High-Temperature-Resistant Interconnections Formed by Using
Transcription
High-Temperature-Resistant Interconnections Formed by Using
Kato et al.: High-Temperature-Resistant Interconnections (1/6) [Technical Paper] High-Temperature-Resistant Interconnections Formed by Using Nickel Micro-plating and Ni Nano-particles for Power Devices Noriyuki Kato, Suguru Hashimoto, Tomonori Iizuka, and Kohei Tatsumi Graduate School of Information, Production and Systems, Waseda University, 2-7 Hibikino, Wakamatsu-ku, Kitakyushu, Fukuoka 808-0135, Japan (Received August 6, 2013; accepted November 14, 2013) Abstract The improvement of interconnection technology is becoming a top priority for the operation of SiC devices at high temperatures. We proposed a new interconnection method using nickel electroplating to form bonds between chip electrodes and substrate leads. We also newly proposed low-temperature nickel nanoparticle sintering to form die bonding connections. SiC devices assembled with these new connection methods operated successfully in a high-temperature environment of about 300°C. We confirmed that these methods had adequate potential as an advanced heat resistant package in comparison with conventional interconnections. Keywords:High Temperature Resistant Packaging, Power Device, Micro Electro-plating Bonding, SiC Device, Nano-Nickel 1. Introduction 2. Experimental Procedure To reduce the size and increase the efficiency of power 2.1 Bonding by nickel micro-plating converters mounted in hybrid vehicles or electric vehicles, We used nickel electroplating to form nickel micro-plat- research on the development of methods to increase the ing bonds. The plating conditions were as follows: Ni plat- output power density using technology such as silicon car- ing bath of Watts solutions, bath temperature of 50°C, bide (SiC) devices is in progress. High temperature SiC electrical current density of 5 A/dm2, and substrate and devices require the materials for packaging capable of lead material of mainly copper. The growth rate of copper working at higher temperature than those for Si devices. plating on a flat plate was ranging from 0.75 μm/min to 0.9 While studies have been conducted on heat-resistant pack- μm/min. aging materials that can replace Al bonding wire or solder To simulate bonding between the chip electrodes and materials having low melting points, various problems in the substrate leads of a power device, we brought a copper alternative methods have been pointed out.[1, 2] wire (diameter: 172 μm) into contact with a copper plate Our approach described in this paper is to form the con- and plated the wire and the plate as shown in Fig. 2 (a). To ductive connection with Ni at relatively lower temperature. simulate die bonding, we brought the gold surface of a sili- [3] For the connection of chip electrodes Ni micro-plating con chip with vapor-deposited gold into contact with a lead bonding (NMPB)[4] was used and die attach connection frame cut into strips, and plated the chip and the lead was carried out by sintering nickel nanoparticles at a low frame as shown in Fig. 2 (b) temperature.[5] An example of the concept for these interconnections is shown in Fig. 1. A䡃-electrode A䡈-electrode A䡃-plang Ni-plang Cu-wire Cu plate Cu-wire 5.0䟚 Cu plate Ni-plang Cu plate Fig. 1 Model of bonded SiC diode specimen. Copyright © The Japan Institute of Electronics Packaging 2.7䟚 Si-chip Au-electrode 5.0䟚 㻿㼕㻯㻌㼐㼑㼢㼕㼏㼑㻌 Nano-Ni lead frame (a) (b) Fig. 2 (a) Copper wire bonded to plate (b) Silicon chip bonded to lead frame. 87 Transactions of The Japan Institute of Electronics Packaging Vol. 6, No. 1, 2013 To evaluate the high-heat-resistance reliability of the 2.7 mm × 2.7 mm and (B) 8.0 mm × 8.0 mm. After applying specimens after micro-plating bonding, we heated them at the solution to chip (A), electrodes were placed on chip various temperatures between 100 and 500°C in an argon (B) with their surfaces facing each other and were heated atmosphere for 60 min and then measured the shear and pressurized in atmosphere to form a bond as shown in strength of the bonds with a shear tester (Rhesca model Fig. 3(b). The bonding conditions were as follows: the PTR-01). The specimens were heated to 500°C for acceler- device stage and the pressurizing head were heated to ated aging testing to evaluate the interface reliability and 300°C, and a constant load of 147 N was applied by the the change in resistivity. pressurizing head. After heating the bonded specimens in To observe the cross section of the bonded portions we an argon atmosphere for 60 min (between room tempera- used a scanning electron microscope (SEM Hitachi model ture and 500°C), we measured their shear strength. S-3400N.). 