Joints Soldered on Electroless Ni Au Surfaces Using Cu-Containing Flux: Strength, Microstructure and Mechanism of Improvement

Transcription

1 Materials Transactions, Vol. 46, No. 11 (2005) pp to 2385 Special Issue on Lead-Free Soldering in Electronics III #2005 The Japan Institute of Metals Joints Soldered on Electroless Ni Au Surfaces Using Cu-Containing Flux: Strength, Microstructure and Mechanis of Iproveent Seishi Kuaoto 1; *, Hitoshi Sakurai 2, Kazuki Ikeda 2 and Katsuaki Suganua 1 1 Departent of Adaptive Machine Systes, Graduate School of Engineering, Osaka University, Ibaraki , Japan 2 Electronics Materials Division, Research and Developent Group, Haria Cheicals, Inc., Kakogawa , Japan The critical issue for soldering on electroless Ni Au surface finishes is fragile solder joints. Previous studies have indicated that these weak joints are a result of the foration of a at the joint interface. In this study, the new flux, which contains a Cu copound, deonstrates iproved solder joint strength for solder ball attachent on electroless Ni Au surface finishes. Cross-sectional analysis revealed that this new flux gives a thinner at the joint interface than that seen with a conventional flux. Moreover, the foration of a uniforly thin Cu Sn interetallic layer on the Ni P plated surface is observed after reflow soldering. We conclude that this Cu Sn layer foration during reflow ipedes extra diffusion of Ni into solder fro the plating surface so that the growth of a can be effectively inhibited and, thus, the joint strength is iproved. (Received May 20, 2005; Accepted July 20, 2005; Published Noveber 15, 2005) Keywords: joint strength, phosphorus-rich layer, flux, electroless nickel phosphorus, lead-free solder 1. Introduction Table 1 Coposition of solder balls (in ass%). High-density packaging has becoe an iportant driver behind the eergence of new sall portable electronic products that depend on the use of sall and light coponents. Chip Size Packages (CSPs) with higher I/O density have thus been widely adopted in cellular phones and portable PCs in preference to Quad Flat Packages (QFPs). One of the concerns when assebling such sall packages is oxidation of the pad surface, which often results in joint failure. Therefore, to ensure less oxidation, substrates for CSPs coonly use an electroless Ni P plating surface finish. However, soe previous studies report that soldering on electroless Ni P plating surface finish tends to give fragile solder joints. 1 6) Weak joints are a result of the foration of a due to excessive diffusion of Ni into the solder at the joint interface. This proble becoes ore serious in sall products due to their icroscopic soldered joints. Several studies on solder coposition and plating technology have been perfored, but have not yet provided a conclusive solution to this proble. 5 10) The present study addresses this issue fro the properties of the flux used for soldering. The newly developed flux in this study, Barrier Flux, includes a copper copound, copper(ii) stearate, to control the interfacial reaction during soldering. This paper reports on (1) solder joint strength, (2) observations of the solder joint interfacial icrostructure and (3) the echanis of iproveent in solder joint strength when using Barrier Flux. 2. Experiental Conditions Two solder ball copositions chosen in this study were Sn 4.0Ag 0.5Cu and Sn 37Pb (ass%), and had a diaeter of 600, as shown in Table 1. The fluxes studied include (1) baseline flux containing no copper copounds, and (2) *Graduate Student, Osaka University Ag Cu Pb Sn Sn 4.0Ag 0.5Cu Bal. Sn 37Pb <0: Bal Table 2 Descriptions of (1) baseline flux and (2) Barrier Flux (in ass%). Flux (1) (2) base resin/rosin activator/stearic acid solvent/hexyl carbitol thixotropic agent/castor wax copper copound/cu(ii) stearate Barrier Flux containing copper(ii) stearate. Copper(II) stearate tends to deposit copper at below 200 C. Table 2 gives a description of each flux. The substrates used had 250 area-arrayed pad openings 0.5 in diaeter. The pads had an approxiately 5 thick electroless Ni P layer containing 6 8 ass% P. The top of the Ni P layer was also covered with about thick Au plating. The procedure for aking the test vehicle was to screenprint flux at 100 thickness, followed by solder ball placeent and reflow in a N 2 atosphere. We used two different teperature profiles: the peak teperature for Sn 4.0Ag 0.5Cu was 250 C while that for Sn 37Pb was 210 C. In the course of an actual package anufacturing process, the packages are subjected to ultiple heating processes. Therefore, after one reflow, we repeated reflow of the test vehicles up to 3 ties under the sae profile conditions for each coposition. Consequently, the test vehicles were retained for about 300 s above the elting teperature. Both shear and heat ball pull (HBP) strength of the solder joint were evaluated using a Dage bond tester (Dage Series 4000P). Shear tests were perfored at a shear speed of 200 /s and a shear height of 5. The pull speed was 300 /s for each

