Abstract

Figure 1: Creep corrosion after accelerated MFG

A failure mechanism known as creep corrosion has been gaining the attention of the electronics industry over the past year. Industry committees have formed to decide which paths are best taken to identify, recreate and eliminate this corrosion process. Though some key elements such as the presence of sulfur and high humidity have been identified as triggers for producing this defect type, the true mechanism and physical nature by which this mass transport controlled process occurs is not fully understood.

Specifically, this paper explores the influencing factors that may contribute to creep corrosion being produced, from the initial board fabrication stages, through the various assembly processes, and ultimately on to the end use of the PWB. The quality of the board fabrication process, overall board design, soldermask registration, and choice of final finish, i.e., OSP, ImAg, HASL, ENIG and ImSn have all been found to be significant factors. Also explored are areas of the assembly process which may induce or accelerate creep corrosion, these include; flux contaminates, thermal exposures and the choice of lead containing and lead free solder pastes.

Finally, the aggressive environments where creep corrosion is frequently observed are simulated in an effort to better understand which factors have the greatest influence on creating creep corrosion. This paper explores the effects of environments where increased levels of sulfur bearing gasses, such as hydrogen sulfide and sulfur dioxide, are present and also where other contaminants such as elemental sulfur dust may be found, on inducing electrical failure in electronic products. The resulting corrosion, its physical nature and the mechanisms whereby its growth becomes accelerated are discussed. Environmental factors such as humidity levels are also explored.

Introduction

Figure 2: Creep corrosion from field failures

In September of 2007, an IPC committee was formed who’s mission was to better understand creep corrosion from a mechanistic point of view, agree on test methodology and to determine the best routes forward to resolve the creep corrosion issue [1]. The initial goal of the committee was to find a controlled test method which would recreate the creep corrosion defect in a controlled laboratory environment. A significant portion of the work that was explored by the committee was Abbott’s investigations of Mixed Flowing Gas and Battelle test environments [2]. This was done to gain a better understanding of the effects of varying Mixed Flowing Gas environments on the corrosion of both copper and silver metals. In parallel to this the committee has also investigated the changes in air quality that has occurred in a large number of developing industrial countries, for example, it has been found that sulfur contamination levels have significantly increased in countries such as China and India [3], leading to an increased level of electronic equipment failures induced by the creep corrosion phenomenon.

To date chemical supplier houses and end users implement a wide array test methods in an attempt to recreate creep corrosion. There is no single industry standard methodology agreed upon. There a definite need for an industry wide test, as it has become clear that slight alterations in the test environment chosen can dramatically affect its ultimate outcome. Previous work has shown that in highly aggressive environments, all final finishes will display creep corrosion and will do so without significant differentiation between the final finishes after testing. A standardized test method is needed for two main reasons; Firstly, such a test will help to accurately reproduce the type and degree of board defect that is observed in a field environment. Once established, this test method can then be employed to investigate all aspects of board fabrication and assembly and can be used to determine their influence on inducing creep corrosion. Secondly, such a test is also needed to better understand how much of a performance improvement is required of the existing, conventional final finishes for the end product to withstand field environments for an acceptable amount of time once deployed in the field. Simply put, without a standardized industry wide test for creep corrosion, it will remain impossible to meaningfully compare and discuss individual attempts and routes to prevent this type of corrosion phenomenon. A group effort is needed, involving both the end users that have encountered equipment failures and chemical supply houses to achieve this goal.

Figure 3: Tire Factory simulation with water

Historically, the creeping defect has been observed in high sulfur bearing environments. Intuitively, research and development teams looked to classic mixed flowing gas regimes in an attempt to mimic the degree of corrosion experienced in field failures. It was quickly found that mixed flowing gas levels had to be pushed well beyond the standard Battelle class 4 levels [4] in order to induce creep corrosion. Hydrogen sulfide concentrations were dramatically increased, as were the temperature and humidity levels used in the test with limits set just belowthe point where the mixed flowing gas equipment would beadversely affected. It took the increase of all three of these factors to create migration of the underlying copper metal,thereby inducing creep corrosion. Even in elevated sulfur environments at elevated temperatures well out of the specifications of standard, and what was considered very aggressive mixed flowing gas (MFG) environments, the degree of corrosion did not mirror that experienced in the field.

