Mehl[
1] introduced the terms upper and lower bainite. He illustrated upper bainite with micrographs of feathery bainite, which originates from grain boundaries. It is composed of very many parallel plates of ferrite with cementite particles in the interspaces. He illustrated lower bainite with acicular units, initiated by intragranular nucleation of ferrite plates. It seems that Mehl, in reality, simply distinguished between bainite nucleated either at grain boundaries or intragranularly. Hultgren[
2] presented a sketch of the start of upper bainite where the plates were nucleated on a grain boundary and somewhat elongated cementite particles formed in the interspaces. Aaronson and Wells[
3] introduced the term “sheaf” for intragranular groups of closely packed parallel plates and proposed that they form by repeated sympathetic nucleation, starting from an initial plate of ferrite. Hillert[
4] observed that the outermost plate of ferrite in feathery bainite was sometimes covered with a more intimate mixture of ferrite and cementite. It could also form in a wider interspace between ferrite plates. He did not identify it as pearlite but proposed that it was a eutectoid mixture that had formed under cooperation between ferrite and cementite although the ferritic constituent came from the bainitic ferrite. He predicted that this mixture should be more common at lower temperatures. Ko and Cottrell[
5] discovered that bainite, similarly to martensite, forms with a surface relief. That inspired Matas and Hehemann[
6] to propose that bainitic ferrite grows very fast, without time for diffusion of carbon. Cementite in lower bainite would then form by a subsequent precipitation of carbon from the supersaturated ferrite. At higher temperatures, there would instead be time for carbon to escape to austenite in the interspaces where it would precipitate as coarse particles of cementite. However, they had measured the lengthening rate of bainite[
7] and Kaufman, Radcliffe and Cohen[
8] reported that their and similar growth rates[
9,
10] were slow enough to be accounted for by carbon diffusion during growth. Goodenow and Hehemann[
11] then explained this apparent discrepancy by proposing that the lengthening of a bainite unit occurs by rapid, diffusionless formation of subunits of limited length and the slower macroscopic growth rate is controlled by slow nucleation of a succession of subunits. Oblak and Hehemann[
12] published micrographs of subunits and a sketch illustrating the morphological differences of Widmanstätten ferrite, upper bainite, and lower bainite. The proposal by Matas and Hehemann,[
6] that cementite precipitates from austenite in upper bainite and from supersaturated ferrite in lower bainite, is still widely accepted as the basis for defining and explaining the morphological difference between upper and lower bainite. A mixture of the two morphologies of cementite can often be observed, and it has been illustrated in sketches,
e.g., by Ohmori
et al.[
13] and Takahashi and Bhadeshia.[
14] On the other hand, Ohmori
et al.,[
13] who also discussed the morphological differences of upper and lower bainite, proposed that the morphological differences in the ferritic constituent should be considered,
i.e., upper bainite consists of lath-shaped ferrite whereas lower bainite consists of plate-shaped ferrite. The definitions for upper and lower bainite will be considered in the discussion. Until then, upper and lower bainite will stand for bainite from higher and lower temperature ranges.
A recent study of the morphology of bainite formation in some Fe-C steels with 0.3 mass pct carbon resulted in two papers. The first one[
15] reported on the effect of grain boundaries on the proeutectoid formation of ferrite. There was a large variation in shape of ferrite particles soon after nucleation but it was observed that their shapes were very similar for each grain boundary. Usually, there was just one kind of shape at a boundary but sometimes two and even more. It was concluded that the large variation of shape, observed between various grain boundaries, was related to the presumed large variation of their crystallographic structure, which is controlled by the relative orientation of the two grains and the direction of the grain boundary relative to the two lattices. Many shapes had facets to both grains of austenite, and a particularly interesting shape was called chevron because it consisted of two legs, one in each austenite grain. Given more time, the legs of a chevron will develop plates and a long series of chevron particles, often covering the whole length of a grain boundary, will develop into a feathery microstructure composed of parallel plates of ferrite in each grain. Sometimes the arrangement of closely spaced, parallel plates only occurred on one side of the grain boundary and they were called semi-feathers. It was concluded that they were formed from other shapes of ferrite particles.
In that work it was realized that, in order to study how the microstructure of bainite develops during growth, it was essential to study a section containing the main growth direction of an object. To increase the chance of finding objects, sectioned in this way, it was decided to study objects starting their growth from a grain boundary. It was then found that one could often observe cases where a plate of ferrite could be followed all the way from start to growth front without any interruption. It was concluded that the lengthening of bainitic ferrite does not depend on repeated nucleation and rapid growth of subunits to a limited length as suggested by Goodenow and Hehemann[
11] and still widely accepted.[
16] The same alloys with 0.3 mass pct carbon were used in the second part of the recent study,[
17] which focused on the second stage of upper bainite formation, in which the thin interspaces between the ferrite plates transform, triggered by the occurrence of cementite. It was stated that chemically this is a eutectoid reaction because cementite stimulates the simultaneous growth of ferrite, although the eutectoid transformation mostly degenerates in upper bainite.
The elongated cementite particles, typical of upper bainite, were predominant in the whole experimental range of temperature but small colonies, typical of lower bainite, appeared in the lower part of the range. This confirmed the observation by Hillert,[
4] who concluded that the eutectoid transformation in the second stage of bainite formation can occur in two modes, a cooperative or degenerate eutectoid reaction. He further speculated that the cooperative eutectoid would be more common at even lower temperatures. In the study of alloys with 0.3 mass pct carbon, an attempt was made to examine how such a transition to lower bainite could occur but the experimental range was limited by the high M
S temperature. The present study was undertaken partly in order to extend the experimental range by depressing the M
S temperature through an increase of the carbon content to 0.7 mass pct carbon. Due to the importance of studying units sectioned along their full length, the main attention was again paid to objects starting from grain boundaries in the plane of polish.