By Abhijit Gurav and Bruce Stacy
Capacitors with matte tin termination finishes were tested for tin whisker growth. The study incorporated two types of configurations: molded case SMT capacitors and multi-layer ceramic chip (MLCC) capacitors with copper metallized endmets. Both copper and nickel under-plates were evaluated on the molded case devices. All MLCC components had a nickel under-plate but two different tin deposits were tested: (1) large grained: tin grain size of 2 – 8 µm (2) small grained: tin grain size = 2 µm.
In an effort to duplicate assembled conditions, all parts were mounted using two common Pb free attachment techniques in addition to traditional eutectic SnPb soldering. To insure whisker growth, a rigorous test condition was used: the boards were thermal cycled from -55°C to +85°C for 500 cycles. This harsh artificial aging causes whiskering on MLCCs predominantly because the difference in CTEs (Coefficients of Thermal Expansion) of the materials of construction imparts significant tin deposit stress. This extreme test cannot predict tin whisker growth on MLCCs during actual product life, but qualitatively it can be used to determine which conditions are most likely to cause whiskering.
The results of this evaluation were revealing. The predictive outcome was that the MLCCs mounted with eutectic SnPb solder would exhibit little to no whiskering, and perform the best among the mounting methods used. We also predicted that the molded SMT components with nickel under-plated lead-frame terminations would exhibit the least degree of whisker formation. The actual results indicated that MLCCs mounted with Pb-free SnCuAg solder exhibited performance comparable to eutectic SnPb solder, and the molded case SMT components exhibited virtually no whisker formation irrespective of the type of under-plate or mounting method used. This paper compares whisker growth densities of the samples tested and offers insights on potential tin whisker mitigation.
An investigation was initiated to determine how capacitors performed using three different board mounting techniques (See Table 1). Investigation of whiskering propensity of lead free board mounting techniques was accomplished by utilizing two configurations of capacitors: molded case SMT capacitors with copper alloy terminations and multi-layer ceramic chip (MLCC) capacitors with copper metallized endmets (See Diagrams 1 and 2). Results were benchmarked against the whiskering propensity of components utilizing SnPb soldering for assembly.
The molded case SMT and MLCC components were mounted on boards with Au/Ni pads using ECA, SnAgCu (96.5Sn/3.0Ag/0.5Cu), and eutectic Sn-Pb pastes with no-clean flux. The ECA boards were cured at 150ºC for 2 hours, the SnAgCu mounted components were reflowed at peak temperatures of 260ºC, and 245ºC and the SnPb mounted components were reflowed at a peak temperature of 245ºC and 215ºC.
To accelerate whisker growth, boards with components mounted by the three different methods were subjected to thermal cycling between -55ºC to 85ºC for 500 cycles. Total cycle time was 20 minutes with 10 minutes residence time at each of the temperature extremes.
The method of determining whisker propensity was based on NEMI2 criteria. The measurables were:
- Whisker density (whisker count over a fixed area)
- Median whisker length
- Maximum whisker length and number (if over 10 µm)
The procedures3 used for whisker inspection were:
- Sputter gold on all mounted samples
- Inspect 5 components per mounting method used on all metallized faces
- At each inspection record:
- Tin nodules, hillocks, odd eruptions, etc (whiskers < 5 µm in length or > 2 µm in aspect ratio4 )
- Estimate, photograph and record length of longest whisker at a magnification of 3000X
- Record whisker density over as an area of 250µm x 250µm (see Figure 1)
- Inspect each sample for tin whiskers by SEM (Scanning Electron Microscopy) at magnification of 300X, and at higher magnifications as needed
Table 2 summarizes various findings of the whisker evaluation of MLCCs with two different tin grain sizes and molded case with lead-frame capacitors evaluated in this study.
Results and Discussion
MLCC Results Lead-free (96.5Sn-3.0Ag-0.5Cu) solder Large Grained: 2 – 8 µm Tin Grain Size
Multi-Layer Ceramic Capacitors (MLCC) were mounted using the SnAgCu lead-free solder and reflowed using two different temperature profiles one with a peak temperature of 260ºC and the other with a peak temperature of 245°C. Irrespective of the peak reflow temperature utilized, these surfaces showed growth of only a few whiskers of < 5 µm. An extensive examination was conducted under the SEM to look for occurrence of whiskers. Only 1 to 3 whiskers were found over an area of 450 µm X 350 µm (one frame of SEM micrograph at 300X magnification). This is equivalent to having only about 1 whisker over a square area of 250 µm X 250 µm. Figures 2 -4 show SEM micrographs of this sample. The longest whisker found overall was ~4 µm long.
