Effect of annealing temperature on titania thin films ... - Springer Link

J Sol-Gel Sci Technol (2010) 55:328–334 DOI 10.1007/s10971-010-2257-y

ORIGINAL PAPER

Effect of annealing temperature on titania thin films prepared by spin coating A. Nakaruk • C. Y. Lin • D. S. Perera C. C. Sorrell

•

Received: 14 March 2010 / Accepted: 22 May 2010 / Published online: 10 June 2010 Springer Science+Business Media, LLC 2010

Abstract Essentially fully dense titania thin films were spin coated on fused quartz substrates under identical conditions and subjected to annealing over the range 750– 900C. The films were of a consistent *400 nm thickness. The anatase ? rutile phase transformation temperature was between 750C and 800C, with first-order kinetics; annealing at 900C yielded single-phase rutile. Silicon contamination from the fused quartz substrate was considered to be critical since it suppressed both titania grain growth (maintaining constant grain size) and the phase transformation (occurring at an unusually high temperature); its presence also was considered to be responsible for the formation of lattice defects, which decreased the transmittances and the band gaps. Keywords Titanium dioxide Spin coating Glancing angle X-ray diffraction Optical properties

1 Introduction During the past decade, titanium dioxide (TiO2, titania) has emerged as an important material for applications such as transparent electrical conductors [1], photovoltaics [2], water purification media [3], self-cleaning surfaces [4], and gas sensors [5]. Of particular interest are the photocatalytic properties of titania for applications such as water splitting for hydrogen production [6]. For these applications, thin A. Nakaruk C. Y. Lin D. S. Perera C. C. Sorrell (&) School of Materials Science and Engineering, University of New South Wales, Sydney, NSW 2052, Australia e-mail: [email protected] A. Nakaruk e-mail: [email protected]

123

and/or thick films are used commonly owing to the advantages of easy preparation, flexibility of use, and low cost in comparison to bulk materials. There are several effective methods for the synthesis of titania films, including sputtering [7], laser ablation [8], electrophoretic deposition [9], anodic oxidation [10], sol–gel [11], screen printing [12], dip coating [13], gel oxidation [14], spray pyrolysis [15], and spin coating [16]. Although spray pyrolysis is one of the most widely used methods for the fabrication of titania thin films owing to rapid growth rates (B100 nm/min [17]), large sample size capacity, and mass production capability [18], it tends to have a low precursor/product yield rate (\5%) during deposition [19]. Consequently, more productive alternative methods, such as spin coating, have been developed [16, 20–22] since they offer the same advantages but with higher yields. Of the three common polymorphs of titania (anatase, rutile, and brookite [23–25]), anatase and rutile are stable at room temperature, although rutile is the only thermodynamically stable phase [26, 27]. There is some uncertainty concerning the suitability of these two polymorphs for photocatalytic applications. The commonly argued basis for this uncertainty lies in the contradiction between the higher transparency, band gap, and surface area of anatase [28–30] in comparison to the lower transparency, band gap, and surface area of rutile [30–32]. A common approach to the resolution of this contradiction is the use of mixedphase anatase ? rutile microstructures, which attempt to combine the effects from the higher transparency and surface area of anatase with the lower band gap of rutile. This can be done by moderation of heating or annealing temperatures in order to control grain growth and the anatase ? rutile phase transformation. The purposes of the present work were to (1) prepare fully dense thin films

J Sol-Gel Sci Technol (2010) 55:328–334

329

using spin coating, followed by annealing at a range of temperatures, in order to obtain mixed-phase films of different anatase/rutile ratios and (2) to characterise these films in terms of the mineralogical, morphological, and optical characteristics.

