Optimization of TLPS Bonding Process and Joint Property using Ni-Sn Paste for High Temperature Power Module Applications

Article information

J Weld Join. 2024;42(2):165-173

Publication date (electronic) : 2024 April 30

doi : https://doi.org/10.5781/JWJ.2024.42.2.3

Da-Gyeong Han ^*

, Dong-Hwan Lee ^**

, Jeong-Won Yoon ^*^,

* Department of Advanced Materials Engineering, Chungbuk National University, Cheongju, 28644, Korea

** Power Module Engineering Design Team, Hyundai Motor Company, Hwasung, 18280, Korea

†Corresponding author: jwyoon@chungbuk.ac.kr

Received 2024 March 4; Revised 2024 March 26; Accepted 2024 April 3.

Abstract

Recently, as interest in eco-friendly vehicles such as electric and hybrid vehicles increases, the demand for power semiconductors, a key component, is also increasing. Power semiconductors convert, distribute, and control power and operate in harsh environments such as high temperature and high pressure. In order to ensure stable reliability in such harsh environments, research on a technology that can stably form joints even at high temperatures is essential. Transitional liquid phase (TLP) bonding was proposed as a high-temperature power semiconductor chip bonding technology, which has the advantage of forming an intermetallic compound (IMC) phase with a high melting point at the joint. However, it takes a long time to convert the joint into full IMC phase. Therefore, in this study, in order to shorten the process time, a paste was manufactured by mixing high-melting point Ni metal powder and low-melting point Sn metal power, and a joint was formed through a TLPS (Transition liquid phase sintering) bonding using the paste. Pastes of different compositions were prepared by adjusting the ratio of Ni and Sn powders. The chip and substrate were bonded through a thermocompression (TC) bonding process, and the highest shear strength was obtained at a bonding temperature of 250 °C for 10 min. Heat treatment was performed at 200 °C for up to 500 h to evaluate the high temperature long-term reliability of the joints. The Ni-Sn TLPS bonded joints remained reliable joints after a long-term aging test at a high temperature of 200 °C.

Keywords: Power semiconductor; TLPS bonding; Ni-Sn paste; Intermetallic compound; Shear strength

1. Introduction

Recently, global awareness of environmental issues has led to higher interest in carbon emission reduction, eco-friendly vehicles, and renewable energy1). The increased interest in eco-friendly vehicles, such as electric and hybrid vehicles, has also increased the demand for power semiconductors that perform power conversion (DC↔AC), distribution, and control2,3). Power semiconductors have been widely used as key components in environments that use high-voltage power. Si semiconductors have been mainly used, but research has been actively conducted on silicon carbide (SiC) and gallium nitride (GaN) semiconductors, which are wide band gap (WBG) semiconductor devices, due to the increasing demand for high-level specifications4-6). Since power semiconductors operate in poor environments (e.g., high temperature, high pressure, and high current density), the development of bonding materials and process technologies for power semiconductors is very important in securing stable characteristics and reliability in such environments7). Thus far, soldering methods based on Pb-free bonding materials have been mainly used as a bonding process technology for power semiconductors, but most Sn-based Pb-free solders had limitations in being applied to high-temperature environments because their melting point is less than 250°C. The Ag sintering method was presented as technology to secure stable reliability at high temperature, but price competitiveness must be secured for its commercialization due to the high price of the Ag powder material and the high cost of nanopowder production8-13). Transient liquid phase (TLP) technology was presented as a new method to address problems with the use of Sn-based solders and metal paste. For TLP bonding technology, a material with a relatively low melting point is inserted between materials with high melting points, and only the material with a low melting point is changed into the liquid phase when heated14). Through the isothermal solidification process, liquid elements diffuse towards the solid base metal, thereby forming intermetallic compounds (IMCs) at the joint15). The joint with IMCs can secure stable reliability at high temperature due to the high melting point. For some compositions, IMCs are known to have higher electrical resistance than pure metals16). In particular, it was confirmed that the joint completely converted into IMC after heat treatment has high electromigration (EM) resistance and excellent lifespan characteristics17,18). Conversion of the entire joint into IMC, however, requires a long period of time. To tackle this problem, transient liquid phase sintering (TLPS) technology, which performs bonding using the paste produced by mixing high-melting point metal powder and low-melting point metal powder, has been presented of late. It can shorten the process time by facilitating atomic diffusion with an increased contact area between atoms compared to TLP bonding19). In this study, after preparing power semiconductor chip bonding paste for response to high temperature, the joint characteristics were analyzed. After preparing five types of paste by adjusting the contents of Ni powder, Sn powder, and flux, the optimization process was performed. The TLPS bonding process was performed using the selected optimal paste, and the joint microstructure, mechanical strength, and reliability at high temperature were evaluated over time.

