Effect of the Added Polyethylene Glycol Molecular Weight and Calcination Heating Rate on the Morphology of TiO₂ Films Formed by Sol-Gel

Abstract

This research investigated the changing morphology of silicon (Si)-supported titanium dioxide (TiO₂) thin films with different heating rates and molecular weights (M_W) of the added polyethylene glycol (PEG). The TiO₂ films were deposited on a Si wafer (100) by sol-gel spin coating with PEG (M_W = 6,000 or 35,000 g·mol⁻¹) as pore generating agents. Calcination at 450℃ completely decomposed all the organic residues in the TiO₂ sol, and the resultant films were in the anatase phase. The combustion nature of PEG was found to be the main factor controlling the film's morphology, where the exothermic heat of PEG combustion tended to be higher with increased heating rates and dependent on the type of PEG (extended and folded chain crystal). At heating rates of 10℃·min⁻¹ or higher, the exothermic heat led to localized grain coalescence in the TiO₂ films, which decreased the film porosity. However, this exothermic heat also simultaneously induced pore agglomeration. Hence, the average pore size of PEG-containing films were larger than in films without PEG. In contrast, the heating rate did not significantly affect the morphology films without PEG.

1. Introduction

Silicon (Si)-based devices have been widely used for a solar cell applications because they have a high efficiency of energy conversion. However, they also have a high reflectivity, which can reach approximately 35%R at a 550 nm wavelength.¹⁾ Several approaches have been taken to improve the optical and reflectivity properties of Si for use in solar cells.^2–5) Coating the Si wafer with anti-reflective materials is one way to reduce their reflectivity. Titanium dioxide (TiO₂) thin films have been widely used as an anti-reflective coating for Si solar cells due to their transparency and suitable refractive index of nearly 2.^3–5) The anti-reflective nature depends on several factors, such as the phase, film thickness, pore size and porosity.^3–7) Among the different TiO₂ phases, anatase was found to be appropriate for anti-reflective applications because it has a wider band gap and lower refractive index compared to those of rutile.⁶⁾ Yoldas and Partlow also reported the influence of porosity on the anti-reflective properties, where the refractive index could be reduced to close 2, minimizing light reflection.⁷⁾ However, there exists an optimal pore size range (2–50 nm), where larger pores could cause undesirable light scattering within the film.⁴⁾

Several processing methods have been employed to prepare thin TiO₂ films, including chemical vapor deposition, physical vapor deposition and evaporation. However, these methods yield relatively dense films and are difficult to control the porosity. On the other hand, the simpler sol-gel method allows manipulation of the pore size and porosity by means of using pore-generating agents or calcination conditions. Polyethylene glycol (PEG) is one pore-generating agent that has been widely used in preparing TiO₂ porous thin films by sol-gel.^8–10) The addition amount and molecular weight (M_W) of PEG were reported to be the main parameters in tailoring the film porosity,^9–11) where a higher degree of either parameter tended to improve the film porosity.^9,11) Given that the pore formation is caused by the decomposition of PEG, the characteristics of its thermal decomposition could potentially be another influential parameter affecting the nature of the film porosity. The decomposition of PEG occurs during the heating step of calcination. Common to general polymers, the decomposition kinetics can result in different morphology according to the outward migration. However, there is still a lack of information on the calcination parameters, especially the heating rate, which might affect the thermal decomposition behavior of PEG and so result in different porosity characteristics. Therefore, this study aimed to investigate the thermal decomposition behavior of TiO₂/PEG sols with different M_W PEG as well as the effect of the heating rate on the morphology of the obtained thin films processed by spin coating.

2. Experimental Procedure

In this research, TiO₂ thin films were deposited on Si wafer by a sol-gel spin coating technique with PEG as the pore-generating agent. To examine the effect of the different amounts and M_W of PEG, TiO₂ sols were categorized into the three groups of TiO₂ sols without any PEG (noPEG) and TiO₂ sols with PEG of M_W 6,000 (6kPEG) and 35,000 (35kPEG) g·mol⁻¹, respectively. For the noPEG samples, TiO₂ sols were prepared by mixing 1 cm⁻³ titanium iso-butoxide (Ti(C₄H₈O)₄), 1.2 cm⁻³ acetylacetonate (AcAc; C₅H₈O₂) and 7.5 cm⁻³ n-propanol (n-PrOH; (CH₃)₂CHOH) with stirring for for 1 h. In the case of the 6kPEG and 35kPEG samples, TiO₂ sols were prepared by mixing 1 cm⁻³ Ti(C₄H₈O)₄, 1.2 cm⁻³ AcAc and 5 cm⁻³ n-PrOH and then adding a solution containing 0.03 g of PEG (either M_W = 6,000 or 35,000 g·mol⁻¹) and 2.5 cm⁻³ of n-PrOH into the TiO₂ sols and stirred for 1 h.

