Why commercial off-the-shelf components are incompatible with high-rel processes.

Military manufacturers are employing “off-the-shelf” components or parts in systems designed for military use. Manufacture of these systems and assemblies involves processes not typically used for commercial products such as multiple sequential solder assembly operations, multiple cleaning operations, conformal coating, etc. Commercial parts use is essential, as often equivalent military components are not available, or use thereof would increase cost and cycle time and compromise system performance. However, experience has shown that commercial components are often incompatible with processes used to build military hardware. This frequently results from the requirement that PCBs must be cleaned, typically multiple times, during processing.

Board wash complications. Military electronics assemblies typically are subjected to one or more board cleaning procedures during production. These are commonly carried out using automated inline cleaning systems. This is much less common in commercial enterprises. In fact, there has been a trend in commercial products toward no-clean and low-residue fluxes. However, board cleaning is essential if a PCB is to be conformally coated, and remains common to military production even when coating is not required. COTS parts are built for commercial customers, and typically not designed with exposure to board cleaners in mind.

Raytheon has experienced several failure scenarios resulting from exposure of components to our standard cleaning procedures. While a variety of board cleaner compositions are available, they all have three common features: water, a “saponifier,” and surfactants. Saponification refers to the conversion of water-insoluble fatty acids to soluble salts by exposure to a base (typically sodium hydroxide) to make soap. In electronics, the term is used to describe the conversion of insoluble flux acids to soluble salts by the action of bases. This permits their removal using inline cleaners. The basic nature of these materials can cause several issues, including corrosive failures.

Surfactants lower the surface tension of water, permitting better component wetting. This permits cleaners to enter small openings – gaps under parts; small, unexpected crevices in non-hermetically sealed devices; between individual capacitors of a stacked capacitor, and many others. This becomes problematic because, in most cases, DI water, free of surfactant, is used to remove the cleaners. Because of its high surface tension, the water cannot wash away the cleaner from the areas mentioned. Result: You can get the cleaner in, but you can’t get it out.

While most failures can be resolved by changing cleaning procedures, the result is that the one-size-fits-all approach to manufacturing – all products built using common shop practices – cannot be used in these cases. This often results in downtime, from the initial failure-and-resolution phase to the unexpected increase in man-hours required for production, as special procedures must be implemented for specific cases.

The following are examples in which Raytheon experienced failures as a direct result of component exposure to board wash chemicals. As is common in these cases, the initial supplier response typically was, “We sell these components to several customers, and you are the only one with reported failures.” This inevitably is related to the fact that “other customers” do not expose the parts to the aqueous cleaning common to military production.
Shorting capacitors. Included in the recent trend toward “green” manufacturing is the chemical industry’s effort toward “green” chemistry. One “green” solution to board cleaner chemistry is the use of environmentally benign bicarbonate salts, such as potassium bicarbonate, as saponifiers. This creates unique failure modes, as these are highly ionizable salts and, as such, are strong electrolytes – materials that conduct electricity in the presence of moisture. When these materials are washed into small crevices of boards and not removed by water rinsing, shorting failures at electrically sensitive areas, as well as corrosive failures, are common.

In this example, a stacked capacitor on a COTS product with very tight clearances between the individual capacitors was failing as a result of exposure to a bicarbonate salt cleaner. Failures were immediately noted upon testing of new assemblies. Electrical testing revealed shorting between stacked capacitors to be the cause; a distance of about 0.030˝ separated individual capacitors. In some cases, fiberglass spacers were placed between capacitors, in others not. Step-by-step analysis of the production flow revealed failures occurred only after the assembly’s exposure to board cleaning. Disassembly of a unit that did not have spacers revealed the residue in Figure 1. Masking the component prior to washing proved sufficient to avoid contamination. This did result in more “touch” time for product assembly, however.

