Over the past decade, the Willamette Valley has been a fertile region for the rapid development of microelectronics industries. With the nickname of Silicon Forest, the area has seen the development of several home-grown companies and their off-shoots. These companies, along with several transplants from Silicon Valley, Europe, and Asia, have brought in significant investment and employment to the area. Now Silicon Forest has nudged Oregon's traditional forest out of the role as the largest employment provider.

When Silicon, Gallium Arsenide, and other materials are converted into submicron semiconductor devices through various exotic processes, a whole slew of materials-related problems are encountered. Interestingly, while many of the processes are brand new, we often find that the root causes of the problems are ones we have encountered before.

We have found that our combination of expertise in fields such as corrosion, metallurgy, chemistry, forensic science, physical testing, and microexamination can provide answers to many interesting problems in the microelectronics industry. Some recent projects highlight this...
 
 

Test instruments can often pinpoint the location of an electrical discontinuity in a printed circuit board, but they cannot always identify the cause for such discontinuity. We have found in many cases that an ideal evaluation technique consists of cross sectioning the board so it can be microscopically examined.

Properly preparing specimens for this type of examination without altering their structure requires years of experience and a high level of skill. Basically, it consists of sectioning the specimen using a thin diamond-coated saw a few thousandths of an inch away from the desired inspection location and then carefully sanding and polishing into the affected area.

To protect the delicate structures during the cutting and sanding operations, the components are sometimes encapsulated in a pourable epoxy, which is allowed to harden before cutting the sample.

Next, the sample is mounted in either a thermosetting or cold-curing resin to further protect it during several stages of progressively finer sanding and polishing. Once the sample has been properly prepared, it is examined with either an optical microscope, or if higher magnification is required, a scanning electron microscope.

Our past projects have included the examination of multi-layered circuit boards and through-hole plating quality, and more recently, ball grid arrays (BGA), a self-contained soldering process. Through our examination we can establish the nature and extent of defects such as nonuniformity of plating thickness, holidays, solder wetting problems, solder ball shift, misalignment, and lack of fusion.

Our evaluation can lead to changes in process parameters, component placement, circuitry dimensions, etc. and is a valuable input to manufacturing and packaging engineers.
 

In this project, a client reported that he was having a problem with a particular batch of zinc-plated steel standoff spacers that were purchased from an outside supplier: Within a few hours of being installed on circuit board assemblies, typically about three percent of the spacers would end up fracturing without warning! Clearly, something was seriously wrong, but what?

Our analysis began with a detailed fractographic and metallographic examination, then expanded to include chemical analysis, hardness tests, and some post processing bake-out tests. We found that the failures were the result of hydrogen embrittlement, a time-dependant, brittle cracking phenomenon which occurs under sustained load in high strength steels.

Now, one might ordinarily associate high strength steels with bridges or highrise buildings, rather than electronic circuit boards, but in fact, high strength steels are used for a large number of common products, including circuit board spacers. In this case, the spacers were used as small spring clips, and while the total load on the clips may have been small, the load per unit area (stress) was high, requiring the use of high-strength steel.So why were the spacers failing? We traced the problem to a combination of factors: The carbon content and hardness of the steel were both higher than the material specification allowed; this increases the susceptibility of high-strength steel to hydrogen embrittlement. Additionally, the forming process had left a sharp, tight radius in the spacers, along with small forming cracks; both of these significantly increase the stress concentration, thereby increasing the susceptibility to cracking.

The source of the hydrogen was traced to the zinc plating process. Typically, components such as these will be subjected to a post-forming bake-out treatment, which is intended to remove the hydrogen that is inevitably picked up in the plating process. Through laboratory bake-out testing, we were able to show that in this case, the manufacturer's bake-out treatment had been too short to effectively remove the hydrogen.

With the problem traced to the manufacturer of the spacer, rather than the client's assembly or usage of the spacers, our client was able to go back to the supplier for remedial action.

Fire retardants are most often chlorinated or brominated compounds that decompose to produce free radical quenchers at the high temperatures in a fire. The free radical quenchers inhibit flame formation.

