- Email: [email protected]
Investigation of oxide layer removal mechanism using reactive gases
Microelectronic Engineering 135 (2015) 17–22
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Investigation of oxide layer removal mechanism using reactive gases Hyun-Tae Kim a, Jung-Soo Lim a, Min-Su Kim a, Hoon-Jung Oh b, Dae-Hong Ko b,c, Gyoo-Dong Kim d, Woo-Gon Shin d, Jin-Goo Park a,⇑ a
Department of Bio-Nano Technology and Material Science Engineering, Hanyang University, Ansan 426-791, Republic of Korea BIO-IT Micro Fab Center, Yonsei University, Seoul 120-749, Republic of Korea c Department of Materials Science and Engineering, Yonsei University, Seoul 120-749, Republic of Korea d GEN, Suwon 443-390, Republic of Korea b
a r t i c l e
i n f o
Article history: Received 30 July 2014 Received in revised form 9 February 2015 Accepted 11 February 2015 Available online 17 February 2015 Keywords: Dry etching Buffered oxide etchant (BOE) NH3/NF3 etching Ellipsometry Plasma enhanced chemical vapor deposition (PECVD)
a b s t r a c t In a CMOS technology, the removal of silicon oxide and nitride layer is one of the critical steps as it represents a possible source of high contact resistance and a decrease of gate oxide reliability. In high aspect ratio (HAR), it is very difficult to remove SiO2 with wet etching. In the present study, the effect of the gases such as plasma dry etching of ammonia (NH3) and nitrogen trifluoride (NF3) on the SiO2 and Si3N4 substrates were analyzed and the etch rate was measured. The measurement of the SiO2 and Si3N4 thickness was measured by Ellipsometer. Various factors such as chamber pressure, electrode power and NH3/NF3 gas ratio were affected by the combination and dissociation of NH4F molecules. The existence of the byproduct was analyzed by using a contact angle analyzer and scanning electron microscope, respectively. In this study we have found that, the removal efficiency was mainly dependent on the reaction mechanism and the effect of the by-product. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction SiO2 and Si3N4 are widely used in micro fabrication processes as a dielectric and mask material. As pattern size continues to decrease, lithography process required accuracy between layers. It is become harder to make contact hole and trenches into dielectric layers. Commonly, SiO2 is the widely used dielectric material and Si3N4 has been used as a passivation layer [1]. In CMOS technology, removal of these films is a critical step as it makes a possible source of high contact resistance and a decrease of gate oxide reliability [2,3]. Possible over etch in the nitride processing may result in damages of a thin oxide and an underlying Si substrate through imperfections of the oxide [3]. The ability to achieve selective etching of SiO2 and Si3N4 is becoming an increasingly important requirement. Silicon nitride is used as a passivation layer that protects circuits from mechanical and chemical attack, or as an etch stop layer, enabling the fabrication of certain damascene and self-aligned contact (SAC) structures. Selective SiO2 and Si3N4 etching have been demonstrated in several systems [4–9]. High aspect ratio (HAR) silicon trench etch is a key process to remove the silicon oxide on the pattern. During the etching process, a chemical reaction between the chemical etchants and ⇑ Corresponding author. Tel.: +82 31 500 5226; fax: +82 31 400 4729. E-mail address: [email protected] (J.-G. Park). http://dx.doi.org/10.1016/j.mee.2015.02.025 0167-9317/Ó 2015 Elsevier B.V. All rights reserved.