BGA device: the problem with “weep” holes. The previous example focused on the presence of strong electrolytes in board cleaners leading to shorting conditions. More commonly, board cleaners do not contain such salts. For example, ethanolamine (an organic base) is perhaps the most common saponifier found in these products. While use of such cleaners can alleviate failures directly attributable to salt residues deposited by board cleaning, any aqueous solution, and even pure water, also can cause catastrophic failures if trapped in devices.

This example concerns a particular BGA device design that led to cleaner entrapment. Severe corrosion was evident in many of these devices as a result. Additionally, as the chemicals in the cleaner are hygroscopic (attract moisture from the environment), they represented a long-term reliability concern.

Figures 2 and 3 show the device construction. A central chip, topped with thermally conducting grease, is surrounded by capacitors and covered with an epoxy bound lid (take particular note of the “weep hole” – the top of the lid in Figure 3 – built into the design). Discussions with the manufacturer revealed the lid was placed over the device for mechanical support, while the weep holes (others not shown) were incorporated to vent gasses assumed created from soldering heat. It should also be noted the device in Figures 2 and 3 suffered from cleaner entrapment, which accounts for the overall appearance and presence of deposits. In this case, the cleaner is composed of 75% water and 25% cleaner concentrate.

As indicated in the images, cleaner entrapment led to deposit formation, which ultimately led to device failure. When the lids were removed from devices, in most cases, droplets of cleaner were obvious if the PCBs were less than about a week old. This created, in effect, a condition in which many of the exposed components were akin to being submerged in the cleaner.

In some cases, boards were not immediately powered up and as a result, components were permitted to dry significantly prior to use. In other cases, boards were powered up soon after production; the applied voltage in this environment led to some interesting phenomena. For example, selective electromigration of the metals in the solder (SnPb) was evident between many capacitors. As well as causing corrosion, this led to leakage pathways.
Figure 4 depicts two distinct phenomena related to board cleaner exposure: corrosion of copper under the solder mask leading to blistering, and electromigration of metals between the capacitor leads. In Figure 4, the dark side of the capacitor is lead rich, lead corrosion products forming the black color, while the brighter side is tin rich. (Scanning electron microscopy/energy dispersive spectroscopy analyses were employed.) Copper is also observed over much of the capacitor. This phenomenon was more pronounced in boards that were electrically tested soon after exposure to board wash.

While electrically testing components submerged in cleaning solution creates obvious failure scenarios, long-term reliability is also a concern in this case. Analysis of the failure mechanisms revealed the following sequence of events occurs with these devices:

1. Water wash is entrapped after entry through the weep holes upon exposure to board cleaning. This is not removed by DI rinse for reasons mentioned.
2. Native no-clean flux mixes with the wash solution, forming gelatinous residues concentrated around the capacitors. (The capacitors were soldered using a rosin-based no-clean flux by the supplier.)
3. The more volatile components of the cleaner evaporate over time (the base saponifier, an organic alcohol, and some of the water).
4. Nonvolatile components of the cleaner remain indefinitely.

The initial phases, when the more-corrosive components of the cleaner are present, can lead to extreme corrosion (Figure 5), even without powering on the devices. Corrective actions, including instituting post-clean bakes, removing the lids, etc., have proven to mitigate these issues; however, long-term consequences of this exposure are still worrisome.

Figure 5 is illustrative of a COTS issue faced by military customers that use board-cleaning operations. Discussions with the supplier revealed that, to their knowledge, we were the first customer to experience such failures. This resulted from the fact that none of the other customers, which are overwhelmingly commercial, exposed these components to board washing. Interestingly, a military build was available from this supplier that would have avoided this issue; however, it was not yet available in the BGA models employed for the projects in question.

These examples illustrate typical problems associated with insufficient removal of board cleaner components from assemblies post-processing. While such issues are certainly not new, they are increasingly becoming a point of concern in military applications. This results from the trend of increasing use of COTS parts designed for no-clean commercial assembly procedures. 

W. John Wolfgong, Ph.D., is a chemist at Raytheon Network Centric Systems (raytheon.com); wolfgong@raytheon.com.