To examine the thermal decomposition behavior, powders were obtained by drying the sols at 120℃ for 24 h. Thermo-gravimetric analysis (TGA) and differential scanning calorimetry (DSC; NETZSCH model STA449F3) were used with different heating rates (1, 3, 5, 10 and 20℃·min⁻¹) under an O₂ atmosphere. Decomposition of the organic residues after heating were detected by Fourier transform infrared spectroscopy (FTIR; BRUKER Optics Vertex70).

The TiO₂ films were fabricated by spin-coating the sols on a (100) Si wafer at 2,000 rpm for 30 s, and then aged at 120℃ for 1 h before being calcined at 450℃ for 1 h at a heating rate of either 1 or 10℃·min⁻¹. The phasic nature and morphology were characterized by X-ray diffraction (XRD; model D/Max 2200P/C) and Field-emission scanning electron microscopy (FESEM; modal JSM-7100F). Grain and pore sizes were measured by hand-counting based on characteristic length method. Pore area fraction (film porosity) was calculated using the Image J software.

3. Results and Discussion

3.1 Determination of the calcination temperature and the effect of heating rate on the PEG thermal decomposition

Initially, the appropriate calcination temperature to decompose the PEG was evaluated by TGA. Figure 1 shows representative thermal degradation profiles of the noPEG and 35kPEG samples along with that for pure PEG (M_W = 35,000 g·mol⁻¹) as a reference. The noPEG sample yielded a two stage weight loss, with the first stage (90–150℃) ascribed to the dehydration of adsorbed water and the second (300–440℃) to the dissociation of Ti-AcAc complexes.^12,13) This range was in agreement with that reported previously (350℃) for the decomposition of the Ti-AcAc complex.¹³⁾ The 35kPEG sample showed three stages of thermal decomposition. The first was water dehydration at 95–120℃ the second at 215–280℃ was associated with the decomposition of PEG, since it was similar to that found in the pure PEG sample. The last stage at 285–410℃ likely originated from the decomposition of the Ti-AcAc complex. A constant residual weight was obtained after 450℃ in both samples, and so subsequent calcinations were performed at 450℃ to eliminate unwanted residues.

Fig. 1

Representative (of 3 samples) TGA profiles of the pure PEG (M_W = 35,000 g·mol⁻¹) and the noPEG and 35kPEG TiO₂ sol films.

To further confirm the complete decomposition of organic and inorganic impurities, FTIR was used to examine the samples calcined at 450℃ for 1 h (Fig. 2). The strong absorption bands at 490 and 657 cm⁻¹ were indexed to be from Ti-O and Ti-O-Ti, respectively.¹⁴⁾ The peaks at 3420 and 1640 cm⁻¹ were attributed to O-H bending and H-O-H stretching, respectively,¹⁵⁾ while CO₂ contamination due to instrumental error resulted in the peak at 2345 cm⁻¹.¹⁵⁾ No observable organic peaks were detected. Therefore, coupled with the TGA results, calcination at 450℃ could eliminate all organics in the samples.

Fig. 2

Representative (of 3 samples) FTIR spectra of the noPEG, 6kPEG and 35kPEG TiO₂ sol films after being calcined at 450℃ for 1 h at a heating rate of 20℃·min⁻¹.

