Flip Chip Assembly on PCB

Estimated reading time: 11 minutes

This article focuses on using different assembly options such as dip fluxing, flux jetting and reflow encapsulate for 200- to 250-µm pitch lead-free (SnAgCu) flip chips on FR-4 substrates. The impact of PCB surface finishes (OSP and ENIG) was investigated from an assembly perspective.

By Todd Castello, Dave Geiger, Dongkai Shangguan and Jonas Sjöeberg

lip-chip assembly is a key capability to enable product miniaturization on handheld products and modules, improve electrical performance, and in some cases, lower total product cost because the package itself is excluded. Previous studies have investigated several flip-chip interconnection types, including anisotropic conductive film or paste, Au/Au thermosonic bonding and eutectic SnPb flip chips on FR-4 substrates. This study focuses on the assembly of 200- to 250-µm pitch lead-free SnAgCu flip chips on FR-4 substrates.

Test Vehicle

PCB design. The test vehicle is designed as a module measuring 28.57 x 22.22 mm (Figure 1). The module is panellized in a 127- x 177.8-mm panel with 20 modules on each panel (Figure 2). The PCB has six layers with a total thickness of 0.5 mm. Board material is laser-drillable FR-4 (Tg = 140°C; CTE X,Y = 12 to 16 ppm/°C). Surface finishes are organic solder preservative (OSP) and electroless nickel with immersion gold (ENIG), with 3.81 to 5.08 µm for Ni thickness and 0.0762 to 0.203 µm for Au thickness over the nickel. Components are on a single side, allowing a single SMT process step; however, the module was designed such that it can accommodate solder spheres on the backside.

Figure 1. Module layout.

For the flip chip, two types of land patterns were used - one for the perimeter-array type of flip chips (PB8 and PST-01) and the other for the full area-array flip chip (FA10). For the area array package (250-µm pitch), a specific land pattern was designed. This land pattern will work only for this daisy-chain sample; if a real part is to be used, then the copper would need to be changed to an individual pad and would need to use a microvia connected to the pad to route. For the perimeter array package, a trench-type design was used. This land pattern was used for the PB-8 (200-µm pitch) and the PST-01 (250-µm pitch). By using this type of design, PCB cost is kept fairly low because the line and spacing is moderate when using this design. The main challenge for PCB manufacturing is solder-mask registration.

Flip Chip Dies. The components selected for this study can be seen in Table 1. The flip-chip components are daisy-chained dies, allowing continuity checks to be conducted throughout the process. Die passivation is nitride on all flip-chip devices. For the PB-8 flip chip, UBM is 102-µm in diameter and bump is 120-µm in diameter and 95-µm high. The FA-10 and the PST-01 flip chips have a UBM diameter of 102 µm, bump diameter of 135 µm with a height of 120 µm. All solder bumps in this study are 95.5Sn/3.8Ag/0.7Cu with a melting temperature of 217°C. As reference, assemblies also were made with 63Sn/37Pb solder bumps on the same substrate type.

Flux and Underfill Material Selection

Flux Selection. Over the years, different fluxes have been tested during the development of an SnPb flip-chip process. These fluxes were used for the SnAgCu assemblies in this study, and have been found to leave minimal residues and to be compatible with most underfills used in previous studies. Two types of fluxes and one reflow encapsulate were used for this study. One is a traditional tacky gel flux (Flux A) applied by dipping the flip-chip bumps into a controlled thickness of flux on the pick-and-place equipment with a drum fluxer. The second flux is a liquid flux (Flux B) that is applied by jetting the flux onto the substrate in a separate machine. Both fluxes used are no-clean formulations. The third “flux” is a reflow encapsulate or “no-flow underfill” (Flux D, also named Underfill D in the underfill section) designed for lead-free soldering. In the reflow encapsulate, flux is mixed with underfill epoxy to avoid two separate processes and the need for nitrogen in reflow.

Figure 2. Panellized modules.

Underfill Selection. Underfill A was selected from previous studies to compare lead-free combinations with SnPb. The combination of Flux A with Underfill A was demonstrated to be compatible with the eutectic SnPb solder from a previous experiment. Underfill A is an acid anhydrate-based material that requires special handling and permission to be used in some countries. In many countries, the material requires special ventilation, and extra care should be taken during transportation, adding cost when using this underfill material. This prompted the investigation of two underfill materials that do not use acid-anhydrate chemistry. Underfills B and C are non-acid anhydrate materials. Underfill C is a potential reworkable underfill with a lower Tg. The reflow encapsulate (Underfill D) is an acid anhydrate-based material, and a commercially available material designed for lead-free flip chips.

Figure 3. Bridging on PB8 with SnPb solder bumps on ENIG, with an old non-acid anhydrate underfill material.

In previous studies with SnPb flip chips, one non-acid anhydrate underfill was used, but this material allowed the solder to bridge after about 1,000 air-to-air thermal cycles between -55° and 125°C (Figure 3). The cause of this was due to a void in the underfill that formed because of the trench design in which the underfill did not flow into the trench area. Solder was extruded during temperature cycling and proceeded to the other solder joint through cracks in the underfill. The first failure for any flip-chip device due to this bridging phenomenon was seen at 1,000 cycles.

Assembly Information

Process Flow. Three different process flows were used; traditional dip fluxing, spray/jet fluxing and reflow encapsulate. Cleaning of flux residues after reflow was not performed on any of the processes. Even though PCBs were vacuum-packed from the supplier, they were baked at 125°C for four hours prior to assembly to ensure that no moisture was trapped in the PCBs.

Fluxing. The thickness used for the dip-fluxing process was 40 µm (about ¼ to ½ bump height on the die). Previous studies show no significant difference in the wetting behavior for different flux thicknesses, as long as reflow is done in nitrogen. However, a thinner film (approximately 25 µm) would give less flux residues, but also would limit the process window with regard to getting an even flux film on the flux drum, and in some cases, small lines of missing flux can be seen on the flux drum (Figure 4). Therefore, a flux film of 40 µm was used for this study.

Figure 4. Lines of missing flux on the flux drum at 20-µm thick flux film.

The flux-jetting process was optimized to give the least amount of flux possible, minimizing flux residues, but still retaining enough flux to ensure wetting and tackiness throughout the complete assembly process. The reflow encapsulate process used an underfill dispenser to apply material to the PCB. The machine was set up in the same way as for underfill dispensing of other underfill materials.

Placement. Standard SMT equipment with specified accuracy (better than ±40 µm at three sigma) was used during the flip-chip assembly. The actual accuracy was measured with a modified version of the IPC-9850. The machine also requires a vision system capable of detecting the bumps and space between the bumps. Most cameras used by machine suppliers require 4 to 5 pixels in both X and Y direction within the feature, bump or space, to be able to detect it in a stable way.

After pick-and-place, all flip-chip devices were inspected with X-ray for alignment to the pads, and no misalignment larger than 25% off-pad was seen. Other than a slightly dull appearance of SnAgCu solder bumps, there is no difference between SnPb and SnAgCu solder bumps during placement. No assembly issues were seen with dip-fluxing and flux-jetting processes. The reflow encapsulate process requires more fine-tuning with regard to selecting the correct placement nozzle, placement pressure, dwell time at placement and others. During the setup, issues with getting reflow encapsulate material on top of flip chips were seen (Figure 5).

Figure 5. Reflow encapsulate on top of the FA10 flip chip.

Reflow. A typical lead-free reflow profile was used with a peak temperature of 243°C and 50 s above 217°C. Both nitrogen and air atmospheres were tested. Due to poor yields from reflow in air, it was decided to stop testing with reflow in air. These results correspond well to results with SnPb solder bumps; and no differences were seen with different fluxes when nitrogen was used.

Figure 6. Fillets on the dispense side and the opposite side showing self-filleting behavior.

