- Email: [email protected]
The effect of thermal annealing sequence on amorphous InGaZnO thin film transistor with a plasma-treated source–drain structure
Thin Solid Films 517 (2009) 6349–6352
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
The effect of thermal annealing sequence on amorphous InGaZnO thin film transistor with a plasma-treated source–drain structure Hyun Soo Shin a, Byung Du Ahn a, Kyung Ho Kim a, Jin-Seong Park b, Hyun Jae Kim a,⁎ a b
School of Electrical and Electronic Engineering, Yonsei University, 262, Seongsanno, Seodaemoon-ku, 120-749, Seoul, Republic of Korea Corporate R&D Center, Samsung SDI Co., Ltd, 428-5, Gongse-Dong, Kiheung-Gu, Yongin-Si, Gyeonggi-Do 449-902, Republic of Korea
a r t i c l e
i n f o
Available online 9 March 2009 Keywords: a-IGZO Thermal annealing Plasma treatment Sheet resistance
a b s t r a c t In this paper, the effects of thermal annealing and the plasma treatment sequence on the performance of amorphous InGaZnO (a-IGZO) thin-film transistors (TFTs) without conventional source/drain (S/D) layer deposition were investigated. We fabricated TFTs using two different processes, one where S/Ds were plasma-treated after thermal annealing, the second where the S/Ds were plasma-treated before annealing. The performance of the former exhibited a linear mobility of 4.97 cm2/V s, an on/off ratio of 4.6 × 106, a Vth of 2.56 V, and a subthreshold slope of 0.65 V/decade. However, the TFT parameters of the latter sample were reduced to a linear mobility of 0.07 cm2/V s, an on/off ratio of 1.5 × 105, a Vth of 2.33 V, and a subthreshold slope of 3.54 V/decade. It was shown that the sheet resistance of plasma-treated S/D areas increased after thermal annealing by about three orders of magnitude. As a result, the increase of the sheet resistance caused a decrease of TFT performance. © 2009 Elsevier B.V. Al rights reserved.
1. Introduction Amorphous Si (a-Si) TFTs have been widely used as a backplane in active matrix liquid crystal display (AMLCD) [1–3], because a-Si TFTs can be fabricated with lower manufacturing cost compared with the conventional low temperature polycrystalline silicon (LTPS) [4,5]. However, for active matrix organic light emission diode (AMOLED), LTPS TFTs have been used as switching and driving TFTs, because LTPS TFTs have better stability performance than a-Si TFTs [5]. However, LTPS TFTs require more mask process steps, at least 7 masks, to produce a backplane and are more costly. Although AMOLED displays have a better image quality than AMLCD, the increase of manufacturing costs resulting from the use of LTPS TFTs is an obstacle for AMOLED to replace AMLCD in the display industry. For this reason, much recent research has focused on oxide semiconductors such as ZnO [6,7], In2O3 [8], InGaO (IGO) [9], ZnSnO (ZTO) [10], InZnO (IZO) [11], InGaZnO (IGZO) [12–15] as channel layers. These studies have shown that oxide TFTs can substitute for LTPS TFTs in oxide semiconductors as a channel layer for AMOLED driving devices. In fact, since Hosono et al. presented high performance TFTs with amorphous IGZO (a-IGZO) semiconductors, many studies have shown that oxide TFTs are the best candidates for AMOLED, since these oxide TFTs have shown a high mobility of N10 cm2/V s and an excellent subthreshold gate swing of 0.20 V/decade even in the amorphous phase [15]. Subsequently, many researchers have focused on improving the electrical performance of a-IGZO TFTs [16–19]. They have shown that many fabrication process ⁎ Corresponding author. E-mail address: [email protected] (H.J. Kim). 0040-6090/$ – see front matter © 2009 Elsevier B.V. Al rights reserved. doi:10.1016/j.tsf.2009.02.071
factors can affect TFT performance, such as the oxygen partial pressure [16,17], the sputtering power [16], the post-annealing condition [13,16,17], the composition of each element [12], and the contact material for source and drain (S/D) metals [18–20], among others. The plasma treatment procedure was adopted to gain higher TFT performance before forming S/ D metal [21], but post-annealing to generate carriers in the channel layer, could cause unintentional degradation of TFT performance. Until now, however, most research groups have focused on the conventional S/D structure, and thus TFTs have been fabricated using plasma treatment on an S/D area to form low resistivity of the S/D without S/D metal, which were firstly realized and investigated in our group [22,23], as shown Fig. 1 (b). The main difference of our structure compared to the conventional S/ D structure is the absence of additional S/D metal. After fabrication of TFTs, the effects of thermal annealing sequence on a-IGZO TFTs were investigated. 2. Experimental details Amorphous IGZO TFTs were fabricated using an inverted staggered test structure. As illustrated in Fig. 1(a) and (b), a 200-nm-thick layer of MoW as a gate metal was deposited using DC magnetron sputtering on a glass substrate with A 300-nm-thick SiO2/100-nm-thick SiNx bilayer as a buffer, and patterned via photolithography. A 200-nmthick SiNx thin film was formed on the patterned gate metal with plasma enhanced chemical vapor deposition (PECVD) at 330 °C as a gate insulator (G/I) layer. The 50-nm-thick a-IGZO channel layer (aIGZO target; In2O3:Ga2O3:ZnO = 1:1:1 mol%) was deposited via rf magnetron sputtering at room temperature and patterned. Amorphous IGZO channel deposition was performed using a circular 8 in
6350
H.S. Shin et al. / Thin Solid Films 517 (2009) 6349–6352
Fig. 1. (a) Optical microscope image of plasma-treated S/D structure, (b) schematic diagram of cross-sectional view, (c) the TFT fabrication process of the plasma-treated after thermal annealing (PTaTA) sample, and (d) the TFT fabrication process of the plasma-treated before thermal annealing (PTbTA) sample.
