Journal of ELECTRONIC MATERIALS, Vol. 37, No. 7, 2008 DOI: 10.1007/s11664-008-0397-4 Ó 2008 TMS Regular Issue Paper Adhesion and Reliability of Anisotropic Conductive Films (ACFs) Joints on Organic Solderability Preservatives (OSPs) Metal Surface Finish HYOUNG-JOON KIM 1,2 and KYUNG-WOOK PAIK 1 1. Nano Packaging and Interconnects Laboratory (NPIL), Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 373-1, Guseong-dong, Yuseong-gu, Daejeon 305-701, Republic of Korea. 2. e-mail: kim.hyoungjoon@gmail.com The effect of final metal finishes of Cu electrodes on the adhesion and reliability of anisotropic conductive film (ACF) joints was investigated. Two different metal surface finishes, electroless Ni/immersion Au (ENIG) and organic solderability preservatives (OSPs) coated on Cu, were selected in this study for ACF bonding. The adhesion strength of ACF/OSP joints was higher than that of ACF/bare Cu and ACF/ENIG joints. The fracture sites of the ACF/bare Cu and ACF/ENIG joints were ACF/metal interfaces, while those of ACF/OSP joints were inside the ACF. Transmission electron microscope (TEM) and Fourier-transform infrared (FT-IR) analyses showed that the OSP coating layer on the Cu electrodes reacted with the epoxy resin of the ACFs but still remained at the bonding interface. According to the in-depth X-ray photoelectron spectroscopy (XPS) analysis, additional C-N bonds formed after the OSP-epoxy reaction and the outermost nitrogen of the OSP layer participated in curing of the epoxy resin of the ACF. Therefore, the OSP layer acted as an adhesion promoter to ACFs. Furthermore, this role of the OSP layer enhanced the reliability of the ACF/OSP joints under high temperature and humid environments, as compared to the ACF/ENIG joints. Key words: Anisotropic conductive film (ACF), organic solderability preservative (OSP), adhesion, surface finish, flexible substrate, flex-on-board (FOB) INTRODUCTION An electroless Ni/immersion Au (ENIG) layer has been used as a common metal surface finish of printed circuit boards (PCBs) for solder bonding as well as anisotropic conductive film (ACF) bonding. However, ENIG has a higher processing cost, and it has reliability problems. One such problem is known as the black pad, which causes a brittle failure at ENIG/solder joints. Therefore, various alternative metal surface finish technologies such as an organic solderability preservative (OSP), direct immersion Au (DIG), electroless Ni/electroless Pd/immersion Au (ENEPIG), and others have been proposed to (Received February 27, 2007; accepted January 18, 2008; published online March 29, 2008) replace the ENIG system. Essentially, OSP is a thin organic layer coated on the surface of Cu electrodes. It can protect the surface of Cu electrodes from oxidation and tarnishing. Although the protective function of OSP disintegrates at elevated temperatures due to the instability of OSPs at high temperatures, the use of OSP has increased recently. The simple processes, environmental considerations, and its lower cost are the driving forces in the growing use of OSPs. The OSP process provides more than a 50% cost reduction versus the ENIG process. 1 Moreover, the need for coplanar surface mount (SMT) surfaces and the advent of chip scale packages and ball grid arrays (BGAs) expand the need for OSPs due to its good co-planarity. 2 Benzotriazole was one of the earliest OSP formulations. However, the copper benzotriazole 1003
1004 Kim and Paik complex coating layer showed very poor oxidation resistance under thermal cycles, indicating that the heat resistance of OSP needed to be enhanced. Therefore, benzimidazole has been used as a base material of OSP to improve the thermal stability of OSP, resulting in an OSP layer that can survive several thermal cycles. 3 In fact, benzimidazole has been used extensively in industry as a good corrosion inhibitor for transition metals and their alloy surfaces, as it forms a protective layer, particularly in copper. 4 7 The protective layer is formed initially through a complexing reaction with copper to form an organometallic bond, followed by a build-up of the benzimidazole copper complexes. The greater thermal stability of the benzimidazole copper complex layer allows OSP to be Pb-free compatible. Many studies on the interfacial phenomena and reliability issues of various Pb-free solders on an OSP-finished copper surface have been reported. 