Effects of an alternating field in field-aided lateral crystallization process for low temperature poly-silicon

Thin Solid Films 503 (2006) 236 – 240 www.elsevier.com/locate/tsf

Effects of an alternating field in field-aided lateral crystallization process for low temperature poly-silicon Sung-Hwa Choi, Sung Bo Lee, Young-Woong Kim, Chang Kyung Kim, Duck-Kyun Choi * Department of Ceramic Engineering, Hanyang University, 17 Haengdang-dong, Seongdong-gu, Seoul 133-791, South Korea Received 26 January 2005; received in revised form 15 July 2005; accepted 3 November 2005 Available online 28 December 2005

Abstract The effect of the alternating field (AC voltage) instead of the static field (DC voltage) was investigated in the field-aided lateral crystallization process, which is one of the low temperature crystallization processes for the amorphous silicon films. Using a photolithography process, a 5-mmwide bar-shaped photoresist (PR) pattern was formed on the a-Si. On the PR-patterned a-Si, a 2 – 3-nm-thick Cu catalyst layer was deposited by a DC sputtering, and then, the Cu layer on the PR pattern was lifted off. The silver electrodes were pasted at the opposite sides of the Cu-free bar pattern. Then, the patterned specimen was annealed at 500 -C in N2 ambient for 5 h with the application of various AC fields (ranging from 1 to 5 V/cm) along with a DC field of 30 V/cm. As compared with the case of a DC field of 35 V/cm only, the specimen from a mixed field of 30 V/cm DC and 5 V/cm AC resulted in 1.5 times faster crystallization rate, regardless of experimental frequency values ranging from 10 Hz to 50 MHz. Presumably, the enhancement of the crystallization rate under the combined field is associated with an increase in the flux of the crucial diffusion species, Cu atoms, which govern the overall crystallization rate due to the effect by the AC field. D 2005 Elsevier B.V. All rights reserved. PACS: 61.43; 85.70.K Keywords: Field-aided lateral crystallization; AC field; DC field; Polycrystalline silicon thin-film transistors; Low temperature crystallization

1. Introduction Polycrystalline silicon thin-film transistors (poly-Si TFTs) have recently attracted considerable attention for their high field-effect mobility and response velocity [1,2]. However, poly-Si TFTs, which are the essential part of liquid crystal displays, are processed over 600 -C so that the use of the high temperature durable expensive quartz substrate is inevitable for the high performance TFT fabrication. If it is possible to fabricate poly-Si TFTs below the softening point of low price commercial glass, we can integrate all the functions such as voice, display, information processing, memory, input and output in a very inexpensive way on the glass substrate. Thereby, it can be widely applied to laptops, personal digital assistants, mobile phones, and desktop computers. The important considerations for manufacturing low temperature poly-Si include not only the development of a low temperature process but also the guarantee of high quality poly-Si [3,4]. * Corresponding author. E-mail address: [email protected] (D.-K. Choi). 0040-6090/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2005.11.007

Therefore, to obtain poly-Si below 500 -C, hot-wire chemical vapor deposition (CVD) or very high frequency glow discharge CVD was studied [5– 7]. But, the contamination in deposited poly-Si or the poor quality of the film is hampering the extensive study. Consequently, more studies are focused on the way to crystallize the amorphous Si (a-Si) film below 500 -C after depositing a-Si film at a temperature below 300 -C. Among a variety of crystallization techniques available to obtain poly-Si at low temperatures, the field-aided lateral crystallization (FALC) process is known to be induced by an influence of the electric field toward a specified direction after the silicide phase formation by a reaction between a metal catalyst and a-Si [8,9]. It is reported that the crystallization rate by the FALC process is much faster than that by the metalinduced lateral crystallization (MILC) process [10,11]. It is also reported that the undesirable metal pollution in the channel region can be greatly reduced by the FALC process [12]. Up to now, all the reports related to the FALC process are dealing with a static field (DC voltage) to induce the crystallization. It would also be interesting to investigate the crystallization aspect by an alternating field (AC voltage). The

S.-H. Choi et al. / Thin Solid Films 503 (2006) 236 – 240

aim of this study is to examine effects of AC field on the crystallization of a-Si in the FALC process. Compared with the case of the FALC process with a DC field only and the MILC process, the crystallization rate was increased in the FALC process with a mixed field of AC and DC. The potential role of the AC field on the crystallization was suggested. 2. Experimental details Silicon oxide of 500 nm thick was grown on Corning glass 1737 for the surface passivation. Then 80 nm of amorphous silicon (a-Si) was deposited on the passivated surface by plasma-enhanced chemical vapor deposition at 280 -C using Si2H6 and H2 as source gases. The wafer was cut into 3  3 cm2 square specimens for the application of the electric field. To effectively remove the impurities on a-Si film, RCA cleaning was conducted [13]. The organic impurities were removed using a 1:2:7 NH4OH/H2O2/deionized water solution in a temperature range between 50 -C and 60 -C for 10 min. Then, the specimen was immediately dipped into a diluted HF (10:1) solution to remove the native oxide in the a-Si thin film. After cleaning the specimen, 5 mm bar-shaped photoresist (PR) patterns were formed on the a-Si using a photolithography process. A thin Cu catalyst layer, about 2– 3 nm, was deposited

Fig. 1. Schematic diagram of experimental system for FALC process: (a) crosssectional view and (b) top view.

