Effect of rf plasma nitriding time on electrical and optical properties of ZnO thin films

Effect of rf plasma nitriding time on electrical and optical properties of ZnO thin films

ARTICLE IN PRESS Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357 www.elsevier.com/locate/jpcs Effect of rf plasma nitriding time on e...

213KB Sizes 2 Downloads 24 Views

ARTICLE IN PRESS

Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357 www.elsevier.com/locate/jpcs

Effect of rf plasma nitriding time on electrical and optical properties of ZnO thin films S.H. Mohameda,, A.M. Abd El-Rahmana, A.M. Salemb, L. Pichonc, F.M. El-Hossarya a

Physics Department, Faculty of Science, South Valley University, 82524 Sohag, Egypt Physics Division, Electron Microscopy & Thin Film Department National Research Center, Dokki, Cairo, Egypt c Laboratoire de Metallurgie Physique, Universite de Poitiers, 86960 Futuroscope, France

b

Received 23 January 2006; received in revised form 8 May 2006; accepted 27 May 2006

Abstract ZnO thin films were prepared by thermal oxidation of metallic Zn films and nitrided by an inductively coupled rf plasma. The effects of successive plasma processing cyclic times on structural and optical properties as well as electrical resistivity were examined by different characterization techniques. A small amount of nitrogen was detected at the film–substrate interface. The grain size decreased slightly as the treatment time increased. The surface roughness of examined films increased while the thickness decreased with increasing plasma treatment time. The electrical resistivity decreased about four orders of magnitude when the sample nitrided for 15 min. However, the transmittance increased as the plasma treatment time increased. The optical band gap increased from 2.76 to 3.02 eV with increasing plasma treatment time from 0 to 15 min. r 2006 Elsevier Ltd. All rights reserved. PACS: 16.10.Nz; 74.25.Gz; 73.61.r; 81.70.Jb; 81.65.Mq; 61.65.b Keywords: D. Optical properties

1. Introduction ZnO is promising as an economical transparent conducting oxide with high optical transmittance and low electrical resistivity. High transmittance and good electrical conductivity ZnO thin films have potential use in a variety of technological applications including optoelectronic devices [1], gas sensors [2], photovoltaic devices [3], solar cells [4], and transparent electrodes [5]. ZnO can be grown on low-priced substrates such as glass at relatively low temperatures. Several techniques can be used to grow ZnO thin films. Spray pyrolysis [6], magnetron sputtering [7], reactive plasma deposition [8], metalorganic chemical vapor deposition [9], femtosecond pulsed-laser deposition [10], and thermal oxidation [11] are among them. Many studies have been made on annealing process to obtain high structural quality films [12,13]. Corresponding author. Tel.: +20 9346 097 87; fax: +20 9346 011 59.

E-mail address: [email protected] (S.H. Mohamed). 0022-3697/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jpcs.2006.05.048

Relieving accumulated strain energy, diminishing defects and enlarging grain size are the main annealing effects. The annealing process is used to flatten the surface of lowtemperature buffer layers in fabricating high-quality epitaxial films [14]. It is also adopted to treat the stoichiometry deviations through thermal oxidation [15,16]. Interestingly, when the annealing process is carried out at relatively high temperatures (4600 1C), evaporation of lattice constituents takes place at ZnO surface simultaneously with grain growth [17,18]. Furthermore, the annealing process can cause a drastic increase in the electrical resistivity of ZnO thin films [11,16,19], and occasionally a slight decrease in optical transmittance [12]. The last two effects are not desirable especially when ZnO thin films are being applied as transparent electrodes or as windows in solar cells. The goal of this manuscript is to study the influence of rf plasma nitriding on properties of ZnO thin films prepared via thermal oxidation of metallic Zn films. Glow discharge optical spectroscopy (GDOS), profilometer,

ARTICLE IN PRESS 2352

S.H. Mohamed et al. / Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357

X-ray diffraction (XRD) and spectrophotometer are used to study the effect of rf plasma cyclic times on the structural and optical properties, and electrical resistivity of the as-oxidized and nitrided films. More focus is being put on the discussion of the electrical resistivity and the optical transmittance.

using a Profilometer (Dektak II) for 3 mm scan length at medium speed. The surface roughness has tested for both longitudinal and transverse directions of the film to calculate the average surface roughness.