8 11 Recently, modular approach has been the trend of handheld products manufactured for their higher functionality and smaller size. Each functional module, on flexible substrates, is connected to the main organic rigid substrates. Organic rigid substrate flexible substrate (RS FS) bonding using ACFs is one of the most promising module assembly methods. As a result, ACF bonding of a flexible substrate on an OSP-finished rigid substrate has become more important. However, the effect of an OSP coating on the adhesion and reliability of the ACF/OSP joints has not been investigated. Furthermore, the feasibility of ACF bonding on the OSP-coated surface should be evaluated. MATERIALS AND EXPERIMENTS Test Vehicle Preparation An ACF material for PCB bonding application was used in this study. The details of the ACF material are listed in Table I. In general, the ACFs for PCB bonding applications use Ni metal balls as conductive particles instead of metal-coated polymer balls, due to the need for high current-carrying capability and a high surface roughness of the metal electrode on rigid substrates. The FR-4-based rigid substrate had a 12 lm-thick Cu metal electrode. The final metal surface finish conditions were ENIG with a 5-lm-thick electroless Ni and 0.3-lm-thick immersion Au layer, and OSP on copper surface. The pitch of the Cu metal electrodes was 400 lm, consisting of Cu metal Table I. Details of the ACF Material Base Resin Type Bisphenol A Type Epoxy T g ( C) 108.5 Thickness (lm) 40 Width (mm) 2.5 Conductive particle 6-lm-diameter Au-coated Ni ball electrodes with a 220 lm width and a 180 lm gap between adjacent Cu metal electrodes. A polyimidebased casting-type flexible substrate (EspaNex TM ) with 25-lm-thick polyimide (PI) and 12-lm-thick Cu metal electrodes was used. The final metal finish of flexible substrates was 0.5 lm Ni/0.1 lm Au. Top views of the rigid and flexible substrates are shown in Fig. 1. The contact structures of ACF/OSP and ACF/ENIG joints were observed using a focused ion beam (FIB). For the contact resistance measurement, Kelvin-patterned flexible and rigid substrates were bonded using the ACF. To assemble the test vehicles, conventional thermocompression ACF bonding was performed by applying heat and pressure simultaneously for 20 s. The in situ temperature at the ACF layer was measured by using an adhesive-type thermocouple, and the measured temperature profile is shown in Fig. 2. Bonding pressure was 7.2 MPa. The Characteristics of the ACF/OSP Reaction The morphologies of an OSP coating layer before and after ACF bonding were observed by FIB and a transmission electron microscope (TEM). These observations were crucial for understanding the role of the OSP coating layer during ACF bonding. In order to investigate the reactivity between the epoxy resin of ACF and the OSP, an ACF without conductive particles and a latent curing agent was prepared. This ACF was applied onto each of the three type of Cu foils (bare Cu, ENIG finish, and OSP finish), and these three foils were heated to 190 C on a hot plate. Then, they were cooled to room temperature. After heating and cooling, the ACFs on each Cu foil were removed with acetone. Because these ACFs did not contain a curing agent, no curing reaction occurs even at a high temperature of 190 C. However, if there were any interfacial reactions between the epoxy resin of ACF and the surface finish materials, the residues or products of chemical reactions would remain at the Cu foil surface even after acetone cleaning. For this reason, the differences in the Cu foil surfaces were initially checked optically. And then, the surfaces were analyzed using a Fourier-transform infrared (FT-IR) spectroscopy and an X-ray photoelectron spectroscopy (XPS). XPS was performed in an ultrahigh-vacuum (UHV) system at a base pressure of 10-10 Torr. Photoelectrons were excited by nonmonochromatized Mg Ka (1253.6 ev) radiation. The binding energies were calibrated by setting the instrument work function to give an Ag3d 5/2 line position at 368.3 ev. A wide scan was taken in order to survey all spectra, and high-resolution spectra were also analyzed. Adhesion Test Three types of Cu foils (bare Cu, ENIG finish, and OSP finish) and two copper-clad laminates (CCLs) with an ENIG finish and an OSP finish were prepared in order to investigate the effect of the OSP
