmission spectra has been described before.²¹ Contact angles of selected glass samples were determined by measuring the spread of a 1.0 mm³ droplet of water.

The photocatalytic activity of the samples was assessed by testing their ability to destroy an overlayer of a test organic chemical—stearic acid—on a 4 cm × 4 cm section of glass coated with TiO₂. Stearic acid was chosen as the test chemical because it can be readily formed into an even layer by a spin coating procedure. It also has an easily identifiable IR signature in the C–H region (in which the glass substrate is transparent) and is essentially non-volatile at the operating temperature of the photocatalysis measurements. The stearic acid was applied by dropping 7.5 mm³ of a 0.4 mM methanol solution onto the glass surface, which was spun at 1500 rpm during the dropping procedure. The infrared spectrum of the stearic acid overlayer was quantified over the range 3000–2800 cm⁻¹. The glass coated with stearic acid was irradiated for 15 min with 254 nm radiation from a BDH germicidal lamp (2 × 8 W).

Table 1 Film deposition conditions and analysis for the APCVD reaction of TiCl₄ and ethanol

Reactor T/°C	TiCl4 bubbler T/°C	EtOH bubbler T/°C	TiCl4 carrier flow/dm^3 min^−1	EtOH carrier flow/dm^3 min^−1	N2 flow/dm^3 min^−1	Run time/s	Contact angle/°	Stearic acid degradation^a (%)
a Percentage stearic acid degradation after 15 min irradiation of an overlayer on the film with 254 nm light.
650	68	42	0.1	0.1	11.0	10	3	25
650	68	42	0.2	0.2	10.8	10	5	62
650	68	42	0.3	0.3	10.6	10	3	65
650	68	42	0.4	0.4	10.4	10	2	94
625	68	42	0.2	0.2	7.5	30	39	55
600	68	43	0.4	0.4	7.1	30	6	72
550	68	43	0.4	0.4	7.1	180	1	67
500	68	43	0.2	0.2	7.5	30	9	60

---

Results

The reaction of anhydrous titanium(IV) chloride with the oxygen sources methanol, ethanol, isopropanol and water under APCVD conditions at 500–650 °C led to coatings of titanium dioxide on the substrates (Table 1). All the films prepared from the APCVD of TiCl₄ and a co-oxygen source were uniform, colourless and impervious to 48 h immersion in common solvents (water, ethanol, acetone, toluene) and mineral acids. The films passed the Scotch tape test and could not be scratched with a steel scalpel. Visual inspection of all of the films showed that, for deposition times from 10 to 30 s, the films were optically transparent with minimal reflectance. For growth times in excess of 180 s and for films grown with water as a co-reactant, some haze was noted and the films appeared white. Growth rates were typically 0.3 µm min⁻¹ at a substrate temperature of 500 °C, as assessed by scanning electron microscopy (SEM). All of the films have resistivity values in the region of 10³ Ω cm, values typical of fully stoichiometric TiO₂.¹⁵ This is somewhat surprising as TiO₂ synthesised by chemical means at elevated temperatures is known to be oxygen deficient and, hence, has a lower resistivity.^9,12 Presumably, the cool-down stages of the APCVD process and the presence of an oxygen-rich environment ensure that stoichiometric material is obtained.

Glancing angle X-ray diffraction shows that films prepared at 500–650 °C for a deposition time of 30 s or more are crystalline [Fig. 1; tetragonal, a = 3.78(1), c = 9.51(1) Å]. In all cases, only a single phase was observed, and this could be indexed to anatase.²² Deposition times of less than 30 s led to films that were too thin for X-ray diffraction analysis. Line broadening studies of the most intense (1 0 1) peak indicate that crystallite sizes are ca. 400 Å. The anatase phase is the desired form of TiO₂ for photocatalysis.¹⁷

Fig. 1 Glancing angle X-ray diffraction pattern of the film produced from the APCVD reaction of TiCl₄ and MeOH (60 s deposition time at 500 °C). The lines show the expected pattern for anatase TiO₂.

