NEMI Recommends Standard Test Methods to Assess Propensity for Tin Whisker Growth

Estimated reading time: 8 minutes

Many IC suppliers are evaluating and implementing lead-free finishes as the industry moves toward lead (Pb) elimination in electronic assemblies.

By Nhat Vo, Irina Boguslavsky and Peter Bush

While Sn-based finishes provide good corrosion protection and a solderable surface, they also have the potential for whisker growth. Now that Sn and its alloys with low content of the alloying element are being considered to replace SnPb, standard test methods will help the industry move more quickly in evaluation and development of Sn-based Pb-free finishes. Standard tests also will permit meaningful comparison of whisker propensity for different plating systems and processes, provide a consistent inspection protocol for tin whisker examination, and provide a standard method to compare and report results.

The National Electronics Manufacturing Initiative (NEMI) formed the Tin Whisker Accelerated Test Project in 2001 to identify an accelerated test method for whiskering. More than 40 companies have participated in this project. A benchmark study collected all existing methods for growing whiskers. The team also identified and discussed theories behind whisker formation to compare test methods to whisker growth mechanism and fundamentals. A separate project — the Tin Whisker Modeling Project — was formed to evaluate different whisker growth theories.

null

The Tin Whisker Accelerated Test Project has evaluated four principal test methods reported to successfully grow whiskers on some, but not all, Sn-plated samples. These test methods are storage at ambient office conditions; storage between 50º and 85ºC; storage at high relative humidity (85 to 95 percent); and air-to-air temperature cycling (-55ºC/85ºC).

NEMI recommends three test methods — two storage conditions and one temperature cycling condition — to evaluate the propensity of Sn-based plating finishes to grow whiskers. The project team has submitted a test method document to JEDEC for release.

In Phase 1 of testing, samples — brass coupons and eight-lead small-outline integrated circuit packages (SOIC) — were prepared with bright Sn along with SnPb alloy as a control. Bright Sn was chosen because the literature indicates that it is more prone to whisker growth. The samples then were subjected to assorted combinations of the four environments identified. Whiskers formed only on the bright Sn-plated coupons, and there were far fewer than expected. Some odd-shaped eruptions formed on the eight-lead SOICs, but no confirmed whiskers. Because the samples were plated in a laboratory environment, the level of impurities and contamination is thought to be low, thus helping to retard whisker growth. A second hypothesis is that when the terminations of the eight-lead SOICs were formed, the plating finish cracked, reducing stress in the finish and helping to retard whisker growth.

(a) Column

null

(b) Hillock or nodule

null

(c) Flower or Odd-shaped Eruption (OSE)

null

(d) Needles

null

(e) Needle growing from nodule

null

(f) Needle growing from OSE

null

(g) Striations on whisker and consistent cross-section

null

(e) Rare branched whisker

null

(f) Kinked whiskerFigure 1. SEM photos illustrating different whisker types and characteristics.

null

The results of the Phase 1 study were inconclusive; therefore, the team performed an additional study of the test methods with packages (eight-lead SOICs) plated at an assembly house using production baths.

Two IC suppliers volunteered to plate eight-lead SOIC samples for the Phase 2 study. Supplier A provided samples plated with either a methane sulfonic acid (MSA) bath or a sulfate-based electrolyte. Supplier B provided samples plated with a second MSA bath. Thick (10 to 12 µm) and thin (2 to 3 µm) matte Sn samples, as well as SnPb samples, were included in the evaluation. The samples tested are listed below, and the different environment combinations are presented in Table 1.

• A = 2 to 3 µm, matte Sn (sulfate) on OLIN194 Cu SOIC molded/singulated
• B = 10 to 12 µm, matte Sn (Sulfate) on OLIN194 Cu SOIC molded/singulated
• C = 2 to 3 µm, bright Sn on brass coupon
• D = 10 to 12 µm, 90Sn/10Pb on OLIN194 Cu SOIC molded/singulated (control)
• E = 2 to 3 µm, matte Sn (MSA) on OLIN194 Cu SOIC molded/singulated
• F = 10 to 12 µm, matte Sn (MSA) on OLIN194 Cu SOIC molded/singulated

Air-to-air temperature cycling equipment capable of cycling from -55º (0, -10ºC) to 85ºC (10ºC, -0) was used for the thermal cycling procedure. A temperature humidity chamber capable of 60ºC/95±5 percent RH and 30ºC/90±5 percent RH environments was used for temperature/humidity experiments. Ambient storage was defined as ~23ºC and ~30 to 60 percent RH conditions (air-conditioned office environment). A scanning electron microscope (SEM) was used for whisker inspection.

For inspection purposes, the following definition of a whisker was developed: "A spontaneous columnar or cylindrical filament, which rarely branches, of mono-crystalline tin emanating from the surface of a plating finish." Tin whiskers also may have an aspect ratio (length/width) >2; can be kinked, bent and twisted; a consistent cross-sectional shape; and may have striations or rings around it.