Adhesion and Reliability of Anisotropic Conductive Films (ACFs) Joints on Organic Solderability Preservatives (OSPs) Metal Surface Finish 1005 Fig. 1. Top views of a rigid substrate and a flexible substrate. C) o In-situ ACF temperature ( 240 220 200 180 160 140 120 100 80 60 40 20 Temperature profile during ACF bonding 0 0 5 10 15 20 Bonding time (sec) Fig. 2. In situ ACF temperature profile during ACF bonding. finish on the adhesion of ACF joints. ACFs were applied on CCLs, and prebonding was performed at 70 C. After removing the releasing paper of ACFs, each Cu foil was bonded on CCLs using ACFs, as shown in Fig. 3. Following this step, a 90 deg peel test was performed with a 10 mm s -1 test speed. After the peel test, the fractured sites of the CCLs and Cu foils were observed by using a scanning electron microscope (SEM). Reliability Test After the Kelvin-patterned RS FS bonding using ACFs, a pressure-cooker test (PCT) was performed as a reliability assessment. Test conditions for the PCT were 121 C, 100% relative humidity (RH), and 2 atm. During the test, the changes in the contact resistances were measured every 24 h for a total test time of 144 h. After the test, delaminations or cracks were observed using a cross-sectional SEM. RESULTS AND DISCUSSION Comparison of ACF Joint Structure and Contact Resistance Figure 4 shows cross-sections of the ACF joints and the deformations of the conductive particles in each surface-finished sample. As shown in Fig. 4, Fig. 3. ENIG Cu foil-enig CCL and OSP Cu foil-osp CCL combinations, prepared for the adhesion test. the Ni conductive particles were well squeezed between both the Cu electrodes of the rigid and the flexible substrates. Essentially, the contacts between the Cu electrodes and the conductive particles were mechanically established by the contraction of the ACF resin. Therefore, the electrical conduction was established through particle contacts, and initial contact resistances were nearly identical for both ENIG and OSP surface-finished samples, as shown in Table II. OSP Layer Observation before and after ACF Bonding Figure 5a shows an OSP-finished Cu electrode surface before ACF bonding. In this figure, a rectangular hole was formed on the OSP-finished Cu electrode, resulting in sputtering of Ar ions in the FIB chamber. Figure 5b shows the existence of the OSP coating layer on a Cu electrode. The thickness of the OSP coating layer was about 70 nm to 100 nm, and a small amount of thickness deviation was caused by the roughness of the Cu electrode. To observe the change in the OSP layer after ACF bonding, an ACF/OSP joint was cross-sectioned using the FIB. As shown in Fig. 6a, it was difficult to distinguish the OSP layer from the ACF layer by SEM observation, therefore a TEM analysis was performed. As described in Fig. 6b, the initial OSP coating layer remained after the ACF bonding, and a well-defined ACF/OSP interface was observed.
1006 Kim and Paik Fig. 4. The deformation of conductive Ni particles in ACFs at the ENIG-finished RS FS bonding and the OSP-finished RS FS bonding ACF joints by FIB cross-sectional analysis. Table II. Measured Initial Contact Resistances (40 Measurements for Each Sample) Sample Type Contact Resistance (mx) ENIG-finished RS FS 14.74 ± 0.59 OSP-finished RS FS 15.45 ± 0.59 This result indicates that the OSP coating layer is not totally decomposed or removed during ACF bonding. Moreover, the total thickness of the OSP layer decreased from the as-coated 100 nm to 20 nm after ACF bonding. Therefore, based on these thickness changes, it is obvious that the OSP layer reacts with the epoxy resin of the ACF during the ACF bonding process. Characterization of the OSP/ACF Reactivity To investigate the reactivity between the OSP and the ACF during ACF bonding, an ACF without a curing agent was prepared. This ACF was then applied to three types of Cu foils (bare Cu, ENIG finish, and OSP finish) and these ACF-applied Cu foils were heated to the ACF curing temperature and then cooled to room temperature. Figure 7 shows the differences between each Cu foil surface after the thermal treatment. It was found that no reaction took place at the ACF/bare Cu and ACF/ENIG interfaces. Therefore, the ACFs on both bare and ENIG-finished Cu foils were completely removed by acetone cleaning. However, as shown in Fig. 7b, a stained area where the ACF was applied was observed on the OSP-finished Cu foil surface, even after an acetone clean. This result implies that a certain chemical reaction between the epoxy resin of ACF and the benzimidazole of the OSP takes place at the ACF curing temperature, causing the reacted residue to remain even after acetone cleaning. For an in-depth surface analysis, FT-IR and XPS were performed on the stained area. Figure 8a shows the FT-IR analysis result on an as-received OSP-finished Cu foil surface. This graph is in good agreement with the reported result for Fig. 5. Observation of a top view and a cross-section of the OSP layer on Cu electrode surface before ACF bonding using a FIB.
