Solid State Bonding of Tin and Copper by Metal Salt Generation Bonding Technique Using Citric Acid

Abstract

In the previous studies, the surface modification effect of formic acid on the bonding strength of tin and copper has been investigated. As a result of previous investigations, it was found that the lowering of the bonding temperature is achieved by removing the oxide film on the bonding surfaces with formic acid, forming a metal salt film, and thermally decomposing the film. However, it has been pointed out that formic acid is toxic and irritating and difficult to handle. Therefore, in this study, we decided to investigate the effects on the removal of the oxide film on the bonding surface and the bonding strength using citric acid, which is relatively harmless to the human body. In addition to observing the fractured surfaces and the bonded interfaces of the solid-state bonded tin and copper by SEM, thermal analysis of the compound produced by surface treatment with citric acid was carried out. The metal salt coating treatment using citric acid was carried out by boiling the bonding surfaces of tin and copper in citric acid for 300 s. Solid-state bonding was performed in a vacuum chamber under the following conditions; 383–473 K for bonding temperature, 7 MPa for bonding pressure and 1800 s for bonding time. Regardless of the metal salt coating treatment, the bond strength of the joint increased with the increase of the bonding temperature. When the metal salt coating treatment using citric acid was applied, the bonding temperature was reduced by 70 K to fabricate a joint with the base metal strength of tin. However, when the metal salt-coated surface was exposed to the atmosphere, the bonded surface was again covered with an oxide film and the bond strength decreased.

Effect of surface modification with citric acid on the joint efficiency between tin and copper.

1. Introduction

Currently, there is a demand for miniaturization of electronic devices, and as a result, the connection method for electronic components is shifting from conventional fusion bonding to solid state bonding. Since some electronic components are made of resins with low heat resistance (e.g. Organic Light Emitting Diode) and low mechanical strength (e.g. electromechanical components), the practical application of solid-state bonding methods is required. It is important to perform solid state bonding at low temperature and low pressure without damaging these electronic devices. Furthermore, solid state bonding at low temperature and low pressure can solve the problem of residual stress due to mismatch of thermal expansion in dissimilar material joints. However, the bonded surface is composed of fine irregularities, processed layers and oxide films that inhibit bonding.^1,2) Therefore, in order to form a strong connection, at least the oxide film on the bonding surface has to be removed. Ultrasonic vibration^3–5) and plasma processing^6–8) have been studied as joining methods for solid phase bonding while removing the oxide film on the bonding surface, and some of them have been put to practical use. Previous studies have shown that when the metal salt generation bonding technique using formic acid is applied to the solid phase bonding in various material combinations of joints, the bonding strength is greatly improved.^9–13) On the other hand, while formic acid is difficult to handle, citric acid is not harmful to the human body and is environmentally friendly. Therefore, the purpose of this study is to investigate the effect of surface modification using citric acid on the solid state bonding strength of tin and copper. In addition, by exposing the metal salt-coated surface to the atmosphere for a predetermined period of time, the durability of the effect of the metal salt coating treatment on the bond strength was evaluated.