Inspection Procedure

Leaded Packages. Three packages randomly chosen from the test sample were mounted in upright, inverted and inverted rotated positions. Carbon tape or paint created conductive paths, or carbon was evaporated on the parts if inspection was made immediately. At 300X magnification, three randomly located fields from a) underside of lead, b) top of lead, and c) side of lead (Figure 2) were inspected. The stage was tilted to 45º for the latter two images. The fields were selected as representative of the overall condition of the parts as determined by the first inspection.

Coupons. On each of three coupons selected from the test samples, at least three images were collected at 300X and data was recorded as above (Figure 3). Similarly, images at 3,000X were collected for grain size determination.

Figure 2. Schematic of package lead.

null

All whiskers in the three fields were counted and recorded. The longest whisker in each field was measured and recorded, using higher magnification if necessary. Finally, one image at 3,000X from an undisturbed region of plating was collected. The estimated grain size range of the deposit was reported. The average number of whiskers from the three fields and the length of the longest whisker found in those fields were reported. The date of plating, whisker test conditions (duration, temperature, humidity, number of cycles, etc.) and the date of inspection were included.

null

Each plating bath was maintained at the optimum conditions specific for each process. Level of metallic contamination was measured by atomic absorption analysis prior to plating (t0) and after the plating was complete (tf) (Table 2).

Phase 2 Test Results

The results of the eight-lead SOIC whisker test for three plating processes (two MSA-based and one sulfate-based plating chemistries) and for two deposit thicknesses (2.5 and 10 µm) subjected to all test conditions were analyzed. The thin samples from Supplier A were found to exhibit significantly higher whiskering under all tested conditions than the other two samples. However, thick samples plated with both MSA-based processes show comparable whisker performance. What is surprising is that thick samples from Supplier B have more whiskers than thin ones plated with the same process. For thick samples, test conditions that included thermal cycling followed by both ambient and 30ºC/90 percent RH storage produced significantly higher whiskering than either ambient or 60ºC/95 percent RH conditions. Deposits produced with a sulfate-based process, both thin and thick, did not produce significant whiskers. SnPb (control) samples showed no whisker growth in all test conditions.

Conclusion

The environmental stress conditions evaluated in the Phase 2 study were sufficient to create whiskers on eight-lead SOICs. More whiskers grew with the -55ºC/85ºC temperature cycle method. Ambient and the 60ºC/95 percent RH and 30ºC/90 percent RH storage methods also grew whiskers, but were not as effective as the temperature cycle methods.

The addition of temperature and humidity exposure did not add significantly to the whisker length or frequency when temperature cycle was performed first. There also were no differences observed in whisker performance within the deposit thickness range tested.

Figure 3. Schematic of test coupon.

null

The results suggest that the bath chemistry/plating process has the most significant influence on whisker growth. There appears to be substantial difference between the two MSA-based processes from the suppliers, and the sulfate-based chemistry tested generally seems to have better whisker performance than one of the MSA-based baths, but only slightly better performance over a good-practice MSA bath.

Whiskers formed on the Sn-plated samples under all environmental conditions tested in the Phase 2 experiments, indicating that the proposed test methods are adequate for evaluation of whisker propensity of Sn deposits.

Techniques for tin whiskers described here have been submitted to JEDEC for consideration as industry standards. Three test conditions are recommended for future evaluations to continue to build a whisker database: temperature cycling (-55º to 85ºC, approximately three cycles/hour) and temperature humidity tests at 60ºC/93 percent RH and ambient storage (air-conditioned facility). Additionally, NEMI is collaborating with the Japan Electronics & Information Technology Industries Association (JEITA) and a European semiconductor consortium (E3) to form unified whisker test methods.

A Phase 3 experiment to verify the results of the Phase 1 and 2 evaluations should start in the fourth quarter of 2003. One goal is to show that the three recommended tests can produce whiskers with similar samples consistently. Another is to confirm that it is applicable for other Sn-based (SnBi and SnCu) finishes and possibly other sample types. The tests will be carried out to extended durations to determine appropriate endpoints for each test method. The endpoints will be defined by the growth saturation.

Originally published in the Proceedings of the SMTA International Conference, Chicago, Ill., September 21-25, 2003.

References

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

Nhat (Nick) Vo, senior process engineer, may be contacted at Motorola Final Manufacturing Technology Center, Austin, TX 78729; (512) 996-7927; E-mail: n.vo@motorola.com. Irina Boguslavsky, Ph.D., CEO of EFECT LLC, may be contacted at (631) 912-9630; E-mail: boguslavsky@cwo.net. Peter Bush, director, may be contacted at South Campus Instrumentation Center at SUNY Buffalo, Buffalo, NY 14214; (716) 829-3561; E-mail: pjbush@acsu.buffalo.edu.