Aerospace Rework in a Lead-Free World
The global transition to lead-free (LF) products is having a significant impact on the electronics industry in many areas, including aerospace. Even though the aerospace industry is specifically excluded from the mandates of the European Union's RoHS (restriction on hazardous substances) directive, the industry will still feel deleterious effects from the transition.
This is because many aerospace products are designed and produced by manufacturers that supply to commercial customers worldwide. For economic reasons, these suppliers may deliver LF products to all of their customers, regardless of RoHS exclusions. Therefore, aerospace prime contractors will be forced to use products that contain LF solder alloys in military or commercial aircraft, whether required to do so by environmental legislation or not. This brings up several important concerns of which the aerospace industry needs to be aware, including the issue of rework and repair of conventional tin-lead (SnPb) components using LF solder.
Aerospace electronics-both commercial and military-differ from consumer electronics in that rework and repair in the field are critical to maintaining operational aircraft. For the military in particular, this can mean performing rework at a distant base or aboard a ship, situations in which space and the materials and equipment available for repairs are often limited. As a result, with LF becoming increasingly widespread, the rework and repair issues concerning long-term reliability of electronics assemblies must take into account the very strong possibility of mixing LF and SnPb solder alloys at some point in the life span of the assembly.
Solder joint performance in the wake of these field repairs is especially critical, because the operating conditions for aerospace assemblies are harsher from an environmental and mechanical standpoint than many of those in the commercial or consumer sectors. For instance, the shear stress on a solder joint is very high during takeoff of military aircraft, especially for naval aircraft taking off from a carrier deck.
A number of studies have already investigated the influence of lead on LF solders. However, most of these studies have focused on the incorporation of lead into LF solders as a result of having both types of solders on the same assembly. This approach results in a small amount of lead in the LF solder, but does not investigate the conditions present during a repair operation in which a significant amount of SnPb solder remains on an existing board or component.
The objectives of a recent study conducted at the University of Missouri-Rolla were to study the soldering processes and resultant metallurgies of various mixed LF and SnPb solder alloys under rework and repair conditions. The metallurgical microstructure, composition and strength of mixed LF and SnPb solder joints were also studied in order to begin correlation to real-world aerospace conditions.
A single SnPb solder and a single LF solder were used in the investigation. The SnPb solder was 0.032-inch diameter 63Sn37Pb solder wire with an integrated no-clean flux. The LF solder was a commercially available Sn-Ag-Cu (SAC) no-flux wire with a composition of 3 percent silver and 0.5 percent copper, with the balance made up of tin.
Mixtures of the two solders were done on the board using hand soldering. Preheating of the soldering iron prior to processing was 200 C for the SnPb alloy and 235 C for the SAC alloy. A rosin flux was used with the SAC alloy and excess solder was removed with a solder sucker. Residual SnPb solder was left on the board to simulate a repair process. Typically, the amount of SnPb solder in the final mixture was between 5 percent and 10 percent.
There were two different types of test boards used in the study. The first was a custom-made FR4-based board with 2-by-2-millimeter copper pads 1.5 millimeters apart that were suitable for hand soldering 1206 surface mount capacitors.
The other type of board used was an RF module test kit sample with through-hole devices and eutectic SnPb solder joints. Solder was removed from the through-holes using a solder sucker and a soldering iron heated to 210 C.
A Thermotron ATS-320-V-10-705- CO2 system was used to temperature cycle boards between -55 and 125 C, with a transition time of approximately 2 seconds and a 10-minute soak at either extreme. Samples were pulled after 100, 300, 625 and 805 temperature cycles.
Shear testing of the surface mount capacitors was done using a Romulus IV universal testing system outfitted with a shear test module. The programmed load rate during shear testing was 8 pounds of force per second, but the actual time to shear the component to failure was approximately 5 seconds. This equates to a net load rate of approximately 2 pounds per second.
Chemical and microstructural characterization of the solder joints was done using optical and scanning electron microscopy (SEM) of polished cross sections. The microscopes used were a Hitachi S-570 LaB6 filament and a Hitachi S-4700 field emission SEM, each equipped with energy dispersive spectroscopy (EDS) units for chemical analysis. A Hirox digital video-microscope imaging system was used to record the optical images.
Results from the shear testing of the as-prepared (zero temperature cycles) and temperature-cycled surface mount 1206 capacitors with the different solders are presented in the table titled "Shear Strength Test Results." It is worth noting that because these assemblies were soldered by hand to simulate repair, the amount of solder on each component was not the same, although care was taken to try to achieve as consistent a solder volume as possible.
The as-prepared (zero temperature cycles) samples required 12 to 15 pounds of shear force to remove them from the test boards, with the standard SnPb joints being the strongest. The standard deviation of the test data was approximately 2 to 4 pounds. The limited data set prohibits meaningful statistical analysis, but generalized data trends can be discerned.