2. Experimental Details

As shown in Fig. 1, the specimen shape is a block of 15 mm × 15 mm × 5 mm, cut from an ingot of 99.9% tin. Since tin is easily deformed plastically, Cu is bonded as the gripping part during tensile testing. For the purpose of removing the machined layer from the bonded surface of tin (15 mm × 15 mm), electrolytic polishing was applied in a solution containing 5% perchloric acid in 10% ethyl glycol monobutyl ether and 85% ethyl alcohol after mechanical polishing. The surface roughness Ra of the tin after electrolytic polishing was 0.33 µm. On the other hand, the bonded surface of copper block (15 mm × 15 mm × 15 mm, 99.9%) was mechanically polished using SiC polishing paper (#4000 grade). The surface roughness Ra of the copper after mechanically polishing was 0.07 µm. The metal salt coating process was carried out by immersing the tin and copper surfaces in citric acid heated to about 373 K for 30–1500 s. The products on the surface of tin and copper after metal salt coating treatment with citric acid were identified using an X-ray photoelectron spectrometer (XPS: Shimadzu/Kratos AXIS-HS) with monochromatized X-ray excitation of the AlKα line operating at 140 W. To evaluate the persistence of the metal salt coating treatment effect, the treated surfaces were heated to 323 K in air and held at 3.6 ks or 10.8 ks, and then the specimens were bonded to examine the bond strength. After the metal salt coating treatment, the specimens were placed in a vacuum chamber within 180 s to avoid re-oxidation of the coated surface and change in properties due to moisture absorption. Solid state bonding was carried out in a vacuum chamber by varying the bonding temperature from 383 to 473 K. The bonding load and time were set to 7 MPa and 1.8 ks, respectively. The bonding load was applied from before the start of heating until the end, and the temperature rise rate was kept constant at 0.35 K/s. After solid state bonding, the bonded specimen was cut into three pieces for observation of the bonding interface and strength test. Observation and elemental analysis of the bonded interface and fracture surface after tensile testing were performed by scanning electron microscope (SEM: Shimadzu SSX-550) and energy dispersive X-ray analysis (EDX: Shimadzu SEDX-500). The joint efficiency was investigated by the tensile strength of the joint. The reason for expressing joint strength in terms of joint efficiency is that the higher the joint temperature, the lower the tin height, the stronger the effect of plastic restraint of the copper, and the greater the apparent joint strength. For the tensile test, the specimens were cut into three pieces and then further cut into pieces with a cross-sectional area of 3 mm × 3 mm. The tensile test was performed at room temperature with a crosshead speed of 0.017 mm/s.

Fig. 1

Schematic illustration of to-be-bonded specimen (in mm).

3. Experimental Results

3.1 Effect of surface modification on joint efficiency

In order to optimize the treatment time for the metal salt coating, the treatment time was varied and the tensile strength was compared. To optimize the processing time, the processing time was varied from 30–1500 s and the bonding temperature was set to 403 K. Figure 2 shows the relationship between the metal salt coating time and the joint efficiency, and Fig. 3 shows the external views of the fractured surface after the tensile test. The joint efficiency is a value derived from the ratio of the tensile strength to the base metal strength of tin (12 MPa). Therefore, a joint efficiency of 100% means that there is no interfacial fracture and the fracture occurs in the tin matrix. The highest joint efficiency was obtained when the metal salt coating was applied for 300 s. When the metal salt coating time was 30 or 600 s, tin adhesion was observed on the copper surface. When the metal salt coating time was 300 s, the failure occurred in the tin matrix without fracture at the bonded interface. On the other hand, when the metal salt coating time was set to 1500 s, the joint fractured at the bonded interface and little tin adhesion was observed on the fracture surface. These observations indicate that if the metal salt coating time is too short, the oxide film is not completely reduced and removed, and if the coating time is too long, excessive metal salt coating is applied. Based on the results of the above study, it was determined that the optimum metal salt coating time was 300 s. Figure 4 shows the effect of the bonding temperature on the joint efficiency. In the figure, the results without the metal salt coating treatment are also shown for comparison. Regardless of the metal salt coating treatment, the joint efficiency increased with increasing the bonding temperature, and the coating treatment decreased the bonding temperature by 70 K to obtain a joint with 100% joint efficiency. The bonding temperature at which fracture occurred in the tin matrix was T = 473 K without the metal salt coating treatment, but decreased to T = 403 K with the coating treatment. Figure 5 and Fig. 6 show the results of fracture surface observation by SEM and mapping analysis by EDX after tensile test. As shown in Fig. 5, when the metal salt coating was not applied and the bonding temperature was set to 433 K, clear polishing marks were observed on the fracture surface of the copper side. Although Cu adhesion was not observed on the fracture surface of the tin side, the polishing marks on the Cu side were transferred. Since the Cu polishing marks were transferred to the Sn surface, it is considered that adhesion of the bonding surface was sufficiently achieved. As the joining temperature increased, the fracture surface became covered with adhesion. In order to investigate the composition of the adhesions observed on the fracture surface and how they affected the joint efficiency, they were analyzed by EDX. The analysis results are shown in Fig. 5 and Fig. 6. As shown in Fig. 5, the results of EDX analysis when no metal salt coating was applied and the bonding temperature was set to 433 K showed that almost no Cu was detected on the fracture surface of the tin side. However, by increasing the bonding temperature, the amount of Cu detected from the tin side fracture surface increased and was detected from the entire fracture surface. The reason for the detection of Cu on the tin side of the fracture surface is thought to be that the base metal softened as the joining temperature increased, which facilitated the destruction of the oxide film, exposing the new surface and promoting interdiffusion. In other words, if the bonding temperature is increased and the bonding pressure is increased, the oxide film is destroyed and interdiffusion occurs, resulting in high connection strength. As shown in Fig. 6, when the metal salt coating treatment was applied and the bonding temperature was set to 383 K, polishing marks were clearly observed on the copper side fracture surface, although they were not observed on the tin side fracture surface. In addition, traces of tin were not found on the copper fractured surface. Therefore, the reason for the low joint efficiency in spite of the metal salt film treatment is considered to be that the base material was not softened due to the low joining temperature, and adhesion of the bonding surfaces was not achieved. When the bonding temperature was increased to 393 K, the surface roughness of the fracture surface increased and pits and fine particles with a diameter of about 1 µm were observed on the fracture surface of the tin side. The pits and fine particles observed on these fracture surfaces are thought to be generated by the thermal decomposition of tin citrate due to heating during the bonding process. As shown in Fig. 6, when the bonding temperature was 383 K, the EDX surface analysis of the fracture surface showed that Cu was detected from the fine particles observed on the fracture surface of the tin side. As the bonding temperature was further increased, the area of Cu detected on the tin side fractured surface increased, and Cu was detected from the entire fractured surface. From the results of these observations, the Cu distribution detected on the fracture surface of the tin side tended to increase with the increase of the bonding temperature, regardless of whether the metal salt coating was applied or not. However, compared to the case without the metal salt coating treatment, the treatment lowered the bonding temperature by about 60 K, at which the bonding strength started to increase. Therefore, the reason why the joint efficiency is improved from lower temperatures by the metal salt coating treatment is thought to be due to the effect of the adhesion between the atomic surfaces of tin and copper without the oxide film.