the surface layer has been occurred. It is very important to control the various factors affecting the chemical reaction because the etch rate and the surface quality has been changed depending upon this reaction. In general, hydrogen fluoride is used as a chemical for silicon oxide removal [10]. In the SiO2 etching process, a wet process using an HF solution like BOE was employed. The etching rate of wet process was reached several lm/min. But, the usage of chemical solution and DI water rinse process, caused distortion and contamination in patterns [11]. As the decrease in minimum feature size, the removal of the silicon oxide on the pattern becomes difficult, however after the drying process, the etch pattern was collapsed by capillary force of water. Research of dry etching is being carried out to solve these problems. Dry etching is capable of reproducing anisotropic walls and the characteristics are highly reproducible [12]. In this study, the fundamental theory of SiO2 etching of plasma activated NF3/NH3 gas was investigated. Plasma dry etching of ammonia (NH3) and nitrogen trifluoride (NF3) mixtures were employed in the detailed studies. From this process, we could generate ammonium hexafluorosilicate (NH4)2SiF6 as a by-product, which is deposited on the surface and it interrupts the further etching reaction [13–14]. To investigate this problem, the etching of SiO2 using NH3/NF3 reactive gas was analyzed and the etch rate was calculated as a function of NF3 ratio, electrode power and pressure, respectively. From this study we could conclude that the
18
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
removal efficiency mainly depends on the reaction mechanism and the effect of the by-product. 2. Materials and methods A schematic representation of the dry etching reaction chamber is shown in Fig. 1. The NH3/NF3 mixture was excited using pulsed RF plasma (27.12 MHz pulse 3 kW). Helium gas was used as a carrier gas. The temperature was maintained at 35 °C throughout the experiment. The power of plasma was varied from 80 to 160 W, the chamber pressure from 3.75 to 5 Torr, gas ratio of NH3/NF3 from 0.14 to 7.82, and process time from 60 to 180 s, respectively. Thermally grown SiO2 (10000 Å) and low pressure chemical vapor deposition (LPCVD) Si3N4 (1000 Å) were used as a material for the experiments. The size of the sample 20 20 mm was cleaned with dilute SC-1 [NH4OH (25%):H2O2 (38%):DIW = 1:2:50] solution at 60 °C for 10 min before proceeding the dry etching, and then it was loaded into the reaction chamber. After dry etching was over, the annealing process was performed to remove the by-product. The measurement of the SiO2 and Si3N4 thickness was measured by Ellipsometer (M-2000V, J.A. Woollam, USA). The existence of the by-products was analyzed by using a contact angle analyzer (Phoenix, SEO, Korea) and SEM (FE-SEM, MIRA3, TESCAN, Czech), respectively. 3. Results and discussion Fig. 2 shows the schematic of the SiO2 with substrate before proceeding to the etch rate, it was exposed to the plasma NH3/ NF3 gas treatment. During the process by-product such as (NH4)2SiF6 was produced in the chamber, which is deposited on the surface and it was evaporated under the high temperature at 180 °C for 1 min. In this paper, we defined fume as white solid by-product that is easily remove by high temperature treatment or DI water rinse. After the SiO2 were etched without annealing, to find out the chemical composition of the fume, composition of fume was analyzed using Fourier transform infrared spectroscopy (FT-IR). Fig. 3 shows the ATR-FTIR spectrum of etched SiO2 surface and (NH4)2SiF6 powder. Fumes are formed on the SiO2 surface which can be confirmed N–H, NH+4 and SiF2 6 peak. These peaks are representative of (NH4)2SiF6 powder composition. This observation suggested that (NH4)2SiF6 created after NH3/NF3 dry etching process.
