Jump to content

Quenching, Thermal Conductivity, and Liquid Metals


Recommended Posts

I had a thought today and, not being a metallurgical scientist, thought I would open a discussion to see what other folks here think. Is the most direct relation between the speed of a quench and the fluid doing the quenching its thermal conductivity? After doing some research, I found that (pure) water reaches its most thermally conductive state (at atmospheric pressure) at around 130 degrees F. On those few occasions where I do quench in water, and from the recommendation of those who do more regularly, that is coincidentally the same temperature range to pre-heat the water for quenching. Although I cannot locate datasheets for various oils, I presume the same holds true. The question is- is thermal conductivity the most important quality? It would seem that the speed of the quench would be limited by how fast the quench medium can extract energy from the steel which is limited by two things. 1, the heat transfer across the fluid-steel boundary layer interface, and 2, the thermal diffusion across the fluid itself. The faster the thermal conductivity, the I would think the faster both of those things would happen.

 

Now, as I understand, the main advantage of oil over water is that you do not have to worry about the steam jacket forming around the steel. Thus, oil is safer because there is a lesser risk of those unpredictable pressure differences and as a result, less warping and cracking. Depending on how much austenite is ultimately converted to martensite may have a more significant impact on the cracks than the steam displacement, but let's assume that we're talking about a steel that has a maximum martensitic saturation less than that of the lesser amount gained by quenching a higher carbon steel in oil/water. So, if we are trying to gain the maximum evenness and speed from a quench, it would make more sense to use a medium that has a higher thermal conductivity than water (which, despite being the highest of any non-metal liquid, is still very low).

 

This is where it gets interesting. The non-metal liquid part caught my eye. First, let's compare the thermal conductivity of some fluids (W/m*k) all at 25C.

Olive oil- 0.17

Engine oil- 0.15

SAE 50 machining oil- 0.15

(all the relevant oils I could find, which are all very similar)

Water- 0.58

This is an average of 3.7x faster than the oils (again, none of the commercial, specifically quenching oils)

Sea water (could only find at 20C)- .596

At 25C, it would higher still, approximate linear extrapolation being .604

Again, this is concurrent with the practice of adding salt to water to speed up the quench.

 

Liquid metals-

There are far more liquid metals at room temperature than you think. The obvious is elemental mercury, and gallium is close (melts at 30C), but there is an entire class of alloys which are liquid at room temperature. Commonly including Indium, tin, lead, gallium, and bismuth, they were designed to be used in computer cooling applications amongst other things. For now, let's look at a few pure metals that would be liquid at low enough temperatures to use for quenching into.

 

Mercury- k=8.3

This is 14x higher than water! But still on the low end of the spectrum...

Gallium- 40.6 (melts at 29C)

Tin- 67 (melts at 231C, a bit high, but still low enough for martensitic transformation given a large enough volume that the bulk temperature does not rise much higher)

note k=57 at melting point, still 98x higher than water

Indium- 81.8 (melts at 157C, k=42 at 200C)

 

And then there are a handful of manufactured liquid metals which have better thermal conductivity and are undoubtedly outrageously expensive in bulk.

 

So my question is this- is there any reason to try quenching into liquid metals? The uniform density and thermal distribution coupled with possible extreme quenching speeds without vapour jackets forming sounds like a panacea for steels on the lower end of hardenability. If nothing more, it would be a neat experiment to try. Do any of these things even relate? If so, is there a theoretical upper limit to the benefit v speed of quenching? I probably should have asked this first, but is it already done?