The thermal behavior of the TiO₂-based powders was evaluated at different heating rates (1–20℃·min⁻¹), with representative TGA and differential thermogravimetry (DTG) profiles of the noPEG sample shown in Fig. 3(a). The data displays two steps of weight loss in the range of 70–160℃ and 235–440℃, ascribed to the dehydration and decomposition of Ti-AcAc, respectively.^12,13) The TGA and DTG results of the 6kPEG samples are shown in Fig. 3(b), where a three stage weight loss in the range of 70–150, 230–335 and 335–445℃ was evident. The first and third step was the dehydration and Ti-AcAc decomposition, respectively, while the second stage is likely to be the decomposition of PEG molecules because this step was not found in the noPEG samples, while PEG (M_W = 6,000 g·mol⁻¹) has been reported to decompose at 268–351℃.⁹⁾ In addition, PEG tended to arrange into semi-crystalline molecules (extended and folded chain crystals) with increasing molecular weight.^16,17) Indeed, M_W ≥ 4,000 g·mol⁻¹ PEG can simultaneously form both extended and folded chain crystals, where the volume fraction of folded chain crystals increases with increasing molecular weight. In contrast, PEG of M_W < 4,000 g·mol⁻¹ only formed extended chain crystals.¹⁶⁾ Hence, the weight loss at 230–335℃ in the 6kPEG samples is ascribed to the decomposition of semi-crystalline (extended chain crystal) PEG. Figure 3(c) shows the TGA and DTG results of the 35kPEG samples, which contained of four stages of weight loss at 50–210, 200–225, 251–275 and 325–400℃. Dehydration and the decomposition of Ti-AcAc accounted for the first (50–210℃) and fourth (325–400℃) stage, respectively, while the weight loss at 200–225℃ and 251–275℃ were probably related to the decomposition of the folded and extended PEG chain crystals, respectively, given that folded chain crystals are less thermodynamically stable than extended chain crystals.¹⁶⁾

Fig. 3

Representative (of 3 samples) TGA(Left) and DTG (Right) profiles of the (a) noPEG, (b) 6kPEG and (c) 35kPEG TiO₂ sol films during heating to 450℃ at different heating rates.

The maximum decomposition rates characterized by the DTG peaks (Gaussian fitting) were different among the different heating rates and the decomposition steps are shown in Table 1. The temperature of the maximum decomposition rate increased in all samples with increasing heating rates, in agreement with Kissinger's equation,¹⁸⁾ where the energy and temperature of combustion of a polymer increase with the heating rate. The maximum decomposition step in each sample was different and depended on the PEG content (Table 1).

Table 1 Decomposition steps and temperature of the maximum decomposition rate of the TiO₂ sol films calcined at different heating rates.

Sample	Source of decomposition step	Temperature of maximum decomposition rate (℃)
		1℃·min−1	3℃·min−1	5℃·min−1	10℃·min−1	20℃·min−1
noPEG	Ti-AcAc molecules	297.6	317.1	326.7	340.4	353.6
6kPEG	Extended-chain crystal	305.3	327.6	332.1	348.6	354.7
35kPEG	Folded-chain crystal	192.3	205.6	211.8	218.4	225.3

To examine the thermal behavior in terms of input and output of energy during calcination, DSC was employed and representative DSC profiles of the noPEG, 6kPEG and 35kPEG samples are shown in Fig. 4. For the noPEG samples (Fig. 4(a)), the small endothermic peak at 70–200℃ and a broad exothermic peak at 235–400℃ were ascribed to the dehydration of H₂O molecules and the decomposition (combustion) of Ti-AcAc molecules, respectively. The small exothermic peak (approximately 410℃) without any significant weight change was due to the phase transformation of TiO₂ from amorphous to anatase. This peak shifted to a higher temperature with an increasing heating rate up to 10℃·min⁻¹, which is in accord with the reported transformation occurring in the 400–550℃ range.¹⁹⁾ The DSC results of the 6kPEG samples (Fig. 4(b)) revealed broad exothermic peaks in the 235–415℃ range, which could be deconvoluted into two peaks. The first and second peaks could be induced by the decomposition of semi-crystalline PEG molecules (extended chain crystals) and Ti-AcAc molecules, respectively, as previously mentioned in the TGA and DTA results. In case of the 35kPEG samples (Fig. 4(c)), the complicated exothermic peaks were found in the 172–385℃ and involved the combustion of folded chains crystals, extended chain crystals and Ti-AcAc molecules. The highest sharp exothermic peak could be related to the combustion of extended chain crystals even though the decomposition rate of folded chain PEG crystals was higher. Therefore, the combustion of the extended chain PEG crystals could release more energy than the folded chain crystals. This was in agreement with the report that the heat of formation of extended chain PEG crystals was higher than that of folded chain crystals.¹⁶⁾ In addition, the exothermic heat of combustion tended to be higher with higher heating rates as well as the decomposition rate.