2 Experimental procedure Precursors were prepared using titanium isopropoxide (TIP, Reagent Grade, 97 wt%, Sigma–Aldrich) dissolved in isopropanol (Reagent Plus C99 wt%, Sigma–Aldrich) at a titanium concentration of 0.1 M (2.8 g of TIP diluted to 100 mL volume with isopropanol); the solutions were mixed in glass by stirring at 400 rpm for 15 min without heating. Spin coating (Laurell WS-65052) was done by dropping *0.2 mL of solution onto a fused quartz substrate (25 mm 9 25 mm 9 1 mm, Semiconductor Wafer, Inc., Taiwan) spun at 2,000 rpm in air. The film was dried by spinning for 15 s, followed by instantaneous heating at 100C for 5 min. This process was repeated four more times in order to obtain as-deposited thin films of *500 nm thicknesses. Subsequent annealing was done using a tube furnace at a heating rate of 300C/h and soak time of 4 h at temperatures in the range 7508–900C, followed by natural cooling. This process reduced the thickness of all of the films to *400 ± 50 nm. The mineralogy of the films was assessed using glancing angle X-ray diffraction (GAXRD, angle of incidence 1 2h, penetration depth \300 nm, speed 18/min 2h, step size 0.028 2h, Phillips X’pert Materials Research Diffraction). The film thickness was determined using single-beam focussed ion beam (FIB) milling (FEI XP200). In this method, a chromium coating of 20 nm thickness is applied by sputtering, gallium ions (Ga3?) are used to erode a square hole in the film, and an image of the cross-section of the layers is viewed at an angle of 45. The microstructures from the FIB work were examined using field emission scanning electron microscopy (FESEM, 5 kV accelerating voltage, secondary electron emission mode, Hitachi S4500). Optical transmission spectra in the visible region (300–800 nm) were obtained prior to chromium coating using a dual beam UV–VIS spectrophotometer (Perkin Elmer Lambda 35).

3 Results and discussion The GAXRD patterns of the films, which are shown in Fig. 1, show that four types of mineralogies were obtained: (1) as-deposited at 100C: amorphous; (2) annealed at 750C: anatase; (3) annealed at 8008–850C: mixed anatase ? rutile; and (4) annealed at 900C: rutile.

Fig. 1 Representative glancing angle X-ray diffraction (GAXRD) patterns (room temperature) of TiO2 films as a function of annealing temperature

The proportions of anatase and rutile were calculated using the commonly used general Equation 1 [33]. Ia ela qa t ¼ Ir elr qr t

ð1Þ

where Ia (101) is peak intensity (height) of anatase (maximal), Ir (110) is peak intensity (height) of rutile (maximal), la is mass absorption coefficient of anatase (129.41 cm2/g [34]), lr is mass absorption coefficient of rutile (138.87 cm2/g [34]), qa is true density of anatase (3,890 kg/m3 [34]), qr is true density of rutile (4,250 kg/m3 [34]), t is thickness of the film (400 nm). Table 1 summarises the data from the preceding equation. It can be seen that the anatase/rutile ratio decreased with increasing annealing temperature. Figure 2 clarifies that, for these experimental conditions, the anatase ? rutile phase transformation kinetics as a function of temperature was first order. It is assumed commonly that anatase transforms to rutile at *600C in bulk materials [35, 36]. However, the ‘‘actual’’ transformation temperature depends significantly on the method of determination, the presence of impurities/ dopants, and the morphology of the materials [43]. For example, this transformation has been observed in thin films deposited on fused quartz substrates at temperatures as high as 800C in ultrasonically spray pyrolysed films [19] but at 900C in sol–gel dip coated thin films [37]. In the present work, for these experimental conditions, Fig. 1 shows that this temperature, corresponding to the onset of the anatase ? rutile transformation, is between 750C and 800C. It is likely that the apparent increase in the transformation temperature relative to the commonly accepted temperature originated from silicon contamination from the fused quartz substrate [19]. Silicon has been identified as

123

330


Table 1 Summary of analytical data

* 10 nm limit of resolution of field emission scanning electron microscope

Fig. 2 Representative ratios of anatase/rutile phase compositions of TiO2 films as a function of annealing temperature

an inhibitor of this phase transformation [38–45] and it is known that it can diffuse from the substrate to the film at temperatures as low as 600C [41]. Alternatively, contamination from silicon or other inhibitors may have derived from the raw materials, which were of 97 wt% (TIP) and 99 wt% (isopropanol) purities. Silicon contamination has been confirmed in other work by the authors: •

•

Photoluminescence: Nakaruk et al. [44] annealed titania films ultrasonically spray pyrolyzed on fused quartz substrates and observed photoluminescence quenching by silicon contamination in films annealed at C800C. Lattice Expansion: Nakaruk et al. [19] observed glancing angle X-ray diffraction peak shifts in titania films ultrasonically spray pyrolyzed on fused quartz substrates following annealing at C600C.

The surface morphologies of representative titania films are shown in Fig. 3 and the approximate grain sizes are given in Table 1. It can be seen that annealing had the effect of increasing the amount of agglomeration associated with the anatase ? rutule phase transformation

123

[19, 33] with increasing temperature, resulting the apparent generation of a small amount of closed porosity owing to the independent shrinkage of these agglomerates at 900C. Regardless of the annealing temperature, the anatase matrix grain size and rutile agglomerate size showed little change. Although agglomeration as a feature of rutile formation has been reported previously [19, 33], the important observation is that, since the film annealed at 900C consisted solely of rutile, these data demonstrate that it is possible to generate titania consisting of rutile grains of the same size as matrix anatase. However, it is critical to point out that this result may derive from the effect of silicon contamination, which inhibits the anatase ? rutile phase transformation [19, 43–45]. It also is possible that a grain boundary amorphous silica film, generated by contamination from the substrate, may act to suppress titania grain growth [46]. The cross-sectional morphologies of the films, obtained from the FIB images, are shown in Fig. 4. It can be seen that the thickness of all films was consistent at *400 nm, although a variation of ±50 nm both between and within films was possible. It is known that, once recrystallisation of the amorphous film has occurred, the temperature of the annealing process does not affect the thickness [19, 47]. The important observation is that this thickness corresponds to the rutile agglomerate size, which indicates that these films consist of a single agglomerate’s thickness. As mentioned above, previous work by the authors [19, 33] has demonstrated the association of agglomeration and rutile formation. However, in these cases, the films were of thickness *1 lm. Hence, the main difference between the films of those studies and the present is the number of layers of grains present. In the present work, the consistency of the matrix and agglomerate grain sizes indicates that little grain growth occurred. It is possible that the apparent suppression of grain growth resulted from the dimensionality available for grain growth. That is, in the previous studies [19, 33],


331

Fig. 3 Representative field emission scanning electron microscope (FESEM, secondary electron emission) micrographs of TiO2 films as a function of annealing temperature

typical three-dimensional volumetric diffusion and grain growth occurred while, in the present study, only twodimensional lateral diffusion and grain growth could occur. This restriction to two-dimensional mass transfer is likely to have limited the titania necessary for grain growth but it also is likely to have enhanced silicon and/or glass diffusion as the more rapid and shorter grain boundary diffusion transit paths are easily accessible to the substrate [46]. Although these diffusion paths can be expected to be affected by the presence of agglomerates, tending to counteract these effects, examination of the film surfaces in Figs. 3 and 4 shows that the level of asperity generation by agglomeration is fairly consistent for all annealing temperatures.

The transmittances of the films as a function of wavelength are shown in Fig. 5. The transmittance of a film depends principally on the following features: (1) mineralogy, (2) thickness, (3) grain size (viz., mean free path [19]), (4) chemical homogeneity (viz., impurities, dopants, and associated structural defects), (5) surface microstructural homogeneity (viz., smoothness), and (6) bulk microstructural homogeneity (viz., irregularities and inconsistencies). The important observation is that, of these six principal variables, four of these are approximately constant for these films, which indicates that the optical characteristics of these films derive principally from the mineralogy (anatase/rutile ratio) and the chemical homogeneity (silicon impurities). The anatase/rutile ratio should be considered to have been

123

332


Fig. 4 Representative single-beam focussed ion beam (FIB) images of TiO2 films as a function of annealing temperature

calculated relatively accurately since previous work [33] has demonstrated that the peak heights are representative of the corresponding peak areas. The chemical homogeneity can be considered to be affected significantly by contamination from the substrate owing to the shapes of the curves of the films annealed at 8008–900C relative to those annealed at lower temperatures. At the higher annealing temperatures, the absorption edges are not sharp, which is indicative of the presence of atomic and/or microstructural defects [32].

123

Hence, these data provide secondary evidence for the presence of chemical contamination (but not microstructural since the microstructures are consistent) of these films. Silicon contamination from raw materials alone cannot have caused these data as this would be constant for all films. However, contamination from both raw materials and substrates remains a possibility. The data in Fig. 5 have been used to calculate the optical indirect band gap (Eg) of the films. This well known


Fig. 5 Representative UV–VIS spectrophotometry data of TiO2 films as a function of annealing temperature

optical transmittance method [48] uses Equation 2 to convert the transmittance and film thickness data to the absorption coefficient and band gap: 2 1 a ¼ lnðT Þ ffi A hm Eg ð2Þ t where: a is absorption coefficient (unitless), t is film thickness (cm), T is transmittance (%), A* is a constant that does not depend on hm (unitless) [48], h is Planck’s constant (4.135 9 10-15 eV.s), m is frequency (s-1), Eg is optical indirect band gap (eV). As shown in Fig. 6, when (a)1/2 is plotted on the ordinate against hm (photon energy) on the abscissa, then the intercept of the tangent to the absorption edge with the abscissa gives an estimate of the optical band gap energy, which is indirect in the case of anatase and rutile [48–50]. The values obtained from these data are given in Table 1. The data for these films can be summarised as follows: As-Deposited (Amorphous Titania): The optical band gap was 3.35 eV. This value is of limited interest (as a starting point) because it is characteristic of amorphous titania that has not yet recrystallised during annealing.

333

Annealed at 750C (Anatase): The optical band gap was 3.05 eV. This probably is a relatively representative value for well crystallised anatase as it matches well with those values reported elsewhere [48–50] and the data show a sharp absorption edge. However, it is possible that this value is affected to a small degree by the presence of silicon contamination. Annealed at 800, 825, and 850C (Anatase ? Rutile): The optical band gaps were determined to be 2.85, 2.81, 2.72 eV, respectively, on the basis of data showing broad absorption edges. These values merely represent additional reported values for anatase ? rutile mixtures (but likely with relevant silicon contamination) although, in this case, these mixtures are of known relative proportions. In nearly all of other studies, the relative proportions of the two polymorphs were not reported. Annealed at 900C (Rutile): The optical band gap was 2.30 eV. The breadth of the calculated absorption edge indicates the increasing importance of silicon contamination. However, it is pointed out that this value is within the range of those values reported for well crystallised rutile. In the case of the reported optical band gaps, it is difficult to put them into context owing to the wide scale of values reported, which range between 2.40 and 3.50 eV for anatase [32, 50] and between 2.30 and 3.34 eV for singlephase rutile and mixed phase anatase ? rutile [32, 51]. Even wider ranges of values that have been reported individually have been summarised by Serpone [52]. Kang et al. [32] have suggested that modification of the oxide crystal structure, phase transformation, and morphology are responsible for the observation of wide ranges of optical indirect band gaps. Further, as shown in Table 1 and Fig. 5, the concurrent low transmission percentage and broad absorption edge increase the imprecision of the graphical analytical technique used to calculate the indirect band gap owing to obvious geometrical ramifications [32, 53–55]. Consequently, it is problematic to make comparisons between data when any of the six factors that affect the transmittance listed above are not identical.

4 Summary

Fig. 6 Representative optical indirect band gap data of TiO2 films as a function of annealing temperature

Essentially fully dense titania thin films were coated on fused quartz substrates using spin coating. The films asdeposited at 100C were amorphous, with no visible grain morphology. After annealing in the temperature range 750–900C, both the agglomerate size and film thickness were * 400 nm, indicating that the films were of a single agglomerate’s thickness. Since annealing at 750C yielded films consisting of pure anatase while annealing at 800C yielded anatase ? rutile, the phase transformation

123

334

temperature for these experimental conditions was in the range 750-800C. Annealing revealed first-order kinetics for the anatase ? rutile phase transformation. Annealing at 900C yielded single-phase rutile. These data are affected significantly by silicon contamination from the fused quartz substrate. This element is known to suppress both titania grain growth (hence the constant matrix and agglomerate grain size) and the phase transformation (hence the increased temperature); its presence also was responsible for the formation of structural defects that decreased both the transmittances and the band gaps. Acknowledgments The authors are grateful for the financial support of Austral Brick Co. Pty. Ltd., the National Hydrogen Materials Alliance, and the Australian Research Council, which have allowed this and other developmental work to be undertaken.

References 1. Lansa˚ker PC, Backholm J, Niklasson GA, Granqvist CG (2009) Thin Solid Films 518:1225 2. Kroeze JE, Savenije TJ (2004) Thin Solid Films 451–452:54 3. Pihosh Y, Goto M, Kasahara A, Tosa M (2009) Appl Surf Sci 256:937 4. Yaghoubi H, Taghavinia N, Alamdari EK (2010) Surf Coat Technol 204:1562 5. Sadek AZ, Partridge JG, McCulloch DG, Li YX, Yu XF, Wlodarski W, Kalantar-Zadeh K (2009) Thin Solid Films 518:1294 6. Nowotny J, Sorrell CC, Sheppard LR, Bak T (2005) Int J Hydrogen Energy 30:521 7. Wu KR, Hung CH (2009) Appl Surf Sci 256:1595 8. Chang WT, Ciou JY, Wei MK, Wong MS (2008) J Vac Sci Technol A 26:898 9. Abdullah HZ, Sorrell CC (2007) Mater Sci Forum 561–565:2163 10. Abdullah HZ, Sorrell CC (2007) Mater Sci Forum 561–565:2159 11. Yang HM, Zhang XC, Tao QF (2009) Inorg Mater 45:1139 12. Tsoukleris DS, Kontos AI, Aloupogiannis P, Falaras P (2007) Catal Today 124:110 13. Sokolov S, Ortel E, Radnik J, Kraehnert R (2009) Thin Solid Films 518:27 14. Abdullah HZ, Sorrell CC (2007) Mater Sci Forum 561–565:2167 15. Acik IO, Junolainen A, Mikli V, Danilson M, Krunks M (2009) Appl Surf Sci 256:1391 16. Wang MC, Lin HJ, Yang TS (2009) J Alloys Comp 473:394 17. Sun H, Wang C, Pang S, Li X, Tao Y, Tang H, Liu M (2007) J Non-Cryst Solids 354:1440 18. Aukkaravittayapun S, Wongtida N, Kasecwatin T, Charojrochkul S, Unnanon K, Chindaudom P (2006) Thin Solid Films 496:117 19. Nakaruk A, Ragazzon D, Sorrell CC (2010) Thin Solid Films 518:3735 20. Yu HF, Hu FC (2009) J Sol-Gel Sci Technol 52:158 21. Zhang Y, Shen Y, Gu F, Wu M, Xie Y, Zhang J (2009) Appl Surf Sci 256:85