After limited success with accelerated mixed flowing gas approaches, research tests were then designed to simulate aggressive industrial environments (paper mills, tire factory and automotive clay modeling design rooms, for example) in which creep corrosion had been observed. In these types of tests levels of sulfur, temperature and humidity could be adjusted to almost any intensity without the fear of destroying sensitive equipment. Through trial and error based on some of the knowledge gained from mixed flowing gas testing and some efforts to mimic harsh industrial environments, it was determined that the level of humidity in the test chamber was a crucial factor in creating the desired defect. Going even further, actively causing water condensation to occur on the surface of a test board was shown to be an on/off switch for the mass transport controlled corrosion process to occur. Figures 3 and 4 detail the differences typically observed in the degree of creep corrosion induced when high levels of humidity are and are not present in an industrial environment simulation test. Figure 3 displays significant creep corrosion on all pad edges. Figure 4 does not. This paper examines the mechanism whereby the migration of copper can occur across a planar surface, resulting in corrosion and ultimately equipment failures being observed in field environments.

Figure 4: Tire Factory simulation without water

Initially, for the purposes of printed circuit board manufacture, the final finish applied; ImAg, OSP, lead freeHASL, ImSn, etc., was used solely as a solderability preservative. It was designed as a layer to protect the underlying copper from oxidation which will render it unsolderable prior to the assembly process. As technology has advanced demands have been placed on the final finish that are significantly beyond its original mandate as a simple a solderability preservative, such as corrosion resistance.

Many functioning circuit boards contain areas that remain unsoldered. Some of these metal areas are used for additional testing (frequently, for example, as electrical contact test points). End users require the ability to electrically test a finished board, after assembly, after the board has been built into the final product and even after the board has functioned in the end product for extended periods of time, to ensure their products integrity.

Figure 5a and b: Defects in soldermask plugging resulting increep.

Currently, demands are being placed on final finishes to protect the underlying copper substrate during fabrication, throughout the assembly process stages and ultimately,through to the products end of life. This is a challenge that should not be underestimated. As detailed above, any final finish employed must not only withstand high levels of sulfur, but also high levels of humidity at increased temperatures, this was not a demand that was placed on circuit board surface finishes in the past.

In our experience, to minimize creep corrosion not only must a final finish be applied in an optimum manner on to a surface free of contamination but also, soldermask must be well applied and the assembly process adopted must allow for full printing of pasted areas in order to ensure that the quality of the final finish is not compromised. In this paper we will discuss the effects these processing steps can have on the formation of creep corrosion.

Creep corrosion is understood to be induced by a reaction of copper and sulfur [5]. Electron Dispersive Spectroscopy has confirmed this repeatedly for all surface final finishes and also for lead frame finishes [6]. The unique aspect of creep corrosion, from a mechanistic point of view, is the mass transport of the material across laminate and soldermask and although the corrosion film is essentially electrically resistive, over time as the film continues to migrate the board surface and increase in thickness, the electrical resistivity of the film reduces to the point where electrical failure may occur.

An extensive study was carried out at the Sandia National Laboratories to investigate the effects of humidity on copper sulfidation [7]. The Sandia National report explains that the primary corrosion product that forms when copper metal is sulfidized in an atmosphere containing H2S is Cu2S. They also found that the presence of water and atmospheric oxygen are very significant factors as, initially, any exposed copper reacts as detailed below to produce cuprous oxide;

4Cu + O2(g) → 2Cu2O (s)

Cuprous oxide was found to be more reactive than its metallic counterpart and in the presence of sulfur or hydrogen sulfide in the atmosphere the following type of reaction occurs:

2Cu2O (s) + 2H2S → 2Cu2S + 2H2O(s)

In a humid environment, the hydrogen sulfide may dissolve in the water and then dissociate into a number of ionic species, HS-, S2- and H+. Oxygen will also dissolve in the absorbed condensed water. This provides oxidation and transport for the copper sulfidation reaction to occur. The rate of Cu2S formation was found to be dependent on a few key factors:

• the supply of hydrogen sulfide to the surface• the rate at which Cu is able to diffuse
• the rate of chemical reaction both on the surface and through the water

A standard thickness for copper on most circuit boards is at least 1.5 mils thick. The copper thickness in combination with high humidity and sulfur content in aggressive industrial environments will enable the reaction to cycle  continuously. Therefore, high levels of sulfur, a substantial source of copper metal and water provide an idealenvironment for copper sulfidation and metal migration.

