Microstructure and optical properties of nanocrystalline ZnO and ZnO:(Li or Al) thin films

Applied Surface Science 253 (2007) 4593–4598 www.elsevier.com/locate/apsusc

Microstructure and optical properties of nanocrystalline ZnO and ZnO:(Li or Al) thin films A. Yavuz Oral a,*, Z. Banu Bahs¸i a, M. Hasan Aslan b a

Department of Materials Science and Engineering, Gebze Institute of Technology, Gebze 41400, Turkey b Department of Physics, Gebze Institute of Technology, Gebze 41400, Turkey Received 29 July 2006; received in revised form 7 September 2006; accepted 8 October 2006 Available online 13 November 2006

Abstract Zinc oxide thin films (ZnO, ZnO:Li, ZnO:Al) were deposited on glass substrates by a sol–gel technique. Zinc acetate, lithium acetate, and aluminum chloride were used as metal ion sources in the precursor solutions. XRD analysis revealed that Li doped and undoped ZnO films formed single phase zincite structure in contrast to Al:ZnO films which did not fully crystallize at the annealing temperature of 550 8C. Crystallized films had a grain size under 50 nm and showed c-axis grain orientation. All films had a very smooth surface with RMS surface roughness values between 0.23 and 0.35 nm. Surface roughness and optical band tail values increased by Al doping. Compared to undoped ZnO films, Li doping slightly increased the optical band gap of the films. # 2006 Elsevier B.V. All rights reserved. Keywords: Doped zinc oxide; Sol–gel; Thin film; Transparent-conductive oxide

1. Introduction ZnO is a wide band gap semiconductor. It crystallizes in hexagonal wurtzite structure (zincite) [1]. In this structure, Zn atoms are tetrahedrally coordinated to four O atoms, where the Zn d-electrons hybridize with the O p-electrons [2]. The n-type semiconductor behavior is originated by the ionization of excess zinc atoms at interstitial positions and by the ionization of oxygen vacancies forming defect levels approximately 0.01– 0.05 eV below the conduction band [2]. ZnO is used in many applications such as surface acoustic wave devices (SAW), laser devices, gas sensors and MEMS [3]. In addition, there is interest in integrating ZnO with other wide band semiconductors such as AlInGaN due to the lattice match between them [2]. ZnO thin films have been prepared by various techniques such as rf sputtering [4,5], spray pyrolysis [6,7], chemical vapor deposition (CVD) [8–10], pulsed laser deposition [11–13], molecular beam epitaxy [14] and sol–gel processing [3,15–17]. When sol–gel is used, there are two principal routes used to

* Corresponding author. E-mail address: [email protected] (A.Y. Oral). 0169-4332/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2006.10.015

obtain oxide thin films: the alkoxide route using organometallic precursors and the none-alkoxide route using water or alcohol solutions of metal salts [18]. Doped zinc oxide (ZnO) thin films have attracted much attention because of their potential for being used as transparent conducting electrodes after doping with group IIIB elements or fluorine, Furthermore, they can be used as insulating or ferroelectric layers after doping with Li or Mg in optoelectronic devices [19]. Transparent ZnO thin films doped with Al, In, or Ga show good electrical conductivity [20]. Currently, conductive zinc oxides replace indium-tin-oxide (ITO) thin films in the area of transparent conducting electrodes due to their inertness under hydrogen plasma atmosphere [21]. It is generally accepted that doping of ZnO with Al decreases its resistivity contrasted with Li, which is known to increase resistivity in ZnO [22]. Al acts as a donor when it is substitutionally incorporated on zinc lattice sites [23]. Accordingly, Musat et al. [24] fabricated low resistivity Al doped ZnO films when segregation of aluminum (as Al2O3) at the grain boundaries was avoided. Then, substitution of the Al atoms effectively took place in Zn sites of the ZnO structure according to the following equation: Al2 O3 ! 2AlZn þ 2O0 þ 12O2 þ 2e0