The etch rate of SiO2 and Si3N4 was measured as a function of the process time as shown in Fig. 4. The substrates were introduced into the chamber using Helium as a carrier gas with a flow rate of 600 sccm. The etch amount of SiO2 increased gradually with increase of process time. However, the etch rate decreased with an increase in process time. On the other hand, the etch rate of Si3N4 was increased. The results confirm that, the selectivity of SiO2 and Si3N4 was decreased with increase in process time. The maximum selectivity was obtained around 9.3 at 90 s shown in Fig. 5. After excess 90 s., the etch rate of Si3N4 was increased compared with SiO2. Fig. 6 shows the etching behavior of SiO2 in NH3/NF3 mixture of varying ratio. The overall chemical reaction of SiO2 etching involved is normally understood as [15,16]:
SiO2 þ 4HF þ 2NH4 F ! ðNH4 Þ2 SiF6 þ 2H2 O
ð1Þ
The reactions show that mechanism of SiO2 etching in buffered HF (BHF). However, a plasma active NF3/NH3 etching process consists of two steps. Plasma converts NF3 and NH3 to NH4F and NH4FHF (Eq. (2)). These products condense on the SiO2 surface and react with the SiO2 to form solid by-product ((NH4)2SiF6) (Eq. (3)) [16]:
3NH3 þ NF3 ! NH4 F þ NH4 F HF þ N2
ð2Þ
SiO2 þ 6NH4 F ! ðNH4 Þ2 SiF6 þ H2 O þ 4NH3
ð3Þ
The NH4F produced for contributing to the SiO2 etching. At less than 2 (NH3/NF3 ratio) level, the etch rate of SiO2 increases drastically with increasing NH3/NF3 ratio. At greater than 2 (NH3/NF3 ratio) level, the etch rate of SiO2 gradually decreases with increasing NH3/NF3. On the other hand, the etch rate of Si3N4 (40 Å/min) was relatively diminished in above the same condition. The maximum etch rate of SiO2 was 310 Å/min. In order to investigate the effect of recombination and dissociation of etch rate, we altered the chamber pressure and electrode power, respectively. Fig. 7 shows the etch rate of SiO2 as a function of pressure and power. The etch rate was decreased at increasing power. The etch rate of SiO2 was obtained in the range of 330– 240 Å/min at the pressure of (3.75–5 Torr) and the power of (80– 160 W), respectively. At high RF power, the more decomposable gas was produced. However, no reaction gas was observed during the etching. Similarly, the etch rate was decreased as increasing the chamber pressure.
Fig. 1. Schematic of the dry etching reaction chamber.
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
19
Fig. 2. Schematic diagram of the model for removal of SiO2 by plasma NH3/NF3: (a) before etching, (b) plasma NH3/NF3 exposure, (c) (NH4)2SiF6 generation, and (d) annealing.
20
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
Fig. 3. ATR-FTIR spectrum of (a) (NH4)2SiF6 powder and (b) etched SiO2 surface without annealing.
10
Selectivity (SiO2/Si3N4)
9 8 7 6 5 4 3
0
30
60 Time (sec.)
90
Fig. 5. Etch rate of SiO2 and Si3N4 as a function of NH3/NF3 ratio.
Fig. 6. Etch rate of SiO2 and Si3N4 as a function of NH3/NF3 ratio at 35 °C, 4 Torr and plasma power of 70 W.
Fig. 4. Etch rates of SiO2 and Si3N4 as a function of the process time: (a) SiO2, and (b) Si3N4.
Fig. 8 shows the schematic model of fume generation method. To predict the thickness of the fume, a theoretical approach model
21
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
Fig. 9. Etch rate of SiO2, thickness of fume calculated as a function of process time.
interface. In order to, we can calculate the thickness between the SiO2 and (NH4)2SiF6 by the density variation [15]: Molecular weight ðSiO2 Þ Density ðSiO2 Þ Molecular weight ððNH4 Þ2 SiF6 Þ Density ððNH4 Þ2 SiF6 Þ
Fig. 7. Etch rate of SiO2 as a function of (a) pressure at 35 °C and 70 W and (b) power at 35° and 4 Torr when constant flows of 50 sccm of NF3 and 100 sccm of NH3 are supplied.
60:08 2:21 ¼ 178:15 ¼ 0:31 2:01
A stoichiometric relationship was found during the reaction. The difference between the volume variation of SiO2 and (NH4)2SiF6 was around 31%, the SiO2 surface was consumed during the etching process. The consumption of the SiO2 was about 0.31 lm, the fume growth rate was at 1 lm level. Fig. 9 shows the etch rate of SiO2 and thickness of fume was calculated as a function of process time. The fume thickness had a margin of error. The calculated thickness was corresponding with experimental thickness value. Therefore, the thickness of the fume was predicted by the etch rate of SiO2. Fig. 10 shows the SEM images of the by-product. The initial contact angle of the SiO2 surface was 52°. After the etching process and before annealing, the condensation of fume layer was found. This fume formed a white thin layer on the surface. It can be easily soluble in water due to its property; in order to easily detect the difference between by-product and clean surface. If it remains on the surface, it would have become impossible to measure the contact angle because of its hygroscopic property. After removed by high temperature treatment at 180 °C for 1 min, the value of contact angle was turned back to the initial value.