Board Fabrication

Figure 6: Trace running under soldermask

As discussed in previous works board design has a large influence on the initiation of creep corrosion [4]. Today’s technology requires finer line and spaces which dictate areas of soldermask definition and areas that are too tight to allow soldermask to act as an effective barrier. For example, 16-mil pitch quad flat packs do not have soldermask located between each pad. Other areas of a board may have pads that are soldermask defined on one side and are metal defined on the other. These pads may or may not be in close proximity to other board features. It has been found that pads that are soldermask defined appear to have more of a tendency to exhibit creep corrosion due to the relatively smooth, planar topography that is present which offers an ideal surface for corrosion products to migrate across [4]. Some end users have chosen to open up wells around these types of features to contain creep within these wells. It also appears that the high surface roughness of laminate is much harder for creep corrosion products to travel across readily.

Other common board designs have soldermask plugged vias. It has been determined that the technique used to plug vias and the ultimate quality of that technique can greatly influence creep corrosion. Poor via plugging will result in exposed copper areas that have been found to corrode rapidly.

Figure 7: Negative soldermask foot

Figures 5a and b show soldermask plugged vias. The copper like appearance around the vias illustrates the fact that the soldermask in these areas is very thin. After processing though a reflow oven a crack can occur, as detailed above in figure 5a, at the knee of the hole and at the edge of the copper pad, as demonstrated in figure 5b. These sections of exposed, unprotected copper became a location of accelerated creep corrosion. The resulting creep corrosion is particularly excessive in these areas as any contamination present is free to attack these isolated areas exclusively.

Another aspect of soldermask and its effect on creep corrosion frequently discussed is the foot created where the soldermask ends on a trace. It is generally considered that this is the major trigger for creep corrosion because a majority of the initial incidents of creep occurred at the interface of the metal area and the soldermask. A defect known as soldermask interface attack (SMIA) can occur when there is a large negative foot between the soldermask and copper substrate (see figure 7). It can be readily appreciated that plating and other processing solutions may have difficulty getting into or out of these areas. Clearly then the quality of the soldermask, its application and also, how effectively the final finish is applied has great influence over the degree to which this defect can occur.

Figure 8: Top-down of attacked trace

It has been demonstrated that the soldermask exposure and development stages have the greatest effect on the quality and geometry of this area. Many board fabricators push the limits of their exposure and development processes to create a softer soldermask. This is considered desirable for electroless nickel/immersion gold and immersion tin. These two final finishes are very aggressive towards soldermask.The process baths’ high operating temperatures and aggressive chemical components can result in lifting of the soldermask at its edges. To combat this fabricators frequently under expose the mask which then results in decreased cross-linking of the polymer from mask surface to copper substrate. This is simply due to minimal penetration of the UV light to lower levels of the soldermask. The board then has to be over developed to try to remove any under exposed materials from the board’s surface. Inevitably, this results in a crevice between the soldermask and the traces or pads. In stages after soldermask application, this area is not easily reached by processing solutions but if these solutions do penetrate these areas it is unlikely to be readily removed through rinsing. After completion of the board fabrication and assembly this crevice can become a natural reservoir for contamination and moisture to become trapped. This area is usually the first to exhibit creep corrosion.

Enlarge this picture Figure 9: Zygo of attacked trace

One final finish that can exhibit soldermask interface attack is immersion silver. As detailed, the presence of the negative foot becomes an area where immersion silver chemistry can become trapped. The area under the soldermask is quickly depleted of silver ions but as the immersion silver deposition reaction continues to occur elsewhere on the board, copper metal continues to dissolve from the crevice area. The end result is a trace that can be narrowed in width and reduced in depth due to the attack. Figure 8 shows a trace that has suffered soldermask interface attack. The image was taken after the soldermask has been stripped from the board surface. Figure 9 is a zygo representation of the same attacked area.

Figure 10: Top-down image, no attack

Immersion silver plating processes can be altered chemically to significantly reduce SMIA. In the experiment detailed below, a negative foot was present, i.e., the test coupon used was determined to be susceptible to soldermask interface attack by standard immersion silver processes. The plating bath was altered such that the copper was protected and aggressive attack prevented

Figures 10 and 11 show the reduced SMIA, from a “top down” perspective of the trace, as well as a zygo representation.

Enlarge this picture Figure 11: Zygo of trace, no attack

After it was determined that a process change could repeatedly give a dramatic reduction in SMIA, in most cases eliminating almost all cases of the trace reduction, samples were produced and tested for their susceptibility to creep corrosion. They were exposed to a MacDermid in-house developed, tire factory simulation environment for 72 hours. The samples were placed in a jar with sulfur powder and 2-mercaptobenzothiazole (a chemical used during the vulcanization process of rubber used to manufacture tires). The jar was kept at 120°F and cycled to room temperature to ensure condensation twice over the course of the test.