target and sputtered at a pressure of 1 mTorr at a total gas mixing ratio of Ar/O2 = 65/35 and a deposition rf power of 450 W. Only the S/D area was made accessible to plasma treatment using photolithography, while the channel area was covered with photoresist (PR) to protect plasma damage during the S/D plasma treatment. As a result, the PR protected area was determined as the channel size (W / L = 1000 / 100 µm) for extracting TFT parameters as shown in Fig. 1 (a). Two kinds of samples were fabricated separately as indicated in Fig. 1(c) and (d). Fig. 1(c) shows the fabrication process of the plasma treatment of S/D before thermal annealing (PTbTA), and Fig. 1(d) shows the fabrication process of performing the thermal annealing process in advance of the plasma treatment (PTaTA). Thermal annealing was carried out at 350 °C for 1 h in an N2 furnace for both samples. The plasma treatment used only Ar plasma in a reative ion etcher with a power of 150 W at a working pressure of 100 mTorr for 120 s. PR was stripped using commercial stripper at 70 °C for 1 min. Electrical characteristics of TFTs were measured and X-ray photo spectroscopy (XPS) were performed to determine the surface composition of samples. Hall measurement was used for obtaining the carrier concentrations of the samples.
exhibits high In concentration on only IGZO surface as well as uniform In, Ga, and Zn content ratios in IGZO bulk. On the other hand, the sheet resistance of the Ar plasma-treated a-IGZO film before thermal annealing (PTbTA) exceeded the measurement limits of the measuring instrument, and it was remarkably larger than that of Ar plasmatreated a-IGZO film after thermal annealing (PTaTA), as indicated in Table 1. Accordingly, we think that the tremendous increase of the electron concentration in the a-IGZO films exposed to the Ar plasma originated from oxygen deficiency rather than a change in the cation composition. Oxygen peaks of XPS data are shown in Fig. 2, and there is the evidence of oxygen deficiency variation. In an earlier report, it was stated that the binding energy of oxygen 1s (O 1s) located at near
3. Results and discussion To investigate the origin of carriers in Ar plasma-treated a-IGZO thin films, oxygen binding energy in a-IGZO thin film was measured using XPS as shown in Fig. 2. Actually, a few groups have reported that the generation of carrier by Ar plasma results from the deficiency of oxygen [21] and precipitation of In element on surface [24] by using Rutherford backscattering analysis, etc. The damaged layer by Ar plasma was thin (about 3 nm) and had high carrier concentration (over 1020/cm3). It is quite good agreement with our XPS result, which
Fig. 2. X-ray photo spectroscopy data of (a) as-deposited a-IGZO thin film, (b) the PTaTA sample, and (c) the PTbTA sample.
