The short answer- fine grain increases the rate of nucleation in pearlite and thus lowers hardenability.

The long, detailed, answer:

Martensitic steel is the hard stuff, and is our goal in hardening, the more of it we have the harder the steel. Martensite is also very unique among the steel phases we work with in that it is not a matter waiting for carbon to move to accomplish our goals, carbon doesn’t move in martensite formation, the entire crystalline array does, instantaneously. This diffusionless nature of martensite means that it plays no roll in the question you ask Gary, it is the end result of good hardenability, not the cause.

That prelude was to establish a difference between cause and effect to help support the concept I now suggest- the secret to hardening steel is to not let it make anything but the hard stuff.

Every process we work with in this concept is driven by the need to go from states of higher energy to lower energy and thus equilibrium. Steel really prefers stable, low energy, equilibrium so if you allow it to cool at a natural rate the carbon will separate out and make carbide and soft ferrite, this layered separation is what we call pearlite. On cooling, after you reach a bit below 1,200F it is very “unnatural” for the carbon to be in solution and it will seek equilibrium by making the iron and iron carbide separation of pearlite. Also below this there is some diffusional hijinks from upper bainite until the lower temperatures retard diffusion enough to outpace any more transformations.

With normal air cooling of basic carbon steels, these diffusional transformations can be total, so that there is nothing left to form martensite, this is why normalized 1075 or 1080 is soft. So the whole purpose of the quench is to merely defeat these diffusional transformations, not to make martensite; this is why marquenching works- you only need the quench to avoid pearlite and bainite, after that martensite is a done deal.

Now to the question- remember equilibrium and energy states. The pearlite transformation begins at a point of high energy, like a corner where several grain boundaries come together; any sharp corner or lattice defect offers another place where the pearlite transformation can initiate. Large grains= fewer grains, so there are less corners.

Let’s play a game, I will give you several shapes on a paper and your job is to fill them as quickly as possible with one continuous line with your pen, you can curve and wiggle and double back on yourself but you cannot lift the pen… but you can get help. But here is the catch- you can only put your pen down and begin at a corner. The first shape is a circle, and you are done before you begin as there are no corners, it is the lowest energy shape possible with absolute equilibrium. Next comes a triangle and immediately two friends can join in with simultaneous lines, the task is three times quicker. Next a square and by the time you get to something like a decagon, there are so many lines started that the same time that the shape is filled as soon as you start. This is exactly what happens to pearlite formation as you decrease grain size and increase the number of intersections in grain boundaries. So with every order of grain size reduction your required quench speed increases in order to fully avoid pearlite and make as much martensite as possible.

To see it in action on a blade, take one made from 1080 and fully quench it in, let’s say, peanut oil. When you clean it up you may notice a natural hamon 2/3 up the blade. Now turn right around and quench it again and you may notice the line is only ½ up the blade, and repeating the process you will chase it continually closer to the edge. This is due to the increasing points of nucleation (the spot where you can put your pen down) for pearlite in the finer grain requiring faster cooling to be outpaced, but that same cooling limited by the thickness of the bevel.

Generally finer grain size is better for certain properties, hardenability is not one of them, and even good things can be overdone. Hardness is great for a knife, but if we go with the idea that there is no point where enough is enough, we will have chippy and brittle blades. Toughness is great as well, but if we make it our sole focus we will lack strength and edges roll. The reason we have not created unobtanium yet is that everything, particularly with steel, is a give and take and a careful balance of many things to suit our given application. Believe it or not, for machinability and other applications larger grain can be better.

For more information see the work of McQuaid and Ehn, whose names are literally synonymous with grain size as well as Marcus Grossman who developed very specific formulas to determine cross sectional hardenabilites based on grain size alone.

“The Hardenability of Steels” by Siebert/Doane/Breen (ASM)

“Elements of Hardenability” By M.A. Grossman (U.S. Steel Corp)

Kevin,

That was a long & precise answer and I apologize for asking you to go through all of that but that helps a lot in understanding the process.

Thank you for taking the time to explain.

Gary

Kevin: Great reply as usual, and one that certainly makes me better understand what I'm up against. I've been working on some JS performance knives lately with mixed success. I'm using 5160, for now. First few blades failed bend test as I expected they would and gave me a decent starting point. Most of them cut, chop and shave no problem. The bend test is not reproducible. First couple of them went to about 89 degrees and snapped, one I was able to bend back and forth well past 90 ten times before it broke. I thought I was there. I couldn't reproduce it.

The wonderful inter-web has blessed us or cursed us with all sorts of information. I have read as much as I can find on "how to make a blade to pass the JS test". Ask ten people, get ten answers.

I have been edge quenching the blade about 1/4" to 3/8" and it is successful. The cutting, chopping and shaving have proven that. I think my issue is not the edge but what is happening above the edge and resulting in snapping. I had been heat treating at 1525 and tempering at 350. After reading Machinery's Handbook, I dropped the HT temp to 1450. That was the most successful blade yet.

Ah the question? If edge quenching, keep in mind the entire blade has been heated, not just the edge with a torch: edge gets quenched, what's next? Leave it to cool with just the edge in oil, spine in air until cooled completely (which is how I did the better blade)or submerge whole knife in oil after quenching edge and spine has dropped to maybe 1000 deg?

Many people have told me draw the spine. Yup okay. Well there just has to be good and bad ways of doing that as well. Is correct procedure to basically paint the color to a blue/blue purple springy color, or heat that damn spine as hot as you can get it and make that edge dead soft (edge submerged in water of course)?

I think my blade geometry has been good with smooth tapers. I'm starting with 1/4 x 1.5". Spine thickness at three inches seems typically runs 0.1" to 0.11"

I intend to try using some 1084 as soon as Aldo gets some in stock. Maybe me and 5160 just don't get along.

This certainly has been part of the journey and I have learned a lot about what works in my shop.

Thanks for your help.

Hello David. 5160 is used by a lot of test applicants and works well. My recipe is normalize three times, grind to 120, heat to 1500, quench, then temper at 400 two hours, then temper at 400 again. I will then do the finish grind and draw the spine at least twice using a water bath to have the edge in. I use a propane torch for spine drawing, yes it is slower, but you do not have the possibility of over heating the spine as you do with a O/A torch. I have done both edge quench and full quench with this method with good results. What quenchant are you using?

Brion

Thanks Brion:

Maybe I'll try upping my temper to 400. I've been using McMaster-Carr fast quench. My edges are good, it's above the edge I think is the issue.

When you draw the spine are you heating as hot as you can get it as close to the water line as you can get? I've been using a MAPP gas torch, don't have O-A.

I must say I'm a little hesitant to change steel as it has worked so well for so many for so long.

Dave

David, the fast quench should work fine. When I draw the spine, I have a large water container with a limiting plate in it. This means I can adjust the depth the blade goes in. I usually let the water come about 1/2 to 3/4" up the blade edge. I draw the spine to blue grey and the color will come to about a 1/4" above the waterline. I usually draw it three times, cleaning the color off with the grinder between each time.

Brion

Hello Dave, indeed a search of the internet will reveal that advice in this topic is plentiful, there is a whole lot of assumption based on observation or anecdotal evidence, but what is exceedingly rare is an approach based on actual application of the physics involved in the challenge of making such a blade possessing basic properties required.

I have started to write a reply to this post no less than three times, and every previous attempt expanded too easily into a full page that I decided to save them for larger projects like articles that I have planned. So I am going to try to condense this down into the basics of just your questions.

There are two ways of approaching the bend portion of the ABS test, one uses flexibility, the other uses ductility. One allows the blade to go to 90 degrees with the least amount of permanent deformation and returning closer to true, the other relies on ductility to avoid any brittle failure but, by definition, includes a lot of permanent deformation. Essentially one approach makes a spring, the other approach makes a Japanese sword.

The spring approach should draw heavily on geometry and involves adjusting hardness levels of a single steel phase, while the ductile Katana thing involves two zones of different phases (martensite and pearlite) to provide two diametrically opposed properties, strength and ductility, in the same blade. I won’t touch in things like grain size and proper heat treatment, because they should just be a given, if one doesn’t have control of them they really aren’t ready to test.

The first step to success is in steel selection based upon the approach we wish to take. Brion is correct that 5160 has been used for a lot of test applicants, it is best suited for the flexibility approach and many folks like the idea of the blade returning as close to true as possible. I have also found that propane’s slower heating gave a much more consistent and thorough draw on the spine. However, many make the logical fallacy of concluding that because 5160 has been chosen by more people for the test that this means it is the best choice for the test, when all it says is that more people chose it. Despite its popularity 5160 is not the best choice for the ductile approach to the bend, and its inherent chemistry is designed to fight your efforts in that area. Choosing 5160 for the spring approach makes life easier, choosing it for the ductile bend approach creates more of a challenge for yourself.

In order for the ductile approach, via differential hardening, to be maximized you want the greatest contrast possible between the hard edge phase and the ductile spine phase, this is clearly evident in Japanese swords by the hamon. You need the very edge to be hard enough for the 2x4, but you want the main body of the blade to be as completely soft phased as possible. For this it will just make sense that the steel which will make the best hamons will also give you the best strength/ductility contrast, and you will also be able to see your success, before you bend, right there on the side of the blade as clear as if I used my microscope. Thus while 5160 will make a great spring for the flexible approach, 1075, 1080 or 1084, will be a natural for the ductile bend.

5160 has chromium intentionally added for the very purpose of defeating partial hardening, it is literally designed to oppose your efforts to differentially harden it. And when we force such a heat treatment in it, it can throw things in our way that most folks aren’t aware of.

In this IT-curve from my site, you will see that the chrome in 5160 greatly suppresses the formation of pearlite (the soft stuff you want in the spine) from 1,250F to 1,100F but leaves a protruding nose of upper bainite immediately below. Upper bainite (very different from lower bainite), in scientific terms- really sucks! It offers the worst of all properties, low hardness and brittleness. Air cooling 1075 will get you solid pearlite, air cooling 5160 gets you a mixed up mess of several possible phases with less predictable properties.

Full quenching 5160 avoids all of those mixed structures. Then drawing the spine back with the torch produces a more homogenous structure that is hard at the edge and tough at the spine, it is the best choice for the flexible spring approach to 90 degrees. Edge quenching 1075,1080, 1084, readily produces two distinct phases which give us both strength and ductility, for easy bending without breaking, in the same blade.

There is one cautionary point I would give on your current recipe for the 5160. I would not go with that 1450F temp at all, I don’t know if that Machinery's Handbook gave this temp specifically for 5160 or what application it was for if it did (remember that the rest of the industrial world does not think of knives when they talk about 5160). Denying the blade of full solution will indeed help with bending but for optimal edge performance on a real using knife I would not harden from less than 1500F for this alloy. Simply cutting through two 2X4’s will not reveal what the edge is missing over the long haul, I have seen that done with well treated mild steel, but there is a whole lot of ferrite to engage in the hardening process with this steel and you need that heat to do it.

Kevin: Thanks very much, that certainly explains a lot of things. I think I have been treating the 5150 as if it's 108X steel and wondering why it wasn't responding the way I wanted it. Also explains why I couldn't reproduce a successful bend, the 5160 was acting unpredictably the way I was treating it.

I have a much clearer picture now of how I must proceed.

Thanks for your help and also to all other ABS members that make this forum so awesome.

Off to the forge...

Dave

Good luck; David, I should say good forging!

Russell

Kevin, A Great Post!

|quoted:

...air cooling 5160 gets you a mixed up mess of several possible phases with less predictable properties.

Well, I've thought about this for a while and it looks to me like normalizing/grain refining 5160 is going to get a person "a mixed up mess of several possible phases" with a smaller grain size. Is this where quenching for grain refining is viable?

Mike

|quoted:

Well, I've thought about this for a while and it looks to me like normalizing/grain refining 5160 is going to get a person "a mixed up mess of several possible phases" with a smaller grain size. Is this where quenching for grain refining is viable?

Mike

Mike, the thread, or at least my post is in regards to edge quenching 5160 for final hardening rather than for normalization purposes. Edge quenching by definition involves a mixture of phases, but with an alloy that has a pronounced upper bainite field, or at least more accessible than the pearlite field, there is but one more phase added to that mix. For mere grain refinement, the final phase is less relevant than its size, and even less so than the subsequent resulting austenite grain size. For example, most often, in the steels we work with, normalization results in pearlite, a phase that is to be avoided at all costs if we want a fully harden blade. All previous heat treatments are merely setting things up for that final quench and temper, thus the final treatments are most critical based upon the fact that we will have to live with them evermore. Phases are entirely replaced with new heating-cooling cycles, but things such as carbide distribution and nucleation parameters may survive. The main danger I have observed in focusing too tightly on grain size is when it is paid for in sacrificing carbide size and distribution, something much more important to a fine and stable edge.

|quoted:

Mike, the thread, or at least my post is in regards to edge quenching 5160 for final hardening rather than for normalization purposes. Edge quenching by definition involves a mixture of phases, but with an alloy that has a pronounced upper bainite field, or at least more accessible than the pearlite field, there is but one more phase added to that mix. For mere grain refinement, the final phase is less relevant than its size, and even less so than the subsequent resulting austenite grain size. For example, most often, in the steels we work with, normalization results in pearlite, a phase that is to be avoided at all costs if we want a fully harden blade. All previous heat treatments are merely setting things up for that final quench and temper, thus the final treatments are most critical based upon the fact that we will have to live with them evermore. Phases are entirely replaced with new heating-cooling cycles, but things such as carbide distribution and nucleation parameters may survive. The main danger I have observed in focusing too tightly on grain size is when it is paid for in sacrificing carbide size and distribution, something much more important to a fine and stable edge.

Thank you, Kevin... I was thinking the multiple phases would cause differences in grain size (regardless the relative size).

Mike