Enlarge this picture Figure 12: Zygo of soldermask defined pad

The samples were analyzed for presence of creep and the location of its occurrence was noted. After creep evaluation, the samples were stripped of the soldermask and analyzed for soldermask interface attack. It was concluded that those areas which displayed creep did not always correlate to the areas exhibiting soldermask interface attack.

Though no SMIA was observed on the pad edges in Figure 12, the entire outside edge of the pad displayed creep corrosion.

Figure 13: Creep corrosion around pad

It was concluded that though the exposure of copper in the crevice areas was eliminated, the crevices itself was still an initiation site for creep, thereby confirming that the negative foot of the soldermask acts as a trap for contaminates and condensation.

Assembly

Figure 14a Leaded solder

How a sample test board is processed through the steps of assembly is critical to the outcome of any creep corrosion test. Not only does any chemical approach to eliminating creeping corrosion have to withstand the high temperature excursions experienced during the assembly process, which would typically include multiple lead free reflows and a lead free wave cycle but it has been found also that the presence of solder alone changes the kinetics of the creep reaction. As mentioned previously, it has been determined that creep is a migration of copper and sulfur corrosion products across the surface of a board [4, 5, and 6]. Intuitively, one thinks of the galvanic reactions that may accelerate this process. When silver and copper are in electrical contact with each other any exposed copper on the board, in pores or poor plated areas, act as an anode and thereby dissolves, exposed silver behaving cathodically to complete the cell. As detailed above, any dissolved copper metal ions react, in the presence of sulfur, to form copper sulfide. When a ready supply of copper ions, condensed water and sulfur are present this reaction can continue unabated resulting in corrosion product migration across an area of board surface.

Figure 14b: Lead free solder

As mentioned earlier, any finish, whether organic or inorganic, that might be employed to resist creep corrosion must withstand the high thermal exposures experienced during the assembly process. In 2008, the majority of assembly facilities have switched to lead free pastes and solders that are liquefied at much higher temperatures than their leaded counterparts. It has been industry experience that not only do these solders flow less but also the high temperatures necessary to melt them have a deleterious effect on some of the conventional materials and surface finishes used for printed circuit board manufacture.

Figure 15a and b: As Coated OSP and 1 Pb-free reflow respectively

Simple experiments have been carried out to compare leaded solder to lead free solder and these clearly demonstrate that the use of lead free solder does indeed accelerate creep growth. This, clearly, is one major reasonwhy creep corrosion is more prevalent in today’s industry. At first glance, the two figures below may seem very similar in corrosion but with a trained eye the differences are more dramatic. The sample processed through leaded solder (Figure 14 A, below) displays no creep corrosion on the metal defined (left) side of the test BGA and only one or two pads on the soldermask defined (right) side with only minimal creep growth. The lead free counterpart (Figure 14B) displays creep both in the wells of the metal defined pads and a high occurrence of creep also on the soldermask defined side of the test BGA. In fact, in this test every pad on the soldermask defined side displays at least the initiation of creep. Figure 14 a and b;

Figure 16: Partially Pasted BGA

To understand the differences between the degrees of creep corrosion created on leaded and lead free solder treated samples, one must understand the effects of each process on any proposed creep mitigating finish applied. Organic materials such as the classically applied benzotriazole and more unique compounds such as the substituted imidazoles found in organic solderability preservatives (OSP) are compromised during the reflow processes. These organic compounds can experience significant degradation as a result of the high temperatures employed. They can be either reduced in thickness or even compacted as a result of the long exposure to high temperature reflow cycles [8]. The degradation of the surface organic material will leave copper susceptible to corrosion exposed.

Figure 17a: QFP, 17b:Reduced Aperture Stencil

There are metal coatings that are also degraded as a result of reflow processing. The mechanisms involved are different but the high temperatures experienced act as accelerants to the degradation of the coating. One well known example of this is immersion tin. It is well known that tin and copper in contact will migrate together and form two types of intermetallic (IMC) layers. One is Cu6Sn5 and the other Cu3Sn. Though the formation of these IMC layers will happen naturally with time, introducing heat speeds up their formation dramatically. The higher temperatures reached during a lead free reflow process transforms pure tin into an IMC faster than a leaded reflow cycle. The copper and tin migrate together making the pure tin barrier much smaller if it remains at all. The end result of this is that tin no longer acts as an effective protective layer, copper ends up closer to the surface which makes it easier for reaction with sulfur present in the environment to occur.