4594

A.Y. Oral et al. / Applied Surface Science 253 (2007) 4593–4598

where AlZn is Al in zinc lattice site and O0 is oxygen on lattice site. Generally, Li introduces a deep acceptor level and induces ferroelectric behavior [2]. According to Fujihara et al. [25], chemical defect formation due to Li-doping mainly occurs as follows: Li2 O ! Li0Zn þ Lii þ þ O0 where LiZn represents lithium on zinc lattice site and Lii+ lithium in interstitial position. In fact, some Li atoms in interstitial sites may later replace Zn atoms as follows: Lii þ þ ZnZn þ e ! Li0Zn þ Zni þ or Lii þ þ e ! Li0Zn þ V O þ +

where ZnZn represents zinc in zinc site and VO oxygen vacancy on lattice site. The aim of this work is to evaluate the effects of doping (Li or Al) on the microstructure, optical properties, and grain orientation of ZnO thin films prepared by sol–gel spin coating. 2. Experimental The basic solution was prepared by dissolving ZnAc (Zn(CH3COO)22H2O) in 2-propanol and monoethanolamine (C2H7NO, EA) (EA:ZnAc = 1:1). It is generally accepted that

ZnAc transforms to mono acetate in 2-propanol by the reaction above. Then the following reaction was estimated to occur between monoethanolamine and zinc mono acetate (below).

The resulting solution was stirred by using a magnetic stirrer at 50 8C until it became clear. Afterwards, water (H2O:ZnAc = 1:2) was slowly added to obtain optimum wettability between the precursor film and the substrate. Finally, a solution with a concentration of 0.4 M was obtained (Fig. 1). Similarly, Li doping solution was prepared from lithium acetate ((CH3CO2Li), LiAc) with a concentration of 0.2 M

Fig. 1. The flow diagram of the ZnAc precursor solution preparation.

Fig. 2. The flow diagram of: (a) AlCl3 solution preparation; (b) LiAc solution preparation.

(Fig. 2a). Aluminum chloride solution (0.1 M) was prepared by dissolving aluminum chloride (AlCl3) in acetic acid (Fig. 2b). Proportional amounts of doping solutions were added to ZnAc solutions in such a way that final ratio of Zn:(Li or Al) was 99:1, respectively, in each solution. The microscope glass substrates were cleaned with HCl and distilled water. Deposition was carried out in air at three steps with different spinning speeds, which are 2000 rpm for 30 s, 4000 rpm for 30 s, and 6000 rpm for 60 s. A precursor film formed following the spin coating process. The film was then dried at 250 8C for 1 min on a hot plate. After the deposition of the last layer, the resulting films were annealed in air at 550 8C for 1 h (Fig. 3). The crystal structure and grain orientation of ZnO films were determined by X-ray diffraction (XRD-Rigaku Dmax 2200) with Cu Ka radiation. The surface morphology of the films was characterized by field emission scanning electron microscope (FESEM, Philips 30XL SFEG) and atomic force microscopy (Digital Instrument Nanoscope IV). Optical transmission and absorbance spectra of the films were analyzed by using a UV–visible spectrophotometer (Shimadzu-2101PC).

Fig. 3. The flow diagram of ZnO:Me film fabrication.

A.Y. Oral et al. / Applied Surface Science 253 (2007) 4593–4598

3. Results and discussion 3.1. Effect of doping to the grain orientation

Table 1 Lattice parameters and relative intensities of (1 0 0), (0 0 2) and (1 0 1) peaks of doped ZnO thin films Doping type

XRD analysis revealed that both undoped and Li doped ZnO films consist of single phase ZnO with zincite structure. On the other hand, Al:ZnO films did not fully crystallized at 550 8C (Fig. 4). There are intensity increases in the XRD pattern of Al:ZnO films around 2Q values of 34 and 36 corresponds to peak locations of (0 0 2) and (1 0 1). This suggests that crystallization of Wurtzite structure has started but it is in the beginning stage. In fact, prolonged crystallization rate was expected since Al doping is known to prevent grain growth in ZnO [26]. This hindrance of grain growth is believed to be caused by stress formation as a result of the ion size difference between Zn and Al when Al ions replace Zn ions. The highest intensities of the XRD peaks were obtained from Li doped films indicating a better crystal quality. Even though, there is no clear explanation for this phenomenon, it is possible that heterogeneous nucleation is facilitated in the presence of Li+ ions in the ZnO structure. The lattice parameters of the undoped film are very close to the data obtained from zincite structure (JCPDS 36-1451). Li doping slightly increased both a and c parameters (Table 1). Enlargement in the c-axis lattice parameter by Li doping has been previously reported by Chen et al. [27]. If Li+ ions (0.060 nm) with a smaller radius replaced Zn2+ ions (0.074 nm), a decrease in the lattice parameters would be expected. Therefore, the increase in the lattice parameters must be caused by either interstitial incorporation of Li+ ions into the lattice or the formation of electrically inactive Lizn–Lii pairs described by Wardle et al. [28]. Compared to powder diffraction data of zincite structure (JCPDS 36-1451), the XRD patterns of crystallized films indicated enhanced peak intensities corresponding to (0 0 2) plane indicating a preferential orientation along the c-axis. Doping with Li also increased the extent of c-orientation compared to undoped film (Table 1). There are many different ideas about the basis of the c-orientation in ZnO thin films. Bao