The SnPb shear strengths, for example, were typically the highest value of the three solders. If the samples after 100 temperature cycles are excluded, shear strength decreased with an increase in the number of cycles. The reason for the low values after 100 cycles is under investigation.
The SAC solder results are less consistent with temperature cycling, but were usually the lowest value of the three. The data suggests that a mixture of the SnPb and LF solder is preferred over an LF-only solder joint. From a repair perspective, this implies that complete removal of the existing SnPb solder may not be desirable from a shear-strength perspective, because the presence of SnPb in the joint appears to improve the shear strength of the solder.
An optical image of a through-hole after removing most of the original SnPb solder indicates that a thin layer of solder was still present. Therefore, the interface between the copper and the solder after repair with SAC would most likely contain intermetallic compounds along the interface from the original SnPb solder joint, leading to higher shear strength compared to SAC solder joints. Obviously there are many other considerations and more extensive testing is required, but preliminary data indicates a benefit from residual SnPb in the joint.
Characterization of the microstructure and chemical composition of the solder joints was done to identify similarities and differences in the various solder compositions, and evaluate the effectiveness of the repair operations. In order to be as systematic as possible in analyzing a process that relies on human operations, the solder joints were examined after each step of the process. This was done for the SnPb and the SAC alloys as well as the mixture of the two solders. Another variation included in the study was the repair of the original SnPb solder joint with the same composition of SnPb solder. This repair provided a means of characterizing the hand-soldering process and served as a reference microstructure.
As would be expected, the original eutectic SnPb joints and the joints "repaired" with SnPb solder displayed microstructures that were very similar in appearance. In the images, the light-colored areas are lead-rich regions. After temperature cycling, the eutectic lamellae were no longer observed. The chemical composition of the solder, as measured by EDS, was much more homogeneous after temperature cycling.
A more pronounced lamellar microstructure was evident in the SAC alloy prior to temperature cycling than was observed for the SnPb solder. Due to difficulties in preparing samples, a microstructure for the SAC alloy after temperature cycling could not be obtained, but efforts to get a usable sample continue. Inspection of the interface between the SAC alloy and the copper on the board indicated that the copper-tin intermetallic compound was discontinuous.
A number of different samples of a mixed SnPb-SAC solder alloy were analyzed, because this combination reflects the most likely scenario to be encountered in an actual aerospace repair operation-residual SnPb solder on a board or component that is repaired or replaced with an LF solder. As part of the process, the investigation looked at two distinct areas: the center of the joint, which should contain mostly SAC solder, and the area near the interface where the residual SnPb solder would interact with the SAC.
Using this approach, the center of the repaired joints before temperature cycling did not appear to have the lamellar microstructure that either of the unmixed alloys exhibited, and the microstructure after temperature cycling was very similar to the SAC solder; the distribution and size of the secondary phases in solder did not appear to be significantly different than the SAC-only solder.
In other words, characterization of the center of the solder joints did not reveal any information that could be used to adequately explain the higher shear strength measured with a mixed SnPb-SAC solder alloy composition compared to the SAC alloy.
Investigations of the interface of the SnPb and SAC solder with the copper on the board, however, did show a difference. In areas repaired using an SAC alloy after removal of excess SnPb solder, there was a fairly continuous copper-tin intermetallic compound reaction along the interface. Although there are areas that may have a thicker intermetallic compound layer or the intermetallic compound may not be as dense (which was evident by the contrast of the phases), there did not appear to be a break along the interface.
We believe that the residual SnPb alloy on the surface of the copper was a barrier to direct reaction of the SAC repair alloy with the copper substrate. As a result, the shear strength of the joints of a SnPb surface that were hand repaired with SAC was higher than the SAC-only joint.
Additional work is needed to verify the validity and reproducibility of this result. However, the preliminary data indicates that incomplete removal of the SnPb solder during repair with a LF solder may actually be beneficial to the reliability of the joint.
This article is based on a paper presented at the Electronic Circuits World Conference, part of APEX/IPC Printed Circuits Expo 2005. Contributing to the study were Patricia Amick and David Kleine of The Boeing Co. (Chicago), and Steve Vetter and Dale Murry of Northrop Grumman Interconnect Technologies (Springfield, MO). The study represents a single project within a much larger collaboration between the University of Missouri-Rolla (UMR) and Boeing-Phantom Works for the Center for Aerospace Manufacturing Technologies (CAMT) housed at UMR. The work was funded through the Materials and Manufacturing Directorate of the Air Force Research Laboratory at Wright-Patterson Air Force Base (Dayton, OH), through contract no. FA8650-04-C-5704, Dr. Jaimie S. Tiley, program manger.