Adding nickel to the alloy retards copper dissolution almost to the point of SnPb.

The increase in copper dissolution rates using SAC alloys significantly alters the PTH rework process, not to mention the mindset for typical SnPb connector rework. Now, during SAC connector rework (using conventional equipment), it is required to track a board’s cumulative contact time to molten solder to ensure against board overexposure and excessive copper dissolution. Previous studies¹ on a smaller, less thermally massive test vehicle with an OSP surface finish determined that a single connector site on the board could be subjected only to a maximum cumulative contact time equating a one-time connector rework attempt. This is of concern for products that require up to a two-time rework within their expected lifetime. Therefore, alternate solutions are required to increase the operating process window to permit a two-time connector rework using conventional solder fountain methods. Otherwise, alternative rework process approaches may have to be considered, such as hot-gas or infrared techniques. However, these two methods also have limitations, as they would permit only PTH connector removal. Another option to reduce copper dissolution occurrence would be to use a Ni-plated surface finish, as the strong nickel layer decreases dissolution. This may not be feasible for some OEMs, and most EMS providers will still require a process to rework a PTH connector on all surface finishes.

Even larger, more thermally massive card assemblies, such as those found in server and storage-type applications, make PTH connector rework more difficult. The combination of tighter process windows and associated reliability requirements for IPC Class 3 assemblies makes Pb-free PTH rework more challenging.

Background. Past studies focusing on copper dissolution during PTH solder fountain rework using a SAC-based alloy highlighted concerns over the ability to rework PTH connectors without significantly dissolving portions of the copper barrel plating1. Results showed that a simple one-time rework was possible; however, it pushed the process window, defined by the copper dissolution rate and specified remaining copper thickness. There are no industry specifications on post-PTH rework copper thickness.

It has been determined that the PTH barrel’s knee is the most susceptible to dissolution. An established dissolution hierarchy confirmed that the knee location has the highest dissolution rate, followed (in order) by the annular ring (or pad) geometries and the inner barrel wall (Figure 1). Because the knee location has an initial plating thickness thinner than that of the pad, the conclusion is the knee location will come to complete (100%) dissolution (Figure 2) faster than the pad location (in the majority of cases). It is possible, then, for a hidden defect to be present after reworking a PTH connector, which could result in failures in the field.


There has been little research on precise reliability concerns of 100% copper dissolution at the knee. Assuming no bottom layer traces are present, and a good solder joint/fillet exists, it is theoretically possible for an electrical path to continue. The long-term reliability of the joint is unknown. Until this can be characterized, many OEMs are using minimum copper plating thickness specifications2 to ensure PTH joint strength and reliability.

Test vehicle. The test vehicle (Figure 3) selected for this study was a current OEM product card. It was selected because of its thermal nature to characterize copper dissolution on a large, thermally massive PCB. The dimensions are 8.9 x 19.4", 0.096" thick. It consists of a 24-layer, high Tg FR-4 board with 1 oz. copper plating with multiple ground connections and has OSP surface finishes. The board is populated with 32 inline DIMM connectors. All results using this TV will be compared to earlier studies1 performed on a TV less thermally massive in nature (6 x 8", 0.062", two layers of 2 oz. copper, no ground connections). The two TV scenarios will cover the PTH connector rework process representing “consumer” products up to larger “server” products.


Experimental Procedure

Measurement system analysis. An optical microscope with measurement capability was used. To determine the system and operator repeatability, a Gage R&R analysis of the measurement system capability was performed prior to taking any copper thickness measurements. Data analysis confirmed that the measurement system was acceptable, with 98.5% accuracy and a measurement error of ±0.0008".

Phase 1: Copper dissolution DoE (SAC405). To understand significant factors affecting copper dissolution and establish an adequate PTH rework process window when using a SAC405 alloy, a DoE was created and performed. The main factors included were contact time, preheat temperature and pot temperature. The levels selected ranged within typical process settings. The DoE was based largely around center or mid values, with upper and lower axial points (Table 1).


