Date Wednesday, 22 June 2022, at 7:19 p.m.
1,2) I did- same steel, identical composition and the batches are so clean that the variance is a tiny fraction of the actual tolerance - below the tolerance of any listed steel of any type for contaminants by a wide margin.
I can only guess why that is, but Larrin hinted at something. I am focusing on heating to critical and then another full notch past there. I tried heating samples to nonmagnetic and then two big steps after that with O1 - without a soak, the steels that are slightly overheated end up harder, but the overheating is only very quick and it is a small fraction of the time that it would take grain to actually grow. that means I could have more than one shot at heating if a chisel warps too much and not appreciably grow grain. when I mentioned this, Larrin said that it is common for fast induction heating to heat well past where a furnace soak would be. I don't think it's by accident that such a thing is common in industrial induction heating processes.
Here's what larrin also hinted at - you can chase higher hardness in a furnace with a cold terminal temperature (like liquid nitrogen if desirable) to increase hardness some and be safe that you're heating enough, but a higher soak temperature allows more carbon to dissolve in the matrix than - I guess - the eutectoid limit. The result is the matrix loses toughness.
I think this is where the results are gained by forge heat treatment - I don't have to soak the steel in a slow heating furnace to get well past critical, get high hardness - very high out of the quench, and the lack of a longer soak time means my excess carbon remains in the carbides and not in the matrix.
Anyone reading who wants to see what carbides look like separate from the matrix - this is 52100 - I can't find the 26c3 picture, but figure half as many little dots standing proud. this is what an iron looks like after it has planed for a while - it's the back and the cap iron is holding the shaving against the iron tip - the matrix wears away faster than the carbides. 26c3 does show a lot of these, though.
Following what Larrin said, it seems reasonable to try to quench as fast as possible and terminate the quench temperature quickly as low as possible (that's the freezer for me) to maximize hardness, but to do a quick heat to keep excess carbon from dissolving into the matrix.
this is all after-the fact guessing, though. I only sent samples more or less due to critics here. I can take a chisel done with this process and test it in any reasonable test against old and new chisels, and if it betters them, there's no great need. I can also identify hardened plain steel pretty reliably between about 60 and 65 hardness with a washita, india and natural japanese stone. that's how I guessed at 61/62 and 63/64 for O1 and 26c3. I had no idea if my guess was correct, but just suspected it based on the known hardness of novaculite and knowing that overhard japanese chisels can be a real bear on a good natural japanese stone - an impractical bear.
3) I hope it is - what triggered it was trying the steel with one quick hard heat and then quenching. It was good. But it wasn't better than a ward chisel, and wasn't quite as good. After reading about shrinking grain and the idea that hardness could be driven higher with smaller grain and the same toughness, I tried verhoeven's triple quench Idea. I may have read it wrong. Taking the steel to critical and quenching it three times grows the grain a little in files and in other steel, but if grain is really large from forging, it probably shrinks the grain a lot.
I have seen other people talk about descending heat quenches before the final quench, but that seemed fiddly, so I learned to identify a slight color shade prior to when a steel becomes nonmagnetic and quenched. that can be done in air or with quench oil. I am too lazy to wait for air cooling. larrin suggested this may not be doing anything other than modifying carbides. I think what I'm actually doing is heating to about the point where transition starts and then quenches.
then after the last heat, I do the fast high heat and a quench. How high? a mid orange - a temperature that you would never allow the steel to stay at. Snapped samples show no increased grain size. I can take a nicholson file and actually shrink the grain on it, too, even though they're well done at the factory - very well done. The carbides will make the snapped grain on a chisel like this look a little bigger.
Tools get seven total heats.
3-heats - I shape the steel at the forge. I don't heat it to forging temperature, but a little shy of that and hammer it closer to black heat, but it gets reheated each time to critical in the forge, so no damage remains from that. I don't hammer it enough to crack it. that cuts down on the amount of grinding I have to do, and I just like to do it. these are done with descending temp and the last heat doesn't get too much hammering- "cold packing" doesn't lead to an improvement and may do the opposite.
then, i do basic grinding after the steel is cool and forge weld on the bolster.
Once the steel is ready for heat treat, then:
* 3 subcritical heats and quenches
* one fast heat, and the quench process
it sounds like a lot of steps, but it's part of making the chisels and it takes very little time. The subcritical heats don't need to go to full room temp, they just get quenched quickly a good bit past black and then back in the heat. It would take far longer to do it with a furnace and I can keep track of two at a time.
I haven't made O1 chisels like this, so no clue if they would improve. The samples that I sent all got the same heat sequence, but no forging, because I didn't want to fish with the grain and not be able to compare to larrin. I doubt I'm doing enough forging when shaping (halving the thickness at most, but a little less than that) to change the grain orientation.
I don't show any advantage over forge results with O1 or 1095 other than I get higher hardness and slightly less toughness than expected with 1095 - it just needs to be tempered a little further. the cost of 26c3 isn't enough (and the toughness a lot higher) for me to consider using 1095 for anything.
there are a bunch of other little things that drove me toward trying 26c3, but I anticipated that it would be hard and not tough. I was wrong. Fortunately. it's just a good match for the process and the process was backed into just by snapping samples and testing results.