2. Experimental details

3.1. Structural properties

Metallic Zn thin films were prepared by electron-beam evaporation in an Edward’s high vacuum coating unit model 306A at a pressure of 5  106 and 8  105 Torr before and during film deposition, respectively. The films were prepared on ultrasonically cleaned microscopic glasses held at room temperature. The thickness of the films (650725 nm) is controlled using digital film thickness monitor model TM 200 Maxtek. The film deposition rate was 12.570.5 nm/s. Oxidation of the metallic films was carried out in a fully controlled furnace in atmospheric air. The films were oxidized for 1 h at 350 1C. After that the temperature was increased to 550 1C and held for 3 h. The oxidized ZnO films were immersed into pure nitrogen rf plasma at a fixed input plasma power of 225 W for 3 min processing time intervals. The effect of rf plasma processing times was studied for one and up to five time intervals. In each cycle, the film was exposed to plasma for 3 min and then cooled down to room temperature without opening the reactor tube before starting the next treatment cycle for 3 min at the same plasma conditions. The rf plasma system has been described elsewhere [20]. The increasing sample temperature was measured during the rf plasma process by a Chromel–Alumel thermocouple which was attached to the sample holder. For a fixed input rf plasma power of 225 W, the sample temperature at the end of the 3 min treatment was 34375 1C. The crystallinity of the as-oxidized and the plasma nitrided films were examined using a Diano Corporation X-ray diffractometer. CoKa radiation (l ¼ 1.79026 A˚) was used with normal focus. The optical transmittance (T) and reflectivity (R) of the ZnO films were studied using a Jasco 570 double beam spectrophotometer in the wavelength (l) range of 200–2500 nm at normal incidence. The resistivity measurements were carried out using the two-terminal configurations, which are used elsewhere [11,21,22] by applying constant voltage of 570.1 V to the sample and measuring the drown current through it using a Keithley 614 electrometer. The resistivity measurements were done at room temperature. Electrical contacts were made by applying silver paste over part of the film surface with separation distance of 2 mm. The measurements were carried out in a dark environment, atmospheric air, and relative humidity of 40%. Elemental depth profiles of the as-oxidized ZnO and plasma nitrided films were determined by GDOS. In this technique, the quantitative analysis of the atomic elemental concentration as a function of depth depends on the sputtering rate of the examined materials. The surface roughness was measured

3.1.1. Film composition, thickness, and surface roughness The variations of the chemical composition in atomic elements along the depth of the as-oxidized and nitrided samples at successive plasma processing cyclic times are shown in Fig. 1(a–d). The GDOS analysis for as-oxidized sample detects zinc, oxygen and silicon as shown in Fig. 1a. A relative high concentration of silicon and oxygen has been detected in the near surface region (first tens of nanometres), which might be attributed to the contamination and plasma disturbance during GDOS measurements. After that, a supersaturated region is observed which extends up to a depth of 550 nm and then sharply decreases. Fig. 1b shows the elemental concentration as a function of depth for sample treated for one cycle (processing plasma time of 3 min). The film thickness decreases to approximately one half in comparison to the thickness of as-oxidized sample. Low amount from nitrogen is detected near the interface region between the film and the substrate. The concentration of oxygen relatively decreases along the whole depth. Nearly the same behaviour is observed in Fig. 1c and d. One can see that the nitrogen diffuses to a greater depth when the processing plasma time is increased. Fig. 2(a and b) shows the variation of film thickness and surface roughness as a function of treatment time. The film thickness is drastically reduced from 700 to 370 nm after exposing it to nitrogen rf plasma for just 3 min. By increasing the treatment time, the thickness gradually decreases further to 340 and 300 nm for samples treated for 9 and 15 min, respectively. This reduction in thickness can be ascribed to plasma etching. As observed from Fig. 2b, the surface roughness increases with increasing treatment time. The raise of the surface roughness could be due to interactions between the surface and the reactive nitrogen plasma species. For example, the energetic nitrogen ions entering the film migrate towards the inner parts of the film, thus leaving small pipes towards the upper parts of the film, which constitutes the surface roughness. The plasma sputtering (etching) can also increase the surface roughness.

3. Results and discussion

3.1.2. XRD examination Fig. 3(a–d) shows the XRD patterns for ZnO films nitrided for different plasma times. All the patterns reveal the existence of ZnO single-phase with hexagonal wurtzite structure. However, for all the films, there are three main peaks at 37.061, 40.211 and 42.361, which can be identified as (1 0 0), (0 0 2) and (1 0 1), respectively. The yield

ARTICLE IN PRESS S.H. Mohamed et al. / Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357

2353

Fig. 1. Atomic element concentration depth profiles for (a) as-oxidized, (b) treated for 3 min, (c) treated for 9 min, and (d) treated for 15 min ZnO thin films. The symbol ‘‘*’’ refers to the product operation.