 

I'm curious to hear your thoughts on the subject

 

John

Edited by John Page
  • Like 1
Link to post
Share on other sites

Obviously, it isn't all about speeding up the quench. Otherwise, nobody would use oil. Oil is uniform, has no vapor jacket, and quenches slow, but just barely fast enough. Therefore, not stressful (relative to water) and doesn't crack as many blades.
Have you thought about brine? The whole point of brine is to basically have water that doesn't form a steam jacket, and it quenches fast. Like, wicked fast. I've used brine once (10% saltwater brine) and I stuck it in there, and in less than a second I pulled out a cold blade. I haven't used it since because it sounds like screaming. There's a shrill sound for less than a second, and then it's cold. But, even though brine cools steel much, much faster than oil, my brine hardened blades weren't any harder than my oil blades.

Also, what about low-temp salts? What's the thermal conductivity for those? It seems to be a more reliable, more even way of cooling the steel.

Basically, it looks like you're saying, "Why not quench in liquid indium?" and I'm saying, "Why quench in liquid indium, when canola oil does everything we want?" I'm curious about it as well, and it would be an interesting experiment, but I'm failing to see the practicality in it. And also wondering, if it were a good idea, why hasn't somebody tried it before? There are people who dedicate their lives to learning about metal and what it does, people who actually make their living by messing with this kind of stuff. And we already use it for things like cooling computers, but the best we've come up with has been low-temp salts? Just thinking out loud.

Just as a bonus question, what's the thermal conductivity of blood? Just curious if it's actually possible to quench a blade, "In the blood of our enemies.". :mellow:

Link to post
Share on other sites

Safety and cost are going to be big factors here. Mercury is toxic. You might find the sales of it restricted and a spill would be a toxic nightmare. You would also have to protect yourself from the fumes.

 

Tin, as you said, has a bit of a high melting temperature. It's going to be right between Ms and Mf for most steels; more close to the Ms. That would dictate an interrupted or isothermal quench. There's a safety issue with dealing with quenchants that are that hot, even oils.

 

You didn't mention bismuth which has been used as a quenchant but the price of it has risen over the last several years which makes it impracticle.

 

Doug

Link to post
Share on other sites

Collin, I have similar thoughts about salts and brine, and realistically either of those is way more practical. Although realistically, this is all just hypothetical anyway ;) One of the other reasons that led me down this rabbit hole is the use of liquid metals for heating for heat treatment. I've heard of lead being used to bring blades up to tempt, and there are probably some out there who use similar metals for tempering like salts.

 

For the bonus question, I just had to know... Turns out, the thermal conductivity of human blood is reasonably similar to water at .462 +/- .009, and for plasma .570 +/- .010

So theoretically...

 

Doug, you have the right of it. Safety and toxicity make experimentation difficult. I certainly have reservations about plunging hot steel into liquid mercury. I actually misread the temp for tin, the 449 is in Fahrenheit, so 231C. Elemental Bismuth has a melting point higher than Tin at 271C, but its thermal conductivity is very close to Mercury at k=7.97. Out of curiosity, was it blades being quenched in Bi? 271C/520F is hotter than I would temper most things, unless it's a thin, bendy sword.

Link to post
Share on other sites

Conductivity isn't everything.
For water, brine and oil Heat capacity and Heat of vaporization is usually more important.
Liquid ammonia would be interesting in that regard.

Edited by Steffen Dahlberg
Link to post
Share on other sites

Doug and Steffen hit the big issues. Another thing is that if you quench too fast then even a thin object like a blade will because of the thermal stresses between the inside and outside of the blade are too different as the steel itself is limiting the heat exchange. BTW, it is the heat exchange difference causing stresses that make vapor dangerous to blades, not pressure. Heating water really slows down a quench as you get vapor faster and thus overall cooling rate drops.

 

That being said... I met an Austempering expert that swore a 100% lower bainite blade would be ideal. Very tough without tempering and great edge retention (his claims). Tin is just too expensive (as are just about any low-melt-temp metals), but if I had a salt set-up I would really think about giving that a try.

