Spherical wear particles discovered
in microscopic fatigue cracks in the
balls of ball bearings
Spherical particles have been found in microscopic analysis in many different machines. The phenomenon of perfectly round spheres has been a topic of interest since their discovery and a number of theories have been formulated for their origin.
While more than one theory exists, their source can be narrowed down by considering certain factors such as particle size and texture, equipment type, surrounding debris, stage of equipment life and environmental conditions. Table 1 summarizes the predominant theories of the origin of spheres in oil analysis.
Early on, it was thought that spherical wear particles were formed primarily from rolling-contact fatigue wear. In the early 1970s, Douglas Scott, William Seifert and Vernon Westcott describe their theory of spherical wear formation in rolling element bearings.
According to their studies, it starts with the formation of a surface fatigue crack which propagates almost parallel to the surface. These cracks become subjected to high and low pressures as the rolling elements continually track over and off the cracks. This results in the cracks or fissures opening and closing as the raceway flexes under load.
As the roller passes off the crack, the oil is extracted and the process starts over again (Figure 1). According to their theory, the flakes that are generated in the crack (or brought into the crack with the lubricant) are rolled up into a spherical shape after numerous cycles. Figure 2 shows an example of spherical wear particles found in a fatigue crack.
Figure 1. Spherical wear particles discovered in microscopic fatigue
cracks in the balls of ball bearings (5,000x magnification).
Figure 2a. Spherical particles found on a ferrogram slide
along with smooth rubbing wear particles (magnified 2,000x)
Figure 2b. The sphere found in a crack appears to be in the
process of formation. The surface of the sphere does not
appear to have been smoothed out yet (magnified 3,500x)
In 1973, Scott and Mills commented further on spherical wear debris formed in rolling element bearings. Using Seifert and Westcott’s earlier published observations of a jet engine lubricant prior to fatigue failure of the main rolling bearing, Scott and Mills investigated whether the presence of these particles could be used as a diagnostic aid in determining the level of distress in bearings.
According to their findings, at 60 percent of their test duration (time to the rolling bearing failure), they found initial evidence of spherical debris. At 80 percent of their test duration, spherical particles were found regularly on ferrograms. Just prior to failure, there was considerably more wear debris present.
They conclude that in fatigue, there is a gradual buildup of wear that starts as flat, plate-like wear particles and becomes spherical possibly through grinding from cyclic loading. This debris appeared to diminish as fatigue continued; they noted that spherical particles are almost absent at failure.
In 1982, Dan Anderson discussed the significance of these particles. According to him, their presence gives an improved warning of impending trouble because they may be detected before spalling occurs. However, he also pointed out that in some studies, bearings tested at higher loads have fatigued without generating significant quantities of spheres.
So, the possibility of fatigue in rolling bearings cannot be ruled out if no spheres are identified. In identifying spheres, he stated that rolling fatigue typically does not generate spherical particles greater than three microns (Figures 3 and 4) whereas spheres generated by welding, grinding and erosion are commonly greater than 10 microns.
Figure 3. Spherical particles under an optical microscope
Figure 4. Spherical particles shown under SEM
A number of references have been made to sliding as a contributor to spheroid particles. In 1988, Jin Yuansheng and Wang Chenbiao at Beijing’s Tsingjua University researched the occurrence of spherical wear particles in diesel engine applications shortly after the running-in period.
To compare the spheres generated in a diesel engine, they used a Timken sliding wear machine to produce spheres in a non-circulating oil supply. While Yuansheng and Chenbiao do not note sliding as the root cause for generating spheres, they state that the spheres generated in their diesel engine study look similar to those generated in the Timken sliding machine.
In 1977, Ernest Rabinowicz (then a professor at Massachusetts Institute of Technology) also suggested that spherical wear particles trapped in cavities may originate from burnishing in oscillatory uniaxial sliding from fretting wear in addition to the earlier theory of formation within cracks propagated by fatigue.
Also, as reported in a previous Practicing Oil Analysis article by tribologist Dr. Jian Ding, he discusses the generally accepted origins for spherical wear particles. According to Ding, small spherical wear particles less than five microns in diameter are frequently found in applications besides rolling element bearings. He states in diesel engines, small spheres in this size range are generally associated with normal rubbing wear or sliding wear.
Ding also notes that spheres larger than 10 microns in rolling element bearing applications are believed to be generated from surface sliding, welding or ploughing. If these particles are found along with large (50 to 100 micron) laminar particles and/or chunky particles, they are most likely generated during a deep-spalling stage. They typically show signs of overheating or melting (Figure 5).
Figure 6. Filtergram of large spherical particle
Yuansheng and Chenbiao also generated spheres by electric arc sparking. They found that micro sparks could cause debris to collect and form into small droplets of molten steel which took on a spherical shape due to rapid cooling under the action of surface tension.
According to Doyle in 1974, he notes that the original (nonspherical particle) becomes spherical by heat at temperatures below the melting point. He explains, “the self-diffusion rate in the wear particles is high because of their high defect structure. The driving force would be a reduction in surface energy.”