Fig. 2

Relation between joint efficiency and modification time.

Fig. 3

Optical macroscopic images of fractured surfaces of joints after tensile test: (a) t = 300 s and (b) t = 1500 s.

Fig. 4

Effect of surface modification on the relation between joint efficiency and bonding temperature. The bonding pressure and time for all joints were 7 MPa and 1.8 ks respectively.

Fig. 5

SEM micrographs and EDX analysis of the fractured surfaces of joints after tensile test (without surface modification).

Fig. 6

SEM micrographs and EDX analysis of the fractured surfaces of joints after tensile test (with surface modification).

3.2 Observation of bonded interface

In order to investigate the reason why the metal salt coating treatment improved the joint efficiency from low joint temperatures, the bonded interface was observed using SEM. Figure 7 shows the observation results of the junction interface. The line analysis results of the bonded interface are also shown in the figure. Multiple voids of less than 1 µm in diameter were observed mainly on the tin side of the bonded interface with and without the metal salt coating treatment. Furthermore, the distribution density of voids was higher in the case of the metal salt coating treatment than in the case without treatment. Thus, it is suggested that the close contact between tin and copper is mainly determined by the amount of plastic deformation of the tin as the bonding temperature increases. This consideration is supported by the observation in Fig. 7 that the voids decrease as the bonding temperature increases. From the results of this experiment, it was found that the deformation of tin to obtain a joint with 100% joint efficiency was 49.7% without the metal salt coating treatment, but could be reduced to 3.9% by the treatment. In the case of no metal salt coating treatment, as shown in Fig. 7(a), almost no Sn was detected within the Cu side, while a layered structure with a width of about 2 µm was observed with a certain percentage of Cu within the Sn side. On the other hand, as shown in Fig. 7(b), when the metal salt coating treatment was applied, the diffusion layer was hardly observed at the bonded interface. The reason why a diffusion layer was observed at the bonded interface when no metal salt coating treatment was applied was that the bonding temperature had to be increased to soften the base metal and destroy the oxide film in order to increase the joint efficiency. Therefore, the metal salt coating treatment using citric acid was found to be effective in suppressing the reactive layer formation at the bonding interface. As a result of elemental analysis, the reaction layer in Fig. 7(a) was found to be an intermetallic compound consisting of Cu₆Sn₅.

Fig. 7

SEM micrographs and EDX analysis of bonded interfaces: (a) without surface modification (T = 443 K) and (b) with surface modification (T = 403 K).

