8.4 Be Cool!
|It's clear by now that most of the
time the crystal wants to do the opposite of what you want it to do. The
crystal wants big grains, large precipitates, and few dislocations - in other
words, it wants to be a softy.
For sword blades we want the crystal to be hard and tough. Nothing helps but to interfere massively with its preferred life style. The way to do that is to be cool1). And to get your steel to be that too - and presto!
|We know that the
for keeping the crystal from doing its thing is not allowing it enough time for
moving its atoms around.
In other words: we must quench the crystal, we cool it down from some high temperature real fast. If we do that from a temperature above the austenite transition, the crystal has very little time for rearranging its own atoms from fcc austenite to bcc ferrite and in particular for getting most carbon atoms out of the newly formed ferrite in order to make cementite.
If the time is simply too short to produce pearlite and whatever else is called for by the phase diagram, new things must happen.
|Let's look at the extreme: If you
could cool down arbitrarily fast, atoms can't move at all and you would find
the same structure at room temperature that you had at high temperatures.
In the laboratory or in special factories we can do this under special circumstances for special geometries. We can cool a thin piece of any metal down from above 1000 K to room temperature in a fraction of a second. What we get are thin ribbons of amorphous metal that are quite useful for example when it comes to magnetic properties.
Many metals then "freeze-in" the liquid structure they had at the high temperature and are amorphous solids with peculiar properties. The process is called "melt spinning" and you can look it up if you want.
|When we cool down extremely quickly we do not get the nirvana state as called for by the phase diagram but a metastable phase that tries as hard as it can to change to the stable one. Since this takes some energy, metastable phaseslike diamondcan live very long at room temperature. However, as soon as you heat it up to a sufficiently high temperature it will transform to the stable phase.|
|Melt spinning of iron or steel does not produce anything even remotely thick enough to be useful as sword blade. The best you could expect for a real sword blade is to quench it so fast that parts of the austenite structure are "quenched-in", at least in the outer parts of the blade that cool down the fastest. That is the best you can expect with what you have learned so far. If that expected structure would be good for sword blades is not yet clear, however.|
| The "best" you and
everybody else could reasonably expect does
not happen! Plain carbon steel does not
freeze into the austenite structure, no matter how fast we cool it down. Plain
carbon steel still has a big trick up its sleeve. If we cool it real fast, it
produces something completely new, a kind
of metastable phase, that we call martensite,
Martens (1850 - 1914).
Martensite formation is rather strange. It does not happen gradually while we quench. It happens all of a sudden at temperatures as low as just a few 100 centigrade when single atoms can't really move much anymore. But dislocations still can, and martensite formation is a process that involves lots of dislocations.
|Let's look again at the driving force business before we proceed with martensite formation. Let's look at you (and me) for that. You and I are supposed to be good and law-abiding citizens. I'm sure that you strive all the time to achieve the supreme state of being a good, responsible, law abiding, and so on citizen, who reads educational stuff like this article, instead of guzzling beer and munching extremely unhealthy food.while watching scantily clad females on TV doing things that your local moral authority would not approve of.|
|I also do some of that on occasion.
Especially whenever the need to be particular good is not so pressing because,
for example, I still fit into my clothes or my wife is not around.
However, if the deviation from what I should be weight-wise (for crystals it would be nirvana-wise) becomes larger, my attempts to achieve the proper state of being become more severe. Shortly before I would explode, I will take recourse to desperate measures and drink water instead of beer.
|We are talking driving forces here once more. What's weight difference for me, is energy difference for crystals.|
|If you are still in the austenite
state and it is just a little bit below the
temperature, things are not serious yet. The difference in energy between
ferrite and austenite is small.
But being still in the austenite state a lot below the transformation temperature is really, really bad. Ferrite now would have a much lower energy than austenite. This state of being cannot be tolerated anymore.
The need to turn from an fcc austenite crystal into a bcc ferrite crystal gets absolutely peremptory. All your atoms are involved in this because they all need to change their surroundings from 12 next neighbors in the fcc austenite lattice to 8 in the bcc ferrite lattice.
In addition, you also would like to get rid of the carbon in the ferrite or a phase. Ferrite atoms just do not like to have a carbon neighbor. Getting rid of the carbon is a bit less pressing, however, since it concerns only those of your iron atoms that are next to an interstitial carbon atom whereas the need to be ferrite concerns all.
Getting rid of surplus carbon would mean to move individual carbon atoms from wherever they are to somewhere else. But individual atoms can't be moved in a civilian way by diffusion anymore, so you simply cannot make carbon free ferrite grains, period.
The only two options left are to remain (metastable) austenite or to go for martensite as a compromise. It's a lousy thing to do but better than to remain austenite.
|So, what exactly is martensite? A rather special microstructure or arrangement of iron atoms! Even
with our present knowledge of what crystals are and do, fortified with a
phalanx of heavy computers, I doubt that we would have predicted martensite
formation. It seems that crystals are still smarter than we are when driven to
Martensite formation also occurs in other metals and it is a pretty queer thing there too.
|When people like Adolf Martens saw martensite in their optical microscopes, they could not know what they saw and how exactly that stuff had formed. But they could describe it and give it a name. They also figured out that the martensite in steel was extremely hard and thus useful.|
|So once more, what is martensite?|
|The "Mysterious" Formation of Martensite|
|How does martensite form? That's a good question and if we Materials Scientists are perfectly honest, we have to admit that some details still elude us.|
|Martensite formation in detail is so tricky that even modern textbooks are more confusing than elucidating. Let's first look at the standard model that Edgar C. Bain proposed in 1934. It illustrates how martensite formation could have happened. Here it is:|
|Shown are two unit cells of a close
packed fcc crystal. The fcc lattice points are marked with pink or red spheres;
two spheres at the backside are not shown for clarity. The same spheres could
also be taken as symbols for the regular Fe atoms in an austenite fcc crystal.
The green dots mark the possible positions for carbon interstitials in the
The (distorted) bcc lattice co-existing with the fcc austenite lattice is marked with red spheres and red lines
|Bain realized that the face-centered close-packed austenite crystal (with the occasional carbon atom sitting in one of the green positions indicated) actually contains a distorted body-centered lattice as marked by the red spheres in the figure above.|
|It's a perfectly rectangular lattice but not a cube since one side is longer than the other two
(that makes it tetragonal). Your halfway decent general education should enable
you to find out the precise dimensions of the red cuboid.
All that the fcc austenite needs to do for turning into a bcc ferrite crystal is to expand a little bit laterally (x and y direction) and shrink a bit vertically (z direction). Doing that would be a great way to satisfy that urgent desire to become bcc. But how could one do such a transformation all over the crystal in not just one elementary cell of the crystal but in many connected ones? For doing this not only do the atoms need to move, they all need to move precise amounts at the same time.
|Even if we ignore the carbon for a moment, which shouldn't be there after turning fcc austenite to bcc ferrite, the only way of doing this Bain transformation is to move a lot of atoms in lockstep, sort of in military fashion.|
|The regular way, given time, would have been to
move the atoms one by one, each one running around pretty much at random, or in
a civilian fashion.
I didn't just make up those words by the way; they are well known in metal engineering and I have introduced them already some time ago.
|If you, the crystal, now start to
entertain thoughts of " transformation by military movement", you run
into major problems (as always when asking the military in tough times to solve
|There is a little theorem going with this: Normal deformations preserve the anglesbut change the volume. A cube turns into a cuboid. Shear deformations preserves the volume but not the angles. A right-angled cuboid turns into something crooked, for example.|
|The long and short of this is: The Bain transformation as envisioned is a bit too simple minded. Nevertheless, it shows the general direction of what is going on. All we need to do is to figure out how to produce something akin to the Bain transformation by shear and not by "pulling and pushing".|
|Luckily, a nice little theorem in more advanced
math tells (some of) us that every deformation you can produce in one way, e.g.
by pure push / pull, can also be produced in the alternative way (pure shear)
and vice verse. All you need to do is a proper "coordinate
transformation". That involves mathematical things far worse than
we sure don't want to go into this. Let's only note:
|You sure don't want to go into some of the more
sensible questions either; I just list them to give you an idea of the
complexity of the issue:
| It's not possible
for us to work that out for iron (and
plenty of other cases) without a bit of advanced math, so I won't even try.