2 Analyses of Soldering on Electroless Ni Au Pad Using Barrier Flux 2381 HBP test. Failure odes of the test vehicle were deterined with the help of a scanning electron icroscope (SEM). The test vehicles were olded in epoxy resin for grinding and polishing. An SEM equipped with energy-dispersive X- ray (EDX) was utilized to exaine the cross-sectional structure of the interface. The icrostructures of the joint interface were also revealed by using an etchant of 10% HCl and 90% ethanol. This etchant reoves the solder and leaves the interfacial interetallic copounds, allowing the to be characterized by SEM and EDX. 3. Results and Discussion 3.1 Solder joint strength Figure 1 shows the results for shear strength of each flux. Regardless of whether the solder ball coposition was Sn 4.0Ag 0.5Cu or Sn 37Pb, there appeared to be no significant difference in the shear strength between baseline flux and Barrier Flux. Investigation of the failure ode for this test revealed that in all cases the failure occurred inside the solder. This shows that the easured shear strength is a function of the bulk property of the solder; hence the results show siilar strength for each solder coposition. The pull strength data and its failure ode are suarized in Fig. 2. Copared to baseline flux, Barrier Flux showed an approxiately 12% increase in the ean pull strength for both solder copositions. It is also noteworthy that Barrier Flux gave a higher iniu pull strength, especially in the case of Sn 4.0Ag 0.5Cu. These results indicate that the addition of the Cu copound to the flux effectively prevents strength degradation and results in a narrower variation in strength. Results for the failure ode are also arked, as shown in Fig. 3. Here we classified the failure odes in accordance with the definition in Fig. 4. Regardless of solder coposition, the ajor failure was bond failure for baseline flux, whereas Barrier Flux predoinantly showed pad failure. This Frequency (%) (a) Sn-4.0Ag-0.5Cu (b) Sn-37Pb Baseline flux Barrier Flux Baseline flux Barrier Flux Bond failure Pad failure Ball failure Fig. 3 Frequency of the failure odes for each cobination of flux and solder. (a) Bond failure 20 (a) Sn-4.0Ag-0.5Cu (b) Sn-37Pb pull 15 Shear Strength, N 10 5 (b) Pad failure 0 Baseline flux Barrier Flux Baseline flux Barrier Flux pull Fig. 1 Shear strength of solder joints for each cobination of flux and solder. 30 (a) Sn-4.0Ag-0.5Cu (b) Sn-37Pb 25 (c) Ball failure 20 Pull Strength, N pull 5 0 Baseline flux Barrier Flux Baseline flux Barrier Flux Fig. 2 Pull strength of solder joints for each cobination of flux and solder. Fig. 4 Definition of failure odes.

3 2382 S. Kuaoto, H. Sakurai, K. Ikeda and K. Suganua difference indicates that the presence of the Cu copound contributes in soe way to the foration of a stronger joint interface. In the case of the cobination of baseline flux and Sn 4.0Ag 0.5Cu, cross-sectional observation of the bond failure saple revealed that fatigue cracks were initiated and propagated in the interetallic layer at the joint interface and at the interetallic/p-rich interface, as shown in Fig. 5. The foration of a clearly has a significant ipact on the joint strength. Soe previous studies have also reported that the generation of voids in the results in fragile solder joints. 4 6) 3.2 Interfacial icrostructure Figure 6 shows cross-sectional SEM iages of four joint interfaces. A qualitative analysis shows that the dark or blackish bands in each joint interface correspond to the P-rich (a) Cross-section SEM iage (x 200) A B Substrate 200 µ (b) Location A (x 8,000) (c) Location B (x 8,000) Interetallic copound Interetallic copound 5 µ 5µ Fig. 5 Cross-section SEM iages of the bond failure saple for Sn 4.0Ag 0.5Cu ball attachent using baseline flux. (a) Sn-4Ag-0.5Cu, Baseline flux (c) Sn-37Pb, Baseline flux 1µ 1µ (b) Sn-4Ag-0.5Cu, Barrier Flux (d) Sn-37Pb, Barrier Flux 1µ 1µ Fig. 6 Cross-section SEM iages of the joint interface for four cobinations. (a) Sn 4.0Ag 0.5Cu/baseline flux, (b) Sn 4.0Ag 0.5Cu/ Barrier Flux, (c) Sn 37Pb/baseline flux and (d) Sn 37Pb/Barrier Flux (40000).