H.S. Shin et al. / Thin Solid Films 517 (2009) 6349–6352 Table 1 The TFT parameters and the electrical properties of the samples. S-factor μFE Vth (V) (V/decade) (cm2/V s)
On/off ratio
Sheet resistance
Carrier concentration
Conventional IZO S/D structure
8.11
1.62
0.64
9.7 × 106
b85 Ω/□
∼ 1014 cm− 3
Plasma treated after thermal annealing (PTaTA)
4.97
2.56
0.65
4.6 × 106
b 2200 Ω/ □
∼1015 cm− 3
Plasma treated before thermal annealing (PTbTA)
0.07
2.33
3.84
1.5 × 105
–
9.5×1019cm−3
6351
treatment, showed the recovery of the oxygen deficiency peak almost to the level of the as-deposited sample. This result led to an increase in the S/D sheet resistance and the degradation of TFT performance. Fig. 3 shows the transfer characteristics of fabricated samples with W / L = 1000 µm / 100 µm. The on-current of the sample PTaTA was higher than that of the sample PTbTA by an order of 103, and this large on-current appeared as an increase of carrier mobility in the sample PTaTA. The extracted parameters are listed in Table 1. The field-effect mobility in the linear region of TFTs at VDS = 0.1 V was extracted using Eq. (1) IDS = CSiNx μ FE W = LðVGS − Vth ÞVDS ;
ð1Þ
where CSiNx is the gate capacitance per unit area. Vth was linearly fitted value in the saturated region at VDS = 10.1 V using Eq. (2)
532.30 eV is related to the loosely bound oxygen on the surface of the ZnO film, and that the low binding energy of the O 1s spectrum at 530.93 eV is related to O2− ions bound to the Zn2+ ion array, while the binding energy at 531.51 eV is associated with oxygen deficiency in the ZnO film [25]. According to Fig. 2(a), the oxygen deficiency-related peak on the XPS data of the as-deposited sample was smaller than the other two peaks, but for the PTaTA sample, as shown in Fig. 2(b), the oxygen deficiency-related peak increased significantly. However, the other peaks were reduced. These data are consistent with TFT parameters. In contrast, the XPS data of the sample PTbTA in Fig. 2(c), which was added by the thermal annealing process after plasma
The linear mobility of the PTaTA sample was 4.97 cm2/V s, which is smaller than the mobility of the conventional IZO S/D structure of 8.11 cm2/V s, but is much higher than that of the PTbTA sample, which was 0.07 cm2/V s. The reason for the small mobility of the PTbTA sample was due to its very high S/D sheet resistance. As indicated in Table 1, the sheet resistance of the PTbTA sample, which was much larger than that of the PTaTA sample, was out of measurement range and is comparable with the value of the as-deposited a-IGZO thin film. Moreover, the
Fig. 3. The transfer characteristics of (a) the conventional IZO S/D structure, (b) the PTaTA sample, and (c) the PTbTA sample.
Fig. 4. The output characteristics of (a) the conventional IZO S/D structure, (b) the PTaTA sample, and (c) the PTbTA sample.
pffiffiffiffiffiffi IDS = CSiNx μ FE W = 2LðVGS − Vth Þ:
ð2Þ
6352
H.S. Shin et al. / Thin Solid Films 517 (2009) 6349–6352
carrier concentration of the PTbTA sample showed ∼9.5 × 1019 cm− 3 but that of PTaTA showed ∼1015 cm− 3, which is recovered almost to the value of the as-deposited sample and these values correspond to the sheet resistance. The threshold voltages of the PTbTA sample and the PTaTA sample were obtained 2.33 V and 2.56 V respectively. These values were positively shifted about 1 V compared to that of the conventional TFT. The previous research reported by Barquinha et al. showed the threshold voltage was negatively shifted, when the contact resistance was improved in the conventional structure [20]. On the other hand, the reason for positive shift of the plasma-treated TFTs is supposed to be originated from the higher sheet resistance of the plasma-treated TFTs compared to that of the conventional structure. As a result, the higher sheet resistance produced positive shift owing to voltage drop at the S/D surface. The transfer curve of the PTbTA sample was significantly suppressed, especially at VD = 0.1 V, which suggests that the sheet resistance of S/D improves the transconductance of TFTs in our structure. The off-current of the PTaTA sample was ∼10− 10 A, which is similar to that of the conventional structure, but the off-current in the PTaTA sample was less than ∼10− 11 A. Furthermore, the subthreshold slope of the PTbTA sample was 0.65 V/decade, identical to the conventional structure. This means that there was no plasma damage on the interface between the channel layer and the G/I during plasma treatment. In contrast, the subthreshold slope of the PTbTA sample was degraded, but the reason for this degradation is still being investigation. Fig. 4 shows the output characteristics of the samples. Output curves of the PTaTA sample showed fully saturated behavior for all VG similar to the conventional structure, but at the same VG voltage, the saturation VD voltage of the drain current of the PTaTA sample was slightly smaller than that of the conventional S/D structure. These saturation characteristics that appeared at a lower VD than that of the conventional S/D structure are believed to be originated from small contact resistances compared to those of the conventional S/D structure, since our suggested structure does not have heterojunctions. In contrast, the PTbTA sample showed suppressed current and non-saturation behavior for VG voltages as high as 20 V. These characteristics originated from the high sheet resistance of the sample, which obstructed carrier injection from the source electrode to the channel and from the channel to the drain. 4. Conclusions