2. Experimental Method

The Ni-Sn paste optimization experiment was carried out using spherical Ni and Sn metal powders in different sizes. The size of the Ni metal powder ranged from 1 to 3 ㎛ and that of the Sn metal powder ranged from 5 to 7 ㎛. Fig. 1 shows the schematic of the process of manufacturing five paste samples by adjusting the contents of Sn, Ni, and flux. Table 1 shows the mixing ratios used in the experiment. Based on the mixing ratios listed in Table 1, paste #1 to #5 were prepared. In the case of paste #1 and #3 with a flux content of 1 g, metal powders were not mixed well due to the lack of the flux. Fig. 2 shows the samples for bonding strength measurement prepared using paste #2, 4, and 5, which were mixed relatively well. The results of measuring the bonding strength of these samples are shown in Fig. 3. Paste #5 with the highest flux content showed the lowest bonding strength. Paste #4 with the Sn, Ni, and flux ratio of 7:3:2 (wt %) exhibited the highest strength, and it was set as the optimal condition. In the experiment, the direct bond copper (DBC) substrate and 3×3 mm² Cu dummy chip, which were subjected to electroless nickel immersion gold (ENIG) surface treatment, were used. In this study, ENIG surface treatment, which adds Ni(P) and Au layers to prevent the oxidation of the substrate and the excessive diffusion of Cu atoms at the interface, was applied to improve the reliability of the joint. The Ni-Sn paste was applied in a thickness of 80 ㎛ using a stencil mask with an open pad diameter of 4×4 mm². The chip and substrate were bonded through the thermo-compression (TC) process that applies heat and pressure. The schematic of the TC process is shown in Fig. 4. The sample with the Ni-Sn paste applied between the chip and substrate was placed between the top and bottom hot plates, and the 30-second pre-drying process was performed to prevent the paste from moving out. The bonding process was performed under a process temperature of 250°C and a pressure of 6 MPa by varying the bonding time from 30 seconds to 1, 5, and 10 minutes. For long-term reliability evaluation at high temperature, the prepared sample was subjected to heat treatment at 200°C for 250 and 500 hours.

Fig. 1

Schematic of manufacturing process of Ni-Sn paste

Table 1

Mixing ratio of Ni-Sn pastes

Fig. 2

Images of (a) fabricated Ni-Sn pastes and (b) TLPS bonded samples

Fig. 3

Shear strength with different pastes

Fig. 4

Schematic of TC bonding process

To analyze the joint of the sample, cold mounting was performed using epoxy resin. After the polishing process with abrasive paper from 100 # to 2000 #, polishing was performed using 1㎛ and 0.3㎛ alumina suspensions. Sn remained in the joint cross-section without reaction was removed through a Sn etching solution. The Sn etching solution used was prepared by mixing ethanol (C₂H₅OH), nitric acid (HNO₃), and hydrochloric acid (HCl) at a ratio of 95:4:1 (wt %). Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) were used to analyze and observe the cross-sectional images of the specimen joint. The IMC phase that constitutes the joint was analyzed through X-ray diffraction. The chip shear test was then conducted at a shear rate of 200 ㎛/s and a shear height of 100 ㎛ to evaluate the mechanical strength of the joint.

3. Experiment Results

The TLPS joint was formed through the TC process to analyze the bonding characteristics of the Ni-Sn paste. The joint was denser as the process time increased. Through a comparison of the joint images before and after etching in Fig. 5(a) and 5(b), the presence/absence of residual Sn was determined. After the short process time of less than one minute, Sn that did not react was etched and it was marked with circles. When bonding was performed for more than five minutes, it was found that most Sn was converted into IMCs.

Fig. 5

Cross-sectional SEM images of the TLPS bonded joints with bonding time, (a) Before etching and (b) After etching

Through EDX mapping analysis and point analysis, the phase of the joint was analyzed. Fig. 6 shows the joint cross-section mapping results after bonding at 250°C for 5 minutes. Ni layers were observed at both interfaces because the DBC substrate and Cu dummy chip subjected to ENIG surface treatment were used. In addition, P layers were also observed because the reducing agent P was used during the Ni surface treatment process. Fig. 7 provides the cross-section analysis results after bonding at 250°C for 10 minutes. Joint analysis results that were similar to the case of five minutes were observed. Most of the joint was composed of Ni that remained without reacting with the Ni-Sn IMC. In the case of 250°C/5 minutes, however, the Au layer was thicker compared to the case of 10 minutes. This appears to be because the 5-minute bonding time was not sufficient for Au atoms to diffuse and participate in the reaction. In addition, multiple voids were observed at the joint cross-section. This seems to be because Sn atoms moved into the space between Ni atoms, thereby causing densification and volume shrinkage during the sintering, bonding, and phase transformation processes.