Adhesion and Reliability of Anisotropic Conductive Films (ACFs) Joints on Organic Solderability Preservatives (OSPs) Metal Surface Finish 1007 Fig. 6. Observation of the OSP/ACF interface after ACF bonding using the FIB-SEM. The remaining OSP layer can be observed even after ACF bonding by cross-sectional TEM. OSP. 12 In addition, Fig. 8b shows the FT-IR result of the stained region of the Cu foil surface as described in Fig. 7b. The largest peak difference between two graphs was observed in the 2,700 cm -1 to 3,000 cm -1 range. This peak comes from the C-H antisymmetric stretching mode. 13 As shown in Fig. 8a, no peak was observed at 2,700 cm -1 to 3,000 cm -1 from the as-received OSP-finished Cu foil surface, because the benzimidazole, the base material of OSP, consisted of benzene rings and an imidazole structure. Therefore, the origin of this peak at 2,700 cm -1 to 3,000 cm -1 comes from the epoxy resin of ACF. That is, it is clear that the epoxy resin of ACF can react with the benzimidazole of OSP at the ACF curing temperatures. The wide scan of the XPS analysis is shown in Fig. 9. The intensity of the Cu2p 3/2 peaks increased at the stained area. This indicates that the OSP layer was consumed by the interfacial reaction with ACF at the ACF curing temperature. Therefore, the intensity of the Cu background peaks increased after the OSP-ACF reaction. This result agrees well with the cross-sectional TEM result. For in-depth analysis of the chemical bonding changes, C1s peaks were deconvoluted (see Fig. 10). The largest difference before and after the OSP-ACF reaction was the change in the C with N bond portion. As shown in Fig. 10b, the portion of the C-N bond peak increased after the OSP-ACF reaction. This result implies that additional C-N bonds formed as a result
1008 Kim and Paik Bare Cu OSP ENIG ACF w/o curing agent Bare Cu OSP ENIG Stained area Fig. 7. Observation of the reactivity between the ACF and three surface metal finishes before and after heat treatment. of the interfacial reaction between the OSP and the ACF. In other words, due to the similarity of the chemical structure between the benzimidazole of the OSP and the imidazole curing agent of ACF, the outermost nitrogen atom of OSP can open the epoxide ring, and initiate the epoxy curing reaction during ACF bonding. Therefore, strong chemical bonding as well as mechanical adhesive bonding can be formed at the ACF/OSP interface, resulting in an enhancement of the adhesion strength and reliability. The possible reaction site during ACF bonding on the OSP layer is illustrated in Fig. 11. Effect of OSP on Adhesion of ACF Joints The results of a 90 deg peel test are listed in Table III. The average peel strength of the ACFbonded OSP CCL-OSP Cu foil combination showed the strongest adhesion strength among three ACFbonded CCL-Cu foil combinations. In particular, for the ENIG CCL-ENIG Cu foil combination, the delamination of the ENIG layer on the Cu surface significantly reduced the adhesion strength. After the peel test, the fractured surfaces of the CCLs and Cu foils were observed by SEM. According to the failure analysis, the adhesion between the bare Cu surface and ACF was lower than that of the OSP layer and ACF. Therefore, as shown in Fig. 12, the fracture path moved from the bare Cu/ACF As-received OSP finished OSP Cu Cu surface Cu surface After OSP NCF Cu Cu cleaning surface on after after OSP finished OSP-ACF Cu surface reaction Intensity (a.u) Intensity (a.u) 4000 3500 3000 2500 2000 1500 1000 4000 3500 3000 2500 2000 1500 1000 Wavenumber (cm -1 ) Wavenumber (cm -1 ) Fig. 8. Results of the FT-IR spectrum of the as-received OSP-finished Cu surfaces and the stained area of OSP-finished Cu surface reacted with ACF after an acetone cleaning. Intensity (arbitrary unit) C1s N1s O1s B-Cu OSP-Cu OSP-ACF-Cu Cu2p 3/2 Intensity (arbitrary unit) Cu2p 3/2 OSP-Cu OSP-ACF-Cu 0 300 600 900 800 900 1000 Binding energy (ev) Binding energy (ev) Fig. 9. Results of the XPS wide scan and Cu 2p 3/2 peaks of bare Cu surface (B-Cu), OSP-finished Cu surface (OSP-Cu) and the stained area of OSP-finished Cu surfaces reacted with ACF (OSP-ACF-Cu).