3.3 Persistence of the effect of the surface modification

In order to evaluate the persistence of the metal salt coating treatment effect, the coated surfaces were heated in an atmospheric furnace and then bonded to examine the degradation behavior of the joint efficiency. The persistence test of the metal salt-coated surface was performed by warming it in an atmospheric furnace kept at 323 K for a specified time. After the coated surface was warmed up for a predetermined time, it was bonded in a vacuum chamber at a bonding temperature of 403 K, a bonding pressure of 7 MPa, and a bonding time of 1.8 ks. Figure 8 shows the effect of the heating time of the coated surface on the joint efficiency. In order to compare with the case where the coating surface is not heated, the case of 0 s is also shown in Fig. 8. As shown in Fig. 8, the joint efficiency tends to decrease with the increase of heating time of the coated surface in air. When the heating time of the coated surface was 10.8 ks, the value of the joint efficiency approached 0%. In order to investigate the cause of the decrease in joint efficiency due to heating of the coated surface in air, the fractured surface after the tensile test was observed by SEM. The observation results are shown in Fig. 9. As the heating time of the coating surface increased, the polishing marks observed on the Cu and Sn fracture surfaces became more distinct. At the same time, the amount of Cu detected from the fracture surface on the Sn side decreased as the heating time of the coated surface increased. In addition, fine particles, which are considered to be thermal decomposition products of tin citrate, are less likely to be observed from the fracture surface on the Sn side. From the results of these experiments, there are two possible reasons for the decrease in the joint efficiency due to heating the coated surface in air. Firstly, the heating may have caused thermal decomposition of the metal salt coating and re-oxidation of the bonded surface, and secondly, it may have inhibited interdiffusion through the bonded interface during the bonding process.

Fig. 8

Relation between joint efficiency and shelf test time.

Fig. 9

SEM micrographs and EDX analysis of the fractured surfaces of joints after tensile test (with shelf test).

3.4 Analysis of modified surface

It is known that when copper or tin is exposed to the atmosphere, its surface is quickly covered with a natural oxide film. It is also known that the oxide film prevents the improvement of joint efficiency in solid phase bonding. In order to investigate the effect of citric acid on the modification of copper and tin surfaces, 10 µm thick copper foil (30 mg) and 7 µm thick tin foil (12.9 mg) were boiled in citric acid for 300 s and then immediately analyzed with a differential scanning calorimeter (DSC: SII Seiko instrument DSC6200). From the thermal analysis results shown in Fig. 10, endothermic reaction was detected only for tin in the bonding temperature range used in this study (313–403 K). Since tin citrate is known to dehydrate between room temperature and 393 K,^14,15) the endothermic reaction observed in the DSC measurement is presumed to be evidence of the formation of tin citrate. Furthermore, tin citrate is known to thermally decompose into tin oxide, carbon monoxide, and carbon dioxide when heated above 393 K,¹⁵⁾ and a gentle exothermic peak was also observed in Fig. 10. Therefore, it is believed that the tin citrate coating contributed to the improvement of joint efficiency due to the formation of tin in the metallic state, in addition to the suppression of re-oxidation during bonding. In addition, the fine particles observed on the fracture surface of the tin side are considered to be tin oxides formed by the thermal decomposition of tin citrate. The chemical changes of copper and tin surfaces after boiling in citric acid for 300 s were investigated by XPS. The results of the surface analysis using XPS are shown in Fig. 11 and Fig. 12. From the XPS measurement results shown in Fig. 11, the surface of electropolished tin showed a decrease in the peak of tin oxide and an increase in the peak of tin in the metallic state after the surface modification treatment. Similarly, for copper, the surface modification treatment tended to decrease copper oxides and increase copper in metallic state for the surface finished with SiC polishing paper (#4000 grade). Besides, it has been reported that copper oxides react with citric acid to produce copper in the metallic state in addition to copper citrate. The reaction equation is shown below:¹⁶⁾   

\begin{equation} \text{2Cu$_{2}$O} + \text{2[Cit]$^{3-}$} \to \text{[Cu$_{2}$H_$_{2}$Cit$_{2}$]$^{4-}$} + \text{2Cu} + \text{2OH$^{-}$} \end{equation} (1)

After the metal salt film treatment, the samples were washed with distilled water, which is thought to have eluted copper citrate and unreacted citric acid, thus supporting the results of the XPS measurements. Therefore, it is clear that the metal salt coating treatment using citric acid not only produces citrate, but also has a reducing effect on the bonding surface.

