Failure Analysis

Analyzing Contaminants, Stains and Debris – Part 2

Posted on 06. Jul, 2010 by Rob in Contaminant Analysis

Abrasive Contaminants

A gas transmission compressor shaft in a facility in China began showing excessive wear at the bearing journals during initial mechanical testing (spinning) of the compressor. The plant operator suggested the tin based babbitt bearings were the source of the wear. Our analysis of the bearings showed that the babbitt and bond fully met industry specifications. Our analysis also revealed hard abrasive particles, composed of aluminum oxide (alumina) and silicon dioxide (quartz) embedded in the surface of the babbitt (Figures 1 and 2). These materials are consistent with grinding and sanding media, and caused the bearings to act as a cutting tool on the shaft journal surface. Metallographic examination of cross sections taken from the bearing confirmed that these particles were present at the surface only, and not throughout the babbitt layer verifying that they were introduced during the commissioning or initial operation of the compressor rather than during manufacture of the bearings. These results indicated that contamination of the lubrication system feeding oil to the bearings occurred during installation or maintenance by the plant operator’s personnel.

Figure 1 – Embedded silicon dioxide particle in the bearing surface. 450X

Figure 2 – EDS analyses of the embedded particles indicated a mixture of aluminum oxide (Al and O)
and silicon dioxide (Si and O), or quartz (shown above).

Fingerprints

Fingerprints are usually not a source of what we typically think of as contaminants. However, perspiration and the oils from our skin contain chlorine, potassium, sodium and other elements in sufficient concentrations to contaminate and initiate corrosion in sensitive components such as electronic contacts and data storage media. Highly polished parts are also susceptible to cosmetic degradation from fingerprints if no protective clear coat is present. Plating defects such as blisters and poor plating adhesion can result when these “bio-contaminants” are present on parts prior to immersion in the plating bath. This was highlighted in MAI’s recent analysis of blister defects (Figure 3) on a nickel plated exhaust manifold.

Figure 3 – SEM image of blistered nickel plating. 4000X

Analysis of opened blisters by scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) revealed the presence of chlorine, potassium, calcium and sulfur in small, but significant, concentrations. Both the plating and pre-plating rinse bath chemistries were well within optimal parameters. Visual examination of rinsed parts which were staged for plating showed no indications of contamination. Examination of these parts by SEM and EDS, however, revealed extensive fingerprint residue which contained the same elementals as that observed in the opened blisters (Figures 4-6).

Figure 4 – SEM image of fingerprint on rinsed manifold.
Somewhat optically transparent, fingerprints are readily visible in SEM images. 8.0X

Figure 5 – SEM image of dark area on the fingerprint shown in Figure 4 at high magnification
showing potassium chloride salt crystals from exuded perspiration and oils. 7000X

Figure 6 – EDS analysis of crystals shown in Figure 5 composed of sulfur (S), chlorine (Cl at arrow),
potassium (K) and calcium (Ca) in addition to the steel substrate.

Because the plating vendor’s customer set particularly demanding cosmetic standards, manual inspection of the parts had been instituted between each step in the process. Manual inspection of the parts following rinsing, but prior to plating however, nullified an otherwise effective rinsing procedure by introducing residue from the inspectors fingertips. In an attempt to find the source of the problem, the plating vendor initiated a more rigorous manual inspection which only resulted in more defects from greater amounts of handling. Interestingly, the original plating operation with manual inspection presented no problems when started in early January. Blister defects only began to appear as ambient temperatures rose in early summer, reaching a peak in late July when perspiration and exuded oils from the inspectors fingertips increased.

“Micro Soccer Balls”

The root cause in our failure analysis of a gearbox transfer case was straight forward. The unit had been in service for only a short time before failure and was one of several cases which had failed in nearly identical circumstances. Gross abrasive wear of the bearings and gear teeth was obvious. What was less obvious was the source of the contaminant. SEM examination of debris from the gear case contained the expected wear debris from the gears and bearings, considering the degree of damage they exhibited. Also present, however, were numerous spherical particles exhibiting a morphology that resembled microscopic soccer balls (Figure 7).

Figure 7 – One of the many “micro soccer balls” found in the in the gearbox transfer case. 825X

Analysis of the spherical particle by EDS indicated a composition of iron with approximately 2% manganese. Metal will take on a spherical form only when it solidifies from a molten state free of any outside force, including that of gravity when on a horizontal surface. That fact limited our potential sources. Temperatures generated from friction during the failure sequence would not begin to approach the melting point of iron, so that eliminated an internal source. No melting or foundry facilities were present at the manufacturing site where the gearboxes were assembled, furthermore, the particles size and morphology was inconsistent with a foundry source. That left arc welding spatter which, based on our experience with previous analyses, was consistent with the particle size and morphology. The problem here was that no welding was performed in the fabrication of the gearbox and the manufacturing site performed only mechanical assembly operations. The contaminant particles, however, said otherwise. A review of the affected gearboxes production dates, and comparison to plant maintenance records eventually solved the mystery, revealing that a facility upgrade involving welding had been performed near the assembly line on the affected dates.

Unknown Residue

A manufacturer of aluminum high performance automotive pistons noticed a white residue on finished parts during the final inspection process. A review indicated that no changes in the manufacturing process or materials had been implemented, but despite an enhanced cleaning procedure the residue persisted. Piston samples exhibiting the white residue (Figure were sent to MAI to determine its composition and source.

Figure 8 – White residue deposit on the piston at the ring lands. 4.0X

Examination of the residue by SEM revealed a crystalline morphology as shown in Figure 9. Chemical analysis by EDS (Figure 10) indicated phosphorus, potassium and oxygen as the primary components of the residue with trace levels of carbon, sulfur and other elements.

Figure 9 – White residue on the piston exhibited an acicular crystalline morphology. 1000X

Figure 10 – EDS analysis of the residue showed oxygen (O), phosphorus (P) and potassium (K)
as the main constituents. The aluminum (Al) is from the piston substrate.

The oxygen, phosphorus and potassium are characteristic of tribasic potassium phosphate, a common component of soaps and detergents. Analysis of the detergent used to clean the pistons following machining confirmed that this was the source of the residue. Persistent questions to vendor eventually produced an admission that a change had been made to the formulation of this detergent without notification of their customers. This change resulted in a less soluble reside which the original rinse procedure did not completely remove. An extension of the rinse time resolved the problem in the short term and got pistons shipped. An eventual change in detergent brought the rinse time, and production rate, back to the original schedule.

Read part 1 of Analyzing Contaminants, Stains and Debris