Raman analysis of the films shows that anatase TiO₂ was formed in all of the deposition runs. Characteristic bands are observed at 143 (vs), 198 (w), 398 (m), 515 (m) and 640 (ms) cm⁻¹ (Fig. 2). Comparison with literature assignments²³ indicates that the bands should be assigned to E_g, E_g, B_1g, A_1g/B_1g, and E_g modes, respectively. It should be noted that diagnostic Raman spectra could be obtained even from the thinnest coatings grown with only a 10 s deposition time (ca. 0.1 µm thick), whereas X-ray diffraction analysis of these films failed to yield a diffraction pattern. This indicates that Raman spectroscopy can afford useful information where other (standard) techniques fail to provide meaningful data. The Raman spectra did not show any evidence for the presence of hydrogen in the films (no O–H bands).

Fig. 2 Raman spectrum (λ_o = 632.8 nm) of a film produced from the APCVD of TiCl₄ and EtOH on glass (30 s deposition time at 500 °C).

SEM indicates that the films are uniform and have a morphology consistent with an island growth mechanism (Fig. 3). Electron probe line analysis showed a uniform composition across various line scans and energy dispersive X-ray analysis (EDAX) showed a uniform composition across a number of surface spots. Neither technique showed any chlorine incorporation into the films; only titanium and oxygen being observed. Quantification of the oxygen content using these X-ray analytical techniques was complicated because they were operated at the limit of the detection capability (boron cut-off); however, the results are in broad agreement with those obtained from X-ray photoelectron spectroscopic (XPS) analysis.

Fig. 3 SEM micrographs of the films produced from the APCVD of (a) TiCl₄ and ⁱPrOH (15 s deposition time at 500 °C) and (b) TiCl₄ and MeOH (15 s deposition time at 500 °C).

XPS depth profiles showed no chlorine contamination, but all revealed changes in composition with depth. At the surface, some carbon and oxygen were observed, corresponding to adsorbed oxygen and carbon dioxide, which could be removed on the first etching with a molecular beam of Ar⁺ ions. After subsequent etches, a bulk uniform composition was obtained. Elemental compositions were determined by integrating the peak areas from the XPS spectra and applying appropriate significance factors (correlated with standards). The bulk composition titanium-to-oxygen ratio was found to be 1.00∶2.01, consistent with fully stoichiometric TiO₂. The binding energy measurements obtained after depth profiling were typically Ti_2p at 458.6 eV and O_1s at 530.6 eV, values which showed virtually no dependence on film composition. The O_1s binding energies are as expected for a metal oxide (O_1s at 529.5–531.3 eV). Both the Ti_2p and O_1s binding energies agree with values obtained previously for bulk titania and titania thin films.¹¹

The deposition of titania films was found to be very reproducible. Repeat experiments under the same conditions were found to form films that gave identical optical absorption and transmittance/reflectance spectra (Fig. 4). The optical properties of the films were assessed in the region 240–2600 nm by reflectance and transmittance measurements. All of the films grown by APCVD of TiCl₄ and a co-oxygen source with deposition times of up to 60 s showed good transmittance (60–90%) and minimal reflectance, comparable with that of plain glass.

Fig. 4 UV absorption edge reproducibility plots.

The contact angle of the films shows a marked dependence on UV irradiation. Prior to UV irradiation, contact angles were typically in the range 20–40°, slightly lower than for plain glass (40–50°), indicating a relatively weak hydrophobic surface. However, after only 30 min irradiation with 254 nm light, the contact angles decreased to 1–10°, depending on the sample. That is, the water entirely wets the film, a phenomenon known as super-hydrophilicity. The contact angles increased to the pre-irradiated values when the samples were kept in the dark for 48 h. The rate of loss of hydrophilicity is markedly less if the samples are stored in natural light; they still showed pronounced hydrophilicity a week after irradiation.

