Rolling to chipping transition hardness

There's a trade off between strength and toughness, and therefore between rolling/flattening and chipping failures. In general, the way to maximize performance is to split the difference because the problems from the larger evil will grow faster than the problems from the smaller evil will shrink. If a blade never chips, added toughness has nothing to improve yet added strength will reduce rolling. If a blade only chips, it's likely that adding hardness will only make things worse. Thus, we want to aim somewhere in the area of "as hard as possible without (exclusively) chipping".

What hardness does this work out to in practice? Which steels can get the hardest while maintaining purely rolling/flattening type failures when overstressed?

I recently made a 1095 blade which scratch tests to about 61-62 and while the sharp tip and heel did chip off, no matter what I do I can't get the edge to chip. It just flattens and/or rolls when I bash it against rocks. Attached is a picture of the edge damage when whacked with a 3mm carbide rod in a mouse trap. How much higher can I expect to get before chipping becomes the dominant failure mode?

Wish I could share data to help, unfortunately not enough experience with steels at different hardness levels, let alone hardening and tempering, but I wanted to ask you about the blade geometry and bevel angle.
Also, yesterday I was setting the edge of a moderately hard 1095 blade for work, and I noticed warping/flatteing/rolling was pretty much the only failure mode for the test work I wad doing (whittling twisted grain, terribly knotty softwood).
I expected this at 56 hrc, didn't expect it was pretty much the same at 60+hrc, makes me wonder what is the point of running the steel so soft at this point

I tested as low as 20 inclusive on this knife (about what is pictured) and as high as 32-35 inclusive. It didn't make any difference here as I couldn't get edge to chip no matter what, but generally lower angles will promote warping and rolling over chipping because a given amount of elongation before failure will result in an edge that is more obviously out of whack.

Cliff got Alvin's 1095 paring knife to roll even at 66hrc, because it was ground crazy thin. It chipped on him too, so apparently for that kind of work with that kind of geometry, the optimum is about as hard as possible. I'm not sure if you make knives, but if so and you're getting deformation failures I'd definitely try this heat treat. There's also the argument that the torsional peak is more important than the charpy results, but I haven't tested to confirm.

I wouldn't have expected you to get flattening against wood though, since there's no leverage for the stresses on the steel to get higher than the stresses on the wood. What's the geometry of this blade, and did you make it? Is it possible that the edge got overheated in sharpening?

Jimmy,
I'm not a knife maker, the knife I was working on was made out of a blank blade, so it was pretty much "commercially heat treated" 1095, I started the testing at zero ground, 3dps so I experienced a fair bit of plastic deformation going against small, very hard pin knots, the damage was gradually reduced going with heavier geometries, ended up about 10/12dps at a little bit less than 0.010bte.

Interesting, I would have imagined even the most ductile steels to be kind of brittle at such an high hardness, I was kind of wrong I guess.

BTW, regards to the "flattening", I should have said denting: the edge was still kind of sharp, but bent to the sides and rippled into the secondary grind.
The primary grind was damaged only in the first runs, silly geometry for the raw testing

Quote

BTW, regards to the "flattening", I should have said denting: the edge was still kind of sharp, but bent to the sides and rippled into the secondary grind.

Ah, that makes more sense. When I said "flattening" I meant that it's not rolling because there's no force pushing to either side, and instead it's just plastically deforming straight in.

For example, here's a test I did today using a carbide rod to impact the edge. It's a little hard to see in the picture, but that "missing" crescent isn't actually missing because it's not a chip. The steel just got pushed out to the side, and the it is polished where the carbide rod whacked into it.

Quote

Interesting, I would have imagined even the most ductile steels to be kind of brittle at such an high hardness, I was kind of wrong I guess.

I am quickly becoming similarly surprised.

That paring knife was so thin that it doesn't take much ductility to roll the edge out of alignment, but the tests I'm doing are showing pretty significant deformation without chipping, and at pretty high hardnesses. The knife pictured, for example, tests about 68hrc on the flats near that divot. I think I overtempered the bevel a bit by aggressive grinding (even though it was water cooled) because I can scratch the bevel with a piece of 65hrc steel. However, the edge can scratch 65hrc back so it ain't *that* overtempered.

It's worth noting that these aren't conventional heat treats though. I'm quenching into liquid sodium at 300-350f and slow cooling from there, which is similar in concept to what Bluntcut was doing in 2016 or so. On the 68hrc blade, I also carburized it to ~1.4%C and attempted to get ultra fine grains as well. I don't know if it worked, but it did turn out a couple points higher than I've ever been able to get 1095 without cryo, and it's not chipping much so maybe it did. I guess it's time to do some control blades with a regular heat treat to make sure I'm not fooling myself somehow.

Well... controls are important. I'm getting the same deformation instead of chipping with conventionally HT 1095 at 65hrc. I guess I don't understand the failure mechanisms that well.

Using a screwdriver and a hammer coming in at an angle, I can reliably chip the edges. The carbide whacker still smooshes things though, even when at an angle.

Not to state the obvious, but wouldn't 52100 and AEB-L be some of the better steels to use for this? And maybe 1095 full hardness?

Maybe hardness only have a minor role in this, as long as the grain size is very small and the carbide content is minimal seems plausible to me that steel will tend to deform rather than crack, which (I think, correct me if I am wrong) usually originate from some kind of inclusion in the steel matrix, like carbide.
If I understand correctly, max hardness 1095 should have most of its carbon content dissolved in the steel matrix, so cementite content should be very low, that could make up a bit for the loss in toughness derived from high hardness.
Just my speculation, I could be totally wrong

You’ll have a tough time defining that hardness for just one steel unless you also fairly narrowly define the use as well. Also I don’t think the torsional toughness test are more important but I do think they’re more repeatable and consistent. Charpy testing higher hardness steels has a lot of scatter in the data.

Quote
Jason
Not to state the obvious, but wouldn't 52100 and AEB-L be some of the better steels to use for this? And maybe 1095 full hardness?

Better in what way? As a useable blade, or a test specimen?

1095 is cheap, I have a bunch, it's simple and doesn't require special protection from decarb when austenitizing, has less to gain from cryo (my dewar is empty at the moment)... It's also quite brittle normally, so an increase in toughness from a better HT should be easier to pick up on. I think it should be tough as any other though, if quenched properly.

I'm not sure what you mean by "full" hardness, exactly, but my control blade was 1475f to brine, freezer, then 320f temper. More or less trying to replicate Alvin's 1095 heat treat, but mine came out at 65hrc.

Quote
me2
You’ll have a tough time defining that hardness for just one steel unless you also fairly narrowly define the use as well. Also I don’t think the torsional toughness test are more important but I do think they’re more repeatable and consistent. Charpy testing higher hardness steels has a lot of scatter in the data.

I was hoping that it'd be fairly insensitive to changes in use because the changes would affect chipping/rolling stresses equally (for a given gemoetry), but it seems not to be the case.

What's the difference between the torsional and charpy tests, anyway? Negative vs zero hydrostatic pressure? I guess that would explain the higher variance in the charpy data and also the relatively large deformation without chipping when the impacts are straight into the edge. I'll have to think some more about how the failure modes would be expected to change in the limits of low and high hydrostatic pressure.

Quote
Millscale
Maybe hardness only have a minor role in this, as long as the grain size is very small and the carbide content is minimal seems plausible to me that steel will tend to deform rather than crack, which (I think, correct me if I am wrong) usually originate from some kind of inclusion in the steel matrix, like carbide.
If I understand correctly, max hardness 1095 should have most of its carbon content dissolved in the steel matrix, so cementite content should be very low, that could make up a bit for the loss in toughness derived from high hardness.
Just my speculation, I could be totally wrong

Right, 1095 should have small carbides and not that many of them, so 1095 should have a fairly high toughness ceiling for a given hardness. In practice it tests quite low because quenches that are fast enough to beat the pearlite nose tend to be fairly harsh. Sodium quenching to near Ms should fix this though.

I've always liked when my edges fail with a bit of deformation and a bit of chipping with the intended geometry. It feels like the steel is optimized.