4595

Li Undoped JCPDS36-1451

a

3.274 3.245 3.244

c

5.229 5.194 5.205

I(1 0 0)

I(0 0 2)

I(1 0 0) + I(0 0 2) + I(1 0 1)

I(1 0 0) + I(0 0 2) + I(1 0 1)

0.022 0.071 0.283

0.907 0.837 0.219

et al. [29] believed that the preferential orientation is caused by the minimization of internal stress and surface energy. Amirhaghi et al. [30] reported that c-orientation might be resulted from facilitated growth along c-axis due to highest atomic density found along (0 0 2) plane. However, preferential growth direction along [0 0 1] of the nuclei does not solely explain why [0 0 1] direction of the nuclei coincides with the caxis of the film to result in c-orientation. We can propose a distinct description for the origin of c-axis orientation in ZnO thin films. If we assume that heterogeneous nucleation (amorphous to crystalline) occurs at the substrate film interface, then the will nuclei grow in their [0 0 1] direction (c-axis of the crystal) due to their facilitated growth as suggested by Amiraghi et al. [30]. If [0 0 1] directions of growing crystals merge, their further growth will be slowed down or hindered. On the other hand, if the [0 0 1] direction of a growing crystal is parallel to surface normal, then it will continue to grow until it reaches the top surface of the film (Fig. 5). In this description, the grains in the vicinity of the substrate should be poorly oriented while the grains at the top surface should be strongly oriented. In a word, orientation in thicker films must to be stronger. In fact, Ohyama et al. [31] observed that orientation becomes stronger as the thickness of the ZnO films increases, which was an observation that is in accordance with the description in this article. 3.2. The morphology of the ZnO films Fig. 6 shows the SEM micrographs of the surface of 10-layer doped and undoped ZnO films annealed at 550 8C. The average grain size of both undoped and Li doped the films were below 50 nm. The low magnification micrographs showed that the film density increased by Al doping. On the other hand, doping with Li ions caused microcracks in the film structure. Atomic force microscope (AFM) was used to measure the surface roughness

Fig. 4. XRD patterns of undoped, Al-doped, and Li-doped films.

Fig. 5. A sketch drawing (not drawn to scale) describing basis of orientation in ZnO thin films.

4596

A.Y. Oral et al. / Applied Surface Science 253 (2007) 4593–4598

Fig. 6. SEM micrographs of: (a) undoped ZnO film at high magnification; (b) undoped ZnO film at low magnification; (c) Al-doped ZnO film at high magnification; (d) Al-doped ZnO film at low magnification; (e) Li-doped ZnO film at high magnification; (f) Li-doped ZnO film at low magnification.

of the films over a 5 mm  5 mm area by contact mode. Compared to undoped films, doping with Al increased surface roughness in contrast to Li doping which did not significantly affect the surface roughness (Fig. 7). Undoped and Li doped films had a very smooth surface with a root mean square (RMS) surface roughness of 0.23 and 0.24 nm, respectively. This value increased to 0.35 nm for Al doped films (Table 2). 3D AFM micrographs indicated that the increase in the surface roughness by Al doping was caused by sharp hill-and-valley structure throughout the film surface (Fig. 7b). 3.3. Optical properties of the ZnO films The optical band gap of the ZnO thin films was estimated by extrapolation of the linear portion of a2 versus hn plots using the relation ahn = A(hn  Eg)n/2, where a is the absorption

coefficient, hn the photon energy and Eg is the optical band gap. For different n values, a good linearity was observed at n = 1 (direct allowed transition) which was found to give the best fit for these films. The measured optical band gap values were very close to each other and slightly larger to band gap of intrinsic ZnO (3.2 eV) [32]. Assuming doping levels are well below Motts critical density, the change in the band gap values can be explained by Burstein–Moss effect. At high doping concentrations, fermi level lifts into the conduction band. Due to filling of the conduction band, absorption transitions occurs between valance band and fermi level in the conduction band instead of valance band and bottom of the conduction band. This change in the absorption energy levels shifts the absorption edge to higher energies (blue shift) and leads to the energy band broadening (DEg), which can be calculated by the following equation [33]: DEg ¼