237

using a DC sputtering system at room temperature. Then, the Cu layer on the PR pattern was lifted off and the 5 mm barshaped Cu-free patterns were left. After the electrodes were formed using silver paste at two opposite sides of the bar pattern, an electric field was applied to the patterned specimen in a tube furnace during thermal annealing by a DC power supply and a DC/AC function generator, as illustrated in Fig. 1(a) and (b). The span between electrodes was 1 cm. In that configuration, the voltage applied (V) and the electric field (V/cm) can be interchangeably used hereafter. The specimen was heated at a rate of 5 -C/min in a tube furnace of N2 ambient and held at a crystallization temperature of 500 -C for 5 h. A mixed AC (up to 5 V/cm) and DC field (30 V/cm) was applied. The frequency range of the AC-field was between 10 Hz and 50 MHz. After the thermal annealing, crystallization behavior and rates were identified by optical microscopy, and the crystallization degree in FALC poly-Si could be estimated by micro-Raman spectroscopy [14]. The laser of 515 nm wavelength was used and its diameter was in the range from 5 Am to 100 Am. 3. Results and discussion Fig. 2 shows the optical images of the patterns of partially crystallized a-Si at 500 -C after depositing the Cu catalyst outside the patterns. Fig. 2(a) shows that the crystallization length reached 4 Am on both sides after the MILC process for 5 h. When a DC field of 35 V/cm was applied, as shown in Fig. 2(b), the crystallization length reached 22 Am on the negative electrode side, whereas the crystallization scarcely occurred on the positive electrode side. At the same crystallization time and temperature, with a mixed electric field of an AC field of 5 V/ cm (here, 100 Hz) and a DC field of 30 V/cm, the crystallization also appearing only on the negative electrode side exhibited a length of 38 Am. This result indicates that the addition of the AC field to the DC field in the FALC process not only induces the directional crystallization toward the metal catalyst-free region as in the case of the FALC process with the DC field only, but also enhances the crystallization rate. To explore the influence of the AC field in depth, we tried various frequencies of the AC field for the crystallization. Fig. 3 shows the results of the crystallization at various crystallization conditions. As seen in Fig. 3, the crystallization rate of the mixed electric field was approximately 1.5 times faster than that of the DC field only (4.5 Am/h) and the frequency dependence was not significant in the present experimental frequency range between 10 Hz and 50 MHz. Considering the maximum value of the AC field (5 V/cm), the range of the mixed field is from 25 V/cm to 35 V/cm, which does not exceed the value of the DC field only case (35 V/cm). Still, the crystallization rate of the a-Si in the case of the mixed field is higher. At present, the most reasonable and acceptable explanation for the FALC phenomenon is based on the combination of three kinds of driving forces: electromigration, potential gradient, and chemical activity difference. Among them, the driving force from the chemical activity difference is known to

238

S.-H. Choi et al. / Thin Solid Films 503 (2006) 236 – 240

Fig. 2. Optical micrographs of the partially crystallized patterns annealed at 500 -C for 5 h (a) without electric field (MILC), (b) with a DC field of 35 V/cm (FALC) and (c) with a mixed field of a DC field of 30 V/cm and an AC field of 5 V/cm.

be the only driving force to induce the MILC phenomenon and was described by Hayzelden et al. [15,16]. On the other hand, in FALC, the other two driving forces related to the electric field are more important [17]. What we have observed consistently from the FALC experiment is that the electromigration overwhelms the other driving forces in most cases. In very rare case where the applied voltage is extremely high (therefore, the electric field intensity is very high but the current level is low due to the saturation of the current beyond certain value), the potential gradient effect could dominate. This case cannot be observed by simply applying high voltage, because the specimen usually burns out. At the F+_ electrode side, the intensity of the pushing force against atoms by electromigration which is always from F _ electrode side to F+_ electrode side surpasses the effect by thermal diffusion due to the chemical activity difference. As a result, in FALC process, the crystallization along the current flow direction significantly increases, while it is suppressed in the opposite direction; this is different from the MILC process where the crystallization rate is identical in all directions. One may be concerned about the stress development due to the fast crystallization in the mixed field condition. Raman spectroscopy analysis gives insight into not only the degree of the crystallization but also the stress in the film. Fig. 4 shows the Raman spectroscopy result of specimens discussed in Fig. 3. Crystalline silicon (c-Si) is known to have a sharp peak at 521 cm 1 in Raman analysis [18], and the peak position shifts depending on the stress. All the specimens turned out to have a sharp peak at 521 cm 1, which indicates that no noticeable stress has developed during the crystallization. Another concern related to the use of a mixed AC field and DC field during the crystallization might be a possibility of enhancement in the crystallization rate arising from Joule heating caused by current increase in the a-Si on the glass substrate. We measured the current by an amperemeter which was connected in series to the power supply during heat treatment and the temperature on the specimen by a thermocouple which is attached to the specimen. There was a