4. Conclusion
Fig. 8. Schematic model of fume generation.
was used to calculate the thickness. The growing fume process involves the chemical reactions, which consume the SiO2 and also produced the fume. The fume was generated at the SiO2 – fume
The effect of the plasma dry etching of ammonia (NH3) and nitrogen trifluoride (NF3) on the SiO2 and Si3N4 substrates were analyzed and the etch rate was measured, which is strongly related to NH4F concentration and not to HF concentration. The NH4F molecules were produced by chemical reaction of NH3 and NF3. Various factors such as chamber pressure, electrode power and NH3/NF3 gas ratio could have been effected by the combination and dissociation of NH4F molecules. This result confirms that it could lead to a change in the etch rate. The maximum etch rate of SiO2 was obtained at NF3 and NH3 for a 2:1 ratio that exhibited the greatest amount of NH4F concentration.
22
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
Fig. 10. SEM and contact angle images of the surface after NH3/NF3 exposure: (a) an initial SiO2 surface, (b) a high density plasma oxide surface after NH3/NF3 exposure, (c) a thermal silicon oxide surface after NH3/NF3 exposure, and (d) silicon oxide surface after annealing at 180 °C, 1 min.
Acknowledgements This work was supported by the IT R&D program of MOTIE (10043438, Development of key cleaning technology for 10 nmsemiconductor and 8th generation display using damage free technology) and by the Future Semiconductor Device Technology Development Program #10045366 funded by MOTIE – South Korea (Ministry of Trade, Industry & Energy) and KSRC – South Korea (Korea Semiconductor Research Consortium). References [1] S.H. Choi, K.J. Jeong, J.S. Choi, T.Y. Cho, H.G. Chun, J. Korean Phys. Soc. 33 (1998) S99. [2] B.E.E. Kastenmeier, P.J. Matsuo, G.S. Oehrlein, J. Vac. Sci. Technol. A 17 (6) (1999) 3179. [3] M. Barklund, H.O. Blom, J. Vac. Sci. Technol. A 11 (1993) 1226.
[4] J. Givens, S. Geissler, J. Lee, O. Cain, J. Marks, P. Keswick, C. Cunningham, J. Vac. Sci. Technol. B 12 (1994) 427. [5] K. Ueno, T. Kikkawa, K. Tokashiki, J. Vac. Sci. Technol. B 13 (1995) 1447. [6] H. Hayashi, K. Kurihara, M. Sekine, Jpn. J. Appl. Phys. Part 1 35 (1996) 2488. [7] S. Fang, C. Chiang, D. Fraser, B. Lee, P. Keswick, M. Chang, K. Fung, J. Vac. Sci. Technol. A 14 (1996) 1092. [8] Y. Zhang, G.S. Oehrlein, F.H. Bell, J. Vac. Sci. Technol. A 14 (1996) 2127. [9] B.E. Deal, C.R. Helms, Handbook of Semiconductor Wafer Cleaning Technology (1993) 274. Chap. 7. [10] H. Nishino, N. Hayasaka, H. Okano, J. Appl. Phys. 74 (2) (1993) 1345. [11] K.H. Shim, H.Y. Yang, Y.H. Ki, H.D. Yang, J.H. Yang, W.K. Hong, S. Kang, T.S. Jeong, T.S. Kim, Electron. Mater. Lett. 8 (4) (2012) 423–428. [12] K. Yamakawa, M. Hori, T. Goto, S. Den, T. Katagiri, H. Kano, J. Appl. Phys. 98 (2005) 013301. [13] H. Ogawa, T. Arai, M. Yanagisawa, T. Ichiki, Y. Horiike, J. Appl. Phys. 41 (2002) 5349. [14] H. Kikyuama, N. Miki, K. Sak, J. Takano, I. Kawanabe, M. Miyashita, T. Ohmi, IEEE Trans. Semicond. Manufact. 4 (1) (1991) 26. [15] Y. Song, S. Dhar, L.C. Feldman, J. Appl. Phys. 95 (2004). [16] Cor L. Claeys, ECS Trans. 11(6) (2007) 215–222.