Figure 18: Immersion silver with reduced paste after reflow and sulfur exposure

Within assembly more metals are introduced on the surface of the PWB via the solder and as a result, more galvanic drivers now play a part in the overall creep corrosion reaction. Specifically, a large quantity of tin is present along with other smaller concentrations of additional metals such as copper, silver, nickel, bismuth and indium to name a few. The galvanic potential differences on the board’s surface become greater (even more so when an unsoldered pad is connected to one with solder). Figure 16, below is animage of a test BGA with soldermask defined pads. There is a ground plane of copper connecting all the pads. The farthest right row in the image has been printed with lead free solder paste. After reflowing the sample and then exposing it to 72 hours in a high sulfur, high humidity environment creep corrosion has formed. From the image the creep growth is clearly more excessive on the pads that are in closer, more direct contact to those with solder paste. It would appear that as the distance from a soldered pad to anon-soldered pad is reduced the tendency for creep corrosion to occur increases.

Figure 19: High Temperature OSP with reduced paste after reflow and sulfur exposure

Another example of this effect is observed when a pad is only partially pasted with solder. The solder may not spread to the pad edges as a result of reducing the stencil aperture. Due to fine spacing on circuit boards, printed solder paste is reduced over small areas. This will prevent bridging from occurring. With lead free solders, the actual spread of the solder is dramatically less when compared to its leaded counterpart. The diminished paste spread leaves the combination of exposed final finish and solder in direct contact with each other. This area has been found to be highly susceptible to creep corrosion (Figures 18 and 19).

Field Environments

Figure 20: EDS analysis of creep corrosion of an OSP treated test board

As mentioned above creep corrosion has been observed in a variety of industrial environments as diverse as tire manufacturing plants, paper mills and automotive design studios. In developing countries, such as China and India, amongst many others, a marked rise in airborne pollution has occurred in line with industrial growth experiences in these regions. In fact, the levels of airborne contamination, such as S2-, SO2, NO2 and S bearing particular matter has risen to such a degree that electronic equipment used even in domestic settings must survive increasingly aggressive environments, where all the components necessary to create, creep corrosion are present.

Figure 21: EDS analysis of creep corrosion of an ImAg treated test board

Previous publications have clearly demonstrated that high sulfur containing atmospheres are essential to create creep corrosion and have also detailed that the sole corrosion product, regardless of the final finish utilized, is copper sulfide.

Few, if any, publications have emphasized strongly enough the fact that the level of humidity present in a field or test environment is an absolutely critical component which very strongly influences the rate at which creep corrosion can be induced.

Figure 22: Imm Ag with low levels of humidity in the test chamber

As explained earlier, water, or more correctly, condensation plays a major role in the overall reaction of the Cu2S formation and migration. Exposing a circuit board to high sulfur bearing environments alone will not create significant creeping corrosion in the absence of high humidity levels. A tire factory simulation study (as described earlier) was run on immersion silver samples for 72 hours after being subjected to standard lead free assembly conditions. One set of panels was placed in a test chamber with low humidity levels and an identical set was placed in a test chamber with a high relative humidity level. The samplesexposed to these simulated tire factory environments exhibited widely different results. In the absence ofsignificant humidity levels, despite high levels of S containing compounds being present, creep corrosion could not be produced. When humidity levels are high much increased levels of creeping corrosion are observed.

Conclusion

Figure 23: Imm Ag with high levels (close to saturation) of humidity in the test chamber

PCB final finishes were initially designed to function primarily as solderability preservatives between fabrication and assembly processes. These are now at risk of creep corrosion in harsh sulfur bearing, humid environments.

The factors that significantly influence the rate at which this form of corrosion is produced range all the way from the quality of board fabrication processes employed, through the assembly process and finally onto the harsh working environments that today’s electronics are commonly exposed to.

The formation of creep corrosion is driven by factors such as; exposed copper substrate areas, the presence of multimetal galvanic couples creating corrosion cells, airborne sulfur containing compounds and the presence of significant levels of condensation.

Chemical suppliers, board fabricators, assemblers and OEM’s all need to be conscious of their role and the influence that they can bring to bear to minimize or eliminate creep corrosion. Creating an industry standardize test will help to better investigate and identify mitigation practices. This in turn will help bring resolution to this issue.