Atomically Smooth Gallium Nitride Surfaces Generated ... - IEEE Xplore

International Conference on Planarization/CMP Technology・November 19−21, 2014 Kobe

Atomically Smooth Gallium Nitride Surfaces Generated by Chemical Mechanical Polishing with Non-noble Metal Catalyst(Fe-Nx/C) in Acid Solution Yuyu Liu

Li Xu, Guoshun Pan , Chunli Zou, Xiaolei Shi

Multidisciplinary Research on the Circulation of Waste Resources, Graduate School of Environmental Studies, Tohoku University Sendai, Japan [email protected]

State Key Laboratory of Tribology, Tsinghua University Beijing, China Research Institute of Tsinghua University in Shenzhen Shenzhen, China [email protected] Abstract- In this paper, a novel method for preparing atomically smooth gallium nitride (GaN) wafer surfaces which involves chemical mechanical polishing with a non-noble metal catalyst (Fe-Nx) in acidic slurry is presented. It was confirmed that non-noble metal catalyst based slurry could be used for gallium face of GaN. Atomic force microscope images of the processed surface indicate that an atomically flat surface with Ra=0.0518 nm was achieved after planarization and the processed surface has an atomic step-terrace structure. Besides, the rate of removal of the GaN surface was measured to be approximately 66.9 nm/h, more than triple times higher than that nothing was used as catalyst. I. INTRODUCTION

Recently, considerable attention has been focused on III nitride-based (AlN, GaN and InN) materials attributed to their wide potential application in optoelectronics, such as light emitting diodes (LEDs)[1-2], lasers[3], solar cells[4], and detectors[5-6]. However, the full benefit of the nitride devices has not been realized due to the lack of commercially available, high quality native nitride substrates[7-8]. GaN samples have been grown with several different techniques, including high-temperature high-pressure near equilibrium growth[9-10], sodium flux method[11], ammonothermal method[12], metal organic chemical vapor deposition (MOCVD)[7,13] and hydride vapor phase epitaxy (HVPE)[14-17]. Only the HVPE technique has demonstrated the potential of producing large diameter GaN[18-20].At present, the principle substrates on which III nitride-based materials devices are commercially grown and fabricated are sapphire and silicon carbide (SiC) due to their stability in the high-temperature and hydrogen-containing growth environment as well as their availability. In general, however, the surface of the HVPE-grown GaN template is much rougher than that of GaN grown by MOCVD because of the faster growth rate. The surface roughness will be exacerbated when we grow thicker layers. The regular single-step HVPE growth for GaN lms thinner than 10 ȝm exhibits a high dislocation density of 109 cmí2[21], and even using complicated nucleation control including ex-situ air exposure, a thickness of 8 ȝm was needed to obtain a dislocation density of 2 × 108 cmí2[22]. Furthermore, threading dislocations (TDs) have been observed to form

978-1-4799-5557-2/14/$31.00 ©2014 IEEE

237

randomly (a) at the outset of growth of and within the initial III-nitride ‘‘nucleation’’ layer deposited on sapphire substrates[23] and (b) from clusters of point defects at the SiC substrate/ buffer layer interface [24-25]. Dislocations are also generated to accommodate stresses within tilt boundaries formed during the coalescence of three-dimensional GaN islands within the template layer subsequently grown on the nucleation layer on sapphire substrates[26] and ultimately degrade the surface roughness. Thus, the as-grown bulk single crystals have to undergo a wafering and polishing process to form a wafer with complete flat and scratch -free surface so as to used GaN materials as substrates for device fabrication. Chemical mechanical polishing (CMP), uses a combination of chemical and mechanical effects, is applied to the surface as a final treatment to produce atomic level surface flatness by removing the damage on or near the surface due to mechanical polishing. And research concerning the CMP of bulk GaN substrates has been conducted in the latest few years [27-30]. However, gallium face of GaN showed extremely low polishing rate due to their extreme hardness and strong stability against chemical. The devoted efforts and experiences of previous researchers in the development of mechanical processing of sapphire can be applied to the mechanical process of GaN. Although Hideo Aida et al. found that gallium face of GaN was able to polish with colloidal silica slurry[31], the remove rate of GaN in CMP with colloidal silica exhibits a range of 10 to 100 nm per hour due to strong chemical inertness[29,30, 32-33]. Assuming that the depth of the damaged layer to be removed by CMP and the removal rate are 1.6 mm and 40 nm/h, respectively, the processing time will be 40 h, which means this is not a practical method for commercial production. Thus, improving the CMP removal rate for GaN will lead to acceleration in the realization of next generation optoelectronics devices on GaN. In this work, the CMP process gallium face of GaN with colloidal silica based slurry is facilitated by the


presence of non-noble metal materials (Fe-Nx/C) as catalyst to improve the material removal rate (MRR) of GaN. The evidence of damage-free removal of Ga-faced GaN substrate was verified by atomic force microscopy (AFM). And AFM image of the processed surface indicate that an atomically flat surface with Ra=0.0518nm was achieved after etching and the processed surface has an atomic step-terrace structure. Besides, the rate of removal of the GaN surface was measured to be approximately 66.9 nm/h, more than triple times higher than that nothing was used as catalyst. II. EXPERIMENT

2.1 Synthesis of carbon-supported Fe-Nx nanoparticles. The pyrolyzed carbon supported Fe-Nx catalysts were synthesized with selected dopants, TsOH, through a solvent-grinding method followed by heat-treatment at desire temperature. According to the catalyst preparation method described by Li Xu et al.[34]. 2.2 Polishing setup and process control. Catalyst- and catalyst-free-based colloidal silica slurries for CMP were prepared. The catalyst-free-based colloidal silica slurry was from a commercial source. While catalyst-based colloidal silica slurry was made by mixing the catalyst-free-based colloidal silica with different content of catalysts, the pH was 3.0 for all slurry. A 2-inch-diameter HVEP-grown GaN wafer from commercial source was used in this study. The surface roughness of the HVEP-grown GaN was removed by CMP with catalyst. As a reference, the other 2-inch-diameter HVEP-grown GaN template from the same commercial source was also prepared polished with catalyst-free. Both of the two GaN wafers were mounted on a stainless steel plate with an adhesion wax and subject to CMP to remove the surface roughness. A soft polyurethane polishing pad was used for CMP. And the detailed information about slurry and polishing conditions are listed in Table 1. And CMP was continued until the surface morphology was removed. The specimens before and after polishing were ultrasonic rinsed in alcohol for 15 minutes, dried by nitrogen, and then measured by Ohaus DV215CD electron balance to estimate the MRR of GaN as (1)[35]:

Table 1. Polishing conditions for the HVEP-GaN wafer CMP with catalyst CMP without catalyst Polishing pad polyurethane polishing pad Slurry Colloidal silica from commercial source Slurry pH 3.0 3.0 Abrasive particle size / nm 80 80 Abrasive concentration / % 40 40 Applied pressure / psi 3.51 3.51 Platen rotation / rpm 100 100 Stainless steel plate / rpm 50 50 MRR / (nm/h) 66.9 20.5

process were observed on the untreated GaN, as shown in Fig.1 (a). While, hill crocks on the GaN surface were disappeared after CMP 2 hours, as shown in Fig.1 (b) and Fig.1 (c), clearly demonstrated that the CMP process well removed surface defect over a large area of GaN wafer. In contrast, there are a few scratches were observed on the flattened GaN surface which polished with catalyst-free slurry, and the absence of such scratches on the flattened GaN surface with slurry mixed with 8 ppm catalyst, obviously indicted that the catalyst effectively removed surface damages. Furthermore, high definition atomic step-terrace structure acquired from wafer surface polished by CMP with 8 ppm Fe-Nx/C was acquired, as shown in Fig.2. And a mass of residual products were at the edge of terraces. The reason maybe that when the mechanical actions is smaller than chemical actions and the polishing pressure is smaller, the impact force on terraces by single abrasive would also be smaller, so the abrasives with a lower impact force may have insufficient capability to remove the residual products at the edge of terraces. Moreover, the catalyst is beneficial to improve the rate of removal of GaN. which can be seen in Table 1. 3.2 The removal characteristics of GaN wafer processed by using Fe-Nx/C nanoparticles in the slurry. The development of the GaN removal rate on the catalyst concentration is presented in Fig.3. (a)

(c)

(b)

MRR = 4ǻm2 (1) ȡʌd t

Here, ¨m is the weight change after polishing, ȡ is the sapphire density (3.98 g/cm3), t is the polishing time, d is the diameter of sapphires and MRR is the corresponding removal rate of GaN. 2.3 Characterization methods. The surface morphology of the GaN wafers were evaluated by using optical microscopy (OM) and atomic force microscopy (AFM). As for AFM, image covering an area of 1.0 × 1.0 ȝm2 (micro) was obtained and the roughness average (Ra) surface roughness was measured for its observation.

500 um

500 um

Fig. 1 Optical microscopy image of GaN wafer surface (a) before CMP, (b) after 2 h of CMP without catalyst in slurry and (c) after 2 h of CMP with 8 ppm Fe-Nx in slurry.

III. RESULTS AND DISCUSSION

3.1 Characterization of the GaN surface. Fig.1 show differential contrast images of regions of the GaN surface before and after CMP. A great many of hill rocks which were generated during the GaN wafer fabrication

238

500 um

Fig.2

Image of the GaN wafer surface processed by 8 ppm Fe-Nx

catalyst slurry for 2 h (1 × 1 ȝm2 area, roughness Ra: 0.0518 nm)


The removal rate of GaN (nm/h)

75

Fig. 3

60 45 30 20.5365 nm/h

15 0

0

3

6

9

12

The content of Fe-Nx in slurry (ppm)

15

Dependence of the GaN removal rate on the catalyst concentration in the slurry.

The content of catalyst mixed in the slurry affects the polishing rate, indirectly confirming the assumption presented in Table 1 that Fe-Nx/C catalyst is beneficial to improve the MRR of GaN. The polishing rate increases with an increase in the catalyst content. However, when the concentration exceeds 8 ppm, growth impediment is observed. During CMP, electrons in the valence band of the GaN are transferred to the catalyst in the contact region, between the semiconductor and catalyst. The pits generated on the GaN surface enhance oxidation, and the electrons in the catalyst are consumed during the reduction of the oxidation agent in the slurry. The resulting oxides on the GaN surface dissolved in the slurry. However, when the content of catalyst is much more than a threshold value (8 ppm), the rate of removal of GaN became slower. Although, the contact area or contact opportunity between catalyst and GaN is larger, which is beneficial for electrons’ transportation between GaN surface and Fe-Nx/C, but there is different in the rate of electrochemical reaction at the region between catalyst and oxidant and the rate of oxidant migration in the slurry to the catalyst surface. When a chemical species (oxidant) participating in an electrochemical reaction on the catalyst is in short supply, the concentration of this species at the surface decreases causing diffusion, which is added to the migration transport towards the surface in order to maintain the balance of consumption and delivery of that species. In addition, the amount of catalyst added in the slurry is more than 8 ppm, there are amount of catalyst accumulated in the micropores on the polishing pad, rather than flow away with the slurry, and hindered fresh slurry flow into the region between GaN and polish pad, and the pad was polluted resulted in some changes on the pad which is bad for CMP of GaN. The most fundamental and basic material removal model in CMP is the Preston Model, which is application for glass polishing. This equation states that the material removal rate is directly proportion to pressure and relative velocity as (2)

MRR = k p PV

(2) Where MRR is the material removal rate in m/min, P is the down pressure in N/m2, V is the relative velocity between the pad and wafer in m/min and Kp is the Preston coefficient in m2/N. The Preston coefficient depends on various factors that can affect the removal rate such as friction force, chemical reaction, heating and so on. But Park [36] proposed that the removal rate was affected by the pad surface properties during the CMP. For example,

239

MRR increases with the pad surface roughness. Yu et al. [37] considered that the effect of pad surface roughness, and dynamical interaction between the pad and wafer, is based on the asperity theory. Furthermore, Zhao [38] illuminated that abrasive particles can demonstrate a threshold pressure during CMP process, which may have played a critical role in MRR. When the abrasive particles are rolling against the wafer surface under a pressure lower than the threshold value, the removal rate will will be negligible, the removal rete was found to be significant only if the abrasive particles held by the pad were sliding against the wafer surface. In other words, removal rate was found to be negligible if applied pressure is less than the minimum threshold pressure. However, in our work, we found that: (1) The removal rate of GaN obeys Prestonian behavior when the downward pressure is lower than 3.51 psi. Since GaN is chemical inert and physically hard material, the obtained removal rate was extremely low, from 13.8 to 33.4 nm/h under an applied pressure range of 0.5 to 2.5 psi. (2) The upmost removal rate of GaN is 66.9 nm/h when the downward pressure was 3.51 psi, and the MRR decreased when the applied pressure was larger than 3.51 psi, as well as the fact that numerous scratches were produced on the wafer's surface. Maybe this is because of the factor that the interspaces between soft pad and wafer were diminished sharply and the ration of the real contact area was smaller than the situation where with lower pressure. Furthmore, the slurry was hindered and can't flow into the contact area fluently, and resulted in minishing the removal rate of GaN. IV. CONCLUSION

In this work, we present a novel non-precious catalyst (Fe-Nx/C) to improve the removal mass of GaN and Fe-Nx/C showed remarkable catalytic performance towards GaN in colloidal silica slurry. The results demonstrate that the proposed Fe-Nx/C is effective as catalyst to improve the remove rate of sapphire in CMP. The results show that when the slurry contains 8 ppm Fe-Nx/C, the rate of removal of the GaN surface was measured to be approximately 66.9 nm/h, more than triple times higher than that nothing was used as catalyst. AFM image shows that the processed surface has an better atomic step-terrace structure and the Ra is 0.0518 nm. ACKNOWLEDGMENT This work was supported by the International Science & Technology Cooperation Program of China (No. 2011DFA73410), National Key Basic Research Program of China-973 Program (No. 2011CB013102), National Natural Science Foundation of China (No. 91223202) and the Foundation for Innovative Research Groups of the National Natural Science Foundation of China (No.51321092). REFERENCES [1] Y.Y. Zhang, H.Y. Zheng, E.Q. Guo, Y. Cheng, J. Ma, L.C.


Wang, et al, “Effects of Light Extraction Efficiency to The Efficiency Droop of InGaN-based Light-emitting Diodes,” J. Appl. Phys., vol. 113, pp. 14502. January 2013. [2] W.C. Peng, Y. Chung and S. Wu, “Improved Luminance Intensity of InGaN–GaN Light-emitting Diode by Roughening Both The p-GaN Surface and The Undoped-GaN Surface,” Appl. Phys. Lett., vol. 89, pp. 041116, July 2006. [3] G. Fasol, “Room-Temperature Blue Gallium Nitride Laser Diode,” Science, vol. 21, pp. 1751-1752, June 1996. [4] J. Wu, W. Walukiewicz, K.M. Yu, W. Shan, J.W. Ager, E.E. Haller, et al, “Superior Radiation Resistance of In1íxGaxN alloys: Full-solar-spectrum Photovoltaic Material System,” J. Appl. Phys. Vol. 94, pp. 6477. October 2003. [5] A. Vardi, G. Bahir, F. Guillot, C. Bougerol, E. Monroy, S.E. Schacham, et al, “Near Infrared Quantum Cascade Detector inGaN/AlGaN/AlN Heterostructures,” Appl. Phys. Lett. vol. 92, pp. 011112. January 2008. [6] G. Ariyawansa, M.B.M. Rinzan, M. Alevli, M. Strassburg, N. Dietz and A.G.U. Perera, et al, “GaN/AlGaN Ultraviolet/infrared Dual-band Detecto,” Appl. Phys. Lett. vol. 89, pp. 091113, August 2008. [7] S. Nakamura, T. Mukai and M. Senoh, “Candela-class High-brightness InGaN/AlGaN Double-heterostructure Blue-light-emitting Diodes,” Appl. Phys. Lett., vol. 64, pp. 1687-1689. January 1994. [8] S. Nakamura, G. Fasol, S.J. Pearton, The Blue Laser Diode, Berlin: Springer-Verlag, pp. 7, 2000. DOI 10.1007 /978-3-662-04156-7. [9] S. Porowski and I. Grzegory, “Thermodynamical Properties of III–V Nitrides and Crystal Growth of GaN at High N2 Pressure,” J. Cryst. Growth, vol. 178 pp. 174-188, June 1997. [10] T. Inoue, Y. Seki, O. Oda, S. Kurai, Y. Yamada and T. Taguchi, “Growth of Bulk GaN Single Crystals by The Pressure-controlled Solution Growth Method,” J. Cryst. Growth, vol. 229, pp. 35-40, July 2001. [11] M. Aoki, H. Yamane, M. Shimada, T. Sekiguchi, T. Hanada, T. Yao, S. Sarayama, et al, “Growth of GaN Single Crystals from a Na-Ga Melt at 750°C and 5 MPa of N2,” J. Cryst. Growth, vol. 218, pp. 7-12, September 2000. [12] D. R. Ketchum and J. W. Kolis, “Crystal Growth of Gallium Nitride in Supercritical Ammonia,” J. Cryst. Growth, vol. 222, pp. 431-434, January 2001. [13] X.H. Wu, P. Fini, S. Keller, E.J. Tarsa, B. Heying, U.K. Mishra, et al, “Morphological and Structural Transitions in GaN Films Grown on Sapphire by Metal-Organic Chemical Vapor Deposition,” Jpn. J. Appl. Phys. Vol. 35, pp. L1648-L1651, November 1996. [14] X. Sun, R. P. Vaudo and G.R. Brandes, “Fabrication of GaN Wafers for Electronic and Optoelectronic,” Opt. Mater. vol. 23, pp. 1-5, August 2003. [15] H. Aida, S.W. Kim, T. Suzuki, K. Koyama, N. Aota, T. Doi, et al, “Surface Planarization of GaN-on-Sapphire Template by Chemical Mechanical Polishing for Subsequent GaN Homoepitaxy Electronic Materials and Processing,” ECS J. Solid State Sci. Technol. vol. 3, pp. P163-P168, April 2014. [16] H. Ashraf, J.L. Weyher, G.W.G. van Dreumel, A. Gzregorzyck and P.R. Hageman, “Thick GaN Layers Grown by HVPE: Influence of The Templates,” J Cryst. Growth. vol. 310, pp. 3957-3963, August 2008.

240

[17] K. Motoki, T. Okahisa, S. Nakahata, N. Matsumoto, H. Kimura, H. Kasai, et al, “Growth And Characterization of Freestanding GaN Substrates,” J Cryst. Growth. vol. 237-239, pp.912-921, April 2002. [18] M.K. Kelly, R.P. Vaudo, V.M. Phanse, L. Gorgens, O. Ambacher and M. Stutzmann, “Large Free-Standing GaN Substrates by Hydride Vapor Phase Epitaxy And Laser-Induced Liftoff,” Jpn. J. Appl. Phys. vol.38, pp. L217-L219, February 1999. [19] K. Motoki, T. Okahisa, N. Matsumoto, M. Matsushima, H. Kimura1, H. Kasai1, et al, “Preparation of Large Freestanding GaN Substrates by Hydride Vapor Phase Epitaxy Using GaAs As A Starting Substrate,” Jpn. J. Appl. Phys. vol. 40, pp. L140-L143, Decembe 2000. [20] S.S. Park, I.W. Park and S.H. Choh. “Free-Standing GaN Substrates by Hydride Vapor Phase Epitaxy,” Jpn. J. Appl. Phys. vol. 39 , pp. L1141-L1142, September 2000. [21] Y. Golan, X.H. Wu, J.S. Speck, R.P. Vaudo and V.M. Phanse, “Morphology and Microstructural Evolution in The Early Stages of Hydride Vapor Phase Epitaxy of GaN on Sapphire,” Appl. Phys. Lett. vol. 73, pp. 3090, September 1998. [22] D. Martin, J. Napierala, M. Ilegems, R. Butte, and N. Granjean, “High Quality Thin GaN Templates Grown by Hydride Vapor Phase Epitaxy on Sapphire Substrates,” Appl. Phys. Lett. vol. 88, pp. 241914, June 2006. [23] M.A. Moram, C.S. Ghedia, D.V.S. Rao, J.S. Barnard, Y. Zhang, M.J. Kappers, et al, “On The Origin of Threading Dislocations in GaN Films,” J. Appl. Phys. vol. 106, pp. 073513-073519, October 2009. [24] Z. J. Reitmeier, S. Einfeldt, R. F. Davis, X. Zhang, X. Fang and S. Mahajan, “Sequential Growths of AlN And GaN Layers on As-polished 6H-SiC (0001) Substrates,” Acta Mater. vol. 57, pp. 4001-4008, August 2009. [25] Z. J. Reitmeier, S. Einfeldt, R. F. Davis, X. Zhang, X. Fang and S. Mahajan, “Surface And Defect Microstructure of GaN and AlN Layers Grown on Hydrogen-etched 6H-SiC(0001) Substrates,” Acta Mater. vol. 58, pp. 2165-2175, April 2010. [26] X.H. Wu, P. Fini, E.J. Tarsa, B. Heying, S. Keller, U.K. Mishra, et al, “Dislocation Generation in GaN Heteroepitaxy,” J Cryst. Growth. vol. 189-190 pp. 231-243, June 1998. [27] D. Hanser, M. Tutor, E. Preble, M. Williams, X. Xu and D. Tsvetkov, et al, “Surface Preparation of Substrates from Bulk GaN Crystals,” J. Cryst. Growth, vol. 305, pp. 372-376. July 2007. [28] H. Aida, T. Doi, H. Takeda, H. Katakura, S.W. Kim and K. Koyama, et al, “Ultraprecision CMP for Sapphire, GaN, And SiC for Advanced Optoelectronics Materials,” Curr. Appl. Phys. vol. 12, pp. S41-S46, September 2012. [29] H. Aida, H. Takeda, K. Koyama, H. Katakura, K. Sunakawa, T. Doi, “Chemical Mechanical Polishing of Gallium Nitride with Colloidal Silica,” J. Electrochem. Soc. vol. 158, pp. H1206-H1212. November 2011. [30] P.R. Taverier, T. Margalith, L.A. Coldren, S.P. Denbaars, D.R. Clarke, “Chemical Mechanical Polishing of Gallium Nitride,” Electrochem. Solid-State Lett, vol. 5, pp. G61-G64. May 2012.