Fig. 6

EDX mapping analysis results of the TLPS joint bonded at 250 °C for 5 min

Fig. 7

EDX mapping analysis results of the TLPS joint bonded at 250 °C for 10 min

EDX point analysis was conducted for the joint. Fig. 8 shows the analysis results for 250°C/5 minutes. The dark gray spheres in the joint were found to be Ni metal powder that remained without reaction. The relatively light gray layer on the chip side was found to be (Ni, Au)₃Sn₄ IMC that contained Au. The joint was composed of Ni powder that remained without reaction and Ni- Sn IMC. Although it was dominated by the Ni₃Sn₄ composition, Ni₃Sn₂ was observed around Ni powder.

Fig. 8

Cross-sectional image and point compositional analysis results of the TLPS joint bonded at 250 °C for 5 min

Fig. 9 shows the cross-sectional image and composition analysis results for 250°C/10 minutes. As with the case of 250°C/5 minutes, the joint was composed of Ni, Ni₃Sn_4, and Ni₃Sn₂. The IMC layer that contained Au near the chip, however, became thinner. The thickness of the (Ni, Au)-Sn layer decreased due to the diffusion of Au atoms into the joint as the process time increased.

Fig. 9

Cross-sectional image and point compositional analysis results of the TLPS joint bonded at 250 °C for 10 min

The shear test was conducted to evaluate mechanical characteristics according to the bonding time, and the results are shown in Fig. 10. When bonding was performed for 1 to 5 minutes, there was a slight reduction in strength. After 10 minutes, however, strength values higher than 30 MPa were measured. The sample bonded for 5 minutes showed the lowest strength value. To analyze the cause of strength degradation, fracture surface analysis was conducted. The fracture surface analysis results were divided into a part composed of Ni-Sn IMC and a part composed of Ni-Sn and Au-Sn IMCs under the 250°C/5 minute condition.

Fig. 10

Shear strength variation of the TLPS joints with bonding time

In the EDX point analysis results, a high content of Au was confirmed at point 1 in Fig. 11 compared to point 2. For a comparison with the 250°C/10 minute condition that exhibited higher strength, fracture surface mapping analysis was conducted under each condition and the results are shown in Fig. 12. The distribution of Au was found to be a factor that affected the strength difference. A high Au distribution was observed under the 5 minute condition, and fracture occurred in the IMC layer with a high Au content.

Fig. 11

Fracture image and point analysis results of the TLPS joint bonded at 250 °C for 5 min

Fig. 12

Fracture surface mapping images; (a) 250 °C/5 min and (b) 250 °C/10 min

It is judged that the joint with different phases has lower shear strength than the joint with a single phase. XRD analysis of the fracture surface was conducted to analyze the joint phase, and the results are shown in Fig. 13. The Ni₃Sn₄ and Ni₃Sn₂ phases were observed under both 5-minute and 10-minute conditions. Since Al₂O₃ was used as a ceramic insertion layer of the DBC substrate used in the experiment, its peaks were also observed. In this study, the difference between the 5-minute and 10-minute conditions was determined by the presence/absence of the AuNi₂Sn₄ phase. As the bonding time increased, the AuNi₂Sn₄ IMC phase observed under the 5-minute condition disappeared and only the peaks of Ni₃Sn₄ and Ni₃Sn₂ IMCs were observed in the fracture surface. This indicates that the ternary IMC phase observed under the 5-minute condition acts as a factor that decreases shear strength.

Fig. 13

XRD analysis results of the fractured TLPS joints bonded at 250 °C for 5 and 10 min

For long-term reliability evaluation at high temperature, the TC process sample was subjected to heat treatment at 200°C for 250 and 500 hours. Fig. 14 shows the cross-section mapping results after heat treatment for 250 hours. After 250-hour heat treatment, the joint was composed of residual Ni, (Ni, Au)₃Sn₄, and Ni₃Sn₄. Fig. 15 shows the cross-section mapping results after heat treatment for 500 hours. The mapping images after two different heat treatment periods were found to be similar. The EDX point analysis results in Fig. 16 and 17, however, showed that the joint subjected to 500- hour heat treatment was composed of residual Ni and Ni₃Sn₂. It was analyzed that Sn-rich Ni₃Sn₄ changed into Ni-rich Ni₃Sn₂ due to the additional reaction with the remaining Ni metal during the long-term heat treatment at high temperature. It was found that this phase change affects the shear strength change. Fig. 18 shows the shear strength of the sample subjected to heat treatment. As the heat treatment time increased, the shear strength of the joint increased. After 500-hour heat treatment, a high shear strength of approximately 55 MPa was measured. As heat treatment continued, phases that constituted the joint were converted into the single phase of Ni₃Sn₂ through the phase change, resulting in an increase in shear strength. Consequently, the joint composed of the Ni₃Sn₂ compound layers formed through additional reactions between Ni and Ni-Sn IMC exhibited very high bonding strength.