Link to post
Share on other sites

Interesting, thanks for chiming in! I can totally see how heat capacity would matter, but for metals, it would essentially be eliminating the possibility of flashing to vapour. Having seen enough of what happens inside nuclear reactor cores, I totally get the catastrophic effects of water flashing to steam :ph34r: I suppose that the differential heat transfer would be a much lesser threat in oil too, but would it still be gone entirely in metals? That is sort of what I'm seeing, but as you say Jerrod, the internal neat transfer would definitely be the limiting factor with these sorts of liquids. The purely bainite blade is curious. Might be something to try if I happen to be at someone's shop who has salts...

 

Anyway, thanks again for the input! This is why I ask :)

  • Like 1
Link to post
Share on other sites

Another thing to think about is that there are 3 ways energy (heat in this case) can be transferred. In addition to the conduction there is convection and radiation. The radiation is too slow to be of concern but for quenching in a liquid, the viscosity of the liquid affects the convection (this is why, as I understand it, warm oil quenches faster than cool oil).

 

I've wondered about the edge holding ability of bainite. The other characteristics, it seem to me, would make a bainite nice for a sword.

 

 

 

Not a metallurgist, just read some books.

ron

Link to post
Share on other sites

liquid Lead was used for years both as a quench, and as a temper medium. As to why this worked don't forget for austenite to form into martinsite it only needs to go bellow around 900DegF to miss the nose. I saw a cooling curve for Parks 50 and the interesting thing was the non linear rate of cooling (I would imagine this is true of all oils) the rate is very fast until about 800degF and then becomes much much slower exactly what you would want to harden a shallow hardening steel and lower the risk of cracking.

 

I would think that salts are the best choice for what you are asking about, salts work so well that they are still used in industry even with safety and environmental hassles that they entail.

Mp

Link to post
Share on other sites

I would think that you want fast and uniform heat extraction in the first few seconds of the quench but you only need that to be fast enough to get under the nose of the TTT curve. Once you have achieved that you want a slow quench to avoid the fast formation of brittle-undeformable-without-breaking-martensite outside while the inside has not transformed yet. That means: cracks. That is the main difference between oil and water to me: water is too fast in this later cooling stage (convection). Or you can design your own quenchant with varying properties using polymers.

Link to post
Share on other sites

I marquench O1 into low temp salt from high temp salt doing batches of 400 small blades every 2 months. I put a PID controller on an old Hamilton Beach roaster, works like a charm. I'd be curious of the thermal conductivity comparison of salt @ 220C, I did the 50C oil quench for years, but it's sure nice to be away from the smell and the residue on blades - and to have such predictable control of the HT process.

Link to post
Share on other sites

And suddenly, interrupted water/oil quenches make so much more sense! After crossing the nose, I just assumed that the martensite was all formed already and that the stress would be at its maximum, but relaxation in autotempering doesn't necessarily mean the steel has to go down below a second temperature at all... Again though, I was thinking more along the lines of steels with lower carbon content which would have trouble reaching full hardness in slower quenches. This is all very interesting.

Link to post
Share on other sites

Good topic!

 

See if you can get hold of Howard Clark, he has done several katana using L6 with a fully bainitic structure using salt pots. They are reputedly indestructible. He used to post here quite a bit, might try IM-ing him.

Link to post
Share on other sites

I love that bainite keeps coming up, something which I do not know much about but am curious to explore. Because of the less rapid cooling of austneite, the bainite returns to ferrite and cementite instead of the martensite of rapid cooling, but in a different way than pearlite. Martensite is about 1.75x harder than pearlite (ferrite and cementite), and while bainite is harder than pearlite, it is softer than martensite. But by how much? For applications where hardness is not the only desired property (flexibility in swords), it makes sense that bainite would be great, as it does not have the same lamella patterns as martensite and thus has (as I understand) a less destructive distribution of an already inherently lower internal stress. Based on my loose understanding of forcing functions as they relate to thermodynamics, quenching into a higher temperature would (assuming similar heat transfer properties) cause a slower cooling of the steel.