electrical properties of the PTbTA sample was the increase of this sample's S/D sheet resistance after thermal annealing. However, the PTaTA sample showed sufficiently good TFT performance enough to substitute for the conventional S/D structure, which was caused by low resistance with generated carriers during Ar plasma exposure. XPS measurements consistently supported the transfer and output characteristics of the TFTs. Therefore the plasma treatment process should be carried out after the thermal annealing process to prevent the recovery of oxygen vacancies generated during plasma exposure. Acknowledgment This work was supported by the Korea Science and Engineering Foundation (KOSEF, R0A-2007-000-10044-0 (2007)). References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
In this paper, newly structured a-IGZO TFTs were fabricated without conventional S/D metal. Additional S/D metal was substituted with low resistivity Ar plasma-treated a-IGZO thin film. The linear mobility of the PTaTA sample was 4.97 cm2/V s, its on/off ratio was 4.6 × 106, its subthreshold slope was 0.65 V/decade, and its Vth was 2.56 V. The parameters of the PTbTA sample were a mobility of 0.07 cm2/V s, a subthreshold slope of 3.84 V/decade, an on/off ratio of 1.5 × 105, and a Vth of 2.33 V. The reason for the degradation of the
[20] [21] [22] [23] [24] [25]
P.G. LeComber, W.E. Spear, A. Ghaith, Electron. Lett. 15 (1979) 179. T.L. Credelle, Int. Display Res. Conf., 1988, p. 208. W.E. Howard, J. SID 3 (3) (1995) 127. C.W. Kim, Y.B. Park, D.G. Kim, H.S. Kong, S.S. Kim, Proc. 1st Int. Display Manufacturing Conf., 2000. J. J Lih, C.F. Sung, C.H. Li, T.H. Hsiao, H.H. Lee, SID 03 Digest, vol. 18, 2003, p. 1504. R.L. Hoffman, B.J. Norris, J.F. Wager, Appl. Phys. Lett. 82 (2003) 733. E. Fortunato, P. Barquinha, A. Pimentel, A. Gonçalves, A. Marques, R. Martins, L. Pereira, Appl. Phys. Lett. 85 (2004) 2541. G. Lavareda, C. Nunes de Carvalho, E. Fortunato, A.R. Ramos, E. Alves, O. Conde, A. Amaral, J. Non-Cryst. Solids 352 (2006) 2311. R.E. Presley, D. Hong, H.Q. Chiang, C.M. Hung, R.L. Hoffman, J.F. Wager, Solid-State Electron. 50 (2006) 500. H.Q. Chianga), J.F. Wager, R.L. Hoffman, J. Jeong, D.A. Keszler, Appl. Phys. Lett. 86 (2005) 013503. E. Fortunato, P. Barquinha, A. Pimentel, L. Pereira, G. Gonçalves, R. Martins, Phys. Status Solidi 1 (1) (2007) R34 (RRL). T. Iwasaki, N. Itagaki, T. Den, H. Kumomi, K. Nomura, T. Kamiya, H. Hosono, Appl. Phys. Lett. 90 (2007) 242114. H. Hosono, K. Nomura, Y. Ogo, T. Uruga, T. Kamiya, J. Non-Cryst. Solids 354 (2008) 2796. M.K. Kim, J.H. Jeong, H.J. Lee, T.K. Ahn, H.S. Shin, J.S. Park, J.K. Jeong, Y.G. Mo, H.D. Kim, Appl. Phys. Lett. 90 (2007) 212114. K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, H. Hosono, Nature 432 (2004) 288. H.Q. Chiang, B.R. McFarlane, D. Hong, R.E. Presley, J.F. Wager, J. Non-Cryst. Solids 354 (2008) 2826. A. Suresh, P. Gollakota, P. Wellenius, A. Dhawan, J.F. Muth, Thin Solid Films 516 (2008) 1326. Y. Shimura, K. Nomura, H. Yanagi, T. Kamiya, M. Hirano, H. Hosono, Thin Solid Films 516 (2007) 5899. M.K. Kim, J.S. Park, J.K. Jeong, J.H. Jeong, T.K. Ahn, H.W. Yang, H.J. Lee, H.J. Chung, Y.G. Mo, H.D. Kim, IMID 07 Digest, 2007, p. 640. P. Barquinha, A. Vilà, G. Gonçalves, L. Pereira, R. Martins, J. Morante, E. Fortunato, IEEE Trans. Electron Devices 55 (2008) 954. J.S. Park, J.K. Jeong, Y.G. Mo, H.D. Kim, S.I. Kim, Appl. Phys. Lett. 90 (2007) 262106. B.D. Ahn, H.S. Shin, H.J. Kim, J.S. Park, J.K. Jeong, Appl. Phys. Lett. 93 (2008) 203506. B.D. Ahn, H.S. Shin, G.H. Kim, J.S. Park, H.J. Kim, Jpn. J. Appl. Phys. (in press). J.C. Park, I.H. Song, S.I. Kim, S.W. Kim, C.J. Kim, J.C. Lee, H.I. Lee, E.H. Lee, H. Yin, K.K. Kim, K.W. Kwon, Y.S. Park, Appl. Phys. Lett. 93 (2008) 053501. S.M. Park, T. Ikegami, K. Ebihara, Thin Solid Films 513 (2006) 90.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
The effect of thermal annealing sequence on amorphous InGaZnO thin film transistor with a plasma-treated source–drain structure Hyun Soo Shin a, Byung Du Ahn a, Kyung Ho Kim a, Jin-Seong Park b, Hyun Jae Kim a,⁎ a b
School of Electrical and Electronic Engineering, Yonsei University, 262, Seongsanno, Seodaemoon-ku, 120-749, Seoul, Republic of Korea Corporate R&D Center, Samsung SDI Co., Ltd, 428-5, Gongse-Dong, Kiheung-Gu, Yongin-Si, Gyeonggi-Do 449-902, Republic of Korea
a r t i c l e
i n f o
Available online 9 March 2009 Keywords: a-IGZO Thermal annealing Plasma treatment Sheet resistance