The photocatalytic behaviour of the films was tested by examining their capacity to destroy a test organic overlayer. All of the TiO₂ films showed a good photocatalytic response to stearic acid overlayers, which were often completely destroyed after 30 min of irradiation. Notably, the hazy films were also photocatalytically active. Direct correlations between changes in the intensity of the infrared bands of stearic acid at 2950 and 2900 cm⁻¹ with irradiation time are shown in Fig. 5. The normalised peak areas quoted represent the total quantified area of the C–H bands relative to the initial IR response of the film before irradiation. All of the photoactivity tests were performed with 254 nm light. This monochromatic source was chosen because it has a well-defined emission profile (99.9% of emitted energy at 254 nm), is inexpensive and gave fast destruction rates of stearic acid overlayers. Notably, photo-degradation experiments could be completed within 1 h. The 254 nm light source did not remove any stearic acid on the plain glass used as a control, even after 10 h illumination. For detailed photocatalytic measurements, which are beyond the scope of this work, 365 nm lamps, Hg lamps and filters could be used.

Fig. 5 Rate of degradation of stearic acid overlayers on TiO₂ anatase films during irradiation at 254 nm.

Discussion

APCVD of TiCl₄ and an oxygen source provides a facile method for forming TiO₂ films on glass substrates. The films show excellent surface coverage, adhesion and purity. In all cases, anatase was deposited free of chlorine contamination. This is useful as sometimes thin films prepared by APCVD from metal chloride precursors have chlorine in the product.²⁴ Film thicknesses were uniform at growth temperatures of 500 °C and could be directly controlled by varying the deposition time. Notably, the films were free of pin-hole defects. The co-oxygen source did have a marked effect on the nature of the films: the use of water as a co-reactant source yielded specular/hazy films, whereas the use of isopropanol, methanol or ethanol gave uniform transparent films. It is probable that significant reaction occurred in the gas phase when water was used as a co-reactant, and this led to the formation of large particles rather than a uniform coating. The hazy specular films had similar self-cleaning action and low contact angles compared to the other transparent films prepared in this study.

The contact angle measurements show that the films are initially fairly hydrophobic. Typically, the contact angles (20–40°) are dramatically reduced to 1–10° after UV irradiation. Samples maintain low contact angles as long as they are stored in the light, but not when kept in the dark for longer than 48 h. It is widely accepted that the TiO₂ surface differs from the bulk due to incomplete coordination at the surface; whereas TiO₂ in the bulk contains six-coordinate Ti and three-coordinate oxygen atoms, TiO₂ at the surface has five-coordinate Ti and two-coordinate O atoms. Upon UV irradiation, conversion of the surface Ti⁴⁺ sites to Ti³⁺ sites may occur.²⁵ The latter favour water dissociation at the surface, forming hydroxyl species that dramatically increase the hydrophilicity. On long-term storage in the dark, oxygen from the air can replace the hydroxylated surface and, as a consequence, the contact angle increases.

Notably, under all experimental conditions employed in this work, it was the anatase form of TiO₂ that was produced. This phase is formed kinetically in a number of processes, although rutile is thermodynamically the more stable form. Conveniently, anatase TiO₂ is the most desirable form for self-cleaning purposes as it has the most effective photocatalytic response and a low contact angle. Previous low pressure CVD experiments have shown that anatase is routinely formed at deposition temperatures below 650 °C from a variety of titanium alkoxide precursors,²⁴ whereas rutile is encountered at higher substrate temperatures. This is in contrast to dip- and spin-coated films, which often show mixtures of anatase and rutile phases.²⁴

The growth rates of the titania films were found to be dependent on the flow rate of TiCl₄ through the system and film thickness varied linearly with deposition time (ca. 0.3 µm min⁻¹). This is a much faster growth rate than conventional low pressure CVD (typically 0.2 µm h⁻¹).¹² No deposition could be achieved in the system in the absence of an oxygen precursor. Four oxygen precursors were tried—methanol, ethanol, isopropanol and water. All of the alcohol precursors produced similar uniform films, with identical growth rates, surface coverage and film properties. The identical growth rates for the different alcohol precursors indicate that the reaction is not mass transport limited.²⁶ It is possible that the reactions all proceed by a protic exchange mechanism, however, further speculation is unwarranted at this stage. Films grown using water as a precursor had a white appearance associated with larger particles. This is a common phenomenon in CVD and is a consequence of excessive gas phase reaction of the precursors.²⁴ Notably, the use of alcohol precursors did not introduce detectable amounts of carbon into the films. Furthermore, the fact that uniform layers of TiO₂ were formed from alcohols indicates that the degree of gas phase reaction is much lower than for water and that the chemistry occurs on the substrate surface.

All of the films showed good photocatalytic activity with respect to the destruction of a stearic acid overlayer and activities were comparable with those achieved with sputtered anatase films.⁷ The rate of photocatalytic destruction of stearic acid is related to the film thickness, thicker films showing the greater rate of destruction. This is, to our knowledge, the first paper to report a CVD film functioning as an active photocatalyst.^27,28

Conclusion

Atmospheric pressure chemical vapour deposition of TiCl₄ and an oxygen precursor, an alcohol or water, at 500–650 °C produces anatase films of good quality on glass substrates. The films are optically transparent, hydrophilic and function as effective photocatalysts for the destruction of stearic acid overlayers. The rate of photodestruction of a test organic compound, stearic acid, was found to be related to film thickness. The films all show light-induced hydrophilicity. Dual-source APCVD is thus shown to be an ideal method for depositing uniform films of anatase. Notably, despite using a chloride-containing starting material, no chlorine was detected in the products. Alcohols have been shown to be effective sources of oxygen in the CVD process. Even when the alkyl chain length of the alcohol was increased from methyl to isopropyl, no carbon contamination was detected in the CVD films. The surface coverage of the films on a static coater system was excellent, giving uniform coatings over the entire substrate, indicating that the process is not transport limited. The combination of low contact angle and extremely fast photocatalytic response means that these TiO₂ surfaces are suitable for self-cleaning applications.

Acknowledgements

I. P. P., R. J. H. C. and S. O. N. thank the EPSRC for grants GR/M95095 and GR/M82592. A. M. and N. E. thank the EPSRC for grant GR/M95042/01. Drs K. Sanderson and S. Hurst of Pilkington plc are thanked for substrate glass and advice.

