A Personal Commentary on “Transformation of Austenite at Constant Subcritical Temperatures”
- 855 Downloads
KeywordsMartensite Cementite Bainite Pearlite Metallic Glass
The first recorded citation of the word “bainite” is in an article published 12 years after Davenport and Bain introduced isothermal transformation as a technique to study the progressive evolution of phases as a function of time and temperature. This long interval is a reflection of the fact that although the phase was named in 1934, it was not used even by Bain and his coworkers in articles published subsequently, referring instead to a dark etching acicular aggregate somewhat similar to martensite.
Anyone who reads the 1930 article, I think, would agree with me that such modesty, although laudable, was not needed given the breathtaking quality of the experimental work and the number of insights revealed, some of which have now become legend.
The sheer amount and depth of experimental work reported in the article is impressive. Take, for example, the dilatometry. In order to avoid complications resulting from the expansion of the dilatometer rods themselves, a substantial part of the equipment was immersed into the molten lead bath to achieve the desired temperature. The specimen grips were designed to be self-aligning because the hot sample in its austenitic condition had to be introduced quickly before the assembly again was immersed into the molten metal.
The dilatometric measurements were sufficiently accurate to note that during the formation of bainite or martensite, “the volume change is not necessarily uniformly reflected in all dimensions. Indeed, the thickness of flat disk specimens actually decreases as the volume increases.” The observation was not explained at the time, but we now know that this is a consequence of anisotropic transformation plasticity resulting from crystallographic variant selection. Indeed, the exploitation of such anisotropy is the basis of methods for designing welding alloys capable of compensating for the thermal contraction that leads to pernicious residual stresses.
Davenport and Bain achieved their goal of introducing a “time factor to the iron-carbon [equilibrium] phase diagram.” The resulting representation, known now as a time-temperature-transformation (TTT) diagram, formed the basis of numerous studies, which today, have extended to all materials including, for example, polymers and the crystallization of metallic glasses. It led to the development of many kinds of dilatometric instruments that are now a standard feature of a well-founded metallurgical laboratory. But more so, the diagrams, because they dealt with constant temperature transformation, became the focus of theoretical studies on the nucleation and growth mechanisms of the phases that form in steels. A further tranche of theory, which admittedly has had limited success, was established to convert TTT diagrams into those appropriate for continuous cooling transformation. It also became popular to look at metallographic samples in their partially transformed condition so that morphology could be studied prior to its modification by impingement.
There are a number of TTT diagrams in the original article, but those for steels containing chromium are particularly revealing (Figures 11 and 12). Transformation rates in these alloys are retarded sufficiently to avoid overlap between different reactions. They provide the first clues that the diagram really consists of two sets of C curves, one for ferrite or pearlite, which form at high temperatures, followed by a gap until a temperature is reached in which bainite forms. This is a consequence of changes in atomic mobility leading to different transformation mechanisms as the temperature is reduced. Zener in 1946 recognized this in his classical theoretical article on the kinetics of transformations in steels in which his idealized TTT diagram consisted of two sets of separated C-curves in the manner described. This two-curve representation forms the basis of a vast number of quantitative treatments of steels and emphasizes the distinction between transformation mechanisms.
The second feature evident in the original TTT diagrams is that the bainite curves (Figures 11 and 12) have flat tops; this means that there is an incomplete transformation at a given temperature and that the temperature must be reduced in order to achieve further reaction. This feature is related to what is known now as the T 0 curve, which defines the limit of how much bainite can be obtained at a given temperature. It is the most important design criterion for modern bainitic steels, whose properties rely on transformation plasticity and which are used extensively in the automotive industries.
This brings us to one of the most important instinctive conclusions reached by Davenport and Bain. They were aware from the acicular appearance that the “bainite” was not related to pearlite and that it etched darker than virgin martensite. They could not resolve the carbides in those days but felt that there must be some cementite precipitation associated with bainite; after all, it was a little softer than martensite. The evidence in their possession led them to suggest that the allotropic change occurs in advance of the carbide precipitation. In other words, the transformation is like martensite, but the product soon afterward tempers to precipitate cementite.
To me, this is an incredibly clever piece of deduction, which accounted well for the diverse data they had collected given the limited experimental methods available at the time. Of course, there has been a lot of trauma since this original proposal for the mechanism of transformation, but I think it is safe to believe that the mechanism stands, and indeed, it is only this mechanism that permits the quantitative design of steels.