The following factors may play a key role in providing such robust performance of MLCCs against the tendency to whisker growth during thermal cycling while using Pb-free for mounting:
- The peak temperatures employed during Pb-free solder reflow is 260°C and 245°C. Pure tin melts at 232°C. The higher temperature employed during reflow of Pb-free solder may relieve any residual stresses in the tin deposit and facilitate extensive recrystallization of tin.
- Employment of plating conditions resulting in low residual plating-stress, pure matte finish tin, and Sn grain size of ~2 to 8 µm.
Small Grained: = 2 µm Tin Grain Size
MLCC components with pure tin finish exhibiting a finer Sn grain structure of =2 µm were mounted using SnAgCu paste but reflowed only at the single peak temperature of 260°C. These were tested simultaneously through 500 thermal cycles in the same environmental chamber as the other samples. Whisker size was equivalent to the deposits with the larger grain size, but the whisker density was twice as much. Figures 5 – 7 are SEM micrographs showing the microstructure and whiskers found these chips.
Standard Sn-Pb Solder (63Sn/37Pb Eutectic) Large Grained: 2 – 8 µm Tin Grain Size
MLCC components mounted using the standard Sn-Pb eutectic solder and reflowed at peak temperatures of 245°C and 215°C ± 5°C were then examined after 500 thermal cycles. SEM examination indicated that the samples reflowed at the lower temperature had a higher whisker density than the SnPb or Pb-free chips mounted using the peak temperatures of 260°C and 245°C. The length of whiskers or tin nodules found was considerably smaller than 25 µm. As shown in figures 8 -12 the majority of the whiskers were ~10 µm in length or shorter. The average length of whiskers was ~8.5 µm, and the longest whisker found had a length of ~19 µm (Photo 13). Figures 16 and 17 are identical images, but the 16 was taken utilizing secondary electron mode and 17 back-scatter mode of the SEM. These two images indicate that the tips of whiskers were Pb-rich (note the lighter color on Photo 17). Whiskers grew out of the tin surface and through the layer of Sn-Pb solder, developing a crown of the Pb-rich phase. The lower portion of the whisker was predominantly tin, which primarily consists of tin atoms diffusing into the whisker from the tin surface underneath. This indicates that because the peak reflow temperature (215ºC) used was less than the liquidus of tin (231ºC) the tin deposit didn’t entirely reflow and/or alloy with the Pb. These remaining areas of non-stress relieved, pure electrodeposited tin grew whiskers.
This situation may be mitigated by utilizing a reflow profile that has a higher peak temperature of 245°C. SnPb soldered chips reflowed at the higher peak temperature demonstrated an equivalence in whiskering propensity to the Pb-free soldered capacitors reflowed at 245°C and 260°C (See Figs. 14 – 16).
Small Grained: = 2 µm Tin Grain Size
MLCC components with pure tin finish exhibiting a finer Sn grain structure of =2 µm mounted using eutectic SnPb solder at a peak reflow temperature of 215°C were tested after 500 thermal cycles. Figures 18-21 are SEM micrographs showing the microstructure and whiskers found on these samples. As seen in Fig. 22, the longer whiskers (>10µm) grew out of either small grains of tin or near the boundary between a recrystallized grain and an area of finer grained deposit. In contrast, the larger recrystallized areas did not grow long whiskers, but only a few low aspect ratio tin nodule structures.
ECA (silver adhesive)
MLCC samples were mounted using ECA (silver adhesive), cured at 150°C for 2 hours, and then tested through 500 thermal cycles from -55°C to +85°C. There were ~50 to 60 whiskers over an area of ~450 µm X 350 µm (one frame of SEM micrograph at 300X) which is equivalent of having ~20 to 25 whiskers over a square of 250 µm (e.g., Figure 18). The average length of whiskers after 500 thermal cycles was about 10-12 µm, and maximum length was ~17 to 18 µm (Figs. 23-26). This mounting technique had the highest whiskering propensity as the tin deposit wasn’t relieved by reflow or alloying solder.