a b s t r a c t In this paper, the effects of thermal annealing and the plasma treatment sequence on the performance of amorphous InGaZnO (a-IGZO) thin-film transistors (TFTs) without conventional source/drain (S/D) layer deposition were investigated. We fabricated TFTs using two different processes, one where S/Ds were plasma-treated after thermal annealing, the second where the S/Ds were plasma-treated before annealing. The performance of the former exhibited a linear mobility of 4.97 cm2/V s, an on/off ratio of 4.6 × 106, a Vth of 2.56 V, and a subthreshold slope of 0.65 V/decade. However, the TFT parameters of the latter sample were reduced to a linear mobility of 0.07 cm2/V s, an on/off ratio of 1.5 × 105, a Vth of 2.33 V, and a subthreshold slope of 3.54 V/decade. It was shown that the sheet resistance of plasma-treated S/D areas increased after thermal annealing by about three orders of magnitude. As a result, the increase of the sheet resistance caused a decrease of TFT performance. © 2009 Elsevier B.V. Al rights reserved.
1. Introduction Amorphous Si (a-Si) TFTs have been widely used as a backplane in active matrix liquid crystal display (AMLCD) [1–3], because a-Si TFTs can be fabricated with lower manufacturing cost compared with the conventional low temperature polycrystalline silicon (LTPS) [4,5]. However, for active matrix organic light emission diode (AMOLED), LTPS TFTs have been used as switching and driving TFTs, because LTPS TFTs have better stability performance than a-Si TFTs [5]. However, LTPS TFTs require more mask process steps, at least 7 masks, to produce a backplane and are more costly. Although AMOLED displays have a better image quality than AMLCD, the increase of manufacturing costs resulting from the use of LTPS TFTs is an obstacle for AMOLED to replace AMLCD in the display industry. For this reason, much recent research has focused on oxide semiconductors such as ZnO [6,7], In2O3 [8], InGaO (IGO) [9], ZnSnO (ZTO) [10], InZnO (IZO) [11], InGaZnO (IGZO) [12–15] as channel layers. These studies have shown that oxide TFTs can substitute for LTPS TFTs in oxide semiconductors as a channel layer for AMOLED driving devices. In fact, since Hosono et al. presented high performance TFTs with amorphous IGZO (a-IGZO) semiconductors, many studies have shown that oxide TFTs are the best candidates for AMOLED, since these oxide TFTs have shown a high mobility of N10 cm2/V s and an excellent subthreshold gate swing of 0.20 V/decade even in the amorphous phase [15]. Subsequently, many researchers have focused on improving the electrical performance of a-IGZO TFTs [16–19]. They have shown that many fabrication process ⁎ Corresponding author. E-mail address: [email protected] (H.J. Kim). 0040-6090/$ – see front matter © 2009 Elsevier B.V. Al rights reserved. doi:10.1016/j.tsf.2009.02.071
factors can affect TFT performance, such as the oxygen partial pressure [16,17], the sputtering power [16], the post-annealing condition [13,16,17], the composition of each element [12], and the contact material for source and drain (S/D) metals [18–20], among others. The plasma treatment procedure was adopted to gain higher TFT performance before forming S/ D metal [21], but post-annealing to generate carriers in the channel layer, could cause unintentional degradation of TFT performance. Until now, however, most research groups have focused on the conventional S/D structure, and thus TFTs have been fabricated using plasma treatment on an S/D area to form low resistivity of the S/D without S/D metal, which were firstly realized and investigated in our group [22,23], as shown Fig. 1 (b). The main difference of our structure compared to the conventional S/ D structure is the absence of additional S/D metal. After fabrication of TFTs, the effects of thermal annealing sequence on a-IGZO TFTs were investigated. 2. Experimental details Amorphous IGZO TFTs were fabricated using an inverted staggered test structure. As illustrated in Fig. 1(a) and (b), a 200-nm-thick layer of MoW as a gate metal was deposited using DC magnetron sputtering on a glass substrate with A 300-nm-thick SiO2/100-nm-thick SiNx bilayer as a buffer, and patterned via photolithography. A 200-nmthick SiNx thin film was formed on the patterned gate metal with plasma enhanced chemical vapor deposition (PECVD) at 330 °C as a gate insulator (G/I) layer. The 50-nm-thick a-IGZO channel layer (aIGZO target; In2O3:Ga2O3:ZnO = 1:1:1 mol%) was deposited via rf magnetron sputtering at room temperature and patterned. Amorphous IGZO channel deposition was performed using a circular 8 in
6350
H.S. Shin et al. / Thin Solid Films 517 (2009) 6349–6352
Fig. 1. (a) Optical microscope image of plasma-treated S/D structure, (b) schematic diagram of cross-sectional view, (c) the TFT fabrication process of the plasma-treated after thermal annealing (PTaTA) sample, and (d) the TFT fabrication process of the plasma-treated before thermal annealing (PTbTA) sample.
target and sputtered at a pressure of 1 mTorr at a total gas mixing ratio of Ar/O2 = 65/35 and a deposition rf power of 450 W. Only the S/D area was made accessible to plasma treatment using photolithography, while the channel area was covered with photoresist (PR) to protect plasma damage during the S/D plasma treatment. As a result, the PR protected area was determined as the channel size (W / L = 1000 / 100 µm) for extracting TFT parameters as shown in Fig. 1 (a). Two kinds of samples were fabricated separately as indicated in Fig. 1(c) and (d). Fig. 1(c) shows the fabrication process of the plasma treatment of S/D before thermal annealing (PTbTA), and Fig. 1(d) shows the fabrication process of performing the thermal annealing process in advance of the plasma treatment (PTaTA). Thermal annealing was carried out at 350 °C for 1 h in an N2 furnace for both samples. The plasma treatment used only Ar plasma in a reative ion etcher with a power of 150 W at a working pressure of 100 mTorr for 120 s. PR was stripped using commercial stripper at 70 °C for 1 min. Electrical characteristics of TFTs were measured and X-ray photo spectroscopy (XPS) were performed to determine the surface composition of samples. Hall measurement was used for obtaining the carrier concentrations of the samples.
exhibits high In concentration on only IGZO surface as well as uniform In, Ga, and Zn content ratios in IGZO bulk. On the other hand, the sheet resistance of the Ar plasma-treated a-IGZO film before thermal annealing (PTbTA) exceeded the measurement limits of the measuring instrument, and it was remarkably larger than that of Ar plasmatreated a-IGZO film after thermal annealing (PTaTA), as indicated in Table 1. Accordingly, we think that the tremendous increase of the electron concentration in the a-IGZO films exposed to the Ar plasma originated from oxygen deficiency rather than a change in the cation composition. Oxygen peaks of XPS data are shown in Fig. 2, and there is the evidence of oxygen deficiency variation. In an earlier report, it was stated that the binding energy of oxygen 1s (O 1s) located at near
3. Results and discussion To investigate the origin of carriers in Ar plasma-treated a-IGZO thin films, oxygen binding energy in a-IGZO thin film was measured using XPS as shown in Fig. 2. Actually, a few groups have reported that the generation of carrier by Ar plasma results from the deficiency of oxygen [21] and precipitation of In element on surface [24] by using Rutherford backscattering analysis, etc. The damaged layer by Ar plasma was thin (about 3 nm) and had high carrier concentration (over 1020/cm3). It is quite good agreement with our XPS result, which
Fig. 2. X-ray photo spectroscopy data of (a) as-deposited a-IGZO thin film, (b) the PTaTA sample, and (c) the PTbTA sample.