 

It seems that most liquids have a peak thermal conductivity at some temperature after which the curve steeply drops off, but for salts I imagine it is a little more stable? using salt water as an approximation is not even worth doing, but for metals, the bainite range is still well within a good range. Specific heat capacity seems to do just the opposite. At some temperature below what we would be quenching into, the specific heat is lowest after which it continues to rise towards some asymptotic maximum value as temperature goes to infinity. Maybe that's the balance that makes high temperature quenching possible? Having lesser heat transport (forced convection from hand movement probably outbalances natural internal convection) but a higher specific heat capacity would...act as a natural damper? Not too sure... Hope I can get Howard's insight in here too :)

Link to post
Share on other sites

Don't forget that you need to still pass the nose of the TTT curve to get bainite (so you still need to be fast at the begining) and martensite doesn't even start to form until you get to the Ms temp. Ideally you cool just fast enough to miss the curve and no faster. If we take a look at the 5160 TTT diagram (the best one I had scanned in and ready) we would want to get from about 1425 F to 900 F in about 3.75 seconds. Then from 900 F to 450 F in about 60 seconds. Once you cross the Ms line time is irrelevant for formation of martensite, so go from 450 F to room temp (or colder) as slowly as possible. Again, that is just the ideal. If you could go from 1425 F to 500 F salt (or whatever) then take it out and put it in 300 F oil and just let that cool on it's own that would be a pretty sweet quench cycle. Alternatively, if you left it in the 500 F salt for a bit over 10,000 seconds (3 hours) then you would have 100% lower bainite.

 

5160_TTT.jpg

Link to post
Share on other sites

Thanks for the graph! That makes this a lot easier to follow.

 

I thought that you formed martensite or bainite, not bainite from martensite? Or is there no way to form bainite directly from quenching like you do martensite? Do the acicular bainite crystals grow out of the decomposing (for lack of a better word) martensite? Since bainite is more or less just acicular ferrite with cementite filling the empty space, does that mean the grains of the tetragonal martensite lattice break down and become each of the bainite nucleations? Also, is the bainite line the 50% dashed one?

 

At the 500 degree mark, it seems like a liquid tin bath would be perfect (improbabilities aside) for bainite. But salts would be much more realistic...

 

Doing a quick comparison, it looks like the specific heat of water (140F), 4.29 kJ/kg*K is about 18x higher than liquid tin (.24), and lead (.14) and bismuth (.15) are both quite a bit lower. Not sure what do do with those numbers, but there they are.

Link to post
Share on other sites

You go from austenite to either ferrite/pearlite, bainite, or martensite. It is possible to have a mixture of all of the above, including retained austenite. Bainite does not turn into martensite nor the other way around. The dashed 50% line is the indication that you have changed 50% of your austenite into whatever is to the right of it, either upper pearlite, lower pearlite, upper bainite, or lower bainite. There is no indication of where it changes from pearlite to bainite on that chart. Generally you get bainite below the nose and pearlite above it, sometimes you have to go way below the nose, but just above the Ms line.

 

Tin would work well if not for the cost associated with it. Last time I bought tin (I want to say back in '08 or '09) it was about $9.50/#. Also, if you are going for bainite that tends to mean soaking for a while. That means diffusion may occur and tin is really bad in steel. Though just a handful of hours at 500 F shouldn't be an issue as any diffused tin should be so shallow it would come off during final grind/polish.

Link to post
Share on other sites

Very interesting, Dan. That's something I wish I could see!

Jerrod, that makes much more sense now, thanks! It hadn't occurred to me to think about tin migration (or whatever metal). At these temps, I figured it would more or less just gild the steel, which as you say, grinds right off. But diffusing into the matrix is another story. I assume salts don't have as much of an issue with this? Cost does seem a bit prohibitive, although it should be a one time cost, as it doesn't break down and hopefully is not effervescent. Oxidation might be tricky, but skimming off the dross shouldn't reduce the volume that much. I may have to try this sometime on a small scale :ph34r:

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...