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Investigation of oxide layer removal mechanism using reactive gases Hyun-Tae Kim a, Jung-Soo Lim a, Min-Su Kim a, Hoon-Jung Oh b, Dae-Hong Ko b,c, Gyoo-Dong Kim d, Woo-Gon Shin d, Jin-Goo Park a,⇑ a
Department of Bio-Nano Technology and Material Science Engineering, Hanyang University, Ansan 426-791, Republic of Korea BIO-IT Micro Fab Center, Yonsei University, Seoul 120-749, Republic of Korea c Department of Materials Science and Engineering, Yonsei University, Seoul 120-749, Republic of Korea d GEN, Suwon 443-390, Republic of Korea b
a r t i c l e
i n f o
Article history: Received 30 July 2014 Received in revised form 9 February 2015 Accepted 11 February 2015 Available online 17 February 2015 Keywords: Dry etching Buffered oxide etchant (BOE) NH3/NF3 etching Ellipsometry Plasma enhanced chemical vapor deposition (PECVD)
a b s t r a c t In a CMOS technology, the removal of silicon oxide and nitride layer is one of the critical steps as it represents a possible source of high contact resistance and a decrease of gate oxide reliability. In high aspect ratio (HAR), it is very difficult to remove SiO2 with wet etching. In the present study, the effect of the gases such as plasma dry etching of ammonia (NH3) and nitrogen trifluoride (NF3) on the SiO2 and Si3N4 substrates were analyzed and the etch rate was measured. The measurement of the SiO2 and Si3N4 thickness was measured by Ellipsometer. Various factors such as chamber pressure, electrode power and NH3/NF3 gas ratio were affected by the combination and dissociation of NH4F molecules. The existence of the byproduct was analyzed by using a contact angle analyzer and scanning electron microscope, respectively. In this study we have found that, the removal efficiency was mainly dependent on the reaction mechanism and the effect of the by-product. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction SiO2 and Si3N4 are widely used in micro fabrication processes as a dielectric and mask material. As pattern size continues to decrease, lithography process required accuracy between layers. It is become harder to make contact hole and trenches into dielectric layers. Commonly, SiO2 is the widely used dielectric material and Si3N4 has been used as a passivation layer [1]. In CMOS technology, removal of these films is a critical step as it makes a possible source of high contact resistance and a decrease of gate oxide reliability [2,3]. Possible over etch in the nitride processing may result in damages of a thin oxide and an underlying Si substrate through imperfections of the oxide [3]. The ability to achieve selective etching of SiO2 and Si3N4 is becoming an increasingly important requirement. Silicon nitride is used as a passivation layer that protects circuits from mechanical and chemical attack, or as an etch stop layer, enabling the fabrication of certain damascene and self-aligned contact (SAC) structures. Selective SiO2 and Si3N4 etching have been demonstrated in several systems [4–9]. High aspect ratio (HAR) silicon trench etch is a key process to remove the silicon oxide on the pattern. During the etching process, a chemical reaction between the chemical etchants and ⇑ Corresponding author. Tel.: +82 31 500 5226; fax: +82 31 400 4729. E-mail address: [email protected] (J.-G. Park). http://dx.doi.org/10.1016/j.mee.2015.02.025 0167-9317/Ó 2015 Elsevier B.V. All rights reserved.
