As has been mentioned many times before in heat treatment discussions, at room temperature iron atoms are arranged in a body center cubic (BCC) stacking. But when the metal is heated to 1335°F, or above, that arrangement shifts to a face centered cubic (FCC) stacking. FCC is actually a more efficient configuration for iron atoms and thus take up less space for the same number of atoms. So while steel heating in a fire does indeed expand as it absorbs the energy, when it reaches approximately 1335°F that expansion is interrupted as the steel actually contracts from the more efficient stacking. This is just one of the reasons that the “critical temperature” (or Ac1) is also known as an arrest point, heating and expansion are arrested as the crystalline shift occurs. Ac1 stands for the French words for “arrest in heating.” When the shift in the crystalline stacking is complete, heating and expansion will continue as before.

This first paragraph was necessary to understand the volume shift in steel as transformations occur, because on cooling the iron atoms will resume the BCC stacking and will expand. Normally these expansions and contractions are gradual enough that all but the most drastic changes in cross sectional shapes are unaffected. But when heating or cooling is more rapid, thinner areas will reach temperatures at different rates than thicker sections and, if this differential is extreme enough, there can be distortion due to differential rates of expansion or contraction. This is one good reason that you can often find “presoak” heats in industrial recommendations. Not all parts are as simple as a knife blade in shape and so a little hold time, just before pushing through the transformation, in order to help cross sectional temps to equalize, can help in reducing distortion.

On cooling this can be even more extreme, and at around 1000°F there will be an expansion as the steel comes out of austenite solution and forms BCC pearlite. But since this occurs in relatively slow cooling, such as still air or slower, it also is not too drastic, and that is why normalizing can help keep things straight after forging. However, if we cool fast enough to avoid that transformation at 1000°F a new atomic stacking could be in store for the iron atoms. Having the crystalline lattice distorted by trapped carbon atoms from rapid cooling forces it into a very expansive body centered tetragonal (BCT) stacking, which takes up the most space of all. This is why steel will clean itself of scale when it is quenched hardened, as the metal expands under the rigid crust of iron oxide scale, it pops off leaving clean gray metal underneath.

This massive expansion to BCT is where most of our distortion woes come from. While the cross section of the average knife blade is simple enough to handle the gradual back and forth from BCC to FCC, the rapid (the BCT transformation happens near the speed of sound) and extreme expansion to BCT is more apparent. This is why we like to use the gentlest quenchant we can in order to obtain full hardness, why we agitate and take other measures to ensure the cooling is as even as it can possibly be. This is also why proper quench oils are formulated to slow their cooling curve in the temperature range the BCT transformation will occur.

When a clayed blade is quenched in water the cooling speed will allow the thicker spine to complete the FCC to BCC transformation at 1000°F before the edge can reach the 400°F mark where the BCT transformation can begin. But you will also have the natural contraction of the more rapidly cooling edge contributing to the effect. You can see this in videos such as the one Ethan linked to. There will be a slight drop of the point of the blade as the spine expands to BCC, that is immediately followed by a sudden recurve as the edge converts to BCT. Always remember that the shift to BCC in pearlite formation at 1000°F is an exothermic reaction, so that once it initiates, it has the ability to retard cooling and facilitate its own reaction; this is why we see the bright band of recalescence when we air cool hot steel.

Th key here is what part of the blade becomes the anchor point as the other part expands. With water, the action of recalescence is overcome and the fresh pearlitic spine becomes the anchor point for the massive BCT expansion and the blade gains curvature, or what I would call positive “sori” in Japanese terms.

Oil has a different cooling curve and the spine struggles to complete the cooling to BCC before the edge begins to convert to BCT. As the edge expands the partially austenitic spine has the ability to move with it, rather than anchor it. Then, when the spine finally catches up, it will convert to BCC but the already expanded edge will not be able to compensate, or worse, as it could become the anchor, in which case you would even get reverse, or negative, “sori.”

Then there are unimaginable numbers of variables that can get added to this situation, from agitation, to clay placement and thickness, and there are still influences by the timing of the natural thermal expansion rates of the metal aside from the crystalline shifts in stacking.

When I first started studying this, years ago, what I found even more curious was the effects of fully quenching un-clayed blades. I found certain blade profiles with longer, narrow, single edges, like a Scottish dirk, would almost always curve into the edge, giving a sickle shape rather than a straight blade, when quenched vertically in oil. I reasonably surmised that the pearlite formation, at the spine, before the edge could fully convert was the culprit. So I started quenching such blades horizontally and spine first, which seemed to solve the problem. But then I had an O-1 blade of the same configuration sickle on me?? I knew the 0-1 wasn’t making any pearlite, so something else had to be going on. That is when it occurred to me that the plastic nature of austenite was also playing a role in the phenomenon of anchor points. The edge of the O-1 was converting to BCT at a quicker rate than the spine. In any case I found many of these issues would go away with marquenching, and other measures to counter the effects of differential cooling and expansion.

Golly, thank you for taking the time to explain all that!

I missed out on the original question of whether or not you should put the time in to explain this effect, but, apparently, you thought you should, regardless.

Good thing you did.

Thanks for your time, Kevin.

Thank you Kevin! It make sense. Amazingly I actually followed that. Good stuff!

Brion

Kevin thanks for the explanation and your time! That's some interesting stuff right there.

great post, thank you Kevin.

MP