Fig. 10

DSC curves of samples modified by citric acid (modification time = 300 s): (a) tin foil and (b) copper foil.

Fig. 11

XPS spectra of the Sn surface: (a) non-modified sample and (b) modified sample.

Fig. 12

XPS spectra of the Cu surface: (a) non-modified sample and (b) modified sample.

4. Conclusions

The following conclusions can be drawn from the present study:

1. (1)    By replacing the oxide film on the bonding surface with a metal salt coating and pyrolyzing that coating during the bonding process, the atomic surface was exposed and the bonding strength was improved.
2. (2)    Due to the substitution effect of the metal salt coating, a base metal breaking joint was obtained at a joint temperature 70 K lower than that without the substitution.
3. (3)    It was found that the metal salt coating was re-oxidized by exposure to air, reducing the effect of the bond strength improvement.

Acknowledgement

This work was supported by JSPS KAKENHI Grant Number JP20K04592.

REFERENCES

• 1)   S.  Koyama,  M.  Takahashi and  K.  Ikeuchi: Ceram. Trans. 198 (2007) 461–466.
• 2)   S.  Koyama,  S.K.  Ting,  S.  Amari,  K.  Matsubara and  I.  Shohji: Mater. Trans. 54 (2013) 1975–1980. doi:10.2320/matertrans.M2013101
• 3)   N.  Watanabe and  T.  Asano: Jpn. J. Appl. Phys. 42 (2003) 2193–2197. doi:10.1143/JJAP.42.2193
• 4)   J.  Lee,  H.  Kim and  C.D.  Yoo: J. Electron. Mater. 34 (2005) 96–102. doi:10.1007/s11664-005-0185-3
• 5)   F.  Wang,  L.  Han and  J.  Zhong: Sens. Actuators A 149 (2009) 100–105. doi:10.1016/j.sna.2008.11.006
• 6)   A.  Shigetou,  T.  Itoh and  T.  Suga: J. Mater. Sci. 40 (2005) 3149–3154. doi:10.1007/s10853-005-2677-1
• 7)   E.  Higurashi,  T.  Imamura,  T.  Suga and  R.  Sawada: IEEE Photonics Technol. Lett. 19 (2007) 1994–1996. doi:10.1109/LPT.2007.908642
• 8)   T.  Imamura,  E.  Higurashi,  T.  Suga and  R.  Sawada: IEEJ Trans. Sens. Micromach. 128 (2008) 266–270. doi:10.1541/ieejsmas.128.266
• 9)   S.  Koyama,  Y.  Aoki and  I.  Shohji: Mater. Trans. 51 (2010) 1759–1763. doi:10.2320/matertrans.MJ201019
• 10)   S.  Koyama,  N.  Hagiwara and  I.  Shohji: Jpn. J. Appl. Phys. 54 (2015) 030216. doi:10.7567/JJAP.54.030216
• 11)   S.  Koyama and  T.  Tsuneto: Procedia Mater. Sci. 12 (2016) 95–99. doi:10.1016/j.mspro.2016.03.017
• 12)   S.  Koyama and  V.P.  Nguyen: Key Eng. Mater. 741 (2017) 31–35. doi:10.4028/www.scientific.net/KEM.741.31
• 13)   S.  Koyama,  K.  Ozawa and  T.  Muraoka: J. JILM 69 (2019) 548–552. doi:10.2464/jilm.69.548
• 14)   M.  Bhagwat,  P.  Shah and  V.  Ramaswamy: Mater. Lett. 57 (2003) 1604–1611. doi:10.1016/S0167-577X(02)01040-6
• 15)   R.S.  Ningthoujam and  S.K.  Kulshreshtha: Mater. Res. Bull. 44 (2009) 57–62. doi:10.1016/j.materresbull.2008.04.004
• 16)   J.  Eskhult,  C.  Ulrich,  F.  Bjorefors and  L.  Nyholm: Electrochim. Acta 53 (2008) 2188–2197. doi:10.1016/j.electacta.2007.09.032