[31] H. Aida, S.W. Kim, T. Suzuki, K. Koyama, N. Aota and T. Doi, et al, “Surface Planarization of GaN-on-Sapphire Template by Chemical Mechanical Polishing for Subsequent GaN Homoepitaxy”. ECS J. Solid State Sci. Technol. vol. 158, pp. H163-H168. April 2014. [32] L. Zhou, V. Audurier, P. Pirouz and J.A. Powell, “Chemomechanical Polishing of Silicon Carbide,” J. Electron. Soc. vol. 144, pp. L161-L163, February1997. [33] C.L. Nelson, W.C. Mitchel and R.L. Hengehold, “Effects of Process Parameter Variations on The Removal Rate in Chemical Mechanical Polishing of 4H-SiC,” J. Electron. Mater. vol. 30, pp. 1271-1275. October 2001. [34] L. Xu, G. Pan and X. Liang. “Nitrogen/sulfur Co-doped Non-noble Metal Material As An Efficient Electrocatalyst for The Oxygen Reduction Reaction in Alkaline Media,” RSC Adv. vol. 4, pp. 19756-19765. April 2014. [35] U. Phatak, S. Bukkapatnam, Z. Kong and R. Komanduri, “Sensor-based Modeling of Slurry Chemistry Effects on The

241

Material Removal Rate (MRR) in Copper-CMP Process,” Int. J. Mach. Tool. Manu. vol. 49, pp. 171-181, February 2009. [36] K.H. Park, H.J. Kim, O.M. Chang and H.D. Jeong, “Effects of Pad Properties on Material Removal in Chemical Mechanical Polishing,” J. Mater Process Tech. vol. 187-188, pp. 73-76, June 2007. [37] T.K. Yu, C.C. Yu and M. Orlowski, “A Statistical Polishing Pad Model For Chemical-Mechanical Polishing,” IEEE In Electron Device Meeting Technical Digest, USA, pp. 865-868, December 1993. DOI: 10.1109/IEDM.1993.347263 [38] B. Zhao, “A New Fundamental Insight on Chemical Mechanical Polishing in IC Processes,” In Proceeding of The Fourth International Chemical-Mechanical (CMP) Planarization for VLSI Multilevel Inter-connection Conference, USA, pp. 12-22, February 1999.