4 Analyses of Soldering on Electroless Ni Au Pad Using Barrier Flux 2383 Table 3 Thickness of the (n). Solder Sn 4.0Ag 0.5Cu Sn 37Pb Flux (1) (2) (1) (2) location location location location location AVG (1) Baseline flux (2) Barrier Flux layer. This is generally assued to be coposed of Ni 3 PorNi 3 P þ Ni. According to EDX analysis, in the case of an Sn 4.0Ag 0.5Cu ball joint using Barrier Flux, the contained 88 at% Ni and 12 at% P, suggesting that this is ore likely to be Ni 3 P þ Ni than Ni 3 P. For each cobination of flux and solder ball, the ean thickness of the was calculated using the easured thicknesses at five different locations in the P-rich layer. As shown in Table 3, Barrier Flux gave a thinner than the baseline flux. It is presued that the reduction in thickness contributes to both the prevention of low joint strength and the low frequency in joint failure ode. In the case of Sn 4.0Ag 0.5Cu solder ball joints, the coparison of interetallic icrostructure was ade by SEM analysis. Prior to the SEM observations, the test vehicles were treated with an etchant of 10% HCl and 90% ethanol to reove the solder, finally leaving the interfacial interetallic copounds. As shown in Fig. 7, baseline flux produced nuerous granular interetallics about 1 in diaeter. On the other hand, in the case of Barrier Flux, we found that needle-shaped interetallics about 0.2 in width had fored. This shows that the addition of Cu copounds to flux led to the foration of fine interfacial interetallics that ipeded the diffusion of Ni into the solder. Table 4 Qualitative analyses of the interfacial interetallic copounds (in ass%). Solder Sn 4.0Ag 0.5Cu Sn 37Pb Flux (1) (2) (1) (2) Sn Cu Ni (1) Baseline flux (2) Barrier Flux Our qualitative analyses of the interfacial interetallic copounds are suarized in Table 4. We found that there was no significant differences in the interfacial interetallic coposition between fluxes for Sn 4.0Ag 0.5Cu solder ball joints; however, in the case of Sn 37Pb ball joints, each flux yielded different interetallic copounds. In contrast to soldering by baseline flux, where the foration of the Ni Sn interetallic copound was observed at the joint interface, Barrier Flux yielded interfacial Sn Cu Ni copounds rich in Cu but with relatively sall aounts of Ni. The presence of Cu at the joint interface prevented excessive diffusion of Ni into the solder, consequently aking the thinner. 3.3 Foration of the Cu layer at the joint interface When soldering with Barrier Flux, regardless of solder ball coposition, Sn 4.0Ag 0.5Cu or Sn 37Pb, the interfacial interetallic copounds contained Cu. To gain a better understanding of this phenoenon, we conducted the following verification experient. Barrier Flux was screenprinted on a test substrate with electroless Ni P plating pads. The test substrate was then subjected to reflow without solder ball placeent in a N 2 atosphere. After one reflow, the flux residue was washed away with organic solvent, followed by Auger electron spectroscopy (AES) analysis of the Ni P pad surface obtained. The presence of Cu on the pad surface was observed as shown in Fig. 8. It is assued that this Cu is derived fro the Cu copound in Barrier Flux. (a) Sn-4Ag-0.5Cu, baseline flux (b) Sn-4Ag-0.5Cu, Barrier Flux 5µ 5µ Fig. 7 SEM iages of the icrostructure of interfacial interetallics. (a) Sn 4.0Ag 0.5Cu/baseline flux and (b) Sn 4.0Ag 0.5Cu/ Barrier Flux.

