2017 Volume 58 Issue 8 Pages 1217-1222
The contact resistance and microstructure of the flip-chip joints processed using anisotropic conductive adhesive (ACA) were characterized on flexible printed-circuit-board (FPCB) and stretchable FPCB/polydimethylsiloxane (FPCB/PDMS) substrates. On the FPCB substrate, the contact resistance was remarkably reduced from 56.9 mΩ to 13.3 mΩ when the bonding pressure increased from 10 MPa to 200 MPa. However, at a bonding pressure of 300 MPa, it substantially increased to 74.8 mΩ with an excessive deviation of ±37.7 mΩ. On the more compliant FPCB/PDMS substrate, the contact resistance decreased from 43.2 mΩ to 31.2 mΩ when the bonding pressure increased from 10 MPa to 50 MPa. Severe distortion of the FPCB/PDMS substrate occurred at bonding pressures above 50 MPa because of the softness of the PDMS. A more compliant substrate has a lower appropriate bonding pressure for the flip-chip process.
Recently, much attention has been focused on wearable devices for various new applications that could not be possible with conventional rigid electronics.1–9) To ensure the sufficient functionality, wearable comfortability, and aesthetical acceptability of wearable devices, stretchable electronic packaging technologies with an island-bridge or archipelago configuration, which consists of mechanically disparate soft and hard materials in a single structure, have been proposed.5,8,10–14)
Flip-chip bonding has been used in the electronic packaging industry for high-density and fine-pitch chip interconnection on flexible printed circuit boards (FPCBs) and rigid substrates, such as printed circuit boards (PCBs) and liquid-crystal-display glass substrates.15–17) However, it is notably difficult to find any reports of the flip-chip process on a stretchable substrate that consists of polydimethylsiloxane (PDMS), which is considered the most suitable elastomer to use as the substrate material for stretchable electronic packaging because of its superior stretchability, flexibility, biocompatibility, easy processability, low curing temperature, low dielectric constant, low loss tangent, and high dielectric strength.18–22)
The flip-chip process can be accomplished by using either solder-bump reflow or an anisotropic conductive adhesive (ACA). Compared to the flip-chip technology with solder bump reflow, the process using an ACA or anisotropic conductive film (ACF), where conducting particles are randomly distributed in an adhesive, has the significant advantages of low-temperature bonding and fluxless bonding.15–17) PDMS is deformed by solvent-induced swelling and is not recommended for use at temperatures above 200℃.23–26) The flip-chip process using ACA is more applicable for stretchable packaging using the PDMS substrate than the flip-chip process using solder reflow because the reflow temperatures of typical Pb-free solders, such as Sn-Ag-Cu and Sn-Ag, are above 200℃, and solvent cleaning for flux removal is required after the solder reflow.15)
To develop chip interconnection technology for stretchable electronic packaging from the established technologies for conventional and flexible electronic packages, the flip-chip bonding behavior on substrates of different stiffness values must be compared. We sequentially measured the average contact resistance of flip-chip joints processed on the rigid Si substrate, flexible FPCB substrate, and stretchable and compliant FPCB/PMDS substrate. In this paper, which is the second part of our study, we investigated the flip-chip bonding behavior on the flexible FPCB substrate and stretchable FPCB/PMDS substrate.
The fabrication process of a Si chip to mount on the FPCB and FPCB/PDMS substrates was described in detail in our previous report regarding the flip-chip behavior on rigid Si substrates.27) The 78-μm-thick flexible FPCB substrate consisted of 45-μm-thick polyimide, 18-μm-thick Cu with the daisy-chain pattern, and 15-μm-thick photosolder-resist (PSR). Forty-two pads of 135-μm-diameter and 600-μm-pitch were fabricated by treating the Cu surface, which was exposed through the PSR, with the 4-μm-thick electroless Ni and 30-mm-thick immersion gold (ENIG).