2.3 Chips used for bonding strength evaluation 2.2 Low-temperature nickel nanoparticle sintering The chips used for bonding were silicon dummy chips We used a 5 wt.% nickel nanoparticle solution as the with aluminum film electrodes, and SiC-SBD(Shottky bar- bonding material. The nickel nanoparticles had a diameter rier diode) chips (1,200 V, 15 A, 2.7 × 2.7 mm, made by of 20 nm and were dispersed in a solution of isopropyl SiCED Electronics Development GmbH) having an alumi- alcohol (IPA). Since IPA is highly volatile, it evaporated num electrode film on the top surface and a silver film on after application, leaving only the nickel nanoparticles. the back surface. They had a thickness of 365 μm. To We used an electrostatic atomizer (made by Apic study bonding on these aluminum electrode surfaces, we Yamada)[6] to apply the solution. As shown in Fig. 3(a), used chips having an electroless nickel plating film formed this device enabled a selective and uniform application of on an aluminum electrode surface, and chips having vapor- the solution on the specimen surface by means of electro- deposited nickel and gold on aluminum electrode films. static force. We evaluated nickel nanoparticle bonding strength using For the evaluation of bonding characteristics with Ni bonding between pairs of silicon dummy chips (A) and nanoparticles we prepared Si dummy chips with size (A) (B). We applied the nickel nanoparticle solution to the electrode surface and studied on the bonding characteristics between both electrodes. 3. Results and Discussion 3.1 Nickel micro-plating bonding (NMPB) and evaluation 3.1.1 Plating bonding study To simulate bonding between the chip electrodes and the substrate leads of a power device, the copper wire was into contact with a copper plate and plated the wire and the plate, as shown in Fig. 2(a). Figure 4 shows the cross sec- Fig. 3a Schematic diagram for electrostatic atomization. tion of a copper wire (diameter: 172 μm) bonded to the 147N Chip(A) Nano Ni Chip(A) Chip (B) Heating Stage Fig. 3b Bonding steps of dummy chips with using Ni nano-particle atomization. 88 Kato et al.: High-Temperature-Resistant Interconnections (3/6) Cu-wire Ni-plang 5μm Cu plate Fig. 4 Plating cross section of copper wire bonded to copper plate. Fig. 6 Heating temperature-shear strength characteristic for plated bonding specimens of copper wire and copper plate made using different plating times. Si-chip 50μm Ni-plang lead frame Fig. 5 Plating cross section of silicon chip with vapor-deposited gold bonded to lead frame. Fig. 7 Heating temperature-resistance and heating-temperature-resistivity characteristics for specimens of nickel-plated wire on copper. interface between copper and nickel layers, indicating that copper plate by Ni micro-plating. To simulate die bonding, we plated the chip and the lead frame as shown in Fig. 2 (b). Figure 5 shows the observed cross section. Both cross sections indicate no problematic voids or other defects generated by the nickel plating. no deterioration caused by the formation of brittle phases or intermetallics with voiding occurred in the high-temperature environment. 