Adhesion and Reliability of Anisotropic Conductive Films (ACFs) Joints on Organic Solderability Preservatives (OSPs) Metal Surface Finish 1009 Intensity (a.u.) Original peak C combined H C combined N C double bond Intensity (a.u.) Original peak C combined H C combined N 294 292 290 288 286 284 282 Binding energy (ev) 292 290 288 286 284 282 Binding energy (ev) Fig. 10. XPS results of the C 1s peak deconvolution of as-received OSP-finished Cu surface and the stained area of OSP-finished Cu surface reacted with ACF. Fig. 11. A schematic of ACF bonding on the OSP-finished surface. A chemical reaction can occur between the outermost nitrogen of the OSP layer and the epoxide ring of ACF. Table III. Results of the 90 deg Peel Test Sample Type Peel Adhesion Strength (gf) ENIG CCL-ENIG Cu foil 144.7 ± 45.4 OSP CCL-bare Cu foil 944.9 ± 90.6 OSP CCl-OSP Cu foil 1167.4 ± 80.5 interface to inside the ACF layer. The fracture paths of the samples are illustrated in Fig. 13. The change in the fracture path implies that the OSP coating layer has good adhesion with the epoxy-based adhesive, and that the OSP finish can enhance the bondability of ACF joints. This result corresponds with a previous study which reported that the polymeric adhesion promoter, polybenzimidazole, which is similar as the OSP material, showed better adhesion strength as it prevented the surface oxidation of copper. 14 In addition, it is also reported that the benzimidazole can improve the adhesion between polymeric materials such as epoxy or polyimide, and copper. 15,16 Generally, imidazoletype curing agents are widely used for epoxy curing. The chemical structure of the OSP material (benzimidazole) is also based on the imidazole system. Therefore, as described in Fig. 11, the chemical reaction between benzimidazole in OSP and ACF can form strong bonding at the ACF/OSP interface, resulting in an enhanced adhesion. Reliability Test Results According to the PCT results, OSP-finished samples showed better PCT reliability compared with the ENIG-finished samples. As shown in Fig. 14a, the degradation of the ACF joints in the ENIG-finished samples started after 48 h of PCT. After 96 h of PCT, 50% of measurement points exceeded 200 mx, and the degradation of the ACF joints was accelerated as the test time increased. However, the
1010 Kim and Paik Fig. 12. Observation of the fractured Cu foil and CCL sides of OSP CCL-bare Cu foil bonded sample, and (c) Cu foil and (d) CCL sides of OSP CCL-OSP Cu foil bonded sample. Fig. 13. Schematics of the fracture paths of OSP CCL-bare Cu foil, and OSP CCL-OSP Cu foil. changes in the contact resistances of the ACF joints in the OSP-finished samples were very stable up to 120 h of PCT. Actually, the failure rate of the OSPfinished samples after 144 h of PCT was nearly identical to that of the ENIG-finished samples after 72 h of PCT. The main cause of the failure was the delamination of the ACF/PI of the flexible substrate interfaces in both surface-finished samples. However, an additional failure site was observed in the ENIG-finished samples. As shown in Fig. 15a, some additional delaminations and cracks occurred at the ACF/ENIG interface. This defect can act as a potential