The mean grain size (D) can be calculated from the Scherrer formula [23] D¼

0:9l , B cos yB

(1)

where l, yB and B, are the X-ray wavelength (1.79026 A˚), Bragg diffraction angle and line width at half-maximum of (0 0 2) peak at (40.211), respectively. Fig. 4 shows the dependence of the mean grain sizes on treatment time. The particle size slightly decreased from 39 to 37 nm when the sample was treated for 3 min. With further increase in the plasma time process, the particle size remains almost constant at 36 nm. 3.2. Electrical resistivity

Fig. 2. Variations of (a) film thickness and (b) surface roughness as a function of treatment time.

intensities from plasma-treated films are relatively lower than the corresponding values for as-oxidized ones. These lower intensities can be ascribed to the higher surface roughness observed for the plasma-treated films as shown in Fig. 2b.

The variation of room temperature resistivity with treatment time is shown in Fig. 5. It can be seen that the as-oxidized ZnO film has a high resistivity of 3.85  102 O cm. However, plasma treatment for 3 min decreases the resistivity drastically to 2.86  101 O cm. With further increase in the plasma processing time, the resistivity of the sample decreases continuously to reach a value of 2.42  102 O cm at 15 min. In general, nominally undoped ZnO exhibits an n-type conductivity rising from the electrons in the conduction band. It has been widely considered that this n-type

ARTICLE IN PRESS 2354

S.H. Mohamed et al. / Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357

Fig. 3. XRD patterns of ZnO thin films treated at different times.

Fig. 4. Dependence of the mean grain size (D) on treatment time.

conductivity originated from the negative defects such as oxygen vacancies and Zinc interstitials [24]. Also, it has been noted that the conductivity of ZnO films is controlled by intercrystallite depletion barriers [25] and both ionized impurities and grain boundaries contribute significantly to the overall mobility [26]. The high resistivity value obtained for as-oxidized ZnO film, which is usually obtained for post-annealed films, can be possibly ascribed to the high amount of defects, and/or grain boundaries, and to the oxygen reaction with ZnO. During surface treatment, nitrogen ions are able to readily diffuse into the bulk of the film and deactivate the ‘dangling bonds’ associated with the structural defects. This will result in a considerable increase in carrier mobility and hence lifetime of the electrons in the film. 3.3. Optical properties

Fig. 5. Variation of the electrical resistivity as a function of treatment time.

The wavelength dependence of optical transmittance and reflectance spectra of the ZnO films plasma treated for different times are shown in Fig. 6a and b, respectively. It is observed that the transmittance increases and the reflectance decreases with increasing treatment time. The wavelength of the absorption onset shifts to a shorter wavelength with increasing treatment time. The transmittance is expected to depend mainly on three factors: (i) oxygen deficiency, (ii) surface roughness, where the surface scattering reduces the transmission, which in turn depends on the grain size, and (iii) defect centres. We assume that loss due to oxygen deficiency can come only from oxygen vacancies since the as-oxidized films are stoichiometric. It can be seen from Fig. 2b that the surface roughness value

ARTICLE IN PRESS S.H. Mohamed et al. / Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357

2355

Fig. 7. Ln(I/I0) as a function of thickness. I0 was taken as the average of T for the as-oxidized film while I values were taken as the average of T for nitrided samples. Fig. 6. Influence of treatment time on (a) the optical transmittance and (b) reflectance spectra.

increases with increasing treatment time. Therefore, we can conclude that the main cause for the increase in transmittance upon surface plasma treatment is defect centres reduction. The above argument is valid when the thickness of the film is constant. In the present case, the thickness of the ZnO film decreases with increasing treatment time. Thus, the thickness decrease should be considered as another reason for the increase of transmittance upon increasing treatment plasma time. The transmitted intensity of the electromagnetic radiation depends on the thickness (d) and the absorption coefficient ða ¼ 4pk=lÞ through the equation: I ¼ I 0 expðadÞ,

(2)

where I0 can be taken as the average of T for the asoxidized film and I values can be taken as the average of T for the nitrided samples. Thus, the variation in transmittance could be caused by the changes in thickness and/or absorption coefficient. Plotting ln(I/I0) versus d should result in a straight line for constant p which is not fitting with our case, as one can see from Fig. 7. The optical band gap, Eg, values are calculated by assuming a direct transition between the edges of the valence and the conduction bands, for which the variation of the absorption coefficient, a, with photon energy is given by aðhnÞ ¼ Aðhn  E g Þ1=2 .