Table 2 Surface roughness values of undoped, Al-doped, and Li-doped ZnO films Doping type

Surface roughness RMS (nm)

Al Li Undoped

0.35 0.24 0.23

h2 8m

 2=3 3 h2=3 e p

where h is the Planck’s constant, m* the electron effective mass in conduction band and he is the carrier concentration. Al doping did not cause any shift in the band gap value indicating that either it did not form any charged defects or the charged defects formed had been neutralized by other defects. Li doping

A.Y. Oral et al. / Applied Surface Science 253 (2007) 4593–4598

4597

Fig. 7. AFM micrographs of: (a) undoped; (b) Al-doped; (c) Li-doped ZnO films.

slightly increased the band gap (Table 3). Substitutional incorporation of Li atoms first neutralizes the n-type native defects in ZnO. Therefore, this type of incorparation does not increase the carrier concentration. Since the blue shift in the band gap value by Li doping suggest an increase in the n-type carrier concentration, most of the Li ions must be incorporated as interstitial donors into the structure rather than substitutional acceptors. Band tailing can be explained as local perturbation of the band edges by the effect of impurities or any other defect. Density-of-states distribution integrates the number of states at each energy level inside the whole volume and shows that there Table 3 The optical band gaps and corresponding band tails of differently doped ZnO thin films Doping type

Calculated optical band gap (eV)

Calculated optical band tail (eV)

Al Li Undoped

3.31 3.27 3.28

0.269 0.095 0.062

are conduction band states at relatively low potentials and valance band states in high-potential regions [32]. The band tails cause absorption below energy gap and change the adsorption edge from a steeply rising one to exponentially increasing one. Width of the band tail can be calculated by Pankove’s expression: aðhnÞ ¼

3=2 AE0



hn exp E0

 for hn < Eg

where E0 is the parameter describing the width of the localized state in the band gap, a the absorption coefficient, A is a constant, hn the energy of light and Eg is the band gap [34]. Since absorption coefficient is the cumulative effect of all defects (i.e. point, planer, etc.), it is difficult to pinpoint the nature of variation in E0. However, based on Burstein–Moss shift, we can conclude that the difference of free carrier concentration in films is quite low and the E0 change due to free carrier concentration difference is limited. It was observed that the widths of the band tail did not brutally change when the

4598

A.Y. Oral et al. / Applied Surface Science 253 (2007) 4593–4598

compared to undoped ZnO films. Optical band tail values of the films increased by doping.

References

Fig. 8. Transmittance spectra of undoped, Al-doped, and Li-doped ZnO thin films.

films were doped with Li (Table 3). Therefore, the change in the width of band tail values was attributed to the change in grain boundary areas in differently doped films. It was observed that non-crystalline Al doped films had the largest band tail (Table 3). All films showed high transparency in the visible range. However, doping significantly affected the UV transparency (Fig. 8). In the near IR range the energy of the light is well below band gap of the ZnO films and the wavelength of light is much above the grain size, pore size or mean surface roughness value of the films. Therefore, IR radiation does not interact with the films and the loss of transparency is mainly by the reflection of the light. However, the UV transparency is much lower mainly due to electron transitions between the valance and conduction band in all films. In addition, the wavelength of the light in the UV range is shorter and can interact with the smaller defects that are present in all fabricated films. Thus, there was a substantial difference in UV transparencies of differently doped films. ZnO:Al films, which did not contain grain boundaries, cracks or pores had a very high UV transparency. In contrast, Li doped films which had cracks on the surface had the lowest UV transparency. 4. Conclusions We have prepared both doped (Al or Li) and undoped thin films by a sol–gel technique. Both Li doped and undoped ZnO films crystallized in zincite structure. On the other hand, Al:ZnO films did not fully crystallize at the annealing temperature of 550 8C. All films had a very smooth surface with RMS surface roughness values between 0.23 and 0.35 nm. The highest surface roughness value was obtained from Al doped films as a result of hill-and-valley structure of the surface. Li doping slightly increased the optical band gap