Molded Case SMT Capacitors
Molded case SMT capacitors electroplate with pure, matte tin were investigated for tendency towards whisker formation during thermal cycling from -55°C to +85°C. Both nickel and copper under-plates were assessed to determine if whiskering propensity was dependent on under-plate type. Aluminum-organic capacitors, again with lead-frames possessing pure matte tin finish were also included in this study. No whiskers were observed during this thermal cycle testing over the temperature range -55°C to +85°C.on Tantalum and Aluminum capacitors employing any of the three common mounting applications- Lead-free solders, Tin-Lead solders as well as ECA. Figures 27-29 are SEM micrographs of surfaces of pure-tin plated lead-frames of molded capacitors after 500 thermal cycles.
It is understood that the mechanism for whisker growth on pure tin plated components is caused by deposit stress. 12 Stress in MLCCs can be introduced by subjecting them to thermal cycling which causes deposit stress via the mismatch of CTEs (Coefficients of Thermal Expansion) of their materials of construction. Deposit stresses in molded case SMT devices are introduced by promoting inter-metallic growth which is probably accelerated by long term thermal aging, not by thermal cycling.
- Molded case SMT capacitors utilizing pure, matte tin finish exhibited low whiskering propensity when thermal cycled from -55°C to +85°C up to 500 thermal cycles. The tin plated surfaces of terminations in these products remained practically whisker-free irrespective of the mounting method used (Pb-free solder, Sn-Pb eutectic solder or ECA) or an under-plate of nickel or copper was used. Whiskering of tin plated lead frame appears more dependent on stresses induced by intermetallic formation than mismatched CTEs (Coefficients of Thermal Expansion) therefore this test method may not be a fair assessment of the whiskering propensity of these components.
- MLCCs using Pb-free (Sn-Ag-Cu) and SnPb solder reflowed at peak temperatures above the liquidus of tin (=232ºC) performed well during thermal cycling. After 500 thermal cycles whisker density was found to be 1 over an area of 250µm X 250µm, average length of whiskers was ~2 µm, and the longest whisker was only ~4 µm long.
- MLCCs mounted with Eutectic SnPb at 245ºC exhibited less whiskering propensity than those reflowed below the melting point of the tin deposit.
- MLCC’s with average Sn deposit grain size of = 2 um appear to have a higher whisker density and mean whisker length than those with a Sn deposit grain size of 2 – 8 um regardless of the mounting method used when soldered at peak temperatures less than the liquidus of the tin deposit.
- MLCCs mounted with Eutectic SnPb at 215ºC exhibited whiskering on areas of tin not affected by reflow or dissolution by solder.
- Among the three different MLCC attachment methods tested for whisker propensity during thermal cycling, mounting with ECA showed the maximum occurrence of whiskering.
1 Joint Press Release with JEITA and NEMI; Tokyo, June 9, 2003 recommends this method of growing whiskers.
2 Joint Press Release with JEITA and NEMI; Tokyo, June 9, 2003 recommends this method of growing whiskers.
3 These inspection procedures are NEMI based with the length and area being an internal KEMET procedure.
4 KEMET internal criteria.
5 Osterman, M; “Mitigation Strategies for Tin Whiskers,” CALCE-EPSC Publication; August 8, 2002 – NOTE: There is conflicting documentation regarding the ability of fused tin to grow whiskers. Whisker growth may be retarded because the tin is fully alloyed with Ag and Cu from the solder. More data is needed to confirm this.
6 Zhang, et al.: “An Alternative Surface Finish for Tin Lead Solders; Unique Tin Plating Chemistry”: Plating and Surface Finishing, June 1998.
7 Dittes, et al.: “Tin Whisker Formation- Results, Test Methods, and Countermeasures”; IEEE, Electronic Components and Technology Conference Proceedings pp 822 – 826; 2003.
8 Brusse, et al.; “Tin Whiskers: Attributes and Mitigation;” Passive Components Symposium Proceedings, pg 221 -233; Carts Europe, 2002
9 Brusse, J: “Tin Whisker Observations on Pure Tin Plated Ceramic Capacitors,” Proceedings of SurFin 2002, AESF.
10 Zhang, et al: “Whisker Tests, Results, and Prevention,” Proceedings of AESF SurFin ’00, June 2000.
11 Osterman, M; “Mitigation Strategies for Tin Whiskers,” CALCE-EPSC Publication; Release 1.0, August 8, 2002.
12 Ando, et al: Murata: “Stress Analysis and Accelerated Evaluation of Tin Whisker under Thermal Stress,” www.psma.com/HTML/FILES/forums/leadfree/mate_paper.pdf