References

1. (a) N. Fujihira, Y. Satoh and T. Osa, Nature, 1981, 293, 206; (b) D. F. Ollis, Environ. Sci. Technol., 1985, 19, 480; (c) Y. Nosaka, K. Koenuma, K. Ushida and A. Kira, Langmuir, 1996, 12, 736.
2. A. Fujishima and A. Harsta, Nature, 1972, 238, 37.
3. A. Mills and S. LeHunte, J. Photochem. Photobiol., A, 1997, 108, 1.
4. T. Minabe, D. A. Tryk, P. Sawunyama, Y. Kikuchi, K. Hasimoto and A. Fujishama, J. Photochem. Photobiol., A, 2000, 137, 53.
5. K. Suzuki, in Photocatalytic Purification and Treatment of Water and Air, ed. D. Ollis and A. El Akabi, Elsevier, New York, 1993, p. 421.
6. (a) M. C. Marco de Lucas, F. Fabreguette, S. Collin and S. Bourgeois, Int. J. Inorg. Mater., 2000, 2, 255; (b) D. Byun, Y. Jin, B. Kim, J. L. Lee and D. Park, J. Hazard. Mater., 2000, B73, 199.
7. D. Damiriv, A. R. Bally, C. Ballif, P. Homes, P. E. Schmid, R. Sanjines, F. Levy and V. I. Parvulescu, Appl. Catal., B, 2000, 25, 83.
8. X. P. Wang, Y. Yu, X. F. Hu and L. Gao, Thin Solid Films, 2000, 371, 148.
9. (a) M. Gratzel and B. O'Regan, Nature, 1991, 353, 737; (b) K. Kato, A. Tsuzuki, H. Taoda, Y. Torrii, T. Kato and Y. Batsugn, J. Mater. Sci., 1994, 29, 5911.
10. N. Rausch and E. P. Burte, J. Electrochem. Soc., 1993, 140, 143.
11. (a) Y. H. Lee, K. K. Chan and M. J. Brady, J. Vac. Sci. Technol., A, 1995, 13, 596; (b) Q. Chen, Y. Qian, Z. Chen, W. Wu, G. Zhou and Y. Zhang, Appl. Phys. Lett., 1995, 66, 1608; (c) E. K. Kim, M. H. Son, S. K. Min, Y. K. Hon and S. S. Yom, J. Cryst. Growth, 1997, 170, 803.
12. Q. Zhang and G. L. Griffin, Thin Solid Films, 1995, 263, 65.
13. G. A. Battiston, R. Gerbasi, A. Gregori, M. Porchia, S. Cattorein and G. A. Rizzi, Thin Solid Films, 2000, 371, 126.
14. J. D. Deloach, G. Scarel and C. R. Aita, J. Appl. Phys., 1999, 85, 2377.
15. (a) M. A. Gratzel, Nature, 1993, 366, 206; (b) B. Oregan and M. A. Gratzel, Nature, 1991, 353, 737; (c) U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissortel, J. Salbeck, H. Spreitzer and M. A. Gratzel, Nature, 1998, 395, 583; (d) H. Tang, K. Prasad, R. Sanjines, P. E. Schmid and F. Levy, J. Appl. Phys., 1994, 75, 2042.
16. R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni, E. Kojima, A. Kitamura, M. Shimohigoshi and T. Watanabe, Nature, 1997, 388, 431.
17. (a) D. A. Tryk, T. Watanabe, K. Hashimoto and A. Fujishima, Int. Glass Rev., 2001, 2, 34; (b) P. Godwin, Int. Glass Rev., 2001, 2, 29.
18. L. Paz, J. Mater. Res., 1995, 10, 2842.
19. L. Greenburg, US Pat., 6 027 766, Feb 22, 2000.
20. Pilkington Activ^TM self-cleaning glass, Pilkington plc, St Helens, UK (http//:www.activglass.com/index2.htm).
21. A. Newport, C. J. Carmalt, I. P. Parkin and S. A. O'Neill, J. Mater. Chem., 2002, 12, 1906.
22. JCPDS International Centre for Diffraction Data, Card no. 21-1272, 1995.
23. (a) T. Ohsaka, F. Izumi and Y. Fujiki, J. Raman Spectrosc., 1978, 7, 321; (b) L. Burgio and R. J. H. Clark, Spectrochim. Acta, Part A, 2001, 57, 1491.
24. S. Kittaka, K. Matsuno and S. Takahara, J. Solid State Chem., 1997, 132, 447.
25. R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni, E. Kojima, A. Kitamura, M. Shimihigoshi and T. Watanabe, Adv. Mater., 1998, 10, 135.
26. M. L. Hitchman and K. F. Jensen, Chemical Vapour Deposition, Academic Press, London, 1993.
27. (a) J. T. Remillard, J. R. McBride, K. E. Nietering, A. R. Drews and X. Zhang, J. Phys. Chem. B, 2000, 104, 4440; (b) D. Byun, Y. Jin, B. Kim, J. K. Lee and D. Park, J. Hazard. Mater., 2000, 199.
28. J. P. Theil, US Pat., 5 873 203, Feb 23, 1999.

---