Grain Refinement For The Complete Idiot

Michael Kemp

Well-Known Member
Kevin,

As a hobby maker - looking to transition into part-time pro - I am not afraid to expose my ignorance in order to learn. While "grain refinement" is harped on in every knifemaking book I own, none of them give how-to specifics that I can find. And then there is also the issue of carbide control. I am aware that there are diverse attitudes about "the way, the truth, and the light" of heat treatment. I don't want to complicate this post with all that - I would greatly appreciate any response you have time to give on my understanding:

I hope to "keep it simple" by talking a particular steel - I use mostly 5160 at this point. A heat treat kiln is on my to-do list, so any temperatures I mention here are approximations in my current practice.

To normalize I heat 5160 to 1600f - soak for 10 minutes - and air cool. My feeble understanding is that this soak-at-austenizing-temp dissolves the crystalline grains and also somewhat dissolves the carbide structures. The air cooling allows some carbide re-forming (but not too large) and provides about the right cooling rate to create small-ish grain structure.

Is that fair to say?

I see some makers use "flash normalizing" or "thermal cycling" as a way to contain grain growth - going from forging temp into a quench bath for only an instant and then back to forging. Your thoughts?

To anneal (you are going to cringe - so straighten me out) I have been heating to a slightly lower temp (shooting for 1525f) - soak for 10 minutes - and put the blade into dry vermiculite. I get a nice machinable result. I'm assuming this process has created large carbides - sucking the carbon out of the rest of the metal to create more ferrite. I also assume it produces larger grain growth. When I get a programmable kiln I will set it for your recommended annealing transitions.

I have been told that annealing produces small grain size but I have a hard time believing that. Or maybe I'm doing it wrong and need to allow quick cooling to 1300f before sticking it into the vermiculite?

How am I doing so far?

For hardening I heat back to the lower temp (1525f-ish) - soak for about 10 minutes - and quench in canola oil pre-heated to 130f. I wait for the blade to be hand-touchable before taking out of the oil. Obviously the goal here is to create Martensite. I assume that I have also reset the carbide and grain sizes similar to normalizing - but with less grain regrowth - i.e. smaller grain size.

So if you have the patience - I would love to know where I am on the mark and where I am off base.

Thanks! Michael

p.s. and then a couple of tempering cycles for 2 hours at 350f or 375f - but I assume this has zero effect on grain growth or carbides.
 
Last edited:

me2

Well-Known Member
Kevin,

As a hobby maker - looking to transition into part-time pro - I am not afraid to expose my ignorance in order to learn. While "grain refinement" is harped on in every knifemaking book I own, none of them give how-to specifics that I can find. And then there is also the issue of carbide control. I am aware that there are diverse attitudes about "the way, the truth, and the light" of heat treatment. I don't want to complicate this post with all that - I would greatly appreciate any response you have time to give on my understanding:


I hope to "keep it simple" by talking a particular steel - I use mostly 5160 at this point. A heat treat kiln is on my to-do list, so any temperatures I mention here are approximations in my current practice.

To normalize I heat 5160 to 1600f - soak for 10 minutes - and air cool. My feeble understanding is that this soak-at-austenizing-temp dissolves the crystalline grains and also somewhat dissolves the carbide structures. The air cooling allows some carbide re-forming (but not too large) and provides about the right cooling rate to create small-ish grain structure.

Looks good to me. A true normalization for something like 5160 will dissolve all the carbides and reform the grains from ferrite/pearlite to austenite. Air cooling indeed keeps the reformed carbides fine. The maximum temperature of the normalizing will play a large roll in determining grain size. The cooling rate works more for the carbides.

Is that fair to say?

I see some makers use "flash normalizing" or "thermal cycling" as a way to contain grain growth - going from forging temp into a quench bath for only an instant and then back to forging. Your thoughts?

To anneal (you are going to cringe - so straighten me out) I have been heating to a slightly lower temp (shooting for 1525f) - soak for 10 minutes - and put the blade into dry vermiculite. I get a nice machinable result. I'm assuming this process has created large carbides - sucking the carbon out of the rest of the metal to create more ferrite. I also assume it produces larger grain growth. When I get a programmable kiln I will set it for your recommended annealing transitions.

This works fine for a steel like 5160. This process creates coarse pearlite and ferrite. No grain boundary carbides to be seen. Again, the max temperature will determine your grain size more than cooling rate. This might have a finer grain than your higher temperature normalizing cycle. You probably can't see it without polishing and etching, so don't break any pieces just to look.

I have been told that annealing produces small grain size but I have a hard time believing that. Or maybe I'm doing it wrong and need to allow quick cooling to 1300f before sticking it into the vermiculite?

How am I doing so far?

For hardening I heat back to the lower temp (1525f-ish) - soak for about 10 minutes - and quench in canola oil pre-heated to 130f. I wait for the blade to be hand-touchable before taking out of the oil. Obviously the goal here is to create Martensite. I assume that I have also reset the carbide and grain sizes similar to normalizing - but with less grain regrowth - i.e. smaller grain size.

