Steel Cleanness and Why Measurement Matters

Highly loaded components benefit from clean steels, but careful measurement — from steel processing to final quality — beyond conventional methods is required to achieve a high level of material performance.


In the past, cyclically loaded gearbox components often experienced fatigue failure caused by large oxide inclusions. Gradual improvements to the steelmaking process led to reduced concentrations of oxides and a corresponding shift in failure mode. As a result, common failures were more often due to component design and manufacture (e.g., surface finish). With the next evolution in application performance, enhancements to component manufacturing resulted in superior surface finish and more consistent heat-treated properties.

Today, the demands on mechanical power transmission systems are leading to an ever-increasing emphasis on “power dense” design, focusing on higher power throughput via smaller system components. This trend translates into higher stresses within each component. And with these higher stresses, those upgraded components with improved surface finish, which had previously led to only moderate improvements in performance, are now experiencing a return to increased sub-surface fatigue initiation failures occurring at internal material imperfections. These are essentially non-metallic inclusions which, under most loading conditions, are principally large oxide inclusions.

These highly loaded components benefit from the incorporation of even cleaner steels. The manufacture of very clean (air-melt) steels requires precise control of the steelmaking process to consistently produce the highest performing power-dense components. And key to the required level of process control is measurement.

That said, the application of indiscriminate, unsystematic measurement to steel processing and final quality alone is not enough. A critical yet sometimes overlooked aspect of measurement principles is the need for a fundamental understanding of the measurement being considered; in other words, knowing what attribute of the material is actually being measured, how it is being measured, and whether this attribute actually correlates to the desired performance characteristic.

An illustrative example is total oxygen content, a common metric employed throughout the steel industry to indicate steel cleanness. As a cleanness metric, total oxygen content is often not reflective of steel performance, especially in cyclic contact loading applications such as with bearings or gears. In these cases, macro-oxides or stringers are the prominent inclusion type in fatigue failures. This is readily apparent in a comparison of bearing life and measured total oxygen in the corresponding bearing steels, as shown in Figure 1 [4].

Figure 1: Bearing fatigue life as a function of measured total oxygen content of the steel.

The reason for this apparent inconsistency becomes clear once a deeper understanding of the total oxygen measurement and those steel cleanness features that most affect the desired performance characteristic are provided. First, the interrogated sample size in total oxygen measurements is quite small (typically a mass of about 1 gram). As a result of the small specimen volume, in many of today’s manufactured clean steels, the likelihood of encountering a macro-oxide (those most responsible for contact fatigue failures) in even a series of total oxygen tests is low. It turns out that the total oxygen metric is more accurately considered a measure of the concentration of the uniformly dispersed small oxide inclusions (those having a negligible effect on fatigue).

Because the source of these numerous small oxides is different than the source or cause of the large macro-oxides, there is no clear correlation between the two populations. This does not mean there is no correlation between total oxygen values and fatigue performance. Clearly, (again, referring to Figure 1) as the concentration of large macro-oxides increases, there is a great probability these will be detected in the small total oxygen samples. This is reflected in high total oxygen values and is predictive of poor fatigue performance. Accordingly, total oxygen measurements may be considered a one-sided test; high values are indicative of poor fatigue performance, but low values are no guarantee of good performance. It is important that this aspect of the total oxygen metric is well-understood.

A high-resolution ultrasonic test with its significantly increased interrogated volume provides a better measurement for detection of macro-oxide population and, accordingly, is a better predictor of fatigue performance, as shown in Figure 2.

Figure 2: Bearing fatigue life as a function of macro-oxide population detected by high-resolution ultrasonic measurement.

Circumstances similar to those noted with total oxygen measurement exist with many of the other commonly employed industry standards defining steel cleanness measurements, as noted by Girodin et al. [5], stating that “conventional methods for inclusion rating are not sufficient anymore to evaluate accurately the micro-cleanliness of the current bearing steels…” and, accordingly, are unlikely to correlate with material performance. This point was emphasized by Murakami [6]: “Correlations between cleanliness (as measured by JIS and ASTM methods) and fatigue strength were investigated … but the results were not satisfactory.”

What this means for us today is when using ultra-clean steels in these highly loaded components, the importance of careful measurements — whether in-process signal monitoring or final product quality determination to guarantee a high level of material performance in application — cannot be overstated. After all, one cannot control what one cannot (quantitatively) measure.

In upcoming Materials Matter columns, the subject on the practical worth of industry steel cleanness standards will be discussed in greater detail.


  1. W. Thomson, Popular Lectures and Addresses, Vol. 1, “Electrical Units of Measurement,” May 3, 1883.
  2. H. M. Howe, Metallurgy of Steel, The Scientific Publishing Co., New York, 1890.
  3. A. McLean, 1988 Howe Memorial Lecture; “The Turbulent Tundish – Contaminator or Refiner?”, Steelmaking Conference Proceedings, ISS-AIME, 1988, pp. 3-23.
  4. J. A. Eckel, P. C. Glaws, J. O. Wolfe, and B. J. Zorc, “Clean engineered steels – progress at the end of the twentieth century,” Advances in the Production and Use of Steel with Improved Internal Cleanliness, Atlanta, GA, May 4, 1998, ASTM, pp. 1-11.
  5. Girodin D., Dudragne G., Courbon J. and Vincent A., “Statistical Analysis of Nonmetallic Inclusions for the Estimation of Rolling Contact Fatigue Range and Quality Control of Bearing Steel,” Bearing Steel Technology, ASTM International, STP 1465, (2007), pp. 85-100.
  6. Y. Murakami, Metal Fatigue: Effects of Small Defects and Nonmetallic Inclusions, Elsevier Science Ltd., Oxford, UK, 2002, pp. 75-77.


Previous articleQ&A with Bridget Dubbert
Next articleAdvanced Jiffy Machine Products Inc.
is a senior scientist for materials at TimkenSteel Corporation. He began his career in the steel industry in 1977 and since joining the company in 1987, has held numerous senior technical positions focused on improving steel cleanness and performance. Glaws earned a B.S. in Metallurgical Engineering from Lafayette College and a Ph.D. in Materials Engineering from Carnegie Mellon University. He held a post-doctoral fellowship at the University of Newcastle, NSW, Australia. Glaws can be reached at