A bare board was used for the experiment in this phase, separating individual DIMM locations into the DoE samples (Figure 4). The total sample size was 20, with each sample consisting of 14 barrel locations. This would equate to 28 separate copper thickness measurements taken per sample (at the knee location).


Phase 2: Effect of alloy. Phase 2 work studied the effect of alloy on copper dissolution rates. Three Pb-free alloys were studied (SnCu0.7Ni0.05, SAC305 and SAC405), with each compared to a eutectic SnPb37 alloy as a control. Similar to Phase 1, a bare board was used and separated into location-specific samples. Each sample was exposed to two contact times (30 and 50 sec.) using each alloy. Copper thickness measurements within this phase of work were collected in a similar fashion to Phase 1. Table 2 shows the test matrix.


Solder alloy and contact time were the primary factors varied in Phase 2. All other potential variables – flux type, equipment type, preheat method, preheat time, pot temperature – were set as constants. As each alloy has a different melting temperature, the pot temperature was changed for each to keep the superheat (or difference between melting temperature and operating temperature) constant (Table 3). This is important because alloy temperature does have an impact on dissolution rates. Therefore, keeping each alloy’s superheat consistent was an attempt to remove the variability caused by the differing melting temperatures of each alloy.


Each DoE sample within Phase 1 and Phase 2 was cross-sectioned along the entire length of the sample at a middle row, and copper thickness measurements taken. To correlate findings of copper dissolution rates affected by surface geometry, Phase 1 measurements were taken at three locations (annular ring, knee and barrel wall) similar to earlier studies (Figure 5). Phase 2 measurements were focused at the knee location. Pass/fail criteria were determined on an OEM specification of 0.005" of remaining copper plating.²


Results

The results of the knee, pad and barrel wall copper thickness measurements again indicated the PTH barrel knee experienced a higher degree of copper dissolution. Therefore, all statistical analysis was based on the minimum knee measurement taken per sample.

Phase 1: Copper dissolution DoE (SAC405). DoE statistical results showed contact time to be the most significant factor affecting copper dissolution. Figure 6 shows a main effects plot indicating copper dissolution increases as contact time increases, as expected. The results, however, showed 100% dissolution occurred at the knee after only 47 sec. of contact time.


The results showed that both preheat and pot temperature have little impact or significance on affecting dissolution within the ranges of temperature selected (Figure 7). Pot temperature results are trending in the direction expected: Higher pot (or alloy) temperatures increase copper dissolution.


Based on the statistical results and accounting for the specification of 0.005" of remaining copper thickness, an optimized process window was established for the TV. The process window is:

Contact time (max):     47 sec. (cumulative)

Preheat temperature:    140-150°C

Pot temperature:        260-265°C

The results posed an immediate concern since it is known that the typical (cumulative) contact time required to one-time rework the identical DIMM connector during a SnPb process is near 45 to 50 sec. Attempts to physically rework the same connector using SAC405 alloy were similar in overall time and copper dissolution was observed. It is therefore clear, based on the DoE results, that a two-time rework would be impossible to perform without irreparably damaging the PCB. To increase the process window, further studies were required to characterize the dissolution rates of alternative Pb-free alloys. These findings are discussed in the Phase 2 results.

Phase 2: Alloy DoE effect. Figure 8 shows final copper thickness measurement results for each alloy after being exposed to 30 and 50 sec. contact times. The results show that the SnCu0.7Ni0.05 alloy not only set itself apart from the SAC305/405 alloys with respect to improving copper dissolution, but it also appears to be almost comparable to SnPb. The nickel additive has an impact on reducing dissolution. In addition, the SAC305/405 alloys’ dissolution slopes or rates are much greater than those of the SnCu0.7Ni0.05 and SnPb alloys. This indicates that longer maximum allowable contact times are possible, which may permit successful one-time – and potentially two-time – rework. The results indicate that an alternative Pb-free alloy (other than SAC305/405) may be required to be able to continue to use conventional PTH rework methods.