In a previous newsletter we discussed some of the corrosion problems fire retardants can cause to circuit boards. There, we found that certain bromide salts had formed and subsequently combined with components in the soldering flux to corrode gold plating and solder joints.

We encountered another interesting case involving a similar reaction during an investigation into corrosion of aluminum clean room fixtures which had been mounted using a flexible sealant. A microscopic examination revealed an interesting attack morphology: The aluminum was corroding by localized pitting, with the metal grains spalling off, leaving exposed grain facets at the bottoms of the pits (photo at left).

Our investigation showed that the corrosion was taking place only at locations where the aluminum was both in contact with the sealant and adjacent to galvanized iron strips. This was somewhat surprising because the sealant was water repellant, so there should have been no liquid electrolyte, and we would not have expected corrosion to take place. Yet the aluminum was undergoing rapid intergranular pitting corrosion in these areas.

The corrosion was producing an ugly discoloration of the sealant (photo at right); although unsightly, this wasn't the main concern. Of much greater concern was the possibility that the corrosion could result in airborne corrosion products contaminating the clean room environment.

Our attention next turned to the sealant, which was an organic rubber material. Because it was flammable, it had been doped with chlorinated and brominated fire retardant compounds; this turned out to be the key to the puzzle.

Our analysis revealed that the galvanized zinc coating was combining with the fire retardant in the sealant to produce zinc chloride and zinc bromide. These zinc salts were hygroscopic, so they combined with water vapor in the air to form a salty electrolyte solution, which in turn traveled through the sealant and attacked the grain boundaries of the adjacent aluminum (photo below).

Interestingly, our study suggested that the fire retardant might not have reacted with the zinc, except in this case, a bismuth catalyst had been used in the sealant, and this catalyst had apparently promoted the corrosive attack.

An obvious solution to the problem would have been to remove all the sealant and replace it with new sealant free of fire retardant. However, the zinc chloride and bromide had diffused into the corroded grain boundaries in the aluminum, so even if the bulk of it were removed, small amounts would still remain, and it was feared that this could lead to continued corrosion. Also, replacing the sealant in the completed and installed assembly would be an expensive proposition.

With these concerns, our client turned to us with two more questions: 1) How long could they continue in operation without replacing the sealant at all, and 2) if they were to replace the bulk of the sealant, would the remaining residual sealant continue to suck in moisture and attack the aluminum?
 

We addressed these questions by reviewing the available scientific literature in detail and evaluating the corrosion reactions. From this, we were able to predict the likely changes in the corrosion rate with time. Our calculations showed that the corrosion rate would slow down very quickly with time (in fact, as a cube root function!). Consequently, the clean room integrity would not be compromised during its expected lifetime.

In this project, one of our clients was experiencing severe localized corrosion of the solder pads on several printed circuit boards. Our assignment was to establish what was causing the problem. Unlike some problems, which turn out to have a single root cause, this was a rather complicated interaction of two distinct corrosion mechanisms: stray current corrosion and crevice corrosion.

The stray current corrosion was being driven by battery terminals that were in close proximity to one another. This in itself would not normally be a problem, but in this instance we found that ionic impurities on the solder mask had been sufficient to increase the surface conductivity to the pointwhere the normal voltage differentials were enough to induce rapid stray current corrosion.

The crevice corrosion was a result of poor adhesion and cracking of the via caps, which produced a crevice in the board. Although these would be considered rejectable defects in their own right, the cracks and poor adhesion might not have even been recognized during performance testing if they hadn't been highlighted by the corrosion.

Interestingly, the mere presence of the cracks and poor adhesion normally wouldn't be expected to result in corrosion in this application. However, in this case, a combination of excessive flux residues and cap decomposition products had produced a medium in which the crevice surfaces corroded.

Ironically, the corrosion failures were ultimately considered to be fortunate because they revealed several cleaning and processing problems that could have eventually led to much more serious problems and field failures, had they not been revealed by the unexpected corrosion.