www.elsevier.com/locate/mee Microelectronic Engineering 73–74 (2004) 306–311 Impact of Ar addition to inductively coupled plasma etching of SiC in SF6/O2 Liudi Jiang *, R. Cheung School of Engineering and Electronics, Scottish Microelectronics Centre, The University of Edinburgh, King�s Buildings, West Mains Road, Edinburgh EH9 3JF, UK Available online 18 March 2004 Abstract The influence of Ar addition to SF6/O2 gas mixtures has been investigated for inductively coupled plasma (ICP) etching of SiC with a view to improve both etch rate and etched surface microstructure. SiC etch rates have been studied as a function of Ar concentration, gas flow rates, applied ICP power, chuck power and chamber pressure. SiC etch rate, surface morphology, surface chemistry and etch profiles obtained in SF6/O2/Ar gas mixtures have been compared with those of SiC-etched in SF6/O2 gas mixtures under similar conditions. It was found that, compared with SF6/O2 (4:1) ICP-etched SiC in our studies, smoother surfaces and significant reduction of fluorine-related etch residues can be obtained by optimum Ar addition. SiC etch rate in SF6/O2 gas mixtures can be increased by over 5% with optimum Ar%. The highest SiC etch rate achieved here is approximately 500 nm/min in SF6/O2/Ar gas mixtures. The influence of Ar addition on the SiC etch profile has also been studied. � 2004 Elsevier B.V. All rights reserved. Keywords: Silicon carbide; Inductively coupled plasma; Etch rate 1. Introduction As a promising wide band gap compound semiconductor, silicon carbide (SiC) has attracted significant attention for its beneficial material properties of high thermal conductivity, large breakdown field, high saturated electron drift ve- locity and great hardness and wear resistance. These unique properties mean that SiC is an ex- cellent candidate for electronic devices as well as * Corresponding author. E-mail address:
[email protected] (L. Jiang). 0167-9317/$ - see front matter � 2004 Elsevier B.V. All rights reserv doi:10.1016/j.mee.2004.02.058 for micro-electro-mechanical systems (MEMS) working in extreme environments [1]. Due to the large bond energy between Si and C, and therefore great chemical inertness, plasma-based dry etching is the only practical way to pattern SiC. However, the surface damage and chemistry modifications induced during dry etch patterning of semicon- ductors can significantly affect subsequent elec- tronic device performance [2] and also play an important role in the quality of corresponding MEMS structures [3]. Inductively coupled plasma (ICP) etching of SiC using SF6/O2 gas mixtures has been intensively studied so far because it gives a relatively high etch ed. mail to:
[email protected] L. Jiang, R. Cheung / Microelectronic Engineering 73–74 (2004) 306–311 307 rate [4,5], but much surface damage [2] and etch residue [6] were detected on the etched SiC surface. Plasma etching of SiC using SF6/Ar gas mixture has also been previously reported to result in less surface damage and smoother SiC surface than with the mixtures of SF6/O2 [7,8]. However, in- formation is lacking on the influence of Ar addi- tion to SF6/O2 ICP plasma in SiC etching processes. In this work, the impact of Ar addition in SF6/ O2 ICP etching of SiC has been studied with a view to achieve both a high etch rate and little surface damage. The etch rate at various etch conditions and etch profiles have also been compared and investigated. 0 10 20 30 40 50 60 385 390 395 400 405 410 168 171 174 177 180 D C b ia s (V ) E tc h ra te ( nm /m in ) Ar concentration (%) SF6/O2=4:1 Fig. 1. Etch rate and dc bias as a function of Ar% (ICP pow- er¼ 1000 W, chuck power¼ 120 W, SF6 flow rate¼ 60 sccm, O2 flow rate¼ 15 sccm and pressure¼ 5 mT). 2. Experimental The samples used in this work were bulk 4H- SiC substrates from Cree, Inc. A silicon dioxide (SiO2) layer, approximately 6-lm thick, was de- posited on SiC using plasma-enhanced chemical vapor deposition and then patterned as a mask using optical lithography and Plasmatherm PK2440 reactive ion etching system in a CF4 and H2 gas mixture. A Surface Technology Systems (STS) multiplex ICP system was used to etch the patterned SiC substrates for this study. Since many reports have indicated that SF6/O2 ¼ 4:1 (20% O2) is required to obtain the best etch rate in ICP etching of SiC [5,6,9], in order to focus on studies of the impact of Ar addition in the plasma, SF6/ O2 ¼ 4:1 in the gas mixtures was chosen in all our experiments here as base conditions. SiC etch rate has been studied by introducing Ar with different concentration into the base conditions and also by altering other etch conditions. The surface morphology of the etched SiC surface was analysed using atomic force micros- copy (AFM). The roughness of etched SiC surface has been evaluated by calculating the average RMS roughness from multiple AFM images of every sample. X-ray photoelectron spectroscopy (XPS) has been utilized to obtain information on etch induced surface chemistry modifications using a monochromatic Al Ka (hm ¼ 1486:6 eV) X-ray source operating at 15 kV. According to the cor- responding XPS results, the surface chemical compositions of etched SiC have been calculated using sensitivity factors 0.25, 0.27, 0.66 and 1 for C, Si, O and F, respectively [10]. The comparison of etch profiles was studied using scanning electron microscopy (SEM). 3. Results and discussion Both the SiC etch rate and the resulting dc bias as a function of Ar% in SF6/O2/Ar total gas mix- tures are shown in Fig. 1. The continuous decrease of dc bias with Ar% in the gas mixtures was ob- served. Despite the decrease of dc bias, it was found that a certain amount of added Ar promotes the etch processes, and the etch rate peaks at ap- proximately 30% Ar (SF6/O2/Ar¼ 4:1:2). This re- sult is similar to a previous report [4], where 30% Ar was also found to be the optimum condition for SF6/Ar gas mixtures. Ion-induced etching has been shown to be a dominant mechanism in dry etching processes of SiC [9]. Increasing the Ar concentra- tion in the plasma could not only enhance physical sputtering during the etch processes but also the number of dissociated reactive ions in the cham- ber. Excess addition of Ar also dilutes the reactive fluorine- and oxygen-related active species within the plasma. The requirement of �30% Ar as op- timum concentration in the SF6/O2/Ar total gas mixtures is probably due to the competition of these possible reasons. 0 10 20 30 40 50 60 0.02 0.06 0.10 0.14 0.18 0.22 F /S i r at io Ar concentration (%) SF6/O2=4:1 Fig. 3. The relative ratio of F/Si calculated from the F1s and Si2p spectra of the SiC-etched at different Ar% (ICP pow- er¼ 1000 W, chuck power¼ 120 W, SF6 flow rate¼ 60 sccm, O2 flow rate¼ 15 sccm and pressure¼ 5 mT). 308 L. Jiang, R. Cheung / Microelectronic Engineering 73–74 (2004) 306–311 AFM images in Fig. 2 shows the clear existence of etch-induced residue on etched SiC surface, and the SiC-etched at 30% Ar has the cleanest and smoothest surface. The average surface RMS roughness are 1.7, 0.9 and 2.5 �A for SiC-etched at 0%, 30% and 100% Ar conditions, respectively. It is reported that etch-induced residues on the SiC surface can affect the properties of subsequently fabricated devices such as Schottky diodes [11]. Our previous results [6] show that various fluoro- carbon etch residues are closely related to the F concentration on the etched SiC surface. Fluorine, carbon and silicon photoelectron spectra have been detected in all our SF6/O2/Ar ICP-etched SiC surfaces. The relative ratios of F/Si have been calculated from the corresponding F1s and Si2p spectra [6] and are shown in Fig. 3. The least F- related etch residues are found at 30% Ar etching condition. Ar addition could enhance the physical sputtering as well as the number of dissociated reactive ions during plasma etching which can help to remove the etch induced residues simulta- neously. However, the lack of oxygen species at excess Ar conditions probably results in less ef- fective removal of C atoms and this can also contribute to more fluorocarbon-related residues. Again, this possible competition between the Ar and O2 contribution to the etch mechanism can result in our experimental observations. In order to improve the SiC etch rate under the optimum ratio of SF6/O2/Ar¼ 4:1:2 (�30% Ar), Fig. 2. AFM images of SiC ICP-etched at: (a) 0% Ar, (b) 30% Ar and chuck power¼ 120 W, SF6 flow rate¼ 60 sccm, O2 flow rate¼ 15 scc the influence of SF6 flow rate, ICP power, chuck power and work pressure under this gas ratio has been investigated in the following experiments. By applying different flow rates of SF6 and al- ways keeping the gas mixture ratio of SF6/O2/ Ar¼ 4:1:2 accordingly, SiC etch rate as a function of SF6 flow rate is investigated and the results are shown in Fig. 4. Similar to the flow rate studies under the SF6/O2 (4:1) conditions reported in [6], it is noticed that the highest etch rate is achieved at a mixture of 40 sccm SF6, 10 sccm O2 and 20 sccm Ar. Because all these samples were etched at the same pressure (5 mT) and at the same gas mixture (c) 100% Ar in SF6/O2/Ar gas mixtures (ICP power¼ 1000 W, m and pressure¼ 5 mT). 20 30 40 50 60 320 365 410 455 500 80 W 120W 150W et ch r at e (n m /m in ) SF6 flow rate (sccm) SF6/O2/Ar=4:1:2 Fig. 4. Etch rate vs SF6 flow rate at chuck power of 80, 120 and 150 W, respectively (ICP power¼ 1000 W, pressure¼ 5 mT, O2 flow rate¼ 5, 10 and 15 sccm for 20, 40 and 60 sccm SF6, re- spectively, Ar flow rate¼ 10, 20 and 32 sccm for 20, 40 and 60 sccm SF6, respectively). L. Jiang, R. Cheung / Microelectronic Engineering 73–74 (2004) 306–311 309 ratio SF6/O2/Ar¼ 4:1:2, the densities of the reac- tive species in the plasma are the same. However, different flow rates may affect the residence time of the reactive radicals on the SiC surface being etched and also the speed of the removal of the volatile etch products. Higher flow rates help to remove the volatile SFx and CFx etch products more rapidly. However, when the flow rate is higher than the optimum value, the residence time of the reactive radicals becomes shorter than the time scale inherent to reaction. Therefore, opti- 200 400 600 800 1000 50 150 250 350 450 et ch r at e (n m /m in ) ICP power (W)(a) ( SF6/O2/Ar=4:1:2 Fig. 5. Etch rate as a function of applied: (a) ICP source power and (b Ar flow rate¼ 20 sccm and pressure¼ 5 mT). mum flow rates are required to achieve best SiC etch rate of �500 nm/min here. Figs. 5(a) and (b) show that SiC etch rate in SF6/O2/Ar (4:1:2) gas mixtures can also be im- proved by the increase of applied ICP power and chuck power. Increasing the ICP coil power en- hances the density of the reactive ions and neutrals in the chamber which can result in the increase of the chemical reactions on the etched surface and thereby the etch rate in Fig. 5(a). The increase of chuck power in Fig. 5(b) will enhance the dc bias and elevate the physical ion sputtering effect on the etched surface which will in turn promote the etch rate. Similar trends have also been observed in SF6/O2 ICP etching of SiC [6,9]. However, this is different from previous results in SF6/O2 ICP etching where the SiC etch rate decreases with the increase of pressure [6], Fig. 6 shows that, under the same pressure regime studied here, the increase of working pressure in SF6/O2/Ar (4:1:2) gas mixtures leads to higher etch rate of SiC. The ex- istence of Arþ ions in the plasma under higher pressures can probably promote the physical sputtering on the SiC substrates being etched as well as the dissociation of SF6 and O2 into reactive species, and therefore increase the etch rates. The influence of Ar addition to SiC etch profile is displayed in Fig. 7. There is no obvious profile difference observed between SF6/O2 (4:1) and SF6/ O2/Ar (4:1:2) ICP-etched SiC samples. The mi- crotrenching effect in Fig. 7 indicate that the 0 30 60 90 120 150 0 100 200 300 400 500 600 b) et ch r at e (n m /m in ) Chuck power (W) SF6/O2/Ar=4:1:2 ) chuck power (SF6 flow rate¼ 40 sccm, O2 flow rate¼ 10 sccm, Fig. 7. Profiles of SiC-etched with: (a) SF6/O2 (4:1) plasma (ICP power¼ 1000 W, chuck power¼ 120 W, SF6 flow rate¼ 40 sccm, O2 flow rate¼ 10 sccm and pressure¼ 5 mT) and (b) SF6/O2/Ar (4:1:2) plasma (ICP power¼ 1000 W, chuck power¼ 120 W, SF6 flow rate¼ 40 sccm, O2 flow rate¼ 10 sccm, Ar¼ 20 sccm and pressure¼ 5 mT). 0 2 4 6 8 10 12 100 130 160 190 220 250 80W 120W 150W et ch r at e (n m /m in ) pressure (mT) SF6/O2/Ar=4:1:2 Fig. 6. Etch rate vs pressure at chuck power of 80, 120 and 150W, respectively (ICP power¼ 1000 W, SF6 flow rate¼ 40 sccm, O2 flow rate¼ 10 sccm and Ar flow rate¼ 20 sccm). 310 L. Jiang, R. Cheung / Microelectronic Engineering 73–74 (2004) 306–311 ion-induced etch mechanism is dominant in both cases because it is a result of the impact of directional ions onto the surface being etched. 4. Conclusions We have shown that both the etch rate and the etched surface microstructure of SiC can be im- proved by the optimum addition of Ar in SF6/O2 (4:1) gas mixtures during ICP etching. Compared with the SiC samples etched in the same ICP re- actor using only SF6/O2 (4:1) gas mixtures, AFM and XPS results indicate the smoothest surface with the lowest F/Si ratio can be obtained under SF6/O2/Ar¼ 4:1:2 gas mixture conditions. Under this optimum gas ratio, it has been observed that flow rates of 40 sccm SF6, 10 sccm O2 and 20 sccm Ar correspond to the highest etch rate, and SiC etch rate can also be promoted by the increase of the applied ICP power, chuck power and working pressure. The highest SiC etch rate achieved in our studies here is �500 nm/min. It is also found that introduction of Ar in the SF6/O2 (4:1) plasma does not affect SiC etch profiles significantly. Acknowledgements This work was supported by Engineering and Physical Sciences Research Council (EPSRC) Grant No. GR/R38019/01. We are grateful to R. Brown for his help with X-ray photoelectron spectroscopy measurements. References [1] M. Mehregany, C.A. Zorman, N. Rajan, C.H. Wu, Proc. IEEE 86 (1998) 1594. [2] S.W. Pang, Surface Damage Induced by Dry Etching, Handbook of Advanced Plasma Processing Technologies, in: R.J. Shul, S.J. Pearton, Springer-Verlag, Berlin, Hie- delberg, New York, 2000. [3] Y. Wang, J.A. Henry, A.T. Zehnder, M.A. Hines, J. Phys. Chem. B 107 (2003) 14270. [4] P. Chabert, J. Vac. Sci. Technol. B 19 (2001) 1339. [5] F.A. Khan, I. Adesida, App. Phys. Lett. 75 (1999) 2268. [6] L. Jiang, R. Cheung, R. Brown, A. Mount, J. Appl. Phys. 93 (2003) 1376. [7] M.S. So, S.G. Lim, T.N. Jackson, J. Vac. Sci. Technol. B 17 (1999) 2055. L. Jiang, R. Cheung / Microelectronic Engineering 73–74 (2004) 306–311 311 [8] N. Camara, K. Zekentes, Solid-State Electron. 46 (2002) 1959. [9] N.O.V. Plank, M.A. Blauw, E.W.J.M. van der Drift, R. Cheung, J. Phys. D 36 (2003) 482. [10] D. Briggs, M.P. Seah, in: Practical Surface Analysis, Wiley, New York, 1983. [11] F.A. Khan, B. Roof, L. Zhou, I. Adesida, J. Electron. Mater. 30 (2001) 212. Impact of Ar addition to inductively coupled plasma etching of SiC in SF6/O2 Introduction Experimental Results and discussion Conclusions Acknowledgements References