(3)

By plotting a versus hn and extrapolating the linear region of the resulting curve, Eg can be obtained. The Eg value of as-oxidized ZnO film (2.76 eV) is much lower than the single-crystal value (3.4 eV [27]). This may be ascribed, as Pikus [28], and Srikant and Clarke [29] reported, to the ‘‘growth stresses’’ in the annealed films. The band gap of a stressed Eg can be defined in terms of unstrained E 0g as

Fig. 8. Variation of the optical band gap (Eg) with treatment time.

[28,29]   E g ¼ E c  E v ¼ E v ¼  E 0g þ DE v ,

(4)

where the position of the conduction band (Ec) is taken as a reference, E 0g is the unstressed position of the valence band and DE v is the change in the position of the valence band due to stresses. A negative value of DE v indicates a decrease in the band gap width and a positive value indicates an increase in the band gap. Fig. 8 shows the dependence of the optical band gap of ZnO films on the surface plasma treatment time. It can be seen that Eg increases with increasing surface treatment time. This increase in Eg can be ascribed to the well-known Burstein–Moss shift (BM shift) induced by filling the lowest levels of the conduction band with charge carriers in degenerate semiconductors [30,31]. The variations of refractive index and extinction coefficient, in the visible wavelength range, of ZnO films treated at different plasma processing times are shown in

ARTICLE IN PRESS 2356

S.H. Mohamed et al. / Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357

Fig. 10. Variation of the estimated packing density as a function of treatment time.

Fig. 9. Variations of (a) refractive index (n) and (b) extinction coefficient (k) as a function of wavelength for ZnO thin films treated at different times.

zero. The variation of the extinction coefficient is directly related to the optical transmittance of the films as shown in Fig. 6a. 4. Conclusion

Fig. 9a and b, respectively. It is observed that the refractive index of the films decreases with increasing treatment time. This variation of refractive index may be attributed to the change in the packing density with treatment plasma time. Many attempts have been made to correlate the packing density with refractive index of the material [12,32]. Mecleod [32] summarized the four equations, which relate the packing density with the refractive index. Gupta and Mansingh [12] found that the appropriate equation for ZnO is the one based on the assumption of cylindrical particles. Thus, assuming that the crystallites of the films are of cylindrical form, we can use the following expression [32]: n2f ¼

ð1  PÞn4v þ ð1 þ PÞn2v n2s , ð1 þ PÞn2v þ ð1  PÞn2s

(5)

where nf is the refractive index of the composite film (containing voids and cylindrical particles), ns is the index of the solid material of the film (single crystal value), nv is the index of the void in the film (equals one for air), and P is the packing density. Using the above relation, the packing densities for ZnO films treated at different plasma processing times are determined. The value of the refractive index for the single crystal (ns ¼ 2.004 at 700 nm) is taken from Ref. [33]. The dependence of the packing densities on the treatment time is shown in Fig. 10. It is found that the calculated packing density values decrease with increasing treatment time. The packing density value for as-oxidized film is very close to one. This result indicates that the oxidation technique can produce films with densities comparable to the singlecrystal density. The extinction coefficient also decreases with increasing treatment time. Moreover, the film treated at 15 min has extinction coefficient values very close to

The effects of nitriding using different rf plasma processing times on the properties of ZnO thin films, prepared by thermal oxidation, were studied. The film thickness decreases due to plasma etching. However, it has been found the surface roughness increases with increasing treatment plasma time. The successive plasma processing time has a large influence on ZnO thin films by decreasing the electrical resistivity and increasing the optical transmittance, both are highly desirable for applications of ZnO as a transparent conductor. The rf plasma nitriding possibly improves ZnO TCO properties for transparent electrodes and solar cell windows applications. Acknowledgements We would like to thank our colleagues at the Physics Department (Sohag) for their fruitful discussions. Mr. M. Hamad is acknowledged for his technical support. Prof. Dr. C. Timplier (Universite de Poitiers, France) is acknowledged for his supporting and fruitful discussions. The authors are also grateful to Dr. A. Andres (LawrenceBerkeley, USA) who has read the manuscript and offered his comments. References [1] D.C. Reynolds, D.C. Look, B. Jogai, Solid State Commun. 99 (1996) 873. [2] I. Stambolova, K. Konstantinov, S. Vassilev, P. Peshev, T. Tsacheva, Mater. Chem. Phys. 63 (2000) 104. [3] T. Pauporte, D. Lincot, Electrochem. Acta 45 (2000) 3345. [4] J.C. Lee, K.H. Kang, S.K. Kim, K.H. Yoon, I.J. Park, J. Song, Sol. Energy Mater. Sol. Cells 64 (2000) 185.