Iron, however, found out how to form martensite without math. In what follows I will try to give you an idea of how its done. There is, however, no way to make proper three-dimensional drawings of what is really going on; schematic illustrations in two dimension must suffice:
|The black arrows in the left drawing show what
kind of atom movement by a shear deformation will yield martensite. In
addition, a dislocation in the sheared part
(=pink martensite) has already moved parts of the martensite one step to the
The drawing on the right shows how the bcc crystal formed by shear (=pink martensite) is made to (almost) fit into the host crystal again by moving three dislocations through it on the dotted red planes.
|The lower part of the schematic
crystal is just "sheared" to the right by moving all atoms on a
horizontal plane by the same amount as indicate by the black arrows. The amount
of shear increases with the distance from the phase boundary that must form by
A shear like this is related to what we called "twinning" but we do not need to worry about this here. All we should know is that shear as shown can be produced by moving a hell of a lot of special dislocations (not shown) in a peculiar way. How that works is shown in the twinning link.
|Why is the sheared off part now a bcc lattice? In
the picture this cannot be seen.
Well, for real crystals the shear is not just in the paper plane but has also some amount out of the plane. The whole thing is just impossible to show edge-on. I won't even attempt to draw it three-dimensionally in the proper perspective view; that would be far above my skills. It's probably above anybody's skills since nobody so far has made such a drawing.
It's far easier to write down a few equations that show it all and more. Provided, of course, that you can "see" what linear matrix algebra can "show".
Just believe me that with just shearing as shown schematically, you can make a (very slightly distorted) bcc lattice out of a fcc lattice. That's what the right-hand part of the figure attempts to illustrate.
|After a suitable shear, involving movement of a lot of atoms in military fashion, the crystal still faces a severe stress problem. As the figure clearly shows, the sheared-off part that is now bcc martensite would not fit at all into the still present fcc part surrounding it.|
|The right-hand side of the drawing shows how to
deal with that problem. Run suitable regular
dislocations through it on the dotted red planes, and you produce a
structure as shown. Since those are regular dislocations, in contrast to the
special dislocations producing the original
shear, you can run as many as you like through the martensite and it still
remains bcc martensite.
After enough dislocations did their work (and note that many of them will get stuck on the way!) the bcc martensite part now fits into the "parent" fcc phase, if only with some effort. and some remaining stress.
|Notice that I called the product of all this shearing and deforming "bcc martensite" and not "bcc ferrite"!|
|Why? Because after this whole process of shearing, involving moving a lot of atoms at once in lockstep, i.e. military fashion, by running special dislocations through it, then running a hell of a lot of regular dislocations through the newly formed bcc martensite, then squeezing everything elastically (just changing the distance between atoms slightly) until the martensite fits into the space that its atoms formerly occupied, we still do not have a nice bcc ferrite phase because we also need to consider the carbon!|
|All the carbon that was present in
the fcc austenite is still in the bcc martensite because the shearing did do
nothing to the carbon atoms. They stayed right where they were with respect to
their iron atom neighbors. After the martensite formation happened, they can't
go away anymore either because it's too
cold. The net effect of that is that the bcc martensite is not
really bcc, i.e. body-centered cubic but a
bit elongated in one direction. In other words, the carbon in the martensite is
incorporated in such a way that it induces a small normal deformation and instead of a cube we get a
cuboid as shown in the
For being bcc, it is necessary to be strictly cubic (the "last "c" in bcc stands for cubic). We do not have that anymore, we have a cuboid that the likes of me call a body centered "tetragonal" lattice.
|What that means should be clear: In
pure iron, or iron with very little carbon, there can be no martensite. The
martensite transformation discussed above would just be a messy and complicated
way for turning austenite into ferrite when
there is no better way.
Martensite is only something special if there is carbon involved. What that means is:
|Not all of the austenite can transform to martensite, by the way. After nucleation, the rapidly growing martensite particles, often in the shape of plates, lathes or needles, get into each others way and stop growing. So some austenite might be retained in between, even so it is very unfavorable at room temperature. Also, if the quenching is not all that rapid, e.g. deeper in the bulk, there might be enough time to form a little bit of ferrite / cementite in the usual way or even orderly pearlite. What you get in the end is an awful mess of everything. And this mess is quite hard for reasons that should now be clear.|
|In an TEM electron microscope you can see it all. But even experts have problems unraveling exactly what they see. Below are two pictures that serve only to show one possible general structure and the inside of some martensite,|
|Describing martensitic transformations properly in mathematical terms, while not too difficult for the initiated, does boggle the mind so much that we never attempt it in undergraduate courses. Thus nobody notices that a lot of professors aren't too sure about it either. I will thus not go into this any more, except to summarize by enumerating some important points:|
|So if you quench your whole steel
sword blade, only the outside and the thin
partsthe edgewill cool down so fast that hard martensite is formed.
The parts that cool down fast but not fast enough for martensite, will also be
somewhat harder because cementite or pearlite formation in the normal way is
still rather difficult. What one gets in this case is called
bainite; I'll get to that.
Martensite only forms form the surface down to a certain depth known as
quenching or better
hardness depth. This is easily measured
with a so-called Jominy test, use the ink for
Now you have a hard (and rather brittle) edge and you encased your whole blade in the hard stuff. That's why we call this kind of hardening: "case hardening".
|If you want to avoid having the
whole surface of your blade turning very hard (and brittle), you make sure that
parts of it cool down more slowly - by coating it with some clay or whatever.
Therewith you provide thermal insulation of
the coated parts and reduce the cooling rate to a value where no or only very
little martensite will form.
Here you have, in a nutshell, the Japanese way of making nihontos / katanas.
|If your sword blade or whatever else
you (case) hardened by quenching is now too
hard and brittle, you temper it a bit.
We covered tempering in some detail with the Al-Cu example. When you allow a bit of atomic movement by tempering after quenching, the crystal moves its atoms around in such a way that it gets closer to nirvana. For case-hardened quenched steel the first priority will be to get rid of the martensite by turning it into ferrite, and that is done by casting out the carbon, forcing it to form cementite. Hardness then decreases but ductility increases. Stop tempering at the right moment and you have the compromise between hardness and ductility that is best for you.
|Here is a diagram that shows schematically what you can do by quenching and martensite formation in comparison to slower cooling as a function of the carbon content. Of course, it is only valid for the thin parts that could form martensite|
|With martensite hardening we go way beyond of what could be done with all the other hardening mechanism. The stuff is harder than cementite, glass or granite! It is also perfectly brittle like glass or granite; the first law of economics is still enforced.|
|That explains why in the figure
was measured and not the yield stress. You
can't measure a yield stress if the material never yields but just fractures.
The yield stress numbers were just estimated from the table I gave you long ago so you can compare (approximate) numbers.
|That leaves us with a little puzzle. Put a glass edge on a otherwise simple sword and it is just as good as a true katana? Maybebut how are you going to do that?|
|The great thing about hard but brittle martensite is that it is still nothing but iron and a little bit of carbon. We can make it directly from the steel we use, produce it exactly where we want it, as hard as we want, and attached to the rest as well as could be. Try that with some steel and some glass!|
|Understanding the theory of
martensite formation is not an easy undertaking, even so you can enjoy the
comfort of your easy chair and your favorite beverage while trying to do so.
Hardening a real piece of steel by quenching it in your smithy is not an easy
undertaking either, even if you quench your thirst with all the beer necessary
for that. Fortunately you, the ancient smith (and the modern one) can do that
without the need to go through the easy chair ordeal first.
You do not need to know anything about martensite and what happens inside your steel during quenching. You just need to follow some working recipe. Smiths have quenched steel for the last 2000+ years without knowing a thing about the theory behind it. They made babies the same way.
All smiths knew, however, that quench-hardening is a tricky process that did not always work, for reasons unknown. Baby-making is not all that tricky in comparison but doesn't always work either for unknown reasons until recently. This does not imply that there weren't all kinds of theories around that sought to explain those deficiencies (not enough praying, praying to the wrong / Goddess, moon too full, evil eye, witches, ...), it only implies that all those theories were utterly wrong.
What could go wrong with quenching? Let's make a list:
|This is rather upsetting because
quench-hardening is done after a lot of
masterly work has already been invested in the blade. That all comes to naught
if quench hardening fails since you usually cannot repair what went wrong. It
is therefore small wonder that a lot of mystery and magic developed around
Since quenching did chance the properties of the steel a lot, and since it is hard to conceive that the "nature" of the steel has changed a lot, it was reasonable if wrong to assume that the quenching fluid imparted something crucial to the steel. It is thus small wonder that humankind came up with all kinds of (mostly disgusting) recipes for quenching fluids. The link gives a taste treat.
|In hindsight and armed with materials
science insights is is easy to understand why things could go wrong during
quenching. Here is a list that elucidates what could cause "bad
quenches" and which factors play a major role:
|There you have, in a nutshell, what you
need for successful quenching. Note that two independent factors come together:
|Martensite formation and thermal
stress are interlinked in a complex way. Martensite formation is a phase change
that causes stress, and it only happens when there is a large cooling rate that
unavoidably also produces large thermal stress. How the to stresses add up is
not easy to guess, nor what will happen. In essence, however, you might get one
of the following four general cases:
|I have said quite a lot about the art
of quenching but never even mentioned the quenching fluid, the magic and
mysterious substance our ancestors considered to be of overreaching importance.
Well, they were wrong. Very much so. The quenching fluid has only one function: to establish the cooling speed. It can do that in three ways:
However, various dirt in your liquid might react with the surface of the blade to produce ugly spots (or to prevent this). So it really doesn't matter from which river you took your water as long as it isn't too dirty.
| It was the famous
Réaumur who figured
that out and did away with 1000 years of disgusting superstition.
Here is an
illustration to that.
Water is a pretty good quenching fluid and circulated water is more effective than still water, of course. Circulated oil is more effective than still oil but less effective than still water. All the disgusting stuff is in between.
More than that one does not need to know about quenching fluids.
|Tempering and Slack Quenching|
|Let's say you did properly quench a
medium carbon steel. The outside of your pieces of steel, down to a depth of a
millimeter cooled down fast enough to enforce martensite formation and is now
very hard. And very brittle. Maybe too brittle for the use you have in mind.
What can you do about that?
In essence, you have two options:
|In the first
case you form less martensite and more bainite instead. Depending on
the details of what you want to quench-harden, some particular quenching fluid
then is "just right". That might be one of the reasons why the
obsession of the smiths of old with various strange quenching fluids was not
just superstition. One of those brews might have had just the right thermal
conductivity for your work.
In the second case, known as slack quenching, the outside cools down rapidly to ambient temperature and forms martensite. But you take the piece out while the inside is still hot. That will heat up the outside again, tempering the structure there to some exent. In effect you sacrifice some hardness for improved ductility. What you will get depends, of course, on many parameters like starting temperature, nature an amount of quenching fluid, agitation of fluid or blade, time in the fluid, geometry of blade, thermal isolation like a clay coat on parts of the blade, .... You might make it even more complicated by taking the blade out after some quenching time but re-inserting it after some other time.
Smiths without benefit of a material scientist are likely to consider slack quenching as something powerful but erratic that only works on good days and if all the required magic and prayers are properly done. Consult the TTT science module if you want to know (far) more.
James Joyce in
Finnegans Wake, p. 465:
"... Be bloddysibby. Be irish. Be inish. Be offala. Be hamlet. Be the property plot. Be Yorick and Lankystare. Be cool. Be mackinamucks of yourselves. ..."
© H. Föll (Iron, Steel and Swords script)