failure site during the reliability tests. In contrast, no defects were observed in the OSP finished samples. Consequently, it is considered that the OSP coating layer provides good adhesion with ACFs, and that the improved initial adhesion influences and enhances the reliability of ACF joints. CONCLUSIONS Benzimidazole, the base material of OSP, improved the adhesion strength of ACF joints. According to the results of TEM and FT-IR, the OSP-finished layer reacted with the epoxy resin of the ACF at the ACF curing temperature and enhanced the adhesion. That is, the OSP layer acted as an adhesion promoter. The improved adhesion at the ACF/OSP interface affected the reliability of assembled ACF joints. The ACF joint assembled
Adhesion and Reliability of Anisotropic Conductive Films (ACFs) Joints on Organic Solderability Preservatives (OSPs) Metal Surface Finish 1011 Cumulative percentage (%) 100 90 80 70 60 50 40 30 20 10 0 0 20 40 60 80 100 120 140 160 180 200 Contact resistance (mω) ENIG- PCT 0 hrs 24 hrs 48 hrs 72 hrs 96 hrs 120 hrs 144 hrs Cumulative percentage (%) 100 90 80 70 60 50 40 30 20 10 0 0 20 40 60 80 100 120 140 160 180 200 Contact resistance (mω) OSP- PCT 0 hrs 24 hrs 48 hrs 72 hrs 96 hrs 120 hrs 144 hrs Fig. 14. Results of the changes of contact resistances of ENIG-finished and OSP-finished samples during the pressure-cooker test. Fig. 15. Cross-sections of ENIG-finished and OSP-finished samples after the pressure-cooker test. with OSP-finished rigid substrates showed better reliability than the ENIG-finished case. Besides, no defects such as cracks and contact losses were observed at the ACF/OSP interface. These results implied that the ACF/OSP interface was more stable than the ACF/ENIG interface in high temperature and humid environments. Therefore, OSP is applicable to the final metal finish method for making reliable ACF joints. ACKNOWLEDGEMENT This work was supported by the Center for Electronic Packaging Materials (ERC) of MOST/KOSEF (Grant No. R11-2000-085-08005-0). REFERENCES 1. E. Stafstron, Circ. Assembly 11, 56 (2000). 2. M. Carano, Printed Circ. Fabric. 20, 28 (1997). 3. J.D. Debiase, Surf. Mount Int. 763 (1996). 4. G. Lewis, Corros. Sci. 22, 589 (1982). 5. D.P. Drolet, D.M. Manuta, A.J. Lees, A.D. Katnani, and G.J. Coyle, Inorg. Chim. Acta 146, 173 (1988). 6. Yu.I. Kuznetsov, L.P. Podgoronova, and L.P. Kazanskii, Prot. Metals 40, 130 (2004). 7. G. Xue, J. Ding, P. Wu, and G. Ji, J. Electroanal. Chem. 270, 163 (1989). 8. P. Collier, V. Sunappan, and A. Periannan, Solder. Surf. Mount Technol. 14, 12 (2002). 9. J.W. Nah (Ph.D. Thesis, KAIST 2004). 10. L. Xu and J.H.L. Pang, J. Electron. Mater. 35, 2107 (2006). 11. P.-L. Wu, M.-K. Huang, C. Lee, and S.-R. Tzan, J. Electron. Mater. 33, 157 (2004). 12. M. Frederickson and B. Goers, Circ. World 24, 10 (1998). 13. S. Yoshida and H. Ishida, J. Adhes. 16, 217 (1984). 14. S.M. Song, K. Cho, C.E. Park, H.K. Yun, and S.Y. Oh, J. Appl. Polym. Sci. 85, 2202 (2002). 15. S. Siau, A. Vervaet, E. Schacht, S. Degrande, K. Callewaert, and A.V. Calster, J. Electrochem. Soc. 152, D136 (2005). 16. J. Yu, M. Ree, T.J. Shin, X. Wang, W. Cai, D. Zhou, and K.-W. Lee, J. Poly. Sci. Part B 37, 2806 (1999).