ARTICLE IN PRESS S.H. Mohamed et al. / Journal of Physics and Chemistry of Solids 67 (2006) 2351–2357 [5] K. Nakahara, K. Tamura, M. Sakai, D. Nakagawa, N. Ito, M. Sonobe, H. Takasu, H. Tampo, P. Fons, K. Matsubara, K. Iwata, A. Yamada, S. Niki, Jpn. J. Appl. Phys. 43 (2004) L180. [6] J.-H. Lee, B.-W. Yeo, B.-O. Park, Thin Solid Films 457 (2004) 333. [7] A. Pimentel, E. Fortunato, A. Goncalves, A. Marques, H. Aguas, L. Pereira, I. Ferreira, R. Martins, Thin Solid Films 487 (2005) 212. [8] K. Iwata, T. Sakemi, A. Yamada, P. Fons, K. Awai, T. Yamamoto, S. Shirakata, K. Matsubara, H. Tampo, K. Sakurai, S. Ishizuka, S. Niki, Thin Solid Films 480–481 (2005) 199. [9] Y. Chen, Y. Pu, L. Wang, C. Mo, W. Fang, B. Xiong, F. Jiang, Mater. Sci. Semicond. Process. 8 (2005) 491. [10] E. Millon, O. Albert, J.C. Loulergue, J. Etchepare, D. Hulin, W. Seiler, J. Perriere, J. Appl. Phy. 88 (2000) 6937. [11] S.H. Mohamed, Phys. Stat. Sol. (a) 202 (2005) 1948. [12] V. Gupta, A. Mansingh, J. Appl. Phys. 80 (1996) 1063. [13] M.K. Puchert, P.Y. Timbrell, R.N. Lamb, J. Vac. Sci. Technol. A 14 (1996) 2220. [14] Y. Chen, H.J. Ko, S.K. Hong, T. Yao, Appl. Phys. Lett. 76 (2000) 559. [15] N.R. Aghamalyan, I.A. Gambaryan, E. Kh. Goulanian, R.K. Hovsepyan, R.B. Kostanyan, S.I. Petrosyan, E.S. Vardanyan, A.F. Zerrouk, Semicond. Sci. Technol. 18 (2003) 525. [16] S.H. Mohamed, H.M. Ali, H.A. Mohamed, A.M. Salem, Eur. Phys. J. Appl. Phys. 31 (2005) 95. [17] I.W. Kim, S.J. Doh, C.C. Kim, J.H. Je, J. Tashiro, M. Yoshimoto, Appl. Surf. Sci. 241 (2005) 179.

2357

[18] S. Zhu, C.H. Su, S.L. Lehoczky, M.T. Harris, M.J. Callahan, P. McCarty, M.A. George, J. Cryst. Growth 219 (2000) 361. [19] R. Al Asmar, G. Ferblantier, F. Mailly, P. Gall-Borrut, A. Foucaran, Thin Solid Films 473 (2005) 49. [20] F.M. El-Hossary, N.Z. Negm, S.M. Khalil, A.M. Abd El-Rahman, Thin Solid Films 405 (2002) 179. [21] E. Byon, T.H. Oates, A. Anders, Appl. Phys. Lett. 82 (2003) 1634. [22] H. Hosono, M. Miyakawa, H. Kawazoe, K.-I. Shimizu, J. NonCryst. Solids 241 (1998) 190. [23] B.D. Cullity, Elements of X-ray Diffraction, second ed, AddisonWesley, Reading, MA, 1979 (p. 102). [24] D.C. Look, J.W. Hemsky, J.R. Sizelove, Phys. Rev. Lett. 82 (1999) 2552. [25] T.L. Tansley, D.F. Neely, Thin Solid Films 121 (1984) 95. [26] J.J. Robbins, J. Harvey, J. Leaf, C. Fry, C.A. Wolden, Thin Solid Films 473 (2005) 35. [27] O. Madelung, M. Schulz, H. Weiss (Eds.), Numerical Data and Functional Relationships in Science and Technology, Landolt–Bornstein New Series, vol. 17b, Springer, Berlin, 1982, p. 35; O. Madeluung, M. Schulz (Eds.), Landolt–Bornstein New Series, vol. III, Teil 22a, Springer, Berlin, 1987, p. 160. [28] G.E. Pikus, Sov. Phys. Solid State 6 (1964) 261. [29] V. Srikant, D.R. Clarke, J. Appl. Phys. 81 (1997) 6357. [30] E. Burstein, Phys. Rev. 93 (1954) 632. [31] T.S. Moss, Proc. Phys. Soc. Sect. B 76 (1954) 775. [32] H.A. Macleod, J. Vac. Sci. Technol. A 4 (1986) 418. [33] A.N. Pikhtin, A.D. Yas’Kov, Sov. Phys. Semicond. 15 (1981) 81.