[1] A.A. EL-Fadl, G.A. Mohamad, A.B. Abd El-Moiz, M. Rashad, Phys. B 366 (2005) 44. [2] S.J. Pearton, D.P. Norton, K. Ip, Y.W. Heo, T. Steiner, Prog. Mater. Sci. 50 (2005) 293. [3] M. Wang, J. Wang, W. Chen, Y. Cui, L. Wang, Mater. Chem. Phys. 97 (2006) 219. [4] Y. Yoshio, T. Makino, Y. Katayama, T. Hata, Vacuum 59 (2000) 538. [5] K.B. Sundaram, A. Khan, Thin Solid Films 295 (1997) 87. [6] A. Bougrine, M. Addou, A. Kachouane, J.C. Be´rne`de, M. Morsli, Mater. Chem. Phys. 91 (2005) 247. [7] P. Nunes, E. Fortunato, R. Martins, Int. J. Inorg. Mater. 3 (2001) 1125. [8] I. Sakaguchi, H. Ryoken, S. Hishita, H. Haneda, Thin Solid Films 506–507 (2006) 184. [9] P. Nunes, E. Fortunato, R. Martins, Int. J. Inorg. Mater. 3 (2001) 1125. [10] Y. Kashiwaba, F. Katahira, K. Haga, T. Sekiguchi, H. Watanabe, J. Cryst. Growth 221 (2000) 431. [11] Y.W. Sun, J. Gospodyn, P. Kursa, J. Sit, R.G. DeCorby, Y.Y. Tsui, Appl. Surf. Sci. 248 (2005) 392. [12] J. Zou, S. Zhou, C. Xia, Y. Hang, J. Xu, S. Gu, R. Zhang, J. Cryst. Growth 280 (2005) 185. [13] Y.R. Ryu, W.J. Kim, H.W. White, J. Cryst. Growth 219 (2000) 419. [14] C.J. Pan, C.W. Tu, J.J. Song, G. Cantwell, C.C. Lee, B.J. Pong, G.C. Chi, J. Cryst. Growth 282 (2005) 112. [15] A.E.J. Gonzalez, J.A.S. Urueta, Solar Energy Mater. Solar Cells 52 (1998) 345. [16] M. Ohayama, H. Kozuka, T. Yoko, Thin Solid Films 306 (1997) 78. [17] J. Lee, K. Ko, B. Park, J. Cryst. Growth 247 (2003) 119. [18] V. Musat, B. Teixeira, E. Fortunato, R.C.C. Monteiro, P. Vilarinho, Surf. Coat. Technol. 180–181 (2004) 659. [19] S. Fujiara, C. Sasaki, T. Kimura, Appl. Surf. Sci. 180 (2001) 341. [20] B.G. Choi, I.H. Kim, D.H. Kim, K.S. Lee, T.S. Lee, B. Cheong, Y.-J. Baik, W.M. Kim, J. Eur. Ceram. Soc. 25 (2005) 2161. [21] Z.Q. Xu, H. Deng, Y. Li, Q.H. Guo, Y.R. Li, Mater. Res. Bull. 41 (2006) 354. [22] G.A. Mohamed, E. Mohamed, A.A. El-Fadl, Phys. B 308–310 (2001) 949. [23] I. Sieber, N. Wanderka, I. Urban, I. Dorfel, E. Schierhorn, F. Fenske, W. Fuhs, Thin Solid Films 330 (1998) 108. [24] V. Musat, B. Teixeira, E. Fortunato, R.C.C. Monteiro, P. Vilarinho, Surf. Coat. Technol. 180–181 (2004) 659. [25] S. Fujihara, C. Sasaki, T. Kimura, J. Eur. Ceram. Soc. 21 (2001) 2109. [26] Y.M. Chung, C.S. Moon, M.J. Jung, J.G. Han, Surf. Coat. Technol. 200 (20005) 936. [27] L.L. Chen, H.P. He, Z.Z. Ye, Y.J. Zeng, J.G. Lu, B.H. Zhao, L.P. Zhu, Chem. Phys. Lett. 420 (2006) 358. [28] M.G. Wardle, J.P. Goss, P.R. Briddon, Phys. Rev. B 71 (2005) 155–205. [29] D. Bao, H. Gu, A. Kuang, Thin Solid Films 312 (1998) 37. [30] S. Amirhaghi, V. Craciun, D. Craciun, J. Elder, I.W. Boyd, Microelectron. Eng. 25 (1994) 321. [31] M. Ohyama, H. Kozuka, T. Yoko, Thin Solid Films 306 (1997) 78. [32] J.I. Pankove, Optical Progress in Semiconductors, Dover Publications, New York, 1975. [33] H.W. Lee, S.P. Lau, Y.G. Wang, K.Y. Tse, H.H. Hng, B.K. Tay, J. Cryst. Growth 268 (2004) 596. [34] S. Bandyopadhyay, G.K. Paul, S.K. Sen, Solar Energy Mater. Solar Cells 71 (2002) 103.