amorphous

crystalline

cm-1)

(521 cm-1)

9

DC 30V/cm + AC 5V/cm 8

Intensity (arb. units)

Crystallization rate (10-6m/h)

(480

7 6 5 4 3

DC 35V/cm 10 Hz 100 Hz 10 kHz 1 MHz 50 MHz 300

10

100

104 106 5x107

Frequency (Hz) Fig. 3. Crystallization rate at mixed fields of DC (30 V/cm) and AC (5 V/cm) with various frequencies during annealing at 500 -C for 5 h.

400

500

600

700

Raman Shift (cm-1) Fig. 4. Raman spectra during annealing at 500 -C for 5 h at a DC field of 35 V/ cm and at mixed fields of DC (30 V/cm) and AC (5 V/cm) with various frequencies (10 Hz – 50 MHz).

180

DC 35 V/cm 10 kHz 1 MHz

Current (10-6 A)

160 140 120 100 80 60 40

0

60

120

180

240

300

Crystallization rate (10-6m/h)

S.-H. Choi et al. / Thin Solid Films 503 (2006) 236 – 240

239

MILC (1.79 eV) 1

DC 35 V/cm (1.66 eV) 0.1

DC 30V/cm + AC 5 V/cm (1.42 eV) 0.01 15.0

Time (min)

15.5

16.0

16.5

1/kT (eV-1) Fig. 5. Current measurement results from the specimens annealed at 500 -C for 5 h at a DC field of 35 V/cm and at mixed fields of DC (30 V/cm) and AC (5 V/ cm) with various frequencies (10 kHz and 1 MHz).

Crystallization rate (10-6m/h)

difference in current between the DC field only and the mixed electric field of DC and AC (AC + DC) (Fig. 5). Nevertheless, the measured temperature ranged from 500 -C to 504 -C in all conditions. Therefore, it can be concluded that the increase in the crystallization rate in the case of the AC-added FALC process is not an effect of Joule heating. It is worth looking at the effect of the intensity of an alternating field on the crystallization. The amplitude of the AC field at a fixed frequency of 10 kHz was varied between 5 V/ cm and 1 V/cm, and the crystallization experiment was performed at the identical experimental conditions. As one can notice from Fig. 6, with decreasing the amplitude of the AC field from 5 V/cm to 1 V/cm, the crystallization rate gradually decreased from 6.4 Am/h to 6.0 Am/h. However, there is a significant enhancement in the crystallization rate even at the addition of an AC field of 1 V/cm to the DC field as compared with the case of the DC field only (4.5 Am/h). This result confirms the obvious effect of the AC-field again. Crystallization at various conditions would be connected with the different kinetic parameters. Accordingly, the activation energies were evaluated for three different crystallization conditions: MILC, FALC with DC 35 V/cm only (DC FALC), and FALC with 5 V/cm AC-field of 10 kHz along with 30 V/ cm DC-field (AC + DC FALC). The crystallization temperatures were in the range from 425 -C to 500 -C (Fig. 7). The activation energies for the MILC, DC FALC, and AC + DC FALC were 1.79 eV, 1.66 eV, and 1.42 eV, respectively. The 7

AC frequency 10 kHz 6

5

Fig. 7. Activation energies for crystallization of the specimens annealed at 500 -C for 5 h in the MILC process, the FALC process with a DC field of 35 V/cm, and the FALC process with a mixed field of DC 30 V/cm and AC 5 V/cm at a frequency of 10 kHz.

activation energy of the MILC was higher than that of the DC FALC by 0.13 eV, which was, in turn, higher than that of the AC + DC FALC by 0.24 eV. In the present study, the activation energy for crystallization was the lowest in the specimen treated by the FALC process with the mixed field (Fig. 7). For the MILC process, the flux of the diffusing species is governed only by the chemical potential gradient, but for the FALC process, the flux is affected by not only the chemical potential gradient but also the electric potential gradient. The DC field in the FALC process is interpreted as causing the electron wind effect (electromigration), enhancing diffusion of atoms, and consequently, crystallization [17]. The extension of the DC-FALC mechanism to the addition of AC component can be deduced as follows: Although there is an AC component, the resulting voltage swing of DC + AC is from 25 to 35 V. This is similar to the case of small signal capacitance. What is important is that even at the 25 V the current flows from F _ to F+_ direction. Therefore, we can consider this case as the DC field but varies from 25 V to 35 V. One thing which is not yet clear is that the crystallization rate is faster than DC only case (35 V) with a less voltage. That apprehension can be cleared if one remembers that the current level is higher in mixed field case than that in pure DC case, as seen in Fig. 5. Therefore, we expect the higher electromigration effect, which results in a high crystallization rate. What we can claim from our mechanism is the pulsing or jerking effect by AC field along the current flowing direction by electron wind effect (electromigration). The current increase will induce stronger electron wind effect, and consequently, enhance the Cu diffusion, increasing crystallization rate. Further study of the AC field dependence of crystallization in this system is being performed. 4. Conclusions

4

1

2

3

4

5

AC Field (V/cm) Fig. 6. Crystallization rates from the specimens annealed at 500 -C for 5 h at a DC field of 35 V/cm and at mixed fields of DC 30 (V/cm) and AC fields with various amplitudes (1 V/cm, 2 V/cm and 5 V/cm).

Effects of an alternating field on the low temperature crystallization behavior were investigated by adding AC fields to the DC field during the FALC process. As compared with the case of the DC field only (35 V/cm), the specimen treated at a mixed field of the DC and AC fields (30 V/cm and 5 V/cm,

240

S.-H. Choi et al. / Thin Solid Films 503 (2006) 236 – 240

respectively) resulted in 1.5 times faster crystallization rate, regardless of the frequencies used ranging from 10 Hz to 50 MHz. In addition, Raman spectroscopy revealed that the stress in the crystallized film treated at the mixed fields is not different from that at the DC field only. The applied AC field clearly increases current and will strengthen the electron wind effect, leading to the enhancement of the crystallization rate. Acknowledgement This work was financially supported by the Korea Institute of Science and Technology Evaluation and Planning (KISTEP) through the National Research Laboratory (NRL) program and by Korea Research Foundation Grant (KRF2004-005-D00167). References [1] T. Matsuyama, N. Terada, T. Baba, T. Sawada, S. Tsuge, K. Wakisaka, S. Tsuda, J. Non-Cryst. Solids 198 – 200 (1996) 940. [2] I.H. Song, C.H. Kim, W.J. Nam, M.K. Han, Curr. Appl. Phys. 2 (2002) 225.

[3] J. Jang, J.I. Ryu, S.Y. Yoon, K.H. Lee, Vacuum 51 (1998) 769. [4] M. Wong, H.S. Kwok, Microelectron. J. 35 (2004) 337. [5] R.E.I. Schropp, B. Stannowski, A.M. Brockhoff, P.A.T.T. van Veenendaal, J.K. Rath, Mater. Phys. Mech. 1 (2000) 73. [6] J.C. Lee, K.H. Kang, S.K. Kim, K.H. Yoon, J.S. Song, I.J. Park, Sol. Energy Mater. Sol. Cells 74 (2002) 233. [7] K. Ro, K. Nakahata, T. Kamiya, C.M. Fortmann, I. Shimizu, J. Non-Cryst. Solids 266 – 269 (2000) 1088. [8] K.S. Song, J.B. Lee, S.I. Jun, D.K. Choi, S.G. Park, J. Mater. Sci. Lett. 18 (1999) 1209. [9] S.H. Park, S.I. Jun, K.S. Song, C.K. Kim, D.K. Choi, Jpn. J. Appl. Phys. 38 (1999) L108. [10] J.F. Li, X.B. Zeng, X.W. Sun, G.J. Qi, Thin Solid Films 458 (2004) 149. [11] T.K. Kim, G.B. Kim, B.I. Lee, S.K. Joo, IEEE Electron Device Lett. 21 (2000) 347. [12] J.B. Lee, C.J. Lee, D.K. Choi, Jpn. J. Appl. Phys. 40 (2001) 6177. [13] S. Wolf, R.N. Tauber, Silicon Processing for the VLSI Era, vol. 1, Lattice Press, CA, 1999, p. 516. [14] J.E. Gerbi, J.R. Abelson, J. Appl. Phys. 89 (2001) 1465. [15] C. Hayzelden, J.L. Batstone, J. Appl. Phys. 73 (1993) 8279. [16] C. Hayzelden, J.L. Batstone, R.C. Cammarata, Appl. Phys. Lett. 60 (1992) 225. [17] D.-K. Choi, H.-C. Kim, Y.-B. Kim, Appl. Phys. Lett. 87 (2005) 063108. [18] H. Kakinuma, M. Mohri, M. Sakamoto, T. Tsuruoka, J. Appl. Phys. 70 (1991) 7374.