APPLIED PHYSICS LETTERS
VOLUME 74, NUMBER 18
3 MAY 1999
Effects of inductively coupled plasma oxidation on the properties of polycrystalline silicon films and thin film transistors Yong Woo Choi, Sang Won Park, and Byung Tae Ahna) Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, 373-1 Koosung-dong, Yusung-gu, Taejon 305-701, Korea
~Received 21 September 1998; accepted for publication 26 February 1999! We investigated the effects of inductively coupled plasma ~ICP! oxidation on the properties of polycrystalline silicon ~poly-Si! films and thin film transistors ~TFTs!. The ICP oxidation in oxygen plasma passivated the dangling bonds in the poly-Si films, not by oxygen incorporation but by hydrogen incorporation; but the incorporated hydrogen diffused out during the TFT fabrication, so that the effect of the dangling bond passivation was not obtained in the TFT. The ICP oxidation did not remove the intragranular defects such as microtwins and stacking faults, but it reduced the interface trap density and also improved the performance of the poly-Si TFT. The field effect mobility of TFT with an ICP oxide and low-pressure chemical vapor deposited ~LPCVD! oxide double layer was 30.6 cm2/V s, while that of TFT with only a LPCVD oxide was 17.2 cm2/V s. © 1999 American Institute of Physics. @S0003-6951~99!00718-4#
To improve the performance of polycrystalline silicon ~poly-Si! thin film transistors ~TFTs!, it is necessary to reduce interface traps as well as the traps caused by the grain boundaries and intragranular defects, such as dangling bonds, microtwins, and stacking faults.1,2 Thermal oxidation is the best way to obtain a high-quality oxide and a highquality interface on crystalline-Si wafers, but the thermal oxidation rate at low temperature is too low to be applied to the low-temperature poly-Si TFT process, at which a glass substrate can be utilized for large-size liquid crystal display ~LCD! panels at low cost. Several deposition methods such as low-pressure chemical vapor deposition ~LPCVD!,3 plasma-enhanced chemical vapor deposition ~PECVD!,4 and sputtering5 are used for the low-temperature process, but the deposited oxides have a high density of interface traps. The plasma oxidation rate is high at low temperature due to its high-radical density.6 It was reported that the oxide grown by an electron cyclotron resonance ~ECR! plasma on poly-Si films at low temperatures had better electrical characteristics than the thermal oxide. The poly-Si TFTs with ECR-plasma gate oxides showed a good device performance due to both low interface trap density and smooth interface.7,8 In addition, the ECR plasma oxidation passivated the dangling bonds in the poly-Si films by oxygen incorporation, so that the TFTs showed better electrical and thermal stabilities than TFTs passivated by hydrogen incorporation.9,10 However, the ECR plasma system is not suitable for low-temperature poly-Si TFTs because it is hard to make a large volume for ECR systems. Planar-type inductively coupled plasma ~ICP! has a high-plasma density-like ECR plasma11 and a large volume of its system can be easily made because the ICP system has a simple structure.12 Therefore ICP oxidation is a suitable method for low-temperature poly-Si TFTs. In this letter, we report the effects of ICP oxidation on the properties of poly-Si films and TFTs. To investigate the density of Si dangling bonds in a!