Fig. 4

Representative (of 3 samples) DSC profiles of the (a) noPEG, (b) 6kPEG and (c) 35kPEG TiO₂ sol films calcined at 450℃ at different heating rates.

3.2 Effect of the heating rate on phase transformation

Phase transformation of TiO₂ in the films calcined at 450℃ with different heating rates was examined by XRD, with representative examples shown in Fig. 5. All films displayed the dominant TiO₂ anatase phase (JCPDS 78–2486) independent of the PEG M_W. This phase formation was supported by the presence of the Ti-O and T-O-Ti vibrations in the FTIR results, while the DSC results showed a small exothermic peak at 410℃ due to the phase transformation. In agreement, it was previously reported that the TiO₂ anatase was present when TiO₂ sols were calcined at 450℃ for 3 h.²⁰⁾ The crystallite sizes of TiO₂ were calculated by Scherrer's equation and are shown in Fig. 6, where they were found to decrease with an increasing heating rate. This was caused by the shorter duration of grain growth during the calcination cycle. However, the rate of the size reduction in the 6kPEG and 35kPEG films was lower than that in the noPEG films as a function of the heating rate. This could be related to the acquisition of auxiliary heat from the combustion of PEG, which would be higher with higher heating rates as previously discussed in the DSC results in this section.

Fig. 5

Representative (of 12 samples) XRD patterns of the noPEG, 6kPEG and 35kPEG TiO₂ sol films calcined 450℃ for 1 h at heating rates of (a) 1, (b) 5, (c) 10 and (d) 20℃·min⁻¹.

Fig. 6

TiO₂ crystallite size of the TiO₂ sol films calcined at 450℃ for 1 h at different heating rates. Data are shown as the mean ± 1 SD, derived from 12 independent repeats.

3.3 Effect of the calcination heating rate on the TiO₂ film morphology

Morphological variation in the TiO₂ films caused by two different heating rates (1 and 10℃·min⁻¹) in the calcination stage were studied by FESEM (Fig. 7). At a heating rate of 1℃·min⁻¹, there was no noticeable difference in the morphology among the films with and without PEG. This might be due to the low exothermic heat of PEG combustion and that the grains have enough time for coarsening. At a heating rate of 10℃·min⁻¹ the morphology of the noPEG film was quite similar to those calcined at 1℃·min⁻¹, whereas the 6kPEG and 35kPEG films yielded some localized coalescence of grains with a higher degree in the 35kPEG film. This was also suggested to be the result of the additional exothermic heat of PEG combustion leading to localized coalescence of grains in some areas, where this exothermic heat is higher at higher heating rates.

Fig. 7

Representative FESEM images (200,000 × magnification) of the (a), (d) noPEG, (b), (e) 6kPEG and (c), (f) 35kPEG TiO₂ thin films calcined at 450℃ for 1 h at (a)–(c) 1 and (a)–(c) 10℃·min⁻¹. Images shown are representative of those seen from at least 5 such fields of view per sample and 6 independent samples.

Grain size distribution of the films calcined at 450℃ for 1 h at a heating rate of 1 and 10℃·min⁻¹ are shown in Fig. 8. At a heating rate of 1℃·min⁻¹, their grain size distribution profiles were unimodal with a modality of approximately 15–17 nm, with no significant difference between the samples. However, the grain size distribution profiles of the 6kPEG and 35kPEG films changed to bimodal when the heating rate was increased to 10℃·min⁻¹. This supported the previous hypothesis that the exothermic heat of PEG combustion transferred to the adjacent area and led to coalescence of the grains supplying them with auxiliary exothermic heat. Accordingly, the first and second modality represented the normal grains and those grains affected from exothermic heat of combustion, respectively.

Fig. 8

Grain size distributions of the (a), (d) noPEG, (b), (e) 6kPEG and (c), (f) 35kPEG TiO₂ films calcined at 450℃ for 1 h at a heating rate of (a)–(c) 1 or (d)–(e) 10℃·min⁻¹. Data are shown as the mean ± 1 SD, derived from 6 independent repeats.