3.1.3 Copper-nickel alloy resistance measurement Since the electrical resistance of a copper-nickel alloy 3.1.2 Shear testing layer formed in a high-temperature environment may To evaluate whether specimens with plated bonds had increase, we measured the change in resistance caused by high heat resistance, we measured their shear strength alloying. We used the four-terminal method to measure the after heating. We measured specimens in which a copper resistance of copper wire specimens (having a length of 30 wire (with a length of 1 mm and diameter of 172 μm) was mm and diameter of 172 μm) plated for 30 min and then brought into contact with a copper plate and plated for 15 heated in an argon atmosphere. After plating, the wire min or 30 min. The specimens were heated in an argon diameter increased to 200 μm, and the distance between atmosphere and then shear-tested. We carried out shear the terminals used for the wire resistance measurement testing nine times at each heating temperature and found was 12.5 mm. Figure 7 shows the results. Diffusion and the average shear strength of the specimens. As shown in alloying proceeded with an increase in the temperature, Fig. 6, specimens with plating times of both 15 and 30 min causing an increase in resistance.[7] The increase in resis- exhibited high shear strength after being heated to 500°C. tance even for a heating temperature of 500°C (for 1 h) The shear strength of both specimen types increased with was a value of around 10%. The diffusion condition of an increase in the heating temperature. This finding may 500°C for 1 h is equivalent to that of 350°C for 4,580 h, have been the result of increasing copper-nickel alloying assuming that the activation energy for Ni diffusion in Cu caused by diffusion, resulting in greater adhesion and is 225 kJ/mol.[8] This indicates that problematic defects strength at the interface. From the results described in are not considered to be generated in the interconnection 3.1.4, it appeared that no second phase was formed at the during practical use up to 350°C. 89 Transactions of The Japan Institute of Electronics Packaging Vol. 6, No. 1, 2013 (a) Heating condition: no heating (as bonded). (b) Heating condition: heated for 60 min at 300°C. (c) Heating condition: heated for 60 min at 500°C. Fig. 8 SEM images and concentration distribution graphs of cross section of nickel-plated copper plate specimens. 3.1.4 Analysis of copper-nickel diffusion state Cu-Ni system. Further, although Kirkendall voids have To verify the state of copper-nickel diffusion in speci- formed in the copper plate, the results described in Sec- mens, after heating, we used a SEM to observe the cross tions 3.1.2 and 3.1.3 indicate that their effect is not serious. sections of a copper plate covered by Ni plating. We mea- 3.2 Nickel nanoparticle bonding and evaluation sured concentration profiles of each element in the vicinity As described in Section 3.1.1, plating-based die bonding of interfaces using an energy-dispersive X-ray spectros- can be carried out by using a strip or lattice substrate copy (EDX) line analysis. Figure 8 shows the results. The structure, but it is difficult to use for bonding flat plate positions of EDX line analysis are indicated within each surfaces together. It has been shown that flat plates can be SEM image and the concentration profiles are displayed in bonded using low-temperature nickel nanoparticle bond- the graphs. ing[5, 9]; hence, we studied the application of this method A comparison of the graphs of unheated specimens and to die bonding. We brought the electrode surfaces of sili- specimens heated at 300°C in Fig. 8 (a) and (b) reveals con dummy chips that are (A) 2.7 × 2.7 mm and (B) 8.0 × almost no change in the concentration distribution. On the 8.0 mm in size into facing position, bonded them using other hand the concentration distribution of the specimen nickel nanoparticles, and then shear-tested the specimens heated at 500°C in Fig. 8 (c) shows the formation of an after heating. We performed shear testing five times at alloy layer of approximately 5 μm. There were no second each temperature and plotted the average shear strengths. phase observed, as predicted from a phase diagram of Figure 9 shows that a slight drop in the shear strength was 90 Kato et al.: High-Temperature-Resistant Interconnections (5/6) plate at a different constant temperature (ranging between 20 and 325°C). We then measured the diode voltage (VD) and current (ID), var ying the power supply voltage (between 0 and 15 V) using the circuit illustrated in Fig. 10. Figure 11 shows the results for different temperatures. In a high-temperature environment of 325°C, we obtained the diode V-I characteristics and did not obser