Electronic mail:
[email protected]
poly-Si films, electron paramagnetic resonance ~EPR! measurements were performed. The hydrogen and oxygen contents in the poly-Si films were measured with a secondary ion mass spectrometry ~SIMS!. The microstructure of the poly-Si films was observed by transmission electron microscopy ~TEM!. For the EPR and the SIMS measurements, 500nm-poly-Si films were deposited by LPCVD with SiH4 gas at 620 °C on quartz wafers and Si wafers, respectively. The poly-Si films were oxidized in ICP with oxygen gas at 400 °C for 30 min. The thickness of the oxide was about 10 nm. For comparison, the hydrogen plasma treatment was performed in conventional capacitively coupled plasma at 200 °C for 5 h. For TEM observation and TFT fabrication, 100 nm poly-Si films were prepared by the solid phase crystallization of PECVD a-Si films at 600 °C for 48 h in a furnace. We fabricated self-aligned n-channel poly-Si TFTs with a 27 nm ICP oxide and 120 nm LPCVD oxide double layer. For comparison, we also fabricated TFTs with a 150 nm LPCVD oxide layer. The ICP oxidation was performed at 450 °C for 90 min with oxygen gas. The rf power and the oxygen pressure were 2 kW and 40 mTorr, respectively. LPCVD oxides were deposited at 380 °C with SiH4 and O2 gases. Thermally oxidized Si wafers were used as a substrate. For gate electrodes, 300-nm-a-Si films were deposited by PECVD at 300 °C with SiH4 gas, and then crystallized at 600 °C for 48 h. Gate, source, and drain regions were doped by an ion shower method with a He-diluted PH3 gas, at 6 kV of acceleration voltage and 250 °C substrate temperature. No activation was conducted; 500 nm interlayer oxides were deposited by LPCVD at 380 °C with SiH4 and O2 gases. After opening the contact hole, 1-mm-Al films were deposited and patterned. Finally, the samples were sintered at 450 °C for 30 min in 10%H2 /N2 gas. No plasma hydrogenation was performed. Figure 1 shows the EPR spectra of the poly-Si films before, after the ICP oxidation, and after the ICP oxidation and annealing at 600 °C for 40 h in Ar ambient. The g factor
0003-6951/99/74(18)/2693/3/$15.00 2693 © 1999 American Institute of Physics Downloaded 09 Jul 2001 to 143.248.115.51. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp
2694
Appl. Phys. Lett., Vol. 74, No. 18, 3 May 1999
FIG. 1. EPR spectra of the poly-Si films before, after the ICP oxidation at 400 °C for 30 min with oxygen gas, and after the ICP oxidation and annealing at 600 °C for 40 h in Ar ambient.
of the films was 2.0072, which means that the peak is due to the Si dangling bonds.12 The intensity was reduced after the ICP oxidation. This means that the dangling bond density in the poly-Si films was reduced after the ICP oxidation12 but the EPR intensity after the ICP oxidation and annealing at 600 °C for 40 h in Ar ambient was similar to that prior to the ICP oxidation, which means that the dangling bond density was increased after the annealing. Figure 2 shows the SIMS depth profiles of oxygen and hydrogen in the poly-Si films. The hydrogen content increased, even though the oxidation environment was an oxygen plasma state. On the other hand, the oxygen content did not increase. In the infrared spectra, not given here, no oxygen absorption peak at 1106 cm21 was observed. Thus, the reduction of the dangling bond density was due to the hydrogen incorporation, not due to the oxygen incorporation. Nickel, Yin, and Fornash13 also observed an increase of hydrogen content in poly-Si films after an ECR oxygen plasma treatment. However, Lee et al.9 observed an incorporation of oxygen after ECR plasma oxidation. The
Choi, Park, and Ahn
FIG. 3. log ID vs V G curves of the poly-Si TFTs with an ICP oxide/LPCVD oxide double layer and a LPCVD oxide layer. The inset shows the log@ID /(VG2VFB)•V D # vs 1/(V G 2V FB) 2 plot.
diffusion of oxygen at 400 °C in Si films is too slow to incorporate into the poly-Si films.13 In the ECR plasma oxidation, the incorporation of oxygen may be due to the electric field across the poly-Si films. The electric field reduces the potential barrier, and also enhances the diffusion of oxygen,14 but in our case, the substrate was electrically floated so that no electric field was applied and no oxygen was incorporated. Note that the hydrogen content after the ICP oxidation and annealing at 600 °C for 40 h in Ar ambient was similar to that before the ICP oxidation. This indicates that the incorporated hydrogen was diffused out and the increase of the dangling bond density after the annealing as shown in Fig. 1 could be due to the decrease of the hydrogen content. Figure 3 shows the log drain current (I D ) versus the gate voltage (V G ) curves of the poly-Si TFTs, and the inset is the log$ID /(VG2VFB)•V D % vs 1/(V G 2V FB) 2 plot, where V FB is the flat band voltage and V D is the drain voltage. The device parameters are summarized in Table I. The field effect mobility ( m FE) of the ICP/LPCVD TFT is 30.6 cm2/V s, while that of the LPCVD TFT is 17.2 cm2/V s. The threshold voltages (V ths) of the ICP/LPCVD TFT and the LPCVD TFT are 5.8 and 6.6 V, respectively. The device characteristics of the ICP/LPCVD TFT are better than those of the LPCVD TFT. Especially, the m FE is greatly improved by the ICP oxidation. According to the thermionic emission model which considers the channel thickness dependence on gate voltage (V G ), the drain current (I D ) can be expressed as the following equation:15 TABLE I. The device parameters of the TFTs. The threshold voltage was defined as a gate voltage for a 1003(W/L) nA of drain current with a 0.1 V of drain voltage. Field effect mobility ( m FE) was calculated from the measured maximum transconductance ( ] I D / ] V G ) in the linear region with V D 50.1 V. Off current (I off) was defined as a minimum drain current at V D 55 V. (W/L530 m m/30 m m) Parameter
ICP/LPCVD TFT
LPCVD TFT
Threshold voltage (V th) Field effect mobility ( m FE) Off current (I off) Grain boundary trap density (N T ) Preexponential factor of mobility ( m 0 )
5.8 V 30.6 cm2/V s 0.30 pA/mm 7.0131012/cm2 24.3 cm2/V s
6.6 V 17.2 cm2/V s 0.38 pA/mm 5.9231012/cm2 12.0 cm2/V s
FIG. 2. SIMS depth profiles of hydrogen and oxygen in the poly-Si films before, after the ICP oxidation at 400 °C for 30 min with oxygen gas, and after the ICP oxidation and annealing at 600 °C for 40 h in Ar ambient. The depth profile after hydrogenation by capacitively coupled plasma at 200 °C for 5 h was also presented for comparison. Downloaded 09 Jul 2001 to 143.248.115.51. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp
Appl. Phys. Lett., Vol. 74, No. 18, 3 May 1999
Choi, Park, and Ahn
2695
FIG. 4. TEM bright field images of the poly-Si films as ~a! no-treated, ~b! ICP oxidized at 450 °C for 90 min, and ~c! thermally oxidized at 900 °C for 1 min.
I D 5 ~ W/L ! C ox~ V G 2V FB! V D m 0 exp~ 2E a /kT ! ,
~1!
E a 5 @ q 3 t 2oxA~ e Si / e SiO2! / ~ 8• e Si•C ox!# 3 @ N 2T / ~ V G 2V FB! 2 # ,
~2!
W and L are the channel width and length, C ox is the gate oxide capacitance, m 0 is the preexponential factor of mobility, E a is the activation energy, q is the electronic charge, e Si and e SiO2 are the permittivities of Si and SiO2, respectively, N T is the grain boundary trap density. The N T and m 0 were calculated from the slope and the y-axis intercept of log$ID /(VG2VFB)•V D % vs 1/(V G 2V FB) 2 plot ~Fig. 3!, and also summarized in Table I. The N T values of the ICP/ LPCVD TFT and the LPCVD TFT are 7.0131012 and 5.92 31012/cm2, respectively, the m 0 values are 24.3 and 12.0 cm2/V s, respectively, the N T of the ICP/LPCVD TFT is slightly larger than that of the LPCVD TFT, and the m 0 of the ICP/LPCVD TFT is about 2 times as high as that of the LPCVD TFT. Therefore, the improvement of TFT performance was not due to the passivation of dangling bonds, but due to the increase of the preexponential factor of mobility. The reason why the passivation effect was not obtained is that the hydrogen in the poly-Si films diffused out when the device annealed at 600 °C to crystallize the a-Si films for gate electrode formation as shown in Fig. 2. It could also be confirmed by the increase of the EPR intensity after annealing at 600 °C for 40 h in Ar ambient as shown in Fig. 1 that m 0 is closely related to the defects of the grain, such as the intragranular defects and the interface traps.16 Figure 4 shows the TEM bright field images of the poly-Si films as ~a! no-treated, ~b! ICP oxidized at 450 °C for 90 min, and ~c! thermally oxidized at 900 °C for 1 min in a rapid thermal oxidation ~RTO! system. The zone axis is @110# and the stripes are microtwins and/or stacking faults parallel to $111% plane.17 The ICP oxidation did not reduce the intragranular defects such as microtwins and stacking faults, while the thermal oxidation at 900 °C for 1 min reduced them. During thermal oxidation, Si atoms at the Si/SiO2 interface are pushed into the poly-Si films and these excess Si atoms remove the microtwins and the stacking faults.18 Although ex-
pected to create the excess Si atoms, the ICP oxidation did not seem to remove the microtwins and the stacking faults. The oxidation temperature might be too low to remove these defects. Therefore, the increase of m 0 might be due to the reduction of interface trap density and the increase of m FE might be due to the decrease of interface trap density. In summary, ICP oxidation passivated the dangling bonds in the poly-Si films by hydrogen incorporation, not by oxygen incorporation. However, the incorporated hydrogen diffused out during the TFT fabrication so that the effect of the dangling bond passivation was not obtained in the poly-Si TFT, but the ICP oxidation reduced the interface trap density and also improved the performance of the poly-Si TFT. 1
Y. W. Choi, J. N. Lee, T. W. Jang, and B. T. Ahn, IEEE Electron Device Lett. 20, 2 ~1999!. 2 C. A. Dimitriadis, P. A. Coxon, A. J. Lowe, J. Atoememos, and N. A. Zeconomou, IEEE Electron Device Lett. 12, 676 ~1991!. 3 T. Y. Huang, I. W. Wu, A. G. Lewis, A. Ching, and R. H. Bruce, IEEE Electron Device Lett. 11, 541 ~1990!. 4 J. Batey and E. Tierney, J. Appl. Phys. 60, 3136 ~1986!. 5 M. Suyama, A. Okamoto, and T. Serikawa, IEEE Trans. Electron Devices 34, 2124 ~1987!. 6 D. A. Carl, D. W. Hess, and M. A. Lieberman, J. Vac. Sci. Technol. A 8, 2924 ~1990!. 7 J. Y. Lee, C. H. Han, and C. K. Kim, IEEE Electron Device Lett. 15, 301 ~1994!. 8 N. I. Lee, J. W. Lee, S. H. Hur, H. S. Kim, and C. H. Han, IEEE Electron Device Lett. 18, 486 ~1997!. 9 J. Y. Lee, C. H. Han, C. K. Kim, and B. K. Kim, Appl. Phys. Lett. 67, 1880 ~1995!. 10 J. Y. Lee, C. H. Han, and C. K. Kim, IEEE Electron Device Lett. 17, 169 ~1996!. 11 M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Discharges and Materials Processing ~Wiley, New York, 1994!, pp. 19–22. 12 S. Veprek, Z. Igbal, R. O. Kuhne, P. Capezzuto, F. A. Sarott, and J. K. Gimzewsli, J. Phys. C 16, 6241 ~1983!. 13 N. H. Nickel, A. Yin, and S. J. Fornash, Appl. Phys. Lett. 65, 3099 ~1994!. 14 E. J. W. Verwey, Physica 2, 1059 ~1935!. 15 R. E. Proano and D. G. Ast, J. Appl. Phys. 66, 2189 ~1989!. 16 T. Katoh and N. Hirashita, Jpn. J. Appl. Phys., Part 2 28, L2291 ~1989!. 17 J. H. Kim, J. Y. Lee, and K. S. Nam, J. Appl. Phys. 77, 95 ~1995!. 18 Y. Fukushima, T. Ueda, and H. Komiya, Extended Abstracts of the 1993 International Conference on Solid State Devices and Materials ~The Japan Society of Applied Physics, Makuhara, 1993!, pp. 993–995.
Downloaded 09 Jul 2001 to 143.248.115.51. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp
