The Doctor’s first exposure to the subject of non-martensitic transformation products (NMTPs) in steel was many years ago as a young metallurgist who was asked to comment on the microstructure observed in the root of a carburized gear tooth. A “poor quench” was the simple explanation offered for what was observed. NMTPs require a bit more study today, but they are still undesirable and a subject the heat treater should better understand and strive to avoid. Let’s learn more.


Martensitic Transformation

Most of us know that a martensitic structure is the desired outcome of our steel heat-treatment process, but we often fail to fully appreciate the underlying transformation mechanism. We know our quench must be rapid – less than a second or two for a carburized 8620 steel – to avoid the nose of the curve and prevent the formation of NMTPs. Did you ever wonder, however, how a part weighing just a few pounds or a load of parts weighing several thousand pounds can reach these desired quench rates? Or how we can achieve that delicate microstructural balance between the higher-carbon surface of a carburized part and the carbon-gradient-related sub-case and low-carbon core? Both the martensite start (Ms) and martensite finish (Mf) temperatures are a function of carbon and alloy content. A higher carbon level, for example, reduces the Ms point. The secret is in the martensite transformation process itself.

    To begin with, martensitic transformations occur extremely fast (approaching the speed of sound) and have been reported to reach speeds as high as 1,100 m/s (3,280 feet/second) in steel.[1] Furthermore, martensitic transformations are often referred to as diffusionless, meaning they do not rely on diffusion mechanisms (i.e. the long-range diffusion of atoms) for the nucleation and growth of new phases. In these diffusionless transformations, atoms move only short distances (on the order of interatomic spacing) in order to create the new phase. Pearlite formation is an example of nucleation and growth by diffusion.

    Since a martensitic transformation involves the coordinated movement of atoms, it is considered an ordered phase change. This change in crystallographic structure from body-centered cubic (BCC) austenite to body-centered tetragonal (BCT) martensite means that there is a strain associated with the transformation. This is why a heat-treated part that has transformed to martensite is often referred to as being in a strain-induced state. The preferred sites for carbon as an interstitial alloying element are the reason why a tetragonal structure forms. The Bain model or Nishiyama-Wassermann path model can be used to describe this transformation.

    Martensite is not represented on (equilibrium) phase diagrams because it differs from equilibrium in two significant ways. Martensite grows without diffusion, so it retains the same chemical composition as austenite, and the shape of the deformation causes strains that must be accounted for before transformation can happen. Also, the hardness of martensite is a function of the carbon content (since the interstitial carbon atoms hinder dislocation movement and promote solid-solution hardening).

    Martensite formation rarely goes to completion because of the strain associated with the product that leads back to stresses in the parent phase. On cooling of metastable austenite to the Ms temperature, for example, approximately 1% martensite forms, is time independent and is approximately 99% complete at the Mf temperature. It’s a so-called athermal transformation that is a function of the temperature to which the part is quenched and not the time.

    Finally, martensitic transformations are not limited to steels and occur in nonferrous alloys, pure metals, ceramics, polymers, inorganic compounds and even solidified gases. The mechanism of solid-state transformation is similar.


Non-Martensitic Transformation

You may be asking yourself, why do we want to avoid NMTPs? The carburizing and subsequent quench hardening of low-alloy steel gears is an example. Surface microstructures typically contain martensite, retained austenite and in some cases NMTPs (e.g., bainite, ferrite, pearlite and mixed microstructures). The NMTPs are caused by a reduction of near-surface case hardenability due, among other reasons, to a loss of alloying elements (e.g., Cr, Mn, Si) to oxidation.[4] Not only is the surface hardness reduced, the residual surface stresses may become less compressive or even tensile (Fig. 1).

    For example, 8620 with NMTPs on the surface have been shown to have inferior bending-fatigue properties when compared to more highly alloyed steels such as 4615 (with higher nickel and molybdenum) and a lower sensitivity to surface oxidation by virtue of reduced manganese and chromium contents, which do not form NMTPs (Fig. 2). If the hardenability of the steel is sufficient to prevent the formation of NMTPs, surface oxidation has a far lower effect on bending fatigue performance.

    Looking at the bigger picture, we must constantly strive to find ways to improve fatigue performance and lengthen service life. To this end, we want to decrease the surface cyclic tensile stress and/or increase the surface yield stress, thereby increasing the resistance to fatigue-crack nucleation. To achieve this, surface-modification processes based on heat treatment (e.g., carburizing, carbonitriding, laser hardening and induction hardening), non-uniform plastic deformation (e.g., peening and deep rolling) or selected surface-alloy modification (e.g., ion implantation and chemical or physical vapor deposition) can be used singly or in combination with one another. With respect to heat treatment, processes must be carefully controlled and matched to the particular alloy(s) involved to ensure that undesirable features (e.g., incorrect microstructure, microcracks, etc.) are not introduced during manufacture.[5]


Summing Up

The presence of NMTPs in the microstructure should act as an early warning system for the heat treater. The presence of NMTPs signals that potential problems may exist in the loading arrangement, the quenching process or the furnace atmosphere and may be being strongly influenced by the part geometry or choice of material. Companies wishing to avoid issues with NMTPs should specify microstructure in addition to mechanical properties and hardness values so as to minimize any performance-related surprises. IH



1. Bhadeshia, H.K.D.H., “Martensite in Steels,Materials Science and Metallurgy, 2002.

2. Titus, Jack, “Hot Seat,” Gear Solutions, September 2013.

3. Sandoval, Luis and Herbert M. Urbassek, Peter Entel, “The Bain versus Nishiyama-Wassermann Path in the Martensitic Transformation of Fe,” New J. Physics, 11 (2009).

4. Dowling, W., and W. Donlon, W. B. Coppe, C. V. Darragh, The Influence of Heat Treat Process and Alloy on the Surface Microstructure and Fatigue Strength of Carburized Alloy Steel, SAE International, 1999.

5. Matlock, David K., and Khaled A. Alogab, Mark D. Richards, John G. Speer, “Surface Processing to Improve the Fatigue Resistance of Advanced Bar Steels for Automotive Applications,” Mat. Res, Vol. 8 No. 4, 2005.

6. Gear Materials, Properties and Manufacture, Joseph R. Davis (Ed.), ASM International, 2005.