Fig. 14

EDX mapping analysis results of the TLPS joint aged at 200 °C for 250 h

Fig. 15

EDX mapping analysis results of the TLPS joint aged at 200 °C for 500 h

Fig. 16

Cross-sectional image and point compositional analysis results of the TLPS joint aged at 200 °C for 250 h

Fig. 17

Cross-sectional image and point compositional analysis results of the TLPS joint aged at 200 °C for 500 h

Fig. 18

Shear strength variation of the TLPS joints with aging time

4. Conclusion

In this study, Ni-Sn paste was manufactured to form power semiconductor chip joints that can operate in high-temperature environment in a stable manner. In addition, the microstructure and mechanical strength of the transient liquid phase sintering (TLPS) joint were measured according to the process time. Five different paste types were prepared by adjusting the contents of Ni and Sn metal powders. When the Sn, Ni, and flux ratio was 7:3:2 (wt %), the joint with the highest strength was formed. The TLPS process could form intermetallic compounds (IMCs) in the joint within a very short period of time compared to the TLP process due to the rapid atomic diffusion caused by the wide contact area between metal powders. The chip and substrate were bonded through the thermo-compression (TC) process, and the formed joints were compared and analyzed. Residual Sn was observed when bonding was performed for 30 seconds and 1 minute due to the short reaction time, and the highest shear strength was measured under the 250°C/10 minute condition. The lowest shear strength was measured under the 250°C/5 minute condition. The fracture surface analysis results confirmed that the ternary IMC formed in the joint is the cause of strength degradation. For long-term reliability evaluation at high temperature, heat treatment was performed at 200°C for 250 and 500 hours. As the heat treatment time increased, the shear strength of the joint tended to increase. Continued heat treatment caused the phase change of the joint into a single phase, thereby increasing shear strength. The TLPS joint manufactured using Ni-Sn paste in this study was metallurgically stable even after long-term heat treatment at high temperature, and it also exhibited high mechanical strength. Based on these results, it is determined that the TLPS bonding process that uses the Ni-Sn paste attempted in this study is sufficiently applicable to power semiconductor module bonding for electric vehicles, which requires excellent long-term reliability at high temperature.

Acknowledgment

This work was supported by the Technology Innovation Program (20018910) funded by the Ministry of Trade, Industry and Energy and the National Research Foundation of Korea (No. RS-2023-00247545) funded by the Ministry of Science and ICT in 2023.

References

1. Shahzad K, Cheema I. I. Low-carbon technologies in automotive industry and decarbonizing transport. J. Power Sources 5912024;:233888. https://doi.org/10.1016/j.jpowsour.2023.233888.

2. Ghosh A. Possibilities and Challenges for the Inclusion of the Electric Vehicle (EV) to Reduce the Carbon Footprint in the Transport Sector:A Review. Energies 13(10)2020;https://doi.org/10.3390/en13102602.

3. Blaabjerg F, Wang H, Vernica I, Liu B, Davari P. Reliability of Power Electronic Systems for EV/HEV Applications. Proc. IEEE 109(6)2021;:1060. https://doi.org/10.1109/JPROC.2020.3031041.

4. Lee D. H, Heo M. H, Yoon J W. Recent Studies of Transient Liquid Phase Bonding Technology for Electric Vehicles. J. Weld. Join 40(3)2022;:233–241. https://doi.org/10.5781/JWJ.2022.40.3.4.

5. Hu B, Gonzalez J. O, Ran L, Ren H, Zeng Z, Lai W, Gao B, Alatise O, Lu H, Bailey C, Mawby P. Failure and Reliability Analysis of a SiC Power Module Based on Stress Comparison to a Si Device. IEEE Trans. Device Mater. Rel 17(4)2017;:727. https://doi.org/10.1109/TDMR.2017.2766692.