Figure 1 shows the process flow to fabricate the FPCB/PDMS composite substrate by locally attaching the more rigid FPCB to the more compliant PDMS using silicone/acrylic double-coated tape, which consists of silicone pressure-sensitive adhesive coated on one side of a polyester film carrier and acrylic pressure-sensitive adhesive coated on the other side of the carrier. The FPCB part, where the acrylic-adhesive side of the silicone/acrylic double-coated tape was bonded, was attached to the half-cured PDMS using the silicone-adhesive side of the double-coated tape. The half-cured PDMS was formed by mixing the base of Dow Corning Sylgard 184 with its curing agent at a weight ratio of 10:1 and curing at 60℃ for 25 min. Then, the PDMS, where the FPCB was attached, was fully cured at 60℃ for 12 hr to complete the formation of the FPCB/PDMS composite substrate of 40 mm × 18 mm × 1 mm-size. Figure 2 shows the optical images of the flexible FPCB substrate and stretchable FPCB/PDMS substrate.
Fabrication process of a PDMS/FPCB composite substrate: (a) attachment of the double-coated adhesive tape to FPCB; (b) attachment of FPCB to half-cured PDMS; (c) full curing of the FPCB-attached PDMS.
Optical images of the (a) flexible FPCB substrate and (b) stretchable PDMS/FPCB substrate.
After dispensing the ACA, where Au-coated polymer beads of 4-μm-diameter were dispersed as conductive particles, the Cu/Au chip bumps were aligned onto the substrate pads and bonded together by curing the ACA at 160℃ for 150 sec with a bonding pressure of 10–300 MPa. The average contact resistance of the flip-chip joint was analyzed by measuring the daisy-chain resistance. Cross-sectional microstructures of the flip-chip joints were observed using scanning electron microscopy (SEM).
The average contact resistances of the flip-chip joints that formed on the flexible FPCB and stretchable FPCB/PDMS substrates are shown in Fig. 3. For comparison, the average contact resistance measured on the rigid Si substrates27) is also plotted in Fig. 3. On the FPCB substrate, the contact resistance remarkably reduced from 56.9 mΩ to 13.3 mΩ when the bonding pressure increased from 10 MPa to 200 MPa. However, the contact resistance substantially increased to 74.8 mΩ with an excessive deviation of ±37.7 mΩ at the bonding pressure of 300 MPa.
Average contact resistances on the flexible FPCB and stretchable FPCB/PDMS in comparison with the contact resistance on the rigid Si substrate.
Cross-sectional micrographs of the flip-chip joints formed on the FPCB substrate are shown in Fig. 4. The conductive particles became severely deformed with increasing the bonding pressure. However, unlike the case for the rigid Si substrate,27) their intercalation into the Au layer of the chip bump was not observed on the FPCB substrate even at bonding pressures above 100 MPa, which can be due to the cushioning effect of the compliant FPCB substrate. Contrary to other specimens processed at a bonding pressure up to 200 MPa, complete cracking at the ACA resin between the chip and the substrate was observed for four samples bonded at 300 MPa after their mounting and polishing for microstructural observation. The elastic deformation of the compliant FPCB substrate increased at higher bonding pressure and caused more downward deflection at the area under the chip bumps and further upward displacement in other regions. When bonded at a bonding pressure of 300 MPa, the restoring force of the elastically deformed FPCB substrate to its original shape became sufficiently large to cause a large deviation in the measured contact resistance and completely separated the chip from the substrate during the microstructural sample preparation. The downward deflection of the FPCB substrate at the area under the chip bump made the edge of the chip bump touch the FPCB substrate, which hindered the full compression of the ACA conductive particles under the chip bump. Hence, the contact resistance increased with increasing bonding pressure from 200 MPa to 300 MPa, as in Fig. 3.
Cross-sectional scanning electron micrographs of the flip-chip joints formed on the FPCB substrate with a bonding pressure of (a) 10 MPa, (b) 50 MPa, (c) 100 MPa, (d) 200 MPa, and (e) 300 MPa.