The film's porosity and pore size were calculated using the Image J software and hand-counting based on the characteristic length method. The average pore size of all the films calcined at a heating rate of 1℃·min⁻¹ was in a range of 9.2–9.4 nm (with SDs of 2.8, 1.2 and 2.8 for the noPEG, 6kPEG and 35kPEG films, respectively) and was independent of the added PEG content. Because of the sufficient calcination time and relatively low temperature, coarsening mechanism was the dominating factor that caused the pores to grow by Oswald ripening.^21,22) Hence, the added PEG had no significant effect on the film pore size. At a calcination heating rate of 10℃·min⁻¹, the average pore size of the films were slightly decreased due to the shorter (insufficient) period of time for coarsening with the average pore sizes of the noPEG, 6kPEG and 35kPEG TiO₂ sol films being 7.3 ± 2.6, 9.0 ± 3.5 and 8.2 ± 2.9 nm, respectively. The exothermic heat from the PEG combustion caused pore agglomeration, and so the average pore size in the 6kPEG and 35kPEG films was larger than that in the noPEG film. The DSC results showed the combustion of the extended chain PEG crystals released a higher amount of heat than the folded chain crystals. The 6kPEG films had only extended chain PEG crystals, whereas the 35kPEG films contained both extended and folded chain PEG crystals and so the exothermic heat from PEG combustion in the 6kPEG films was higher than that from the 35kPEG films, leading to a higher degree of pore agglomeration and higher average pore size in the 6kPEG films.

The porosity of TiO₂ film was evaluated in terms of its pore area fraction. Those calcined at a heating rate of 1℃·min⁻¹ (7.6%) had a higher porosity than the noPEG film (6.8%), whilst it was highest in the 35kPEG films (8.5%). Generally, the porosity of films increased with a higher M_W of PEG,⁹⁾ because the coarsening is a non-densifying mechanism and so the decomposition of PEG increased the film's porosity.²³⁾ When the calcination heating rate was increased to 10℃·min⁻¹, the porosity of the 6kPEG and 35kPEG films decreased (compared to those calcined at 1℃·min⁻¹) to 6 and 6.6%, respectively. This is presumably due to the exothermic heat of the PEG combustion at 10℃·min⁻¹ leading to localized coalescence grains, as already discussed. These coalescence grains then cause the elimination of some of the pores surrounding that area. Thus, as the 6kPEG and 35kPEG films undergo coarsening and localized grain coalescence, the porosity of these films tended to decrease with an increasing heating rate. In contrast, the noPEG sample experienced only coarsening, a non-densifying mechanism, and so the porosity of the noPEG films calcined at a heating rate of 10℃·min⁻¹ and 1℃·min⁻¹ were similar.

4. Conclusions

The morphology of the spin coated TiO₂ thin films was generally dependent on both the M_W of the added PEG and the heating rate of the calcination. Certain morphological characteristics were closely related to the heat generated by the PEG burnout. The faster the heating rate, the higher the temperature of maximum decomposition rate. This was explained by the heat flash generated at a high heating rate, as characterized by the exothermic peaks in the DSC analysis. At a calcination temperature of 450℃, the TiO₂ remained in the anatase phase regardless of the PEG M_W or the heating rate. A slight reduction in the crystallite size occurred with increasing heating rates, presumably due to the shorter time available for the crystal growth from the heat flash, a process which is predominantly controlled by kinetics. The amount of exothermic heat released from PEG combustion in the 6kPEG films was higher than in the 35kPEG, which led to a higher degree of pore agglomeration and higher average pore size. The configuration of PEG crystals is also suggested to be a factor responsible for the variation in the porosity of the films.

Acknowledgements

The authors were supported by the Department of Metallurgical Engineering, Faculty of Engineering, Chulalongkorn University and the Department of Materials Science and Engineering, Faculty of Engineering and Industrial Technology, Silpakorn University.