the surface layer has been occurred. It is very important to control the various factors affecting the chemical reaction because the etch rate and the surface quality has been changed depending upon this reaction. In general, hydrogen fluoride is used as a chemical for silicon oxide removal [10]. In the SiO2 etching process, a wet process using an HF solution like BOE was employed. The etching rate of wet process was reached several lm/min. But, the usage of chemical solution and DI water rinse process, caused distortion and contamination in patterns [11]. As the decrease in minimum feature size, the removal of the silicon oxide on the pattern becomes difficult, however after the drying process, the etch pattern was collapsed by capillary force of water. Research of dry etching is being carried out to solve these problems. Dry etching is capable of reproducing anisotropic walls and the characteristics are highly reproducible [12]. In this study, the fundamental theory of SiO2 etching of plasma activated NF3/NH3 gas was investigated. Plasma dry etching of ammonia (NH3) and nitrogen trifluoride (NF3) mixtures were employed in the detailed studies. From this process, we could generate ammonium hexafluorosilicate (NH4)2SiF6 as a by-product, which is deposited on the surface and it interrupts the further etching reaction [13–14]. To investigate this problem, the etching of SiO2 using NH3/NF3 reactive gas was analyzed and the etch rate was calculated as a function of NF3 ratio, electrode power and pressure, respectively. From this study we could conclude that the
18
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
removal efficiency mainly depends on the reaction mechanism and the effect of the by-product. 2. Materials and methods A schematic representation of the dry etching reaction chamber is shown in Fig. 1. The NH3/NF3 mixture was excited using pulsed RF plasma (27.12 MHz pulse 3 kW). Helium gas was used as a carrier gas. The temperature was maintained at 35 °C throughout the experiment. The power of plasma was varied from 80 to 160 W, the chamber pressure from 3.75 to 5 Torr, gas ratio of NH3/NF3 from 0.14 to 7.82, and process time from 60 to 180 s, respectively. Thermally grown SiO2 (10000 Å) and low pressure chemical vapor deposition (LPCVD) Si3N4 (1000 Å) were used as a material for the experiments. The size of the sample 20 20 mm was cleaned with dilute SC-1 [NH4OH (25%):H2O2 (38%):DIW = 1:2:50] solution at 60 °C for 10 min before proceeding the dry etching, and then it was loaded into the reaction chamber. After dry etching was over, the annealing process was performed to remove the by-product. The measurement of the SiO2 and Si3N4 thickness was measured by Ellipsometer (M-2000V, J.A. Woollam, USA). The existence of the by-products was analyzed by using a contact angle analyzer (Phoenix, SEO, Korea) and SEM (FE-SEM, MIRA3, TESCAN, Czech), respectively. 3. Results and discussion Fig. 2 shows the schematic of the SiO2 with substrate before proceeding to the etch rate, it was exposed to the plasma NH3/ NF3 gas treatment. During the process by-product such as (NH4)2SiF6 was produced in the chamber, which is deposited on the surface and it was evaporated under the high temperature at 180 °C for 1 min. In this paper, we defined fume as white solid by-product that is easily remove by high temperature treatment or DI water rinse. After the SiO2 were etched without annealing, to find out the chemical composition of the fume, composition of fume was analyzed using Fourier transform infrared spectroscopy (FT-IR). Fig. 3 shows the ATR-FTIR spectrum of etched SiO2 surface and (NH4)2SiF6 powder. Fumes are formed on the SiO2 surface which can be confirmed N–H, NH+4 and SiF2 6 peak. These peaks are representative of (NH4)2SiF6 powder composition. This observation suggested that (NH4)2SiF6 created after NH3/NF3 dry etching process.
The etch rate of SiO2 and Si3N4 was measured as a function of the process time as shown in Fig. 4. The substrates were introduced into the chamber using Helium as a carrier gas with a flow rate of 600 sccm. The etch amount of SiO2 increased gradually with increase of process time. However, the etch rate decreased with an increase in process time. On the other hand, the etch rate of Si3N4 was increased. The results confirm that, the selectivity of SiO2 and Si3N4 was decreased with increase in process time. The maximum selectivity was obtained around 9.3 at 90 s shown in Fig. 5. After excess 90 s., the etch rate of Si3N4 was increased compared with SiO2. Fig. 6 shows the etching behavior of SiO2 in NH3/NF3 mixture of varying ratio. The overall chemical reaction of SiO2 etching involved is normally understood as [15,16]:
SiO2 þ 4HF þ 2NH4 F ! ðNH4 Þ2 SiF6 þ 2H2 O
ð1Þ
The reactions show that mechanism of SiO2 etching in buffered HF (BHF). However, a plasma active NF3/NH3 etching process consists of two steps. Plasma converts NF3 and NH3 to NH4F and NH4FHF (Eq. (2)). These products condense on the SiO2 surface and react with the SiO2 to form solid by-product ((NH4)2SiF6) (Eq. (3)) [16]:
3NH3 þ NF3 ! NH4 F þ NH4 F HF þ N2
ð2Þ
SiO2 þ 6NH4 F ! ðNH4 Þ2 SiF6 þ H2 O þ 4NH3
ð3Þ
The NH4F produced for contributing to the SiO2 etching. At less than 2 (NH3/NF3 ratio) level, the etch rate of SiO2 increases drastically with increasing NH3/NF3 ratio. At greater than 2 (NH3/NF3 ratio) level, the etch rate of SiO2 gradually decreases with increasing NH3/NF3. On the other hand, the etch rate of Si3N4 (40 Å/min) was relatively diminished in above the same condition. The maximum etch rate of SiO2 was 310 Å/min. In order to investigate the effect of recombination and dissociation of etch rate, we altered the chamber pressure and electrode power, respectively. Fig. 7 shows the etch rate of SiO2 as a function of pressure and power. The etch rate was decreased at increasing power. The etch rate of SiO2 was obtained in the range of 330– 240 Å/min at the pressure of (3.75–5 Torr) and the power of (80– 160 W), respectively. At high RF power, the more decomposable gas was produced. However, no reaction gas was observed during the etching. Similarly, the etch rate was decreased as increasing the chamber pressure.
Fig. 1. Schematic of the dry etching reaction chamber.
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
19
Fig. 2. Schematic diagram of the model for removal of SiO2 by plasma NH3/NF3: (a) before etching, (b) plasma NH3/NF3 exposure, (c) (NH4)2SiF6 generation, and (d) annealing.
20
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
Fig. 3. ATR-FTIR spectrum of (a) (NH4)2SiF6 powder and (b) etched SiO2 surface without annealing.
10
Selectivity (SiO2/Si3N4)
9 8 7 6 5 4 3
0
30
60 Time (sec.)
90
Fig. 5. Etch rate of SiO2 and Si3N4 as a function of NH3/NF3 ratio.
Fig. 6. Etch rate of SiO2 and Si3N4 as a function of NH3/NF3 ratio at 35 °C, 4 Torr and plasma power of 70 W.
Fig. 4. Etch rates of SiO2 and Si3N4 as a function of the process time: (a) SiO2, and (b) Si3N4.
Fig. 8 shows the schematic model of fume generation method. To predict the thickness of the fume, a theoretical approach model
21
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
Fig. 9. Etch rate of SiO2, thickness of fume calculated as a function of process time.
interface. In order to, we can calculate the thickness between the SiO2 and (NH4)2SiF6 by the density variation [15]: Molecular weight ðSiO2 Þ Density ðSiO2 Þ Molecular weight ððNH4 Þ2 SiF6 Þ Density ððNH4 Þ2 SiF6 Þ
Fig. 7. Etch rate of SiO2 as a function of (a) pressure at 35 °C and 70 W and (b) power at 35° and 4 Torr when constant flows of 50 sccm of NF3 and 100 sccm of NH3 are supplied.
60:08 2:21 ¼ 178:15 ¼ 0:31 2:01
A stoichiometric relationship was found during the reaction. The difference between the volume variation of SiO2 and (NH4)2SiF6 was around 31%, the SiO2 surface was consumed during the etching process. The consumption of the SiO2 was about 0.31 lm, the fume growth rate was at 1 lm level. Fig. 9 shows the etch rate of SiO2 and thickness of fume was calculated as a function of process time. The fume thickness had a margin of error. The calculated thickness was corresponding with experimental thickness value. Therefore, the thickness of the fume was predicted by the etch rate of SiO2. Fig. 10 shows the SEM images of the by-product. The initial contact angle of the SiO2 surface was 52°. After the etching process and before annealing, the condensation of fume layer was found. This fume formed a white thin layer on the surface. It can be easily soluble in water due to its property; in order to easily detect the difference between by-product and clean surface. If it remains on the surface, it would have become impossible to measure the contact angle because of its hygroscopic property. After removed by high temperature treatment at 180 °C for 1 min, the value of contact angle was turned back to the initial value.