To rephrase Doyle, this natural phenomenon can be compared to water or snow that forms into hail. There are various explanations for the cause of hail; one theory is that hail forms in clouds where freezing water droplets form around dust or snow clumps. As the water solidifies on descent, it also takes on a spheroid shape to form the smallest interfacial surface area with the air.
In other words, because the molten steel cools quickly, at a molecular level, the surface properties allow the water molecules to cling together long enough to solidify in that shape (under the intermolecular forces typically associated with surface tension).
From Yuansheng and Chenbiao’s research, they theorized that the formation mechanism of spheroids in diesel engine applications occurs primarily in a molten state. They ran tests in a diesel engine during a running-in period and compared the generated particles to those formed in the Timken sliding wear machine and by electric sparking.
During running-in testing of a diesel engine, the types of spherical particles found varied by type, size and surface appearance. Smaller, smooth-surfaced spherical wear particles (one to three microns) were present from the beginning to end. They acknowledge that these particles seem to be associated with friction and wear.
Although they state that fatigue as a root cause could not be ruled out, they also propose that the smooth-surfaced particles could also have melted as a result of the high surface flash temperatures during sliding at running-in because these particles often co-existed with rough-surfaced spherical wear particles.
They suggested that the rough-surfaced spherical particles formed primarily due to melting. According to the authors, these spherical particles were similar to the normal melting spheres generated with the Timken sliding wear machine and electric sparking. Based on their metallurgical analysis, the spheres originated from the cylinder liner.
They concluded that high surface flash temperatures occur when asperities on opposing surfaces interact. The local instantaneous temperature on the surface can reach the melting point and cause nearby wear particles to melt.
So, as the cylinder liner became progressively smoother, the number of spherical wear particles decreased. Almost no spherical wear particles were found at the end of the running-in period.
Spheroid particles were also obtained from carbon deposits on the top of the piston. These were more distorted and rough, likely from the corrosive effects of the CO2 and SO2 gases.
In 1978, Magan Patel examined bearings from a diesel engine that had failed after only 200,000 miles of service due to bearing overlay wear. Upon examination of the failed bearings, white spherical particles were found to primarily contain the following additive elements: 50 percent calcium, 15 percent phosphorus, 6 percent sulfur and 20 percent oxygen.
Patel termed these spheres “oil balls” and found the particles ranged from five to 40 microns in size. Further research has shown that glycol (coolant) can react with certain engine oil additives (calcium sulfonate or ZDDP, for example) to form a very hard precipitate.
It is possible that many spheres observed in studies prior to this date were actually oil balls. These oil balls have been found to have a Rockwell C hardness, Rc, of 48, which is about the same hardness as a tool steel. Not only do these particles embed in the bearing surface, but they are hard enough to plough through bearing surfaces (Figure 6).
Figure 6. Oils balls found in bearing overlay
Some origins of spheres are not related to wear debris but are instead external contaminants. In new equipment, it is possible to find glass beads from shot peening (a process that blasts glass beads at machine surfaces to increase fatigue life).
These are easily recognizable because of their glassy appearance. Spherical particles can also ingress from manufacturing processes associated with machining, grinding, cutting and welding (spatter) (Figure 7). Also, fly ash is common in coal-fired power plants (Figure 8).
Figure 7. Weld spatter particle
Figure 8. Nonferrous spheres associated with fly ash
Tribologist Trevor Hunt also noted that spherical particles were found in new oil as an artifact of the formulation and blending process. He characterizes them as “having a distinct color” and being easily collapsed by external heating. According to A.W. Ruff in 1977, he also states that polymers in the lubricating oil can agglomerate smaller spheroid particles.
To aid in determining the source of spherical particles, analytical instruments such as scanning electron microscopes (SEM) with energy dispersive X-ray (EDX) capabilities can be used. They allow the analyst to determine the particle size, shape, texture and elemental composition to help identify the machine component or contaminant source.
Figure 9. Numerous small spheres in a rolling
element bearing oil sample (1,000x magnification)
Although a number of possible origins for the source of spheres have been advanced by researchers, there is a general consensus that the size and quantity of spherical wear debris can reveal the severity of rolling-contact fatigue wear. It is estimated that millions of spherical particles form during rolling element bearing failures.
Because large spherical particles are often the product of high metal-to-metal contact and high frictional temperature, their presence is often considered a supporting symptom for assessing the wear severity levels.
For example, sliding wear with large spherical particles is considered more advanced than similar sliding wear without spherical particles because spherical particles indicate higher temperatures are being reached. For instance, Yuansheng and Chenbiao found that as they increased the load in their diesel engine testing, the number of spheroid particles increased.
However, this is not always the case. Anderson pointed out that in some studies, bearings tested at higher-than-normal operating loads experienced significant fatigue without generating corresponding high quantities of spherical particles.
It is also worth noting that Scott and Mills observed almost no spheres upon failure of the jet engine bearings, and Yuansheng and Chenbiao also stated the spheres almost completely disappeared by the end of the engine running-in period.