Copper dissolution rates and effect of flow rate. Figure 9 shows each alloy’s average and maximum copper dissolution rates. The average dissolution rate was calculated using the mean of the average of all knee measurements taken per sample. The maximum dissolution rate was calculated by using the mean of all minimum knee measurements taken per sample. Each copper thickness measurement was first subtracted from the average baseline copper thickness measurement taken at the knee location, which was measured and calculated to be 0.00174". This would provide the amount of copper dissolved. The maximum dissolution rate would be deemed the worst-case scenario of dissolution. The main point of concern would be the minimum knee thickness measurement or the area that experiences dissolution first.


Figure 9 indicates not only a difference in dissolution rates, but also a variation in the rates of copper dissolution across a given sample. The average standard deviation of copper thickness across a sample was calculated to be 0.0022". This proves that some barrels within a connector can experience complete copper dissolution before others. One reason for this occurrence has been linked to varying solder flow rates across the connector during PTH rework. It has been observed that center pins within a larger connector will experience copper dissolution at a faster rate than pins located at the outer edges. This has been correlated to the fact that center pins are exposed to a higher solder flow rate than end pins (Figure 10). The reason for this is based on typical solder fountain flow well designs, consisting of a single opening in the center through which solder is pumped. The flow of the solder, once making contact with the PCB, is then separated into two horizontal directions of flow running parallel to the PCB. As solder is vertically pumped through the center opening, the center pins will be exposed to a higher turbulent flow of solder compared to the end pins. Another reason for greater dissolution in the center region is due to the center pins’ exposure to slightly higher solder temperatures compared to the end pins. This again is correlated to the solder flow, dictated by the current machine/flow well design. Methods of reducing the solder turbulence at the center region will help reduce the copper dissolution rate.³


Copper dissolution visual indicators. During the course of experimentation, several PTH solder joint observations were used to help identify possible copper dissolution defects during rework. Visual inspection of reworked PTH solder joints can be used to monitor hidden copper dissolution defects located typically at the solder joint knee. Keep in mind that it is possible for copper dissolution to occur only on a handful of pins within a connector. Focusing inspection first on the center pins is recommended.

The sequence of dissolution follows. Figure 11 depicts illustrations of each phase.


Stage 1: Dissolution begins at the knee locations of the barrel. At this point, it is still possible for a fillet to exist because of the presence of the pin and remaining copper pad material. A small dip or indentation in the solder fillet will form when the knee is almost completely dissolved. This will typically be the first visible indication of copper dissolution issues and the potential existence of a hidden defect.

Stage 2: Dissolution continues at the knee and also progresses at the annular ring (pad). Typically, the outer edges of the pad begin to dissolve and feather out. At this point, the extent of dissolution will be fairly evident to a trained eye. The knee will be completely dissolved, and portions of the pad will be dissolved, causing an incomplete solder fillet. It is theoretically possible to have electrical continuity at this stage; however, it is unknown as to whether it will be a reliable joint.

Stage 3: At this point the knee and portions of the annular ring will be completely dissolved. This will cause a break in the solder fillet and will be visible during inspection.

Stage 4: The knee, pad and barrel’s lower portion will be completely dissolved at this stage. Typically this has occurred over 360° of the PTH barrel. Copper dissolution will be obvious at this point, and the PCB irreparable.

Conclusions

The overall conclusions obtained from this study were:

1. A smaller process window exists when using SAC for solder fountain rework due to high copper dissolution rates compared to conventional SnPb.

2. SAC405 PTH rework can barely enable a one-time rework on the thermally massive TV used. Therefore, a two-time rework attempt would be virtually impossible to complete without significant damage to the barrel and PCB.

3. SnCu0.7Ni0.05 alloy shows almost comparable dissolution rates to SnPb (Figure 12).


4. An alternate Pb-free alloy (other than SAC305/405) may be required for PTH rework to increase the process window when using conventional PTH connector rework methods. Otherwise, alternative rework methods may need to be pursued (e.g., hot gas, IR).

5. Changing to an alloy other than the commonly used SAC305/405 at the PTH rework process may also force a change in the primary attach alloy used to match the selected alloy in rework, to aid process controls. Changing the PTH alloy to an alternative formulation will offer other possible complications in contamination, mixing multiple Pb-free alloys, touchup, etc. However, some initial studies indicate mixing various Pb-free alloys will not affect joint reliability.⁴

6. It is known that when using SAC305 or SAC405 alloy, it is required to either replace or coat some stainless steel parts (e.g., pump assemblies, flow wells) within the solder pot as a result of the alloy’s highly corrosive nature. But, based on the selection of a new alternative Pb-free alloy, this “part upgrade” step could be eliminated. This in turn would help reduce initial Pb-free capital changeover costs required.

7. Until the reliability of a PTH joint suffering from extensive copper dissolution is known, a minimum copper thickness specification at the knee is essential to ensure highly reliable PTH solder joints.

8. Copper dissolution could occur only on a few barrels within a connector, indicating variation in the degree or amount of copper dissolved, which is dependent on individual barrel characteristics. Typically, barrels within the connector center are first to be affected. This correlates to an increased turbulent flow rate and alloy temperature within this region.

9. It is possible for 100% dissolution to occur solely at the knee location, which represents a hidden defect difficult to locate during visual inspection. There are visual indicators that occur when dissolution occurs. Being able to locate these indicators during visual inspection will help catch any defective boards and out-of-control processes.

Future Work

Results from this study indicate that to replicate conventional SnPb equipment, methods and techniques, a change from SAC305/405 alloy may be necessary during Pb-free PTH rework. Many alternative Pb-free alloys are available; it is necessary to study a larger number to determine the most suitable option. Further studies will examine several other alternative Pb-free alloys, and each will be compared against results of the SnPb, SAC305, SAC405 and SnCu0.7CuNi0.05 alloys studied in this paper. Various aspects of each alloy will be examined, such as composition, melting temperatures, physical properties and cost. The alloy choice will be based on the final copper dissolution rate results, as well as comparing the above alloy attributes.

Future work will also include performing reliability studies, including ATC and mechanical testing, to further understand the impact of copper dissolution on PCB reliability. This will include both pure and mixed conditions. There is a large amount of data available on solder joint reliability using SAC305/405 alloy, but further testing will be required if an alternative Pb-free alloy is chosen for PTH processes.

Acknowledgments

The authors gratefully acknowledge the participation of IBM in this study. Their involvement and collaboration in this project was invaluable and an important part of its success. A special thank you to the following persons for continued support and commitment to this study: Jim Wilcox, Jim Bielick, Phil Isaacs and Eric Kline at IBM; and Linda Scala, Thilo Sack, Irene Sterian, John McMahon, Mursalin Ahmed, Farrukh Ali, Jeffrey Kennedy, Brian Smith, Ivan Tan, Owen Clarke, Heather McCormick and Simin Bagheri at Celestica.

Ed.: This article was first published at SMTA International in September 2007 and is used here with permission.

References

1. Craig Hamilton, “A Study of Copper Dissolution during Pb-Free PTH Rework,” CMAP, May 2006.
   
   
2. IBM Engineering Specification Second Level Assembly Solder Process, specification # 61X7093 EC H86836, November 2005.
   
   
3. Craig Hamilton, Laminar Flow-well, patent pending, May 2006.
   
   
4. Karl Seelig, “Pb-Free Solder Assembly for Mixed Technology Boards,” Circuits Assembly, March 2006.

Craig Hamilton is process development engineer at Celestica (celestica.com), chamilto@celestica.com. Matthew Kelly is senior engineer at IBM (ibm.com); mattk@ca.ibm.com.  Polina Snugovsky, Ph.D., is engineering consultant at Celestica; polina@celestica.com.

• Prev
• Next