REFERENCES

• 1)  M. Vishwas, S.K. Sharma, K.N. Rao, S. Mohan, K.V.A. Gowda and R.P.S. Chakradhar: Spectrochimica Acta Part A. 74 (2009) 839–842. 10.1016/j.saa.2009.07.018
• 2)  S. Miura, S. Noda, K. Suzuki, M. Inoue, K. Murakami, F. Ohashi and S. Nonomura: Mater. Trans. 56 (2015) 274–276. 10.2320/matertrans.M2014362
• 3)  R. Prado, G. Beobide, A. Marcaide, J. Goikoetxea and A. Aranzabe: Solar Energy Mater. Solar Cells 94 (2010) 1081–1088. 10.1016/j.solmat.2010.02.031
• 4)  M. Vishwas, S.K. Sharma, K.N. Rao, S. Mohan, K.V.A. Gowda and R.P.S. Chakradhar: Spectrochimica Acta Part A. 75 (2010) 1073–1077. 10.1016/j.saa.2009.12.057
• 5)  N. Batra, P. Kumar, S.K. Srivastava, R. Kumar, R. Srivastava, M. Deepa, B.R. Awasthy and P.K. Singh: Mater. Chem. Phys. 130 (2011) 1061–1065. 10.1016/j.matchemphys.2011.08.035
• 6)  L. Miao, P. Jin, K. Kaneko, A. Terai, N. Nabatova-Gabain and S. Tanemura: Appl. Surf. Sci. 212–213 (2003) 255–263. 10.1016/S0169-4332(03)00106-5
• 7)  B.E. Yoldas and D.P. Partlow: Thin Solid Films 129 (1985) 1–14. 10.1016/0040-6090(85)90089-6
• 8)  N. Arconada, Y. Castro and A. Duran: Applied Catalysis A. 385 (2010) 101–107. 10.1016/j.apcata.2010.06.051
• 9)  B. Guo, Z. Liu, L. Hong and H. Jiang: Surf. Coat. Tech. 198 (2005) 24–29. 10.1016/j.surfcoat.2004.10.055
• 10)  B. Guo, Z. Liu, L. Hong, H. Jiang and J.W. Lee: Thin Solid Films 479 (2005) 310–315. 10.1016/j.tsf.2004.11.197
• 11)  S. Bu, Z. Jin, X. Liu, L. Yang and Z. Cheng: Mater. Chem. Phys. 88 (2004) 273–279. 10.1016/j.matchemphys.2004.03.033
• 12)  J. Yu, X. Zhao and Q. Zhao: Thin Solid Films 379 (2000) 7–14. 10.1016/S0040-6090(00)01542-X
• 13)  I.N. Kuznetsova, V. Blaskov and L. Znaidi: Mater. Sci. Eng. B 137 (2007) 31–39. 10.1016/j.mseb.2006.10.001
• 14)  A. Verma, S.B. Samanta, A.K. Bakhshi and S.A. Agnihotry: Solar Energy Mater. Solar Cells 88 (2005) 47–64. 10.1016/j.solmat.2004.10.006
• 15)  M.S. Ghamsari and A.R. Bahramian: Mater. Lett. 62 (2008) 361–364. 10.1016/j.matlet.2007.05.053
• 16)  J.M. Ginés, M.J. Arias, A.M. Rabasco, C. Novak, A. Ruiz-Conde and P.J. Sanchez-Soto: J. Therm. Anal. 46 (1996) 291–304. 10.1007/BF01979969
• 17)  K. Pielichowski and K. Flejtuch: Macromol. Mater. Eng. 288 (2003) 259–264. 10.1002/mame.200390022
• 18)  R.L. Blaine and H.E. Kissinger: Thermochim. Acta 540 (2012) 1–6. 10.1016/j.tca.2012.04.008
• 19)  S. Šegota, L. Curkovic, D. Ljubas, V. Svetlicic, I.F. Houra and N. Tomasic: Ceram. Int. 37 (2011) 1153–1160. 10.1016/j.ceramint.2010.10.034
• 20)  N. Arconada, A. Duran, S. Suarez, R. Portela, J.M. Coronado, B. Sanchez and Y. Castro: Appl. Catal. B 86 (2009) 1–7. 10.1016/j.apcatb.2008.07.021
• 21)  R. German: Sintering: From Empirial Observations to Scientific Principles: (Elsevier Inc., Marquis one, Atlanta, 2014) pp. 227–246.
• 22)  F. Wakai, M. Yoshida, Y. Shinoda and T. Akatsu: Acta Mater. 53 (2005) 1361–1371. 10.1016/j.actamat.2004.11.029
• 23)  L. Dejonghe and M. Rahaman: Handbook of Advanced Ceramics. (Elsevier Inc., Marquis one, Atlanta, 2003) pp. 187–257