H.S. Shin et al. / Thin Solid Films 517 (2009) 6349–6352 Table 1 The TFT parameters and the electrical properties of the samples. S-factor μFE Vth (V) (V/decade) (cm2/V s)
On/off ratio
Sheet resistance
Carrier concentration
Conventional IZO S/D structure
8.11
1.62
0.64
9.7 × 106
b85 Ω/□
∼ 1014 cm− 3
Plasma treated after thermal annealing (PTaTA)
4.97
2.56
0.65
4.6 × 106
b 2200 Ω/ □
∼1015 cm− 3
Plasma treated before thermal annealing (PTbTA)
0.07
2.33
3.84
1.5 × 105
–
9.5×1019cm−3
6351
treatment, showed the recovery of the oxygen deficiency peak almost to the level of the as-deposited sample. This result led to an increase in the S/D sheet resistance and the degradation of TFT performance. Fig. 3 shows the transfer characteristics of fabricated samples with W / L = 1000 µm / 100 µm. The on-current of the sample PTaTA was higher than that of the sample PTbTA by an order of 103, and this large on-current appeared as an increase of carrier mobility in the sample PTaTA. The extracted parameters are listed in Table 1. The field-effect mobility in the linear region of TFTs at VDS = 0.1 V was extracted using Eq. (1) IDS = CSiNx μ FE W = LðVGS − Vth ÞVDS ;
ð1Þ
where CSiNx is the gate capacitance per unit area. Vth was linearly fitted value in the saturated region at VDS = 10.1 V using Eq. (2)
532.30 eV is related to the loosely bound oxygen on the surface of the ZnO film, and that the low binding energy of the O 1s spectrum at 530.93 eV is related to O2− ions bound to the Zn2+ ion array, while the binding energy at 531.51 eV is associated with oxygen deficiency in the ZnO film [25]. According to Fig. 2(a), the oxygen deficiency-related peak on the XPS data of the as-deposited sample was smaller than the other two peaks, but for the PTaTA sample, as shown in Fig. 2(b), the oxygen deficiency-related peak increased significantly. However, the other peaks were reduced. These data are consistent with TFT parameters. In contrast, the XPS data of the sample PTbTA in Fig. 2(c), which was added by the thermal annealing process after plasma
The linear mobility of the PTaTA sample was 4.97 cm2/V s, which is smaller than the mobility of the conventional IZO S/D structure of 8.11 cm2/V s, but is much higher than that of the PTbTA sample, which was 0.07 cm2/V s. The reason for the small mobility of the PTbTA sample was due to its very high S/D sheet resistance. As indicated in Table 1, the sheet resistance of the PTbTA sample, which was much larger than that of the PTaTA sample, was out of measurement range and is comparable with the value of the as-deposited a-IGZO thin film. Moreover, the
Fig. 3. The transfer characteristics of (a) the conventional IZO S/D structure, (b) the PTaTA sample, and (c) the PTbTA sample.
Fig. 4. The output characteristics of (a) the conventional IZO S/D structure, (b) the PTaTA sample, and (c) the PTbTA sample.
pffiffiffiffiffiffi IDS = CSiNx μ FE W = 2LðVGS − Vth Þ:
ð2Þ
6352
H.S. Shin et al. / Thin Solid Films 517 (2009) 6349–6352
carrier concentration of the PTbTA sample showed ∼9.5 × 1019 cm− 3 but that of PTaTA showed ∼1015 cm− 3, which is recovered almost to the value of the as-deposited sample and these values correspond to the sheet resistance. The threshold voltages of the PTbTA sample and the PTaTA sample were obtained 2.33 V and 2.56 V respectively. These values were positively shifted about 1 V compared to that of the conventional TFT. The previous research reported by Barquinha et al. showed the threshold voltage was negatively shifted, when the contact resistance was improved in the conventional structure [20]. On the other hand, the reason for positive shift of the plasma-treated TFTs is supposed to be originated from the higher sheet resistance of the plasma-treated TFTs compared to that of the conventional structure. As a result, the higher sheet resistance produced positive shift owing to voltage drop at the S/D surface. The transfer curve of the PTbTA sample was significantly suppressed, especially at VD = 0.1 V, which suggests that the sheet resistance of S/D improves the transconductance of TFTs in our structure. The off-current of the PTaTA sample was ∼10− 10 A, which is similar to that of the conventional structure, but the off-current in the PTaTA sample was less than ∼10− 11 A. Furthermore, the subthreshold slope of the PTbTA sample was 0.65 V/decade, identical to the conventional structure. This means that there was no plasma damage on the interface between the channel layer and the G/I during plasma treatment. In contrast, the subthreshold slope of the PTbTA sample was degraded, but the reason for this degradation is still being investigation. Fig. 4 shows the output characteristics of the samples. Output