Classic failure analysis (FA) technique generally deals with a single failed component in a complete product to detect the root cause of failure. This approach is inexpensive and less time-consuming. However, this may end up with a wrong interpretation of the product failure or overlooking the synergetic effect of different components related to the failure. 

This article is about a product on the market, which is undergoing a product development process due to observed failures. Due to confidentiality, only limited product details are mentioned in this article.

The product operates in clean hot water environment with temperature around 93 °C. The product is composed of a semi-crystalline polymer – Polyoxymethylene (POM) – housing and PCBA, which is moulded into the housing with polyurethane (PU) and electrodes (Fig. 1).

Fig.1. Top: View of POM housing front and rear side. Bottom: PCBA with its connections and device X-ray view, respectively.

The product owner is a successful small/medium size company with worldwide export and has several times been awarded as Denmark’s fastest growing company in its category. The company offers equivalent products to the market, which have been operating for decades without problem although used in different environments.

Fig. 2 shows a classic example of the failed product. Typical failures observed are: multiple cracks of housing, liquid material protrusion onto surface through cracks, hydrolysed PU, and corroded copper (Cu) tracks of the PCBA.

Fig.2. Top: Crack and liquid material at device surface. Mid: Hydrolysed PU. Bottom: Corroded metal tracks.

A plastic material distributor is used for guidance, and POM is chosen as housing material. In the literature, POM is generally described as strong and rigid, with excellent fatigue, creep, chemical and moisture resistance and thermoplastic. Therefore, it is easy to form and one of the cheapest engineering plastics available. At first glance, this seemed to be exactly what the company needed. Raw material used for production of the POM housing has medium viscosity, is additive free, and a basic resin based polyacetal (another name for POM).

For the resin – one part of the used POM material – the data sheet says: “In case of resistance to hot water, if used for one year, the limit of temperature for copolymer is 90 °C, and if used for ten years the limit is 65 °C.” Limits are based on hot water only as the stress factor. The life expectation for this product is 2.5 years.

POM has many weaknesses. One of them is shown on Fig. 3. During melted mould solidification, specific volume shrinks approximately 29 %. This volume change is compensated in simple geometries by merely pumping more mould into the form during solidification. For complex geometries – as it is in the case here – this option is not valid. As a result, this may cause warpage, surface stress lines, internal stress and voids inside bulk volume (Fig. 3 and 4). High internal stress may later release and cause multiple cracks as shown in Fig. 4.

Fig.3. Top: New POM houses with stress risers, marked with arrows, on the surface. Bottom: Specific volume change from liquid melt to solidified polymer phase. For POM, EHDA line on the graph is valid.
Fig.4. Top image: Freezed-in stress causes micro voids nucleation into macro voids. Bottom image: Multiple crack formation.

When an external load is kept constant, a specimen shows plastic deformation gradually, which is called creep deformation, and POM has this weakness. This is clearly observed at elevated temperatures (T>0.3 Tm), where Tm denotes the melting point of POM and Tm of POM is 166 °C.

The elevated temperature condition, T>0.3 Tm, corresponds to 49.8 °C for POM (0.3 X 166 = 49.8). Considering device operating temperature of 93 °C, high creep deformation is unavoidable. Additionally, POM has very high notch sensitivity, and this may explain why some cracks had initiation point at the sharp corners.

PU is 2-component polyurethane potting material based on polyether- and polyester-polyols and precured aromatic di-isocyanates. Urethanes based on polyester have good temperature resistance properties. Still, they have ester bond in the soft segment, which is susceptible to hydrolysis. Polyether based polyurethane is relatively resistant to hydrolytic attack, but has low temperature resistance. Some ester types are subject to hydrolysis at water immersion.

Although properly compounded polymers will be useful for many years of continuous immersion in water at or below room temperature, any potential application involving continuous immersion in water at temperatures exceeding 50 °C should be carefully evaluated. In hydrolysis of an ester bond, it decomposes into an acid and an alcohol, and the produced acid in the reaction catalyses further ester hydrolysis. Owing to the autocatalytic nature of this degradation, in which at least one of the reactants is also a product, polyester based polyurethanes generally degrade more rapidly than their polyether counterparts. Fig. 2 mid shows a hydrolysed PU mould. The PCBA had a PU based conformal coating, which also showed tendency of hydrolysis.

Corrosion of Cu-tracks were observed at the hydrolysed PU mould areas (Fig. 5). On the image in Fig. 5 corrosion products growing into solder mask were observed. Chemical composition of the corrosion products, solder mask and conformal coating were studied by EDS (Energy Dispersive Spectroscopy) – the results are shown in Fig. 6.

Fig.5. Top: PCBA with hydrolysed PU mould and corroded Cu-track. Bottom: Image of X-section of PCBA along the corroded Cu-track.
Fig.6. SEM micrographs of X-section with EDS spectra of selected elements shown through different layers.

Fig. 6 shows that the corrosion products are copper-chloride (Cu-Cl) compounds. Solder mask EDS analysis were carried out and found to be contain Cl. Cl electron negativity is higher than that of Sulphur (also present in the solder mask). This may explain why corrosion products contain high Cl.

In daily routine, failure often attributes to a single component – i.e. the first failed component – of a product. FA techniques are then used for this single component investigation. This approach is favourable, as it is inexpensive and generally less time-consuming to conduct. However, this approach falls short of the synergetic effect of the different product components, which contribute to the failure and may lead to misinterpretation or inadequate understanding of the failure. This study has shown that by looking at a complete product as a failed item and then conduct the investigation by use of FA techniques for the whole product investigation is crucial to pinpoint solutions that will result in the required product reliability.

More info

For further information on failure analysis, please contact Yavuz Köse, Specialist, Microelectronics.