5 2384 S. Kuaoto, H. Sakurai, K. Ikeda and K. Suganua (a) Before Barrier Flux treatent (b) After Barrier Flux treatent Ni Cu Kinetic Energy, E / ev Kinetic Energy, E / ev Fig. 8 Auger electron spectroscopy analyses of the pad surfaces. (a) Before Barrier Flux treatent and (b) after Barrier Flux treatent. Percentage of deposited Cu (ass %) blank Teperature, t / C Fig. 9 The dependence of Cu deposition on teperature. The aount of deposited Cu was deterined by EDX analysis. The dependence of Cu deposition on teperature was also evaluated. After supplying Barrier Flux to the test substrates using a 100 -thick etal ask, the test coupons were heated on a hotplate aintained at 135, 165, 200 and 200 C for 60 s, respectively. The aount of deposited Cu per pad was deterined by EDX analysis. Figure 9 plots the atoic % of deposited Cu at each teperature. These results showed that Cu starts to be deposited soewhere between 135 and 165 C, with deposition finally saturating at around 200 C. This indicates that the foration of the Cu layer (or the Cu-related reaction layer) on the pad surface is induced prior to solder ball elting. Although the Ni P surface finish is covered with thick Au plating, the presence of Au after heating is less certain due to its trace aount. In the case of reflow Barrier Flux without solder balls on the Ni P plating, the hypothetical behavior of both Au and Cu is as follows. Once the Cu copound is activated during reflow, Au which was originally present as a solid phase starts to substitute with (or elt into) Cu derived fro the Cu copound, after which the Cu is deposited on the pad surface, where it is iediately involved in the foration of interetallics at the joint interface. 4. Conclusions To address the joint strength degradation proble in Sn Ag Cu soldering on electroless Ni P plating, we focused on the functionality of the flux, and as a result were able to develop a novel flux, Barrier Flux, which deposits Cu on the pad surface by eans of substitution during reflow. The results obtained in this study are suarized as follows. (1) The use of Barrier Flux containing a Cu copound which generates Cu at teperatures of 200 C or lower is effective in preventing joint strength degradation in the Sn 4.0Ag 0.5Cu and Sn 37Pb soldering systes. (2) For both Sn 4.0Ag 0.5Cu and Sn 37Pb soldering, the thickness of the fored at the joint interface was decreased by the addition of a Cu copound to the flux. (3) Baseline flux containing no Cu copound yielded granular interetallics (about 1 in width) at the joint interface. On the other hand, we observed the foration of fine, needle-shaped interfacial interetallics (0.2 in width) in the case of Barrier Flux containing a Cu copound. This indicates that the use of Barrier Flux led to decreased diffusion of Ni into the solder. (4) Cu deposition derived fro the Cu copound is induced soewhere between 135 and 165 C. The deposited Cu layer works as a barrier to control the reaction between the Ni P layer and the solder; hence it inhibits the excessive diffusion of Ni.

6 Analyses of Soldering on Electroless Ni Au Pad Using Barrier Flux 2385 REFERENCES 1) S. Sakatani, Y. Kohara, T. Saeki, K. Uenishi and K. F. Kobayashi: 12th Micro Electronics Syposiu, (Japan Institute of Electronics Packaging, Osaka, 2002) ) Z. Mei, P. Gallery, D. Fisher, F. Hua and J. Glazer: Advances in Electronic Packaging 1997, (New York: ASME, 19-2, 1997) pp ) Y. Chonan, T. Koiyaa, J. Onuki, R. Urao, T. Kiura and T. Nagano: Mater. Trans. 43 (2002) ) N. Torazawa, S. Arai, Y. Takase, K. Sasaki and H. Saka: J. Japan Inst. Metals 66 (2002) ) C. W. Hwang, K. Suganua, M. Kiso and S. Hashioto: J. Mater. Res. 18 (2003) ) C. W. Hwang, K. Suganua, M. Kiso and S. Hashioto: J. Electron. Mater. 33 (2004) ) T. Hiraori, M. Ito, M. Yoshikawa, A. Hirose and K. F. Kobayashi: J. Jpn. Inst. Electron. Packag. 6 (2003) ) Y. Chonan, T. Koiyaa and J. Onuki: Surf. Mt. Technol. 54 (2003) ) Y. Nakahara, K. Hira, R. Ninoiya, M. Tagai and M. Sugai: Quart. J. Jpn. Weld. Soc. 21 (2003) ) T. Hiraori, M. Ito, Y. Tanii, A. Hirose and K. F. Kobayashi: 10th Syposiu on Microjoining and Assebly Tech. in Electronics, Yokohaa (2004)