As shown in Fig. 5, the FPCB/PDMS substrate severely distorted with the chip burrowed into it at 50 MPa and 100 MPa because of the softness of the PDMS. Although the contact resistance of the flip-chip joints on the FPCB/PDMS substrate was measured at 50 MPa and 100 MPa, as in Fig. 3, the appropriate bonding pressure for the flip-chip process on this stretchable substrate of soft PDMS can be considered as low as 10 MPa. The contact resistance decreased from 43.2 mΩ to 31.2 mΩ when the bonding pressure increased from 10 MPa to 50 MPa, but it slightly increased to 35.0 mΩ at the bonding pressure of 100 MPa. This contact resistance variation vs. the bonding pressure is similar to that observed on the FPCB substrate, whereas the critical bonding pressure, beyond which the contact resistance increased, decreased to 50 MPa for the softer FPCB/PDMS substrate from 200 MPa for the less compliant FPCB substrate. In Fig. 6, the cross-sectional micrographs of the flip-chip specimens on the FPCB/PDMS substrate show that the local downward deflection of the substrate under the bump prevented the full compression of the ACA particles at the central region of the bump compared to those under the edge of the bump. The comparison of the contact resistance data on the FPCB and FPCB/PDMS substrates with their microstructures shows that the critical bonding pressure, beyond which the contact resistance increases, decreases for the more compliant substrate.
Top-view and side-view optical images of the flip-chip specimens that were processed on the FPCB/PDMS substrate with a bonding pressure of (a) 10 MPa, (b) 50 MPa, and (c) 100 MPa.
Cross-sectional scanning electron micrographs of the flip-chip joints formed on the FPCB/PDMS substrate with a bonding pressure of (a) 10 MPa, (b) 50 MPa, and (c) 100 MPa.
To characterize the deformation behavior of the rigid Si, flexible FPCB, and stretchable FPCB/PDMS substrates as a function of the bonding pressure, the bondline thickness (BLT) was measured for the flip-chip samples on each substrate. Here, the BLT was defined as the actual thickness of the ACA resin that remains between the chip and the substrate.28,29) The BLT is the distance between the chip electrode and the SR layer of the FPCB substrate, but it is the distance from the chip electrode to the top of the Si substrate, where the SR layer is absent. The deviation of the BLTs measured at different positions of a flip-chip sample becomes more noticeable with more non-uniform local deformation of the substrate during the flip-chip bonding. To clarify the difference in BLT on each substrate, the BLTs were measured adjacent to the chip bump, where the local bonding pressure was most severely applied, and at the middle of two chip bumps, where the bonding pressure was least applied, and ΔBLT was calculated.
Figure 7 shows ΔBLT on the Si, FPCB, and FPCB/PDMS substrates as a function of the bonding pressure. On the Si substrate, ΔBLT was notably small and less than 0.35 μm at all bonding pressures, which is affirmed with the cross-sectional microstructures in Fig. 8. On the FPCB substrate, ΔBLT increased from 0.72 μm to 4.07 μm when the bonding pressure increased from 10 MPa to 200 MPa. For the sample bonded at 300 MPa, the BLT could not be measured because of the delamination between the chip and the substrate during the specimen preparation for microstructural observation. The cross-sectional microstructures of the flip-chip specimens bonded at 200 MPa and 300 MPa are compared in Figs. 9(d) and (e), which show the more downward deflection of the substrate under the bump for the specimen bonded at 300 MPa. Thus, ΔBLT is larger with more upward warpage of the substrate between the chip bumps. For the FPCB/PDMS substrate, which has the largest compliance among the three substrates, ΔBLT was 2.4 μm even at 10 MPa and rapidly increased to 11.9 μm and 12.9 μm at 50 MPa and 100 MPa, respectively, with severe deformation of the substrate, as shown in Fig. 10. The elastic moduli of Si, FPCB, and PDMS are 110 GPa, 2.8 GPa, and 1.7 MPa, respectively.19,30) The more significant deformation of the softer substrate during flip-chip bonding increased ΔBLT in the order of FPCB/PDMS, FPCB, and Si substrate, as shown in Fig. 7.
Difference in bondline thickness (ΔBLT) of the flip-chip specimens processed on the Si, FPCB, and FPCB/PDMS substrates as a function of the bonding pressure.
Cross-sectional scanning electron micrographs of the flip-chip specimen processed on the Si substrate at a bonding pressure of (a) 10 MPa, (b) 50 MPa, (c) 100 MPa, (d) 200 MPa, and (e) 300 MPa.
Cross-sectional scanning electron micrographs of the flip-chip specimen processed on the FPCB substrate at a bonding pressure of (a) 10 MPa, (b) 50 MPa, (c) 100 MPa, (d) 200 MPa, and (e) 300 MPa.
Cross-sectional scanning electron micrographs of the flip-chip specimen processed on the FPCB/PDMS substrate at a bonding pressure of (a) 10 MPa, (b) 50 MPa, and (c) 100 MPa.
In our previous work,27) we correlated the average contact resistance of the flip-chip joints on the rigid Si substrate with the average contact area of ACA conductive particles trapped between the chip bump and the rigid glass substrate. However, this correlation cannot be directly applied to the average contact resistance of the flip-chip joints processed on the soft FPCB and FPCB/PDMS substrates because of their cushioning effect. To characterize the different deformation of the ACA conductive particles on the Si, FPCB, and FPCB/PDMS substrates at identical bonding pressures, we measured the gap distance between the chip bump and the substrate pad on each substrate. In Fig. 11, the bump-pad gap on each substrate decreased with increasing the bonding pressure because of the more severe deformation of the trapped ACA conductive particles. On the rigid Si substrate, the bump-pad gap was smaller than those on the FPCB and FPCB/PDMS substrates and became zero at the bonding pressures of 200 and 300 MPa because of the direct contact of the chip bump to the substrate pad. The larger bump-pad gap measured on both FPCB and FPCB/PDMS substrates could be attributed to the cushioning effect of the compliant substrates, which made the deformation of the ACA conductive particles less severe than that on the rigid Si. The variation of the bump-pad gap vs. the bonding pressure was consistent with the change in contact resistance, as shown in Fig. 3.
Gap distance between the chip bump and the substrate pad on the Si, FPCB, and FPCB/PDMS substrates.
On the FPCB substrate, the contact resistance was remarkably reduced from 56.9 mΩ to 13.3 mΩ when the bonding pressure increased from 10 MPa to 200 MPa. However, it substantially increased to 74.8 mΩ with an excessive deviation of ±37.7 mΩ at the bonding pressure of 300 MPa, which could be attributed to the severe cushioning effect of the compliant FPCB substrate. On the stretchable FPCB/PDMS substrate, the contact resistance was reduced from 43.2 mΩ to 31.2 mΩ with increasing bonding pressure from 10 MPa to 50 MPa; then, it slightly increased to 35.0 mΩ at a bonding pressure of 100 MPa. The bonding pressure, beyond which the contact resistance was increased, approached 50 MPa on the FPCB/PDMS substrate because of the larger cushioning effect of the softer but more elastomeric PDMS at 200 MPa on the less compliant FPCB substrate. The deformation behavior of the trapped ACA conductive particles during flip-chip bonding was also differentiated by the stiffness of the substrate because a more compliant substrate exhibited more cushioning effect, which hindered their deformation. The FPCB/PDMS substrate was severely distorted at bonding pressures of 50 MPa and 100 MPa, where the chip burrowed into it because of the softness of the PDMS. Thus, a low bonding pressure is more adequate for flip-chip bonding on a more compliant substrate. The variation of the bump-pad gap distance with the changing bonding pressure on the rigid Si substrate, flexible FPCB substrate, and stretchable FPCB/PMDS substrate is consistent with the contact-resistance-change behavior of the flip-chip joints processed on each substrate.
This work was supported by the ICT R&D programs of MSIP/KEIT of Korea (Project No. B0101-16-0420, Development of Transformational and Slap-on Wearable Device and UI/UX).