123

J Sol-Gel Sci Technol (2010) 55:328–334 22. Kitsuka K, Kaneda K, Ikematsu M, Iseki M, Mushiake K, Ohsaka T (2009) Electrochim Acta 55:31 23. Gra¨tzel M, Rotzinger FP (1985) Chem Phys Lett 118:474 24. Vallet-Regı´ M, Penã J, Martıńez A, Gonza´lez-Calbet JM (1993) J Mater Res 8:2336 25. Freeman CM, Newsam JM, Levine SM, Catlow CRA (1993) J Mater Chem 3:531 26. Zhang H, Banfield JF (1998) J Mater Chem 8:2073 27. Kim SJ, Park SD, Jeong YH, Park S (1999) J Am Ceram Soc 82:927 28. Cˇernigoj U, Sˇtangar UL, Trebsˇe P, Krasˇovec UO, Gross S (2006) Thin Solid Films 495:327 29. Wu PG, Ma CH, Shang JK (2005) Appl Phys A 81:1411 30. Ahmed Yasir V, MohanDas PN, Yusuff KKM (2001) Int J Inor Mater 3:593 31. Huang HS, Huang CC, Huang PC, Yang CF, Hsu CY. J Nanosci Nanotechnol 8:2659 32. Kang SH, Lim JW, Kim HS, Kim JY, Chung YH, Sung YE (2009) Chem Mater 21:2777 33. Nakaruk A, Kavei G, Sorrell CC (2010) Mater Lett 64:1365 34. Hubbard CR, Evans EH, Smith DK (1976) J Appl Cryst 9:169 35. Rao CNR (1961) Can J Chem 39:498 36. Depero LE, Bonzi P, Zocchi M, Casale C, De Michele G (1993) J Mater Res 8:2709 37. Kim DJ, Hahn SH, Oh SH, Kim EJ (2002) Mater Lett 57:355 38. Walters JK, Rigden JS, Dirken PJ, Smith ME, Howells WS, Newport RJ (1997) Chem Phys Lett 264:539 39. Yang J, Huang YX, Ferreira JMF (1997) J Mater Sci Lett 16:1933 40. Yang J, Ferreira JMF (1998) Mater Lett 36:320 41. Okada K, Yamamoto N, Kameshima Y, Yasumori A, MacKenzie KJD (2001) J Am Ceram Soc 84:1591 42. Reidy DJ, Holmes JD, Morris MA (2006) Ceram Int 32:235 43. Hanaor DA, Sorrell CC, Review of Anatase to Rutile Phase Transformation, in preparation 44. Nakaruk A, Reece PJ, Ragazzon D, Sorrell CC (2010) Mater Sci Technol 26:469 45. Su Y, Chen S, Quan X, Zhao H, Zhang Y (2008) Appl Surf Sci 255:2167 46. Kingery WD, Bowen HK, Uhlmann DR (1976) Introduction to ceramics, 2nd edn. Wiley, New York 47. Nakaruk A, Sorrell CC. J Coat Technol Res in press. doi: 10.1007/s11998-010-9245-6 48. Mardare D, Tasca M, Delibas M, Rusu GI (2000) Appl Surf Sci 156:200 49. Ting CC, Chen SY, Liu DM (2000) J Appl Phys 88:4628 50. Won DJ, Wang CH, Jang HK, Choi DJ (2001) Appl Phys A 73:595 51. Miao L, Jin P, Kaneko K, Terai A, Nabatova-Gabain N, Tanemura S (2003) Appl Surf Sci 212–213:255 52. Serpone N (2006) J Phys Chem B 110:24287 53. Miao L, Tanemura S, Watanebe H, Mori Y, Kaneko K, Toh S (2004) J Cryst Growth 260:118 54. Hong YC, Bang CU, Shin DH, Uhm HS (2005) Chem Phys Lett 413:454 55. Wang XH, Li JG, Kamiyama H, Katafa M, Ohashi N, Moriyoshi Y, Ishigaki T (2005) J Am Chem Soc 127:10982