4. Conclusion
Fig. 8. Schematic model of fume generation.
was used to calculate the thickness. The growing fume process involves the chemical reactions, which consume the SiO2 and also produced the fume. The fume was generated at the SiO2 – fume
The effect of the plasma dry etching of ammonia (NH3) and nitrogen trifluoride (NF3) on the SiO2 and Si3N4 substrates were analyzed and the etch rate was measured, which is strongly related to NH4F concentration and not to HF concentration. The NH4F molecules were produced by chemical reaction of NH3 and NF3. Various factors such as chamber pressure, electrode power and NH3/NF3 gas ratio could have been effected by the combination and dissociation of NH4F molecules. This result confirms that it could lead to a change in the etch rate. The maximum etch rate of SiO2 was obtained at NF3 and NH3 for a 2:1 ratio that exhibited the greatest amount of NH4F concentration.
22
H.-T. Kim et al. / Microelectronic Engineering 135 (2015) 17–22
Fig. 10. SEM and contact angle images of the surface after NH3/NF3 exposure: (a) an initial SiO2 surface, (b) a high density plasma oxide surface after NH3/NF3 exposure, (c) a thermal silicon oxide surface after NH3/NF3 exposure, and (d) silicon oxide surface after annealing at 180 °C, 1 min.
Acknowledgements This work was supported by the IT R&D program of MOTIE (10043438, Development of key cleaning technology for 10 nmsemiconductor and 8th generation display using damage free technology) and by the Future Semiconductor Device Technology Development Program #10045366 funded by MOTIE – South Korea (Ministry of Trade, Industry & Energy) and KSRC – South Korea (Korea Semiconductor Research Consortium). References [1] S.H. Choi, K.J. Jeong, J.S. Choi, T.Y. Cho, H.G. Chun, J. Korean Phys. Soc. 33 (1998) S99. [2] B.E.E. Kastenmeier, P.J. Matsuo, G.S. Oehrlein, J. Vac. Sci. Technol. A 17 (6) (1999) 3179. [3] M. Barklund, H.O. Blom, J. Vac. Sci. Technol. A 11 (1993) 1226.
[4] J. Givens, S. Geissler, J. Lee, O. Cain, J. Marks, P. Keswick, C. Cunningham, J. Vac. Sci. Technol. B 12 (1994) 427. [5] K. Ueno, T. Kikkawa, K. Tokashiki, J. Vac. Sci. Technol. B 13 (1995) 1447. [6] H. Hayashi, K. Kurihara, M. Sekine, Jpn. J. Appl. Phys. Part 1 35 (1996) 2488. [7] S. Fang, C. Chiang, D. Fraser, B. Lee, P. Keswick, M. Chang, K. Fung, J. Vac. Sci. Technol. A 14 (1996) 1092. [8] Y. Zhang, G.S. Oehrlein, F.H. Bell, J. Vac. Sci. Technol. A 14 (1996) 2127. [9] B.E. Deal, C.R. Helms, Handbook of Semiconductor Wafer Cleaning Technology (1993) 274. Chap. 7. [10] H. Nishino, N. Hayasaka, H. Okano, J. Appl. Phys. 74 (2) (1993) 1345. [11] K.H. Shim, H.Y. Yang, Y.H. Ki, H.D. Yang, J.H. Yang, W.K. Hong, S. Kang, T.S. Jeong, T.S. Kim, Electron. Mater. Lett. 8 (4) (2012) 423–428. [12] K. Yamakawa, M. Hori, T. Goto, S. Den, T. Katagiri, H. Kano, J. Appl. Phys. 98 (2005) 013301. [13] H. Ogawa, T. Arai, M. Yanagisawa, T. Ichiki, Y. Horiike, J. Appl. Phys. 41 (2002) 5349. [14] H. Kikyuama, N. Miki, K. Sak, J. Takano, I. Kawanabe, M. Miyashita, T. Ohmi, IEEE Trans. Semicond. Manufact. 4 (1) (1991) 26. [15] Y. Song, S. Dhar, L.C. Feldman, J. Appl. Phys. 95 (2004). [16] Cor L. Claeys, ECS Trans. 11(6) (2007) 215–222.