5160 won't have many, if any, carbides as it is quenched. It depends on your austenizing temperature and how much control you have.

So if you have the patience - I would love to know where I am on the mark and where I am off base.

Thanks! Michael

p.s. and then a couple of tempering cycles for 2 hours at 350f or 375f - but I assume this has zero effect on grain growth or carbides.

Temper to desired hardness. I like those temperatures, but then again I'm also in favor of water cooling between temper cycles, so keep that in mind. If nothing else, it saves time. No effect on grain growth, but there is an effect on carbides, but it's a bit beyond what you're trying to accomplish I believe. There are tempering carbides that form, but they don't get talked about much in steels like 5160.
I'm not Kevin, but I'll take a stab at this. I'm in a mood this evening. See comments above in red (not bad, just visible).
 

Kevin R. Cashen

Super Moderator
He, he, he, I want to thank you guys for posting and keeping some activity going on this forum, :3: never mind me I am happy just lurking:biggrin:. I don't do too much here since I mainly got the subforum to help support Knifedogs, so it is good to see folks use it.

Michael, lately I find myself very often telling people not to sweat that they don’t have a salt bath or even a kiln. You don’t need a salt bath, you don’t need a kiln all you need is the knowledge to select to good steel for what you are doing and the knowledge to work it. 5160 is friendly enough to give you want you want with the tools you have at hand. I will even make it simpler- to normalize take it to a low forging temperature but leave the hammer on the bench and let it air cool. With more temperature you can eliminate much of the soak time. I wouldn’t be so quick to say this with 1095 or even 1084, but 5160 is much more tolerant of such heating. The idea here is to equalize grain size and dissolve any large carbide concentrations, but I need to tell you that 5160 is a hypoeutectoid steel (less than .8% Carbon) so it will not have many carbides to deal with at all. Instead, you will have some left over iron to fill with that carbon and leave itty bitty chromium carbides in. However it is very important that you don’t let these carbides have too much of your carbon, the maximum HRC you can expect from 5160 is around 63 so you need every bit of carbon freed up for hardening that you can get. Once again, don’t worry, 5160 is very easy going on these things and will not give you much to worry about.

Quenching during the forging process is entirely unnecessary for the control of grain growth if the forging is done at a correct deformation to heat ratio. When steel is above the recrystallizing temperature the deformation introduces energy that induces new grains to form, this is called dynamic recrystallization. If your rate of deformation matches your heat grain size can be controlled. In any case the grain size from hand forging will be very non-uniform and can benefit from a good normalization anyhow. By far the quickest and most effective way to refine grain is with a quick thermal cycle, the quench is not even necessary all you have to do is cool the steel, in the air, to nonmagnetic (around 950-1000F for 5160) and then reheat. Quenching will speed things up, having about the same effect of two air cools.

On your anneal- No cringing! You are just fine! 5160 is a hypoeutectoid steel it doesn’t form large carbides, or many carbides at all, when slow cooled. The slow cooling/carbide issue is for hypereutectoid steels, those having more than .8% carbon. 5160, 6150, 1070, 1075, 1080, 1084, 15n20 can all be cooled as slow as you like without carbide issues. 1090, 1095, W1, W2, O1 and others with more than .8% carbon should not be slowed cooled if you don’t want problematic carbides. The grain does not have to grow if you keep your heat lower than you normalizing and thermal cycles. True annealing by definition requires recrystallization and slow cooling so I doubt there are many instances of making grain finer, but it can make it more homogenous.

Your hardening sounds good, canola and just about any medium speed oil will work fine for 5160.

Your tempering may be a bit on the low side with the 350F but the 375F should get you around a sweet 59HRC if everything else went all right.

I though I would add this post just because it is my forum, but me2 did just fine and gave you excellent answers.
 

Frank Hunter

Well-Known Member
He, he, he, I want to thank you guys for posting and keeping some activity going on this forum, :3: never mind me I am happy just lurking:biggrin:. I don't do too much here since I mainly got the subforum to help support Knifedogs, so it is good to see folks use it.
Kevin, I've been following your work since I started this endeavor. I liken it to Neil deGrasse Tyson saying “The good thing about science is that it's true whether or not you believe in it.” I forge, I do stock removal. Quenches are in water, oils of varying quality, plate and air. You're a very patient tutor with many of us and I feel I have learned a lot and can apply your experience and documentation to my work to improve it. These recent threads are as informative as always.
 

me2

Well-Known Member
I'm glad you covered this Kevin. I was at a loss for the forge, quench, forge part. I've never really figured out what this is supposed to do.
 

Michael Kemp

Well-Known Member
me2 & Kevin - Thank you for sharing knowledge & experience. It is appreciated.

I'm glad to hear I'm doing OK with the 5160. FWIW I have also forged D2 (which is hypereutectoid - and rather tricky for me to forge - but for some reason I like D2 and expect to forge more) and I am moving into Damascus (I have some 1084 & 15N20 sitting in the corner calling my name) so my experiences will vary.