6. Yoon J. W, Bang J. H, Ko Y H, Yoo S H, Kim J. K, Lee C. W. Power Module Packaging Technology with Extended Reliability for Electric Vehicle Applications. J. Microelectron. Packag. Soc 21(4)2014;:1–13. https://doi.org/10.6117/KMEPS.2014.21.4.001.

7. Roh M. H, Jung J. P, Kim W J. Trasient Liquid Phase bonding for Power Semiconductor. J. Microelectron. Packag. Soc 24(1)2017;:27. https://doi.org/10.6117/kmeps.2017.24.1.027.

8. O'Neal C. B, McGee B, McPherson B, Stabach J, Lollar R, Liederbach R, Passmore B. Advanced materials for high temperature, high performance, wide bandgap power modules. J. Electron. Mater 452016;:245–254. https://doi.org/10.1007/s11664-015-4187-5.

9. Yoon J. W, Jeong S. E. Transient Liquid Phase Sinter Bonding with Tin-Nickel Micro-sized Powders for EV Power Module Applications. J. Microelectron. Packag. Soc 28(2)2021;:71–79. https://doi.org/10.6117/KMEPS.2021.28.2.071.

10. Zhang B, Lu X, Ma H, Wang D, Mei Y. H. Development of Silver Paste With High Sintering Driving Force for Reliable Packaging of Power Electronics. IEEE Trans. Compon. Packag. Manuf. Technol 14(1)2024;:10. https://doi.org/10.1109/TCPMT.2023.3347250.

11. Chen C, Choe C, Kim D, Suganuma K. Lifetime Prediction of a SiC Power Module by Micron/ Submicron Ag Sinter Joining Based on Fatigue, Creep and Thermal Properties from Room Temperature to High Temperature. J. Electron. Mater 502021;:687–698. https://doi.org/10.1007/s11664-020-08410-5.

12. Yu F, Cui J, Zhou Z, Fang K, Johnson R. W, Hamilton M. C. Reliability of Ag Sintering for Power Semiconductor Die Attach in High-Temperature Applications. IEEE Trans. Power Electron 32(9)2017;:7083–7095. https://doi.org/10.1109/TPEL.2016.2631128.

13. Lee N. Y, Lee J. H, Hyun C. Y. Chip Sinter- Bonding Using Ag-Based Paste for Power Semicon- ductor Devices. J. Weld. Join 37(5)2019;:482–492. https://doi.org/10.5781/JWJ.2019.37.5.8.

14. Jung D. H, Sharma A, Mayer M, Jung J. P. A Review on Recent Advances in Transient Liquid Phase (TLP) Bonding for Thermoelectric Power Module. Rev. Adv. Mater. Sci 532018;:147–160. https://doi.org/10.1515/rams-2018-0011.

15. Jung D. H, Roh M. H, Lee J. H, Kim K. H, Jung J. P. Transient Liquid Phase (TLP) Bonding of Device for High Temperature Operation. J. Microelectron. Packag. Soc 24(1)2017;:17–25. https://doi.org/10.6117/KMEPS.2017.24.1.017.

16. Oberst M, Schlegel S, Großmann S, Willing H, Freudenberger R. Impact of the Formation of Inter- metallic Compounds in Current-Carrying Connections. IEEE Trans. Device Mater. Rel 20(1)2020;:157–166. https://doi.org/10.1109/TDMR.2020.2971055.

17. Lin Y. M, Zhan C. J, Jaung J. Y, Lau J. H, Chen T. H, Lo R, Kao M, Tian T, Tu K. N. Electromigration in Ni/Sn intermetallic micro bump joint for 3D IC chip stacking. 61st Electronic Components and Technology Conference (ECTC) ,Florida, USA 2011;:351–357. https://doi.org/10.1109/ECTC.2011.5898537.

18. Satoh T, Wakasugi M, Usui M. Effects of High- Density Current on the Reliability of Ni-Sn Solid- Liquid Interdiffusion Joints with Al Interlayer. J. Electron. Mater 522023;:1132–1144. https://doi.org/10.1007/s11664-022-10059-1.

19. Lee J. H, Jung D. H, Jung J. P. Transient Liquid Phase Diffusion Bonding Technology for Power Semiconductor Packaging. J. Microelectron. Packag. Soc 25(4)2018;:9–15. https://doi.org/10.6117/KMEPS.2018.25.4.009.

Article information Continued

This Journal follow the (http://creativecommons.org/licenses/by-nc/4.0/) Creative Commons Attribution Non-Commercial No Derivatives License

Sample	Sn power	Ni power	Flux
1	6 g	4 g	1 g
2	6 g	4 g	2 g
3	7 g	3 g	1 g
4	7 g	3 g	2 g
5	7 g	3 g	4 g