ve the change in the resistivity caused by the bond deterioration Fig. 9 Heating temperature-shear strength characteristic of during measurements. With using Ni interconnections specimens bonded using nickel nanoparticles. newly introduced in this paper the high temperature operation of the SiC diode was first confirmed. ID 䡎䠍䠙2䡇Ω Variable DC power source SiC diode 0䡚15V VD 4. Conclusions 1. Our findings demonstrated that nickel micro-plating bonding(NMPB) could be used for bonding chips to substrate electrodes and for die bonding to chip Fig. 10 Circuit used for measuring diode V-I characteristics. substrates in power devices. We demonstrated that low-temperature bonding ensured bond reliability 8 in high-temperature environments of over 300°C. 7 We observed no bond deterioration during acceler- ID[mA] 6 ated diffusion tests of bonds formed by copper- 5 nickel plating. 4 3 2. We used nickel micro-plating bonding and low-tem- 2 perature nickel nanoparticle sintering to form 1 bonds between SiC power device chips and sub- 0 0 200 400 600 800 1000 VD[mV] Fig. 11 V-I characteristics of bonded SiC diodes at different heating temperatures. strate electrodes and to form die bonding connections, respectively. We tested device operation in a high-temperature environment of about 300°C. 3. Our findings demonstrated that micro-plating-based chip bonding technology could be considered to found when the specimens were heated at 500°C, but have adequate potential as a practical, simple since the value still remained high (over 10 N). mounting technology that is highly heat-resistant 3.3 High-temperature circuit operation testing and and ensures high reliability and low cost. evaluation using SiC diode chips To evaluate whether the bonding technologies that we References have discussed are practical for use in power device con- [1] H. A. Mantooth, M. M. Mojarradi, and R. W. Johnson, nections, we carried out high-temperature circuit opera- “Emerging Capabilities in Electronics Technologies tion testing on SiC-SBD chips (1,200 V, 15 A, made by for ExtremeEnvironments. Part I - High Temperature SiCED). As shown in Fig. 10, the interconnections were Electronics,” IEEE Power Electronics Society News- created for the specimen by using nickel micro-plating to letter, issue 1, 2006. bond connections between chip electrodes and leads of a [2] L. Coppola, D. Huff, F. Wang, R. Burgos, and D. silver-plated copper substrate joined and bonded by a cop- Boroyevich, “Survey on High Temperature Packaging per wire (using a plating time of 30 min). Nickel nano-par- Materials for SiC-Based Power Electronics,” Proc. ticles were used for die bonding, with a surface-processed aluminum electrode bonded to a silver-plated copper substrate. PESC, Orlando, FL, pp. 2234–2240, 2007. [3] S. Terashima, Y. Yamamoto, T. Uno, and K. Tatsumi, “Significant reduction of wire sweep using Ni plating After bonding the specimens, we performed the electri- to realize ultra fine pitch wire bonding,” Proceedings cal operation tests while heating each specimen on a hot of the 52nd Electronic Components and Technology 91 Transactions of The Japan Institute of Electronics Packaging Vol. 6, No. 1, 2013 Conference, 52, pp. 891–896, 2002. [4] K. Tatsumi and T. Ando, “Plating micro bonding used for Tape Carrier Package,” Proc. NIST/IEEE VLSI PACKAGING WORKSHOP, YORKTOWN HEIGHTS, N.Y. 1993, 10.12. 183–322, 1983. [8] Metal data book, edited by Japan Institute of Metals, Maruzen, 2010. [9] S. Hashimoto, T. C. Lun, K. Tatsumi, A. Nogami, and Y. Sawa, “Study on bonding by using Ni nano particles [5] K. Tatsumi et al., “Electronic component bonding for high temperature packaging,” Proceedings of material, composition for bonding, bonding method, autumn meeting of the Japan Institute of Metals and and electronic component,” PCT/JP2012/065242. Materials 2013, p. 192. [6]http://www.apicyamada.co.jp/pdf/seiden_hunmu.pdf (accessed Aug 2013). [7] C. Y. Ho, M. W. Ackerman, K. Y. Wu, T. N. Havill, R. Suguru Hashimoto H. Bogaard, R. A. Matula, S. G. Oh, and H. M. James, Tomonori Iizuka “Electrical resistivity of ten selected binary alloy sys- Kohei Tatsumi tems,” J. Phys. Chem. Ref. Data, Vol. 12, No. 2, pp. 92 Noriyuki Kato