curves of the PTaTA sample showed fully saturated behavior for all VG similar to the conventional structure, but at the same VG voltage, the saturation VD voltage of the drain current of the PTaTA sample was slightly smaller than that of the conventional S/D structure. These saturation characteristics that appeared at a lower VD than that of the conventional S/D structure are believed to be originated from small contact resistances compared to those of the conventional S/D structure, since our suggested structure does not have heterojunctions. In contrast, the PTbTA sample showed suppressed current and non-saturation behavior for VG voltages as high as 20 V. These characteristics originated from the high sheet resistance of the sample, which obstructed carrier injection from the source electrode to the channel and from the channel to the drain. 4. Conclusions
electrical properties of the PTbTA sample was the increase of this sample's S/D sheet resistance after thermal annealing. However, the PTaTA sample showed sufficiently good TFT performance enough to substitute for the conventional S/D structure, which was caused by low resistance with generated carriers during Ar plasma exposure. XPS measurements consistently supported the transfer and output characteristics of the TFTs. Therefore the plasma treatment process should be carried out after the thermal annealing process to prevent the recovery of oxygen vacancies generated during plasma exposure. Acknowledgment This work was supported by the Korea Science and Engineering Foundation (KOSEF, R0A-2007-000-10044-0 (2007)). References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
In this paper, newly structured a-IGZO TFTs were fabricated without conventional S/D metal. Additional S/D metal was substituted with low resistivity Ar plasma-treated a-IGZO thin film. The linear mobility of the PTaTA sample was 4.97 cm2/V s, its on/off ratio was 4.6 × 106, its subthreshold slope was 0.65 V/decade, and its Vth was 2.56 V. The parameters of the PTbTA sample were a mobility of 0.07 cm2/V s, a subthreshold slope of 3.84 V/decade, an on/off ratio of 1.5 × 105, and a Vth of 2.33 V. The reason for the degradation of the
[20] [21] [22] [23] [24] [25]
P.G. LeComber, W.E. Spear, A. Ghaith, Electron. Lett. 15 (1979) 179. T.L. Credelle, Int. Display Res. Conf., 1988, p. 208. W.E. Howard, J. SID 3 (3) (1995) 127. C.W. Kim, Y.B. Park, D.G. Kim, H.S. Kong, S.S. Kim, Proc. 1st Int. Display Manufacturing Conf., 2000. J. J Lih, C.F. Sung, C.H. Li, T.H. Hsiao, H.H. Lee, SID 03 Digest, vol. 18, 2003, p. 1504. R.L. Hoffman, B.J. Norris, J.F. Wager, Appl. Phys. Lett. 82 (2003) 733. E. Fortunato, P. Barquinha, A. Pimentel, A. Gonçalves, A. Marques, R. Martins, L. Pereira, Appl. Phys. Lett. 85 (2004) 2541. G. Lavareda, C. Nunes de Carvalho, E. Fortunato, A.R. Ramos, E. Alves, O. Conde, A. Amaral, J. Non-Cryst. Solids 352 (2006) 2311. R.E. Presley, D. Hong, H.Q. Chiang, C.M. Hung, R.L. Hoffman, J.F. Wager, Solid-State Electron. 50 (2006) 500. H.Q. Chianga), J.F. Wager, R.L. Hoffman, J. Jeong, D.A. Keszler, Appl. Phys. Lett. 86 (2005) 013503. E. Fortunato, P. Barquinha, A. Pimentel, L. Pereira, G. Gonçalves, R. Martins, Phys. Status Solidi 1 (1) (2007) R34 (RRL). T. Iwasaki, N. Itagaki, T. Den, H. Kumomi, K. Nomura, T. Kamiya, H. Hosono, Appl. Phys. Lett. 90 (2007) 242114. H. Hosono, K. Nomura, Y. Ogo, T. Uruga, T. Kamiya, J. Non-Cryst. Solids 354 (2008) 2796. M.K. Kim, J.H. Jeong, H.J. Lee, T.K. Ahn, H.S. Shin, J.S. Park, J.K. Jeong, Y.G. Mo, H.D. Kim, Appl. Phys. Lett. 90 (2007) 212114. K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, H. Hosono, Nature 432 (2004) 288. H.Q. Chiang, B.R. McFarlane, D. Hong, R.E. Presley, J.F. Wager, J. Non-Cryst. Solids 354 (2008) 2826. A. Suresh, P. Gollakota, P. Wellenius, A. Dhawan, J.F. Muth, Thin Solid Films 516 (2008) 1326. Y. Shimura, K. Nomura, H. Yanagi, T. Kamiya, M. Hirano, H. Hosono, Thin Solid Films 516 (2007) 5899. M.K. Kim, J.S. Park, J.K. Jeong, J.H. Jeong, T.K. Ahn, H.W. Yang, H.J. Lee, H.J. Chung, Y.G. Mo, H.D. Kim, IMID 07 Digest, 2007, p. 640. P. Barquinha, A. Vilà, G. Gonçalves, L. Pereira, R. Martins, J. Morante, E. Fortunato, IEEE Trans. Electron Devices 55 (2008) 954. J.S. Park, J.K. Jeong, Y.G. Mo, H.D. Kim, S.I. Kim, Appl. Phys. Lett. 90 (2007) 262106. B.D. Ahn, H.S. Shin, H.J. Kim, J.S. Park, J.K. Jeong, Appl. Phys. Lett. 93 (2008) 203506. B.D. Ahn, H.S. Shin, G.H. Kim, J.S. Park, H.J. Kim, Jpn. J. Appl. Phys. (in press). J.C. Park, I.H. Song, S.I. Kim, S.W. Kim, C.J. Kim, J.C. Lee, H.I. Lee, E.H. Lee, H. Yin, K.K. Kim, K.W. Kwon, Y.S. Park, Appl. Phys. Lett. 93 (2008) 053501. S.M. Park, T. Ikegami, K. Ebihara, Thin Solid Films 513 (2006) 90.