Don’t maintain the status quo if potential to reduce manufacturing failures exists.

An inquiry led to an investigation on the possible causes of printed wiring board failures, which were becoming increasingly prevalent after SMT manufacturing. Failures were detected by electrical testing, but the location and specific devices causing the failures were undetermined. The failures were suspected to be predominantly in the BGAs located on specific sites on this 16-layer construction. Failure data provided included high resistance shorts occurring in those specified areas. The surface finish was a eutectic HASL, and the solder paste was a water-soluble SnPb.

The diagnostic approach agreed upon included an examination of both the quality of the manufacturing process and the materials used for assembly, as reviewed below.

SMT process. The first order of diagnostics, a manufacturing audit to assess the SMT process, revealed:
Solder paste was properly stored and permitted to reach ambient conditions prior to use. The solder mesh was appropriate for the type of assembly, and the paste was not expired.

Stencils used for paste application were properly proportioned with an aperture size appropriate to the BGA device pitch.

The stencil printing operation revealed no flaws, and paste was applied in a smooth, consistent manner with uniform height and width.

The paste reflowed uniformly and covered leads per IPC-A-610D class 3 specifications. No apparent skips or dewetting were noticed on resistors, capacitors or leaded chip devices.

The reflow oven has four zones and short length, but no issues around reflow quality were seen as a result. The recommendation was made to increase to a seven-zone oven to increase profiling flexibility for various designs.

Reflow profile. The reflow profile was in the specified range for the solder paste of choice (Alpha WS-809). Thermocouple placement was in appropriate locations, and the temperature uniformity among each location reflected an even distribution of heat. The time above liquidus (TAL) varied by only 6 sec. between the highest and lowest time, while the peak temperature varied by 13˚ among locations. There was no evidence of hot or cold spots.

The assembly TAL averaged 90 sec., well within the vendor-recommended TAL range of 40-120 sec. As reflow was clearly achieved, and there was no evidence of cold solder joints, a recommended step was to decrease the TAL from 90 to 60 sec., which would improve wetting without a risk of incomplete solder reflow. However, the present reflow profile is consistent with the recommended parameters prescribed by the solder paste vendor.

Bare board inspection. A visual assessment of the bare boards showed evidence of solder mask overlap into the pad areas, exposed copper under the HASL surface finish, and non-uniform HASL finish in certain cases.
XRF inspection. BGA devices and bare board were analyzed using x-ray fluorescence to determine the alloy composition of the metal (Tables 1 and 2). This is a common method used when the composition of the component alloy or surface finish is in question.

XRF summary. Components indicate a eutectic or near eutectic SnPb composition. Analysis of the HASL finish indicated a non-uniform thickness, and at certain locations, the finish was thin enough to permit the underlying copper to overwhelm the alloy analysis.

X-ray analysis. X-ray analysis was performed on a populated board at the various BGA locations suspected as problematic. No evidence of shorting opens or misalignment was prevalent.

Optical endoscope. The populated assembly was analyzed through an endoscope to see any evidence of incomplete collapse or other observable phenomena such as dewetting or head-in-pillow effect. No unusual occurrences such as excessive solder or flux residue were visible. However, the graininess of the solder balls may indicate the start of an oxidizing surface. This can be due to excessive time in liquidus state. The solder balls seem to be well collapsed and formed.

Wetting balance. The wetting balance test showed significant issues with the wettability of the HASL board at various locations (Figure 1). J-STD-003 criteria suggest time to buoyancy corrected zero, T0 (where thewetting force goes positive), should be less than 1-2 sec. In every sample tested, T0 was greater than 5 sec.

Conclusions and Recommendations

The HASL surface finish is non-uniform, with copper exposure and very thin layers of HASL in some areas.
The wettability is well below the recommended wetting criteria, as specified by IPC specifications, and can be a primary concern during assembly. Residual oxides on the HASL surface may prevent proper wetting, or form weak intermetallic interfaces when reflowed. The type and amount of oxide residue can be ascertained through Sequential Electrochemical Reduction Analysis (SERA).

The x-ray images showed no evidence of poor alignments, opens or shorts.

Optical endoscope images showed no anomalies and good BGA ball collapse.

ENIG should be considered as a board surface finish instead of HASL, if available. It is the most common finish for BGAs, and would eliminate the non-uniform surface finish.

Application of a more uniform HASL surface would ensure complete coverage of the underlying pad and alleviate exposure of the underlying copper, while reducing the risk of oxidation.

The HASL bath should be analyzed for contamination from excessive metals and other residuals from the soldering process.

Reducing the liquidus time by 30 sec. may help alleviate excessive formation of oxides on the solder.
A seven-zone oven would permit more flexibility in shaping the reflow profile around a varied array of designs and constructions.

Lessons Learned

Results of the analysis indicated that, although certain process and manufacturing improvements can be made to reduce both systematic and randomized failures, examination of the raw materials (for example, the PCB) can prevent a significant amount of manufacturing defects.

In this particular case (which is not atypical), a thorough investigation provided the engineer enough data to support the claim that the cause of electrical failure was due to noncompliancy in the surface finish that caused dewets. Establishing documented proof of failure can exonerate the manufacturing process, and prevent excessive and unnecessary expenses.

There is a natural inclination to avoid impugning an established process that has, for the most part, produced a successful product. However, the temptation to maintain the status quo should be avoided if the potential to reduce manufacturing failures exists. Most issues arise when an interaction between non-centered processes result in a multiplicative effect. In this case, the effect of poor distribution of the HASL surface, along with a marginalized reflow process, combined to produce failures that may not have occurred if either of the conditions for HASL or reflow was optimized.

The ACI Technologies Inc. (aciusa.org) is a scientific research corporation dedicated to the advancement of electronics manufacturing processes and materials for the Department of Defense and industry. This column appears monthly.

• Prev
• Next