My "intuition" was that length of time cooling would have more impact on grain size than peak temperature - but from what you've shared I'm getting the message that slowness of cooling is not too important for steel grain size, but it /is/ important for carbide grain size in hypereutectoid steels. Peak temperature - especially if there is any soak time - causes large steel grain growth - - - if I'm understanding correctly.

You refer to "recrystallization" temperature - is this where (on a phase diagram) Austenite starts to precipitate ferrite (hypoeutectoid) or cementite (hypereutectoid)?

Kevin - Re: "thermal cycling" or "flash normalization" (which I assume are two names for the same thing). It sounds like this process can even up the grain size (like true normalization) but you also say it is not needed if forging is done at a proper temp for the steel being forged. But for hand-forging, I'm coming away with the impression that this is a good thing to do.

Annealing - I'm taking your responses as being: this does NOT reduce grain size.

Normalizing - I'm taking your responses as being: the peak temp (& soak time?) determines grain size. High temp & long soak = large steel grain.

I have to admit, I'm still confused: How do you *reduce* grain size?
 
Last edited:

me2

Well-Known Member
me2 & Kevin - Thank you for sharing knowledge & experience. It is appreciated.

I'm glad to hear I'm doing OK with the 5160. FWIW I have also forged D2 (which is hypereutectoid - and rather tricky for me to forge - but for some reason I like D2 and expect to forge more) and I am moving into Damascus (I have some 1084 & 15N20 sitting in the corner calling my name) so my experiences will vary.

Forging D2? You're braver than I. Any chance you'd try stainless?

My "intuition" was that length of time cooling would have more impact on grain size than peak temperature - but from what you've shared I'm getting the message that slowness of cooling is not too important for steel grain size, but it /is/ important for carbide grain size in hypereutectoid steels. Peak temperature - especially if there is any soak time - causes large steel grain growth - - - if I'm understanding correctly.

Soak time has a much smaller effect than temperature w/r to grain growth. There is a certain temperature above which grains will grow very quickly. Below this it is very slow, like exponential increases in time (1 hr, 2 hr, 4 hr, 8 hr, etc) will just give slight changes in the ASTM grain size. This temperature is alloy specific and is also influenced by the grain size you start with, as finer grains grow at lower temperatures, though it's not as severe as some think.

You refer to "recrystallization" temperature - is this where (on a phase diagram) Austenite starts to precipitate ferrite (hypoeutectoid) or cementite (hypereutectoid)?

The recrystallization temperature is not shown on the phase diagram, unless it's added by whoever made the diagram. The temperature you are referring to is a transformation temperature. The recrystalization temperature is when a metal that has been cold worked will reform the grains. When cold working, the grains get stretched out in the direction of the work. At a high enough temperature, they will reform into more uniform grains. As usual, this temperature is variable, and depends on the alloy, and how much cold work was done.

Kevin - Re: "thermal cycling" or "flash normalization" (which I assume are two names for the same thing). It sounds like this process can even up the grain size (like true normalization) but you also say it is not needed if forging is done at a proper temp for the steel being forged. But for hand-forging, I'm coming away with the impression that this is a good thing to do.

Annealing - I'm taking your responses as being: this does NOT reduce grain size.

It can. It depends on what size they are when you anneal. If you normalize at 1650 F then anneal at 1475 F, they will get smaller.

Normalizing - I'm taking your responses as being: the peak temp (& soak time?) determines grain size. High temp & long soak = large steel grain.

I have to admit, I'm still confused: How do you *reduce* grain size?

Grain size can be reduced each time you form new grains. New grains form when there is a phase transformation (austenite to ferrite as above), or during recrystallization of cold worked metal. Basically, you can overheat your steel, blow the grain up, cool it down, and give it a few transformations from austenite to ferrite and back, and you can fix it. There is a point of no return, and high alloy steels respond differently (your D2 vs. your 5160), but you can over shoot the temperature by quite a bit and still fix things. One experiment took some 5150 steel to 1650 deg F then cycled it a couple times above and below the austenite/ferrite transformation temperature and still came up with a very fine grain size. Rapid heating and cooling work best and the finer grain you start with, the more effective the cycling is.
See red comments above. I really feel like I should change, but other colors don't show up for easy reading. I just keep having flash backs of the first report I gave my branch manager. He gave it back for correction and it looked like he'd sacrificed a chicken over it.
 

Michael Kemp

Well-Known Member
me2 - much appreciated. I might try stainless - but it's not on the top of my list - welded cable and bar stock pattern welded steel (aka Damascus) are more at the top of the to-do list.

So if I'm FINALLY getting it - to reduce grain size you heat to just into austenizing temp without hitting the higher grain-growth temp and then back to ferrite/cementite. This thermal cycling - whether fast or slow - is what reduces grain size. I'm imagining this as breaking existing grains down to their smallest size as the transformation occurs at the critical temp (Ac3 - which may or may not correspond to when steