- Email: [email protected]
BCl3 plasma process for high etch rate selective etching of via-holes in GaAs
Vacuum 85 (2010) 452e457
Contents lists available at ScienceDirect
Vacuum journal homepage: www.elsevier.com/locate/vacuum
Study of inductively coupled Cl2/BCl3 plasma process for high etch rate selective etching of via-holes in GaAs D.S. Rawal a, *, V.R. Agarwal a, H.S. Sharma a, B.K. Sehgal a, R. Muralidharan a, Hitendra K. Malik b a b
Solid State Physics laboratory, Lucknow Road, Timarpur, Delhi 110054, India PWPAL, Department of Physics, Indian Institute of Technology, Delhi 110016, India
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 April 2010 Received in revised form 18 August 2010 Accepted 22 August 2010
We have investigated the selective etching of 50 mm diameter via-holes for etch depth >200 mm using 30 mm thick photo resist mask in Inductively Coupled Plasma system with Cl2/BCl3 chemistry. Resultant etch rate/etch profiles are studied as a function of ICP process parameters and photo resist mask sidewall profile. Etch yield and aspect ratio variation with process pressure and substrate bias is also investigated at constant ICP power. The etch yield of ICP process increased with pressure due to reactant limited etch mechanism and reached a maximum of w19 for 200 mm depth at 50 mTorr pressure, 950 W coil power, 80 W substrate bias with an etch rate w4.9 mm/min. Final aspect ratio of etched holes is increased with pressure from 1.02 at 20 mTorr to 1.38 at 40 mTorr respectively for fixed etch time and then decreased to 1.24 at 50 mTorr pressure. The resultant final etch profile and undercut is found to have a strong dependence on the initial slope of photo resist mask sidewall angle and its selectivity in the pressure range of 20e50mTorr. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: GaAs Via-hole ICP Etching Etch yield Aspect ratio
1. Introduction GaAs devices are used extensively in the wireless telecommunications industry, where the high electron mobility of GaAs makes it well suited for high frequency, low noise and high gain applications. Although it has excellent electrical properties, GaAs is a relatively poor thermal conductor, making it difficult to remove heat efficiently from power devices. A commonly used solution is to form via-holes grounds from the wafer backside to the front side circuitry. Such connections provide a good thermal path for heat removal as well as a low impedance ground for RF (radio frequency) devices. This backside via formation is one of the final steps in the device fabrication. After completion of the front side processing, the wafer is mounted face down on a carrier wafer and mechanically thinned to a thickness of approximately 100e200 mm. The back of the wafer is then patterned using photo resist and the holes are etched using reactive ion etching (RIE) through the thinned substrate, stopping on the front side metal. After resist removal the vias are metallized, typically by sputtering a gold seed layer followed by gold electroplating to act as the heat sink/ground connection [1].
* Corresponding author. Fax: þ91 11 23984285. E-mail address: [email protected] (D.S. Rawal). 0042-207X/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.vacuum.2010.08.022
Inductively Coupled Plasma (ICP) etching has been replacing conventional RIE for GaAs backside via etching because of several advantages. Throughput improvement utilizing significantly faster etching rate have already been reported in the literature [2]. Furthermore, the ICP tools provide better control of via size, repeatability and reproducibility [3,4]. The ICP tools produce significantly different via dimension as compared to the conventional RIE tools, if the same size mask used, due to its nature of etching process. The etch profile and surface morphology of viahole grounds is important not only for the inductance consideration but also for the success of backside metallization. The smooth morphology of the etched sidewalls provides reliable and good electrical contact with low resistance. Etching is mainly carried out in chlorine/fluorine plasma. A number of gas combinations CCl2F2, CCl2F2/CCl4, SiCl4/Cl2, BCl3/Cl2/Ar, Cl2/Ar and Cl2/BCl3 have been utilized to fabricate via-holes [4]. Each gas combination has its advantages and disadvantages. The previous work reported using Cl2/BCl3/Ar for etching of GaAs is mainly carried out using RIE for etch depths w100 mm and etch rate reported are much lower 1 mm/min for 3-inch GaAs wafer. Cl2/BCl3 gas mixture with ICP process is being increasingly used for fabrication of via-holes at high etch rates with excellent anisotropy and smooth surface morphology. Generally reported etch depths using ICP for via-hole etching applications in GaAs Monolithic Microwave Integrated Circuits (MMIC) are less than 200 mm using photo resist mask due to lower etch rate and poor mask selectivity. The substrate
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
thickness of 200 mm is preferable as it results in higher yield and throughput due to easier handling of fragile wafers without carriers and also improves the RF performance due to lower electrical losses in MMIC micro strip interconnects on the front side. It is not easy to etch 200 mm deep holes at high etch rates with desired profile, good reproducibility and repeatability as etching depends upon various parameters like component of radicals, ions in plasmas, flow of gases, dc bias and type of mask used. Therefore, it becomes important to understand the effect of these plasma parameters on etch rate, anisotropy, etch profile, aspect ratio, mask selectivity and then develop the process for particular application [3e6]. The etch rate is the GaAs thickness etched in 1 min (mm/ min.), anisotropy is defined as {1 (lateral etch rate/vertical etch rate)}, etch profile refers to the shape of etched feature crosssection, aspect ratio is etch depth/feature’s lateral dimension and selectivity is defined as the etch rate of GaAs/etch rate of photo resist mask. We are reporting the ICP etching process of 50 mm diameter via-holes at relatively high average etch rate for an etch depths of w200 mm, on 3-inch GaAs wafer using positive photo resist mask. As etch depth increases, etch rate reduces drastically, with photo resist mask, affecting process throughput significantly in a production environment. Etch rate variation with ICP process parameters i.e. pressure and substrate bias power, are studied in detail at a constant maximum available ICP coil power to have high plasma density that results in high etch rates. At high plasma density, pressure and bias voltage are the most important parameters to control etch profile and selectivity as they mainly control the ion energy i.e. physical component of etching. Aspect ratio and etch yield are two important parameters that characterise any etch process [7e9]. Variation of aspect ratio and etch yield with process parameters is also studied. As thick photo resist mask is required to etch deep holes, its sidewall profile and selectivity plays an important role in deciding the final etch profile. We are also reporting this effect by varying sidewall angle of photo resist mask [10]. To our knowledge these parameters for an ICP etch process involving etch depths of w200 mm with such high average etch rates have not been reported so far using Cl2/BCl3 chemistry. The main purpose of the study was to develop a relatively high etch rate, production viable ICP etching process to etch w200 mm deep via-holes with acceptable via-hole etch profile, photo resist mask selectivity and etch uniformity for fabricating 50 mm diameter viahole ground connections in GaAs MMICs over a 3-inch wafer.
Fig. 2. SEM cross-section of hole etched upto a depth of more than 175 mm.
positive photo resist AZ 4620 and patterned with 50 mm diameter holes. The patterned photo resist was then post baked at 120 C for variable time to introduce a sloped photo resist profile with improved adhesion. These patterned wafers were then mounted on a carrier wafer with wax to make wafer loading compatible with ICP system and also provide cooling of the wafer during etching process. The patterned wafers were then etched in standard ICP system, one at a time. Plasma of etcher is inductively coupled through a coil at 13.56 MHz, with independent energy control provided by 13.56 MHz RF biasing on the substrate. Helium gas was used to cool backside of the wafer. The substrate temperature was set at 20 C for all test conditions. The etch chemistry was a mixture of Cl2/BCl3 through mass flow controlled process gas lines. The chamber was evacuated to a base pressure of 9e-3 mTorr, by a turbo molecular pump backed by a dry mechanical pump, before initiating the etch process. The etch gases mixture was introduced through an annular region at the top of chamber lid. ICP etching was carried out using only photo resist mask to have less complex process with good etch surface morphology. All the ICP experiments were carried out at near-maximum available coil power and Cl2/BCl3 flow rate ratio of 4:3 for a fixed total flow rate w280 sccm to have high plasma density [11] and increased concentration of reactive Cl species, that
2. Experimental details All test wafers were 3-inch S.I. GaAs wafers with thickness w650 mm these wafers were then coated with in 30 mm thick
8 Et c h R a t e ( u m /m in )
7 6 5 4 3 2 1 0 0
50
100
150
200
Etch depth(um)
Fig. 1. Etch rate as function of etch depth (950 W ICP Power, 65 W Substrate bias and 30 mTorr).
453
Fig. 3. SEM photograph showing deposition on the etched sidewall.
454
D.S. Rawal et al. / Vacuum 85 (2010) 452e457 1.6
A s p e c t R a t io
1.4 1.2 1 0.8 0.6 0.4 0.2 0 0
10
20
30
40
50
60
Pressure(mT)
Fig. 6. Aspect ratio variation with pressure (950 W ICP Power, 80 W Substrate bias). Fig. 4. EDAX spectrum of the etched sidewall etched using thick photo resist mask.
resulted in high etch rates with better etch surface morphology. Higher than 4:3 flow rate ratio increased the etch rate but at the cost of surface morphology, whereas lower flow rate ratio decreased the etch rate significantly. The etch rate, etch depth, etch profile, aspect ratio, etch yield, mask selectivity and surface morphology of etched holes were determined by cleaving through the etched features and examining the sample under scanning electron microscope. Deposition on the etched sidewall was analysed by high resolution INCA EDAX.
3. Results and discussion
hole is very smooth however the sidewalls are rough with some deposition. Fig. 3 shows the SEM photograph of the etched sidewall and Fig. 4 shows the EDAX spectrum of the etched sidewall with peaks corresponding to carbon and chlorine in addition to Ga and As, most probably due to the presence of CClx polymer on the sidewalls. The etching takes place via chemi-sorption of the highly reactive Cl radical on wafer surface, forming volatile molecule and subsequently de-sorption of this volatile molecule. The ions are mainly responsible for bombarding the surface creating damage and increasing reaction probability of the arriving reactive species. The incident ions also assist in the de-sorption of the reaction product via sputtering. The reaction probably occurs in stepwise manner with gallium and arsenic mono-chlorides formation first
3.1. Etch rate variation with etch depth Fig. 1 shows the average etch rate variation with etch depth for ICP process at 950 W coil power, 30 mTorr pressure, 65 W platen power and 45 min. of etch time. It clearly shows that the average etch rate is decreasing with etch depth from 7 mm/min for 70 mm depth to 3.9 mm/min for 175 mm on 3-inch wafer. This is mainly due to the re-deposition of sidewall material and sputtered photo resist on etched hole due to positive slope in etch profile as depth increases. This redepsition on the sidewall promotes sidewall smoothness and anisotropy but at the cost of etch rate. This reduction in etch rate with etch depth is very high in RIE and it is almost impractical to etch depths >100 mm depths with photo resist mask [4]. Therefore, the average etch rate achieved for smaller etch depths is much higher than >100 mm etch depths for same diameter holes. Fig. 2 shows the cross-section of a hole etched up to the depth of more than 150 mm with good anisotropy and surface morphology using photo resist mask. The bottom of the
Et c h r a t e ( u m /m in )
6 5 4 3 2 1 0 0
10
20
30
40
50
60
Pressure(mTorr) Fig. 5. Etch rate variation with pressure for fixed etch time (950 W ICP Power, 80 W Substrate bias).
Fig. 7. (a) Etch cross-section of two holes 150 mm apart etched at 40 mTorr pressure (950 W ICP Power, 90 W substrate bias). (b) Etch cross-section of two holes 150 mm apart etched at 50 mTorr pressure (950 W ICP Power, 90 W substrate bias).
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
455
20 18
Et c h Y ie ld
16 14 12 10 8 6 4 2 0 0
10
20
30
40
50
60
Pressure(mTorr)
Fig. 8. Etch yield calculated as a function of pressure (950 W ICP Power, 80 W Substrate bias).
followed by formation of di-chlorides and tri-chlorides [12] as shown in following equations: GaAs þ 2Cl / GaCl þ AsCl
(1)
GaCl þ Cl / GaCl2
(2)
AsCl þ Cl / AsCl2
(3)
GaCl2 þ Cl / GaCl3
(4)
AsCl2 þ Cl / AsCl3
(5)
3.2. Etch rate variation with process pressure Fig. 5 shows the experimentally observed etch rate variation with process pressure at 950 W coil power and 80 W of substrate bias for an etch time of 45 min. This graph clearly indicates that the etch rate is a strong function of process pressure in the range 20e50 mTorr and is increasing with pressure due to increased density of reactive species, suggesting reactant limited etch mechanism [13] but anisotropy is maintained up to 40 mTorr due to constant ion energy incident on the substrate at 80 W of substrate bias. In other words, physical component of etching is fairly constant up to 40 mTorr. Above 40 mTorr etch rate is increasing but at the cost of anisotropy resulting in large undercut making etch profile unsuitable. Etch rate is increased to 4.9 mm/min at 50 mTorr from 2.9 mm/min at 20 mTorr, for an etch time of 45 min, suggesting that the dominance of chemical etching component with increased concentration of reactive species.
Fig. 10. SEM cross-section of unbaked, 50 mm diameter, vertical photo resist mask profile (sidewall angle: 88 ).
3.3. Aspect ratio with process pressure Aspect ratio in ICP etching is a function of process pressure and time. Fig. 6 shows the aspect ratio variation with pressure at fixed etch time for 50 mm diameter hole. It is seen that aspect ratio is increasing with pressure from 1.02 at 20 mTorr to 1.38 at 40 mTorr of pressure and then starts decreasing and reaches to a value of 1.24 at 50 mTorr pressures mainly due to more of chemical etching at higher pressures, leading to a reduction in anisotropy. Aspect ratio becomes very important when two features to be etched are very close due to lay out constraints. Fig. 7(a) and (b) show the etch cross-section of two 50 mm dia. holes, 150 mm apart etched at 40 mTorr and 50 mTorr process pressure to a depth of greater than 200 mm respectively. At 40 mTorr the holes are etched with high anisotropy that resulted in higher aspect ratio in comparison to holes etched at 50 mTorr, which resulted in merging of two features that could lead to poor device performance. 3.4. Etch yield with process pressure Etch yield can be defined as number of atoms etched per incident ion and is a function of etch rate, density and molecular weight of the material to be etched. The etch yield is proportional to the energy transferred by the incident ions through nuclear collisions
20 19.5 Et c h Y ie ld
19 18.5 18 17.5 17 16.5 0
20
40
60
80
100
120
Substrate bias(W)
Fig. 9. Etch yield variation obtained with substrate bias (950 W ICP Power, 30 mTorr pressure).
Fig. 11. SEM cross-section of etched hole showing resultant etch profile (Etch depth w225 mm).
456
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
Undercut(um )
80 60 40 20 0 40
60
80
100
Mask Sidewall Angle(Degree) Fig. 14. Undercut measured as a function of photo resist mask sidewall angle for 50 mm diameter holes etched under same conditions.
Fig. 12. SEM photo resist profile obtained after oven baking at 120 C, 60 min.
in the material. Its calculation for this ICP etch process is based on measured etch rate and predicted ion flux and is as high as 10e20 [14]. Fig. 8 shows the predicted etch yield variation with pressure for etching 50 mm diameter holes for etch depths up to 200 mm. Etch yield is increasing with pressure as etch rate is increasing and reaches a value of w19 at 50 mTorr. The increase in etch yield with pressure indicates that etching process is limited by mass transfer of reactive species This again suggests that the chemical species required for converting the surface atoms to volatile molecules cannot be provided by the ion itself instead radicals are also equally participating. The radicals are highly reactive species that form volatile product molecules in the presence of bombarding ions as discussed earlier.
3.5. Etch yield with substrate bias Fig. 9 shows etch yield variation as a function of substrate bias. Increasing the bias power from 60 to 100 W at pressure 30 mTorr resulted in reduction in etch yield from 19.5 to 16.8 respectively for an etch time of 45 min. This reduction in etch yield is probably due to higher ion bombardment at 100 W that resulted in sputtering and re-deposition of photo resist on the etched features reducing the average etch rate under these plasma conditions contradictory to the normally observed behaviour.
In other words, etching process is more physically driven at 100 W substrate bias. This reduction in etch yield has in turn resulted into better etch sidewall morphology at 100 W; because of the higher ion bombardment that may sputter the surface evenly regardless of defects present. 3.6. Etch profile variation with mask edge profile An anisotropic etch profile offers the best CD control but practically it is difficult to get a good metal contact in such vertical profiles for via-hole grounds. In order to lower the aspect ratio for ease of metallization, a sloped via etch profile is preferred. Apart from ICP process parameters, initially sloped photo resist mask profile and it’s etch selectivity plays important role in deciding the final via etch profile. Fig. 10 shows the unbaked, 30 mm thick resist profile with 88 sidewall slope and Fig. 11 shows the resultant etch profile for depth >200 mm. This profile is highly anisotropic and unsuitable for metal interconnection. Up to a depth of w100 mm, etched walls are almost vertical as resist erosion is much lower initially with vertical photo resist sidewalls. As photo resist erodes at corners due to longer plasma exposure, its sidewalls become tapered and etch profile starts to broaden as clearly seen in Fig. 11. Also GaAs: photo resist selectivity was found to be w10:1 in this case as there is no resist hardening involved after photolithography process making resist erosion faster during plasma exposure. A baking cycle after photolithography was incorporated to tailor the resist profile by oven baking at 120 C, for 45 min and 60 min. The sidewall angle changed to 78 and 70 for 45 min bake and 60 min bake respectively. Fig. 12 shows the photo resist profile obtained after 60 min bake with w70 sidewall angle. Baking temperature above 120 C has resulted in resist flowing, deteriorating the geometry as well as making its removal difficult after etching. Fig. 13 shows the etch profile corresponding to 70 sidewall photo resist mask. Fig. 14 shows the undercut measured as a function of mask sidewall angle. It is evident from this graph is that undercut is a strong function of photo resist mask profile. It needs to be mentioned that via-hole etch profiles in Figs. 11 and 13 are not suitable for via ground connection. The acceptable profile of 200 mm deep via-hole is almost identical to the profile shown in Fig. 2 and is obtained with resist baking at 120 C for 45 min. Mask selectivity for baked mask profiles was better than 12:1. Resist removal in Cl2/BCl3 chemistry appears to be physically driven e the resist etch rate increased solely with the RF bias power degrading selectivity. The etch uniformity measured for the etched holes over the three inch GaAs wafer is better than 5% for all the experiments indicating good gas flow uniformity in the ICP chamber. 4. Conclusion
Fig. 13. ICP etched via profile obtained with 60 min baked resist profile (Etch depth w229 mm).
We have investigated the selective etching of 50 mm diameter via-holes for etch depth 200 mm using thick photo resist mask in
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
Inductively Coupled Plasma system with Cl2/BCl3 chemistry. Resultant etch rate/etch profiles are studied as a function of ICP process parameters and mask sidewall angle. Etch rate decreased significantly with etch depth and increased with pressure. The etch yield of ICP process increased with pressure indicating etch process is limited by mass transfer of reactive species. Maximum etch yield of 19 is obtained for 200 mm etch depth at 50 mTorr pressure, 950 W coil power, 80 W substrate bias with an etch rate w4.9 mm/min. Final aspect ratio of etched holes is increased with pressure from 1.02 at 20 mTorr to 1.38 at 40 mTorr respectively for fixed etch time and then decreased to 1.24 at 50 mTorr pressure indicating loss of anisotropy. The resultant final etch profile and undercut is also found to have a strong dependence on the initial photo resist mask sidewall angle and its selectivity in the pressure range of 20e50 mTorr. The pressure range of 30e40 mTorr is resulted in desired etch profile of 50 mm dia. holes with relatively high etch rate w4 mm/min and desired anisotropy for 200 mm depths. The etch uniformity obtained for the process is better than 5% over a 3-inch GaAs wafer. Acknowledgements The authors are thankful to DRDO, Government of India for providing financial support to carry out experimental work.
457
Authors are also thankful to Dr. Ashok Kapoor for helping in EDAX measurements.
References [1] Williams RE. GaAs processing technology. 1st ed. Boston: Arctech House; 1985. [2] Bonneau D, Borkowski P, Shelley R, Fortier A, Young MC, Fragos C, et al. GaAs MANTECH technical digest; 2002:113e6. [3] Lim WT, Baek IG, Jung PG, Lee JW, Cho GS, Lee JI, et al. J Electrochem Soc 2004;151:G163e6. [4] Agarwal VR, Rawal DS, Vyas HP. J Electrochem Soc 2005;152:G567e76. [5] Westerman R, Johnson D, Devre M. Chip Magazine 2002;7:29e30. [6] Wang Chih Chang, Lin Yu-Lu, Lin Shun-Kuan, Li Chun-Sheng, Huang Hou-Kuei, Wu Chang-Luen, et al. J Vac Sci Technol B 2007;25:312e7. [7] Golka S, Schartner S, Schrenk W, Strasser G. J Vac Sci Technol B 2007;25: 839e44. [8] Vyvoda MA, Lee H, Malyshev MV, Klemens FP, Cerullo M, Donnelly VM, et al. J Vac Sci Technol A 1998;16:3247e58. [9] Hahn Yoon-Bong, Pearton Stephen J. Korean Jn Chem Eng 2000;17(3):304e9. [10] Clayton F, Westerman R, Johnson D. GaAs MANTECH technical digest; 2002:121e4. [11] Chen YW, Doi BS, Ng GI, Radhakrishnan K, Tan CL. J Vac Sci Technol B 2000;18: 2509e12. [12] Simpson William C, Yarmoff Jory A. Annu Rev Phys Chem 1996;47:527e54. [13] Lee JW, Jeon MH, Devre M, Mackenzie KD, Johnson D, Sasserath JN, et al. Solid State Electronics 2001;45:1683e6. [14] Maeda T, Lee JW, Shul RJ, Han J, Hong J, Lambers ES, et al. Appl Surf Sci 1999;143:174e82.
Contents lists available at ScienceDirect
Vacuum journal homepage: www.elsevier.com/locate/vacuum
Study of inductively coupled Cl2/BCl3 plasma process for high etch rate selective etching of via-holes in GaAs D.S. Rawal a, *, V.R. Agarwal a, H.S. Sharma a, B.K. Sehgal a, R. Muralidharan a, Hitendra K. Malik b a b
Solid State Physics laboratory, Lucknow Road, Timarpur, Delhi 110054, India PWPAL, Department of Physics, Indian Institute of Technology, Delhi 110016, India
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 April 2010 Received in revised form 18 August 2010 Accepted 22 August 2010
We have investigated the selective etching of 50 mm diameter via-holes for etch depth >200 mm using 30 mm thick photo resist mask in Inductively Coupled Plasma system with Cl2/BCl3 chemistry. Resultant etch rate/etch profiles are studied as a function of ICP process parameters and photo resist mask sidewall profile. Etch yield and aspect ratio variation with process pressure and substrate bias is also investigated at constant ICP power. The etch yield of ICP process increased with pressure due to reactant limited etch mechanism and reached a maximum of w19 for 200 mm depth at 50 mTorr pressure, 950 W coil power, 80 W substrate bias with an etch rate w4.9 mm/min. Final aspect ratio of etched holes is increased with pressure from 1.02 at 20 mTorr to 1.38 at 40 mTorr respectively for fixed etch time and then decreased to 1.24 at 50 mTorr pressure. The resultant final etch profile and undercut is found to have a strong dependence on the initial slope of photo resist mask sidewall angle and its selectivity in the pressure range of 20e50mTorr. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: GaAs Via-hole ICP Etching Etch yield Aspect ratio
1. Introduction GaAs devices are used extensively in the wireless telecommunications industry, where the high electron mobility of GaAs makes it well suited for high frequency, low noise and high gain applications. Although it has excellent electrical properties, GaAs is a relatively poor thermal conductor, making it difficult to remove heat efficiently from power devices. A commonly used solution is to form via-holes grounds from the wafer backside to the front side circuitry. Such connections provide a good thermal path for heat removal as well as a low impedance ground for RF (radio frequency) devices. This backside via formation is one of the final steps in the device fabrication. After completion of the front side processing, the wafer is mounted face down on a carrier wafer and mechanically thinned to a thickness of approximately 100e200 mm. The back of the wafer is then patterned using photo resist and the holes are etched using reactive ion etching (RIE) through the thinned substrate, stopping on the front side metal. After resist removal the vias are metallized, typically by sputtering a gold seed layer followed by gold electroplating to act as the heat sink/ground connection [1].
* Corresponding author. Fax: þ91 11 23984285. E-mail address: [email protected] (D.S. Rawal). 0042-207X/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.vacuum.2010.08.022
Inductively Coupled Plasma (ICP) etching has been replacing conventional RIE for GaAs backside via etching because of several advantages. Throughput improvement utilizing significantly faster etching rate have already been reported in the literature [2]. Furthermore, the ICP tools provide better control of via size, repeatability and reproducibility [3,4]. The ICP tools produce significantly different via dimension as compared to the conventional RIE tools, if the same size mask used, due to its nature of etching process. The etch profile and surface morphology of viahole grounds is important not only for the inductance consideration but also for the success of backside metallization. The smooth morphology of the etched sidewalls provides reliable and good electrical contact with low resistance. Etching is mainly carried out in chlorine/fluorine plasma. A number of gas combinations CCl2F2, CCl2F2/CCl4, SiCl4/Cl2, BCl3/Cl2/Ar, Cl2/Ar and Cl2/BCl3 have been utilized to fabricate via-holes [4]. Each gas combination has its advantages and disadvantages. The previous work reported using Cl2/BCl3/Ar for etching of GaAs is mainly carried out using RIE for etch depths w100 mm and etch rate reported are much lower 1 mm/min for 3-inch GaAs wafer. Cl2/BCl3 gas mixture with ICP process is being increasingly used for fabrication of via-holes at high etch rates with excellent anisotropy and smooth surface morphology. Generally reported etch depths using ICP for via-hole etching applications in GaAs Monolithic Microwave Integrated Circuits (MMIC) are less than 200 mm using photo resist mask due to lower etch rate and poor mask selectivity. The substrate
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
thickness of 200 mm is preferable as it results in higher yield and throughput due to easier handling of fragile wafers without carriers and also improves the RF performance due to lower electrical losses in MMIC micro strip interconnects on the front side. It is not easy to etch 200 mm deep holes at high etch rates with desired profile, good reproducibility and repeatability as etching depends upon various parameters like component of radicals, ions in plasmas, flow of gases, dc bias and type of mask used. Therefore, it becomes important to understand the effect of these plasma parameters on etch rate, anisotropy, etch profile, aspect ratio, mask selectivity and then develop the process for particular application [3e6]. The etch rate is the GaAs thickness etched in 1 min (mm/ min.), anisotropy is defined as {1 (lateral etch rate/vertical etch rate)}, etch profile refers to the shape of etched feature crosssection, aspect ratio is etch depth/feature’s lateral dimension and selectivity is defined as the etch rate of GaAs/etch rate of photo resist mask. We are reporting the ICP etching process of 50 mm diameter via-holes at relatively high average etch rate for an etch depths of w200 mm, on 3-inch GaAs wafer using positive photo resist mask. As etch depth increases, etch rate reduces drastically, with photo resist mask, affecting process throughput significantly in a production environment. Etch rate variation with ICP process parameters i.e. pressure and substrate bias power, are studied in detail at a constant maximum available ICP coil power to have high plasma density that results in high etch rates. At high plasma density, pressure and bias voltage are the most important parameters to control etch profile and selectivity as they mainly control the ion energy i.e. physical component of etching. Aspect ratio and etch yield are two important parameters that characterise any etch process [7e9]. Variation of aspect ratio and etch yield with process parameters is also studied. As thick photo resist mask is required to etch deep holes, its sidewall profile and selectivity plays an important role in deciding the final etch profile. We are also reporting this effect by varying sidewall angle of photo resist mask [10]. To our knowledge these parameters for an ICP etch process involving etch depths of w200 mm with such high average etch rates have not been reported so far using Cl2/BCl3 chemistry. The main purpose of the study was to develop a relatively high etch rate, production viable ICP etching process to etch w200 mm deep via-holes with acceptable via-hole etch profile, photo resist mask selectivity and etch uniformity for fabricating 50 mm diameter viahole ground connections in GaAs MMICs over a 3-inch wafer.
Fig. 2. SEM cross-section of hole etched upto a depth of more than 175 mm.
positive photo resist AZ 4620 and patterned with 50 mm diameter holes. The patterned photo resist was then post baked at 120 C for variable time to introduce a sloped photo resist profile with improved adhesion. These patterned wafers were then mounted on a carrier wafer with wax to make wafer loading compatible with ICP system and also provide cooling of the wafer during etching process. The patterned wafers were then etched in standard ICP system, one at a time. Plasma of etcher is inductively coupled through a coil at 13.56 MHz, with independent energy control provided by 13.56 MHz RF biasing on the substrate. Helium gas was used to cool backside of the wafer. The substrate temperature was set at 20 C for all test conditions. The etch chemistry was a mixture of Cl2/BCl3 through mass flow controlled process gas lines. The chamber was evacuated to a base pressure of 9e-3 mTorr, by a turbo molecular pump backed by a dry mechanical pump, before initiating the etch process. The etch gases mixture was introduced through an annular region at the top of chamber lid. ICP etching was carried out using only photo resist mask to have less complex process with good etch surface morphology. All the ICP experiments were carried out at near-maximum available coil power and Cl2/BCl3 flow rate ratio of 4:3 for a fixed total flow rate w280 sccm to have high plasma density [11] and increased concentration of reactive Cl species, that
2. Experimental details All test wafers were 3-inch S.I. GaAs wafers with thickness w650 mm these wafers were then coated with in 30 mm thick
8 Et c h R a t e ( u m /m in )
7 6 5 4 3 2 1 0 0
50
100
150
200
Etch depth(um)
Fig. 1. Etch rate as function of etch depth (950 W ICP Power, 65 W Substrate bias and 30 mTorr).
453
Fig. 3. SEM photograph showing deposition on the etched sidewall.
454
D.S. Rawal et al. / Vacuum 85 (2010) 452e457 1.6
A s p e c t R a t io
1.4 1.2 1 0.8 0.6 0.4 0.2 0 0
10
20
30
40
50
60
Pressure(mT)
Fig. 6. Aspect ratio variation with pressure (950 W ICP Power, 80 W Substrate bias). Fig. 4. EDAX spectrum of the etched sidewall etched using thick photo resist mask.
resulted in high etch rates with better etch surface morphology. Higher than 4:3 flow rate ratio increased the etch rate but at the cost of surface morphology, whereas lower flow rate ratio decreased the etch rate significantly. The etch rate, etch depth, etch profile, aspect ratio, etch yield, mask selectivity and surface morphology of etched holes were determined by cleaving through the etched features and examining the sample under scanning electron microscope. Deposition on the etched sidewall was analysed by high resolution INCA EDAX.
3. Results and discussion
hole is very smooth however the sidewalls are rough with some deposition. Fig. 3 shows the SEM photograph of the etched sidewall and Fig. 4 shows the EDAX spectrum of the etched sidewall with peaks corresponding to carbon and chlorine in addition to Ga and As, most probably due to the presence of CClx polymer on the sidewalls. The etching takes place via chemi-sorption of the highly reactive Cl radical on wafer surface, forming volatile molecule and subsequently de-sorption of this volatile molecule. The ions are mainly responsible for bombarding the surface creating damage and increasing reaction probability of the arriving reactive species. The incident ions also assist in the de-sorption of the reaction product via sputtering. The reaction probably occurs in stepwise manner with gallium and arsenic mono-chlorides formation first
3.1. Etch rate variation with etch depth Fig. 1 shows the average etch rate variation with etch depth for ICP process at 950 W coil power, 30 mTorr pressure, 65 W platen power and 45 min. of etch time. It clearly shows that the average etch rate is decreasing with etch depth from 7 mm/min for 70 mm depth to 3.9 mm/min for 175 mm on 3-inch wafer. This is mainly due to the re-deposition of sidewall material and sputtered photo resist on etched hole due to positive slope in etch profile as depth increases. This redepsition on the sidewall promotes sidewall smoothness and anisotropy but at the cost of etch rate. This reduction in etch rate with etch depth is very high in RIE and it is almost impractical to etch depths >100 mm depths with photo resist mask [4]. Therefore, the average etch rate achieved for smaller etch depths is much higher than >100 mm etch depths for same diameter holes. Fig. 2 shows the cross-section of a hole etched up to the depth of more than 150 mm with good anisotropy and surface morphology using photo resist mask. The bottom of the
Et c h r a t e ( u m /m in )
6 5 4 3 2 1 0 0
10
20
30
40
50
60
Pressure(mTorr) Fig. 5. Etch rate variation with pressure for fixed etch time (950 W ICP Power, 80 W Substrate bias).
Fig. 7. (a) Etch cross-section of two holes 150 mm apart etched at 40 mTorr pressure (950 W ICP Power, 90 W substrate bias). (b) Etch cross-section of two holes 150 mm apart etched at 50 mTorr pressure (950 W ICP Power, 90 W substrate bias).
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
455
20 18
Et c h Y ie ld
16 14 12 10 8 6 4 2 0 0
10
20
30
40
50
60
Pressure(mTorr)
Fig. 8. Etch yield calculated as a function of pressure (950 W ICP Power, 80 W Substrate bias).
followed by formation of di-chlorides and tri-chlorides [12] as shown in following equations: GaAs þ 2Cl / GaCl þ AsCl
(1)
GaCl þ Cl / GaCl2
(2)
AsCl þ Cl / AsCl2
(3)
GaCl2 þ Cl / GaCl3
(4)
AsCl2 þ Cl / AsCl3
(5)
3.2. Etch rate variation with process pressure Fig. 5 shows the experimentally observed etch rate variation with process pressure at 950 W coil power and 80 W of substrate bias for an etch time of 45 min. This graph clearly indicates that the etch rate is a strong function of process pressure in the range 20e50 mTorr and is increasing with pressure due to increased density of reactive species, suggesting reactant limited etch mechanism [13] but anisotropy is maintained up to 40 mTorr due to constant ion energy incident on the substrate at 80 W of substrate bias. In other words, physical component of etching is fairly constant up to 40 mTorr. Above 40 mTorr etch rate is increasing but at the cost of anisotropy resulting in large undercut making etch profile unsuitable. Etch rate is increased to 4.9 mm/min at 50 mTorr from 2.9 mm/min at 20 mTorr, for an etch time of 45 min, suggesting that the dominance of chemical etching component with increased concentration of reactive species.
Fig. 10. SEM cross-section of unbaked, 50 mm diameter, vertical photo resist mask profile (sidewall angle: 88 ).
3.3. Aspect ratio with process pressure Aspect ratio in ICP etching is a function of process pressure and time. Fig. 6 shows the aspect ratio variation with pressure at fixed etch time for 50 mm diameter hole. It is seen that aspect ratio is increasing with pressure from 1.02 at 20 mTorr to 1.38 at 40 mTorr of pressure and then starts decreasing and reaches to a value of 1.24 at 50 mTorr pressures mainly due to more of chemical etching at higher pressures, leading to a reduction in anisotropy. Aspect ratio becomes very important when two features to be etched are very close due to lay out constraints. Fig. 7(a) and (b) show the etch cross-section of two 50 mm dia. holes, 150 mm apart etched at 40 mTorr and 50 mTorr process pressure to a depth of greater than 200 mm respectively. At 40 mTorr the holes are etched with high anisotropy that resulted in higher aspect ratio in comparison to holes etched at 50 mTorr, which resulted in merging of two features that could lead to poor device performance. 3.4. Etch yield with process pressure Etch yield can be defined as number of atoms etched per incident ion and is a function of etch rate, density and molecular weight of the material to be etched. The etch yield is proportional to the energy transferred by the incident ions through nuclear collisions
20 19.5 Et c h Y ie ld
19 18.5 18 17.5 17 16.5 0
20
40
60
80
100
120
Substrate bias(W)
Fig. 9. Etch yield variation obtained with substrate bias (950 W ICP Power, 30 mTorr pressure).
Fig. 11. SEM cross-section of etched hole showing resultant etch profile (Etch depth w225 mm).
456
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
Undercut(um )
80 60 40 20 0 40
60
80
100
Mask Sidewall Angle(Degree) Fig. 14. Undercut measured as a function of photo resist mask sidewall angle for 50 mm diameter holes etched under same conditions.
Fig. 12. SEM photo resist profile obtained after oven baking at 120 C, 60 min.
in the material. Its calculation for this ICP etch process is based on measured etch rate and predicted ion flux and is as high as 10e20 [14]. Fig. 8 shows the predicted etch yield variation with pressure for etching 50 mm diameter holes for etch depths up to 200 mm. Etch yield is increasing with pressure as etch rate is increasing and reaches a value of w19 at 50 mTorr. The increase in etch yield with pressure indicates that etching process is limited by mass transfer of reactive species This again suggests that the chemical species required for converting the surface atoms to volatile molecules cannot be provided by the ion itself instead radicals are also equally participating. The radicals are highly reactive species that form volatile product molecules in the presence of bombarding ions as discussed earlier.
3.5. Etch yield with substrate bias Fig. 9 shows etch yield variation as a function of substrate bias. Increasing the bias power from 60 to 100 W at pressure 30 mTorr resulted in reduction in etch yield from 19.5 to 16.8 respectively for an etch time of 45 min. This reduction in etch yield is probably due to higher ion bombardment at 100 W that resulted in sputtering and re-deposition of photo resist on the etched features reducing the average etch rate under these plasma conditions contradictory to the normally observed behaviour.
In other words, etching process is more physically driven at 100 W substrate bias. This reduction in etch yield has in turn resulted into better etch sidewall morphology at 100 W; because of the higher ion bombardment that may sputter the surface evenly regardless of defects present. 3.6. Etch profile variation with mask edge profile An anisotropic etch profile offers the best CD control but practically it is difficult to get a good metal contact in such vertical profiles for via-hole grounds. In order to lower the aspect ratio for ease of metallization, a sloped via etch profile is preferred. Apart from ICP process parameters, initially sloped photo resist mask profile and it’s etch selectivity plays important role in deciding the final via etch profile. Fig. 10 shows the unbaked, 30 mm thick resist profile with 88 sidewall slope and Fig. 11 shows the resultant etch profile for depth >200 mm. This profile is highly anisotropic and unsuitable for metal interconnection. Up to a depth of w100 mm, etched walls are almost vertical as resist erosion is much lower initially with vertical photo resist sidewalls. As photo resist erodes at corners due to longer plasma exposure, its sidewalls become tapered and etch profile starts to broaden as clearly seen in Fig. 11. Also GaAs: photo resist selectivity was found to be w10:1 in this case as there is no resist hardening involved after photolithography process making resist erosion faster during plasma exposure. A baking cycle after photolithography was incorporated to tailor the resist profile by oven baking at 120 C, for 45 min and 60 min. The sidewall angle changed to 78 and 70 for 45 min bake and 60 min bake respectively. Fig. 12 shows the photo resist profile obtained after 60 min bake with w70 sidewall angle. Baking temperature above 120 C has resulted in resist flowing, deteriorating the geometry as well as making its removal difficult after etching. Fig. 13 shows the etch profile corresponding to 70 sidewall photo resist mask. Fig. 14 shows the undercut measured as a function of mask sidewall angle. It is evident from this graph is that undercut is a strong function of photo resist mask profile. It needs to be mentioned that via-hole etch profiles in Figs. 11 and 13 are not suitable for via ground connection. The acceptable profile of 200 mm deep via-hole is almost identical to the profile shown in Fig. 2 and is obtained with resist baking at 120 C for 45 min. Mask selectivity for baked mask profiles was better than 12:1. Resist removal in Cl2/BCl3 chemistry appears to be physically driven e the resist etch rate increased solely with the RF bias power degrading selectivity. The etch uniformity measured for the etched holes over the three inch GaAs wafer is better than 5% for all the experiments indicating good gas flow uniformity in the ICP chamber. 4. Conclusion
Fig. 13. ICP etched via profile obtained with 60 min baked resist profile (Etch depth w229 mm).
We have investigated the selective etching of 50 mm diameter via-holes for etch depth 200 mm using thick photo resist mask in
D.S. Rawal et al. / Vacuum 85 (2010) 452e457
Inductively Coupled Plasma system with Cl2/BCl3 chemistry. Resultant etch rate/etch profiles are studied as a function of ICP process parameters and mask sidewall angle. Etch rate decreased significantly with etch depth and increased with pressure. The etch yield of ICP process increased with pressure indicating etch process is limited by mass transfer of reactive species. Maximum etch yield of 19 is obtained for 200 mm etch depth at 50 mTorr pressure, 950 W coil power, 80 W substrate bias with an etch rate w4.9 mm/min. Final aspect ratio of etched holes is increased with pressure from 1.02 at 20 mTorr to 1.38 at 40 mTorr respectively for fixed etch time and then decreased to 1.24 at 50 mTorr pressure indicating loss of anisotropy. The resultant final etch profile and undercut is also found to have a strong dependence on the initial photo resist mask sidewall angle and its selectivity in the pressure range of 20e50 mTorr. The pressure range of 30e40 mTorr is resulted in desired etch profile of 50 mm dia. holes with relatively high etch rate w4 mm/min and desired anisotropy for 200 mm depths. The etch uniformity obtained for the process is better than 5% over a 3-inch GaAs wafer. Acknowledgements The authors are thankful to DRDO, Government of India for providing financial support to carry out experimental work.
457
Authors are also thankful to Dr. Ashok Kapoor for helping in EDAX measurements.
References [1] Williams RE. GaAs processing technology. 1st ed. Boston: Arctech House; 1985. [2] Bonneau D, Borkowski P, Shelley R, Fortier A, Young MC, Fragos C, et al. GaAs MANTECH technical digest; 2002:113e6. [3] Lim WT, Baek IG, Jung PG, Lee JW, Cho GS, Lee JI, et al. J Electrochem Soc 2004;151:G163e6. [4] Agarwal VR, Rawal DS, Vyas HP. J Electrochem Soc 2005;152:G567e76. [5] Westerman R, Johnson D, Devre M. Chip Magazine 2002;7:29e30. [6] Wang Chih Chang, Lin Yu-Lu, Lin Shun-Kuan, Li Chun-Sheng, Huang Hou-Kuei, Wu Chang-Luen, et al. J Vac Sci Technol B 2007;25:312e7. [7] Golka S, Schartner S, Schrenk W, Strasser G. J Vac Sci Technol B 2007;25: 839e44. [8] Vyvoda MA, Lee H, Malyshev MV, Klemens FP, Cerullo M, Donnelly VM, et al. J Vac Sci Technol A 1998;16:3247e58. [9] Hahn Yoon-Bong, Pearton Stephen J. Korean Jn Chem Eng 2000;17(3):304e9. [10] Clayton F, Westerman R, Johnson D. GaAs MANTECH technical digest; 2002:121e4. [11] Chen YW, Doi BS, Ng GI, Radhakrishnan K, Tan CL. J Vac Sci Technol B 2000;18: 2509e12. [12] Simpson William C, Yarmoff Jory A. Annu Rev Phys Chem 1996;47:527e54. [13] Lee JW, Jeon MH, Devre M, Mackenzie KD, Johnson D, Sasserath JN, et al. Solid State Electronics 2001;45:1683e6. [14] Maeda T, Lee JW, Shul RJ, Han J, Hong J, Lambers ES, et al. Appl Surf Sci 1999;143:174e82.