Underfill and Underfill Cure. Before underfill dispensing for flip chips, work was done to check the capability of the underfill machine, ensuring consistent underfill results. Cpk measurement on the underfill valve itself was conducted by dispensing Underfill A with 3- and 10-mg dispenses. The 10-mg, ±10% dispense showed a good Cpk value of 2.12, but the Cpk for the 3-mg, ±20% dispense is 1.0. A smaller needle and slower flow speed might give better results, but using traditional needle dispensing at 3 mg and below requires a well-optimized underfill machine and carefully selected underfill material. For flip-chip dies requiring less than 3 mg of underfill, alternative dispensing methods, such as underfill jetting, could be a better solution to ensure a stable process. Underfills A, B and C were dispensed with the same pattern. A single “I-pass” was used on PB8 and FA10; and a single weight-controlled dot was used on PST-01. Underfill D, the reflow encapsulate, was dispensed using weight-controlled dots on all locations. All underfills showed nice fillets on all sides of the flip chips (Figure 6), with Underfills A and D showing a slightly better self-filleting behavior compared to Underfills B and C. Underfills A, B and C were cured in a batch oven with the specified cure schedules from the underfill suppliers. Underfill D was cured in the reflow process after assembly, according to supplier recommendations.

Results and Discussion Figures 7a and 7b. Reflow in nitrogen with 200 ppm O2 (a.) and reflow encapsulate in air. (b.) Poor wetting can be observed with the reflow encapsulate with no wetting out on the pads.

For each panel run, boards were inspected using X-ray prior to reflow, after reflow and after underfill. Continuity tests were done after reflow and after underfill. Poor results on the assemblies reflowed in air, and with the reflow encapsulate Underfill D, were detected as poor wetting and no, or minor, collapse of bumps. The same results were seen with Fluxes A and B and on ENIG and OSP, for reflow in air. Reflow in nitrogen showed full collapse and good wetting with Fluxes A and B on ENIG and OSP (Figures 7 and 8).

Self-alignment properties were studied in a controlled way by misplacing some parts intentionally by 25, 50 and 75% off-pad. Studies indicate that the self-alignment properties with SnAgCu are the same as, or similar to, SnPb in a flip-chip application. The different surface finishes (ENIG and OSP) showed no difference, with both showing self-centering with misplacement up to 50% off-pad with Fluxes A, B or C. However, due to the lack of wetting with the reflow encapsulate (Underfill D), no self-centering was observed on these assemblies. The reflow encapsulate process also showed a strange mixing phenomena. Once the reflow encapsulate came into contact with the applied solder paste on the module, the reflow encapsulate and solder paste started to mix and the metal particles in the solder paste started to flow around in the reflow encapsulate. For mixed assemblies, this could be considered a risk (Figure 9).

Figure 8. A good SnAgCu solder joint on ENIG that wetted fully around the pad. No underfill voids can be seen.

After the assembly was completed, CSAM analysis results showed no voids with any of the underfills (A, B, C, D). Shear test showed no major difference between ENIG and OSP with the SnAgCu solder bumps, and all failures occrred on the flip-chip side.

Figure 9. Solder paste mixed with reflow encapsulate after reflow.

All capillary underfills showed some filler settling. This was most visible with Underfill C. The effect on reliability is yet to be seen, but theoretically we would expect to see first failures on the flip-chip side of the solder joint.

Conclusions

Lead-free solder flip-chip assembly on FR-4 was demonstrated as feasible on both ENIG and OSP surface finishes with reflow in nitrogen. The processes, such as fluxing, pick-and-place, reflow and underfilling were evaluated to determine which combination would provide better yields. It is clear that nitrogen is needed in a flip-chip application to ensure high yields. Self-alignment properties with SnAgCu are the same as, or similar to, SnPb in a flip-chip application. Different surface finishes (ENIG and OSP) showed no difference, with both showing self-centring with misplacement up to 50% off-pad. Dip fluxing and flux jetting are two feasible assembly methods for SnAgCu flip chips. This study shows that the reflow encapsulate evaluated did not work well, as the epoxy gelled before solder wetting was achieved, and work is currently on-going with other reflow encapsulate materials.

REFERENCES

For a complete list of references, please contact the authors.

Todd Castello, staff process development engineer, advanced technology department, Flextronics, may be contacted at (919) 570-1719; e-mail: Todd.Castello@flextronics.com. David Geiger, senior process development engineer, advanced technology department, Flextronics, may be contacted at (408) 576-7000; e-mail: David.Geiger@flextronics.com. Dongkai Shangguan, director, advanced process technology, Flextronics, may be contacted at (408) 428-1336; e-mail: Dongkai.Shangguan@flextronics.com. Jonas Sjöberg, senior specialist, advanced technology department, Flextronics, may be contacted at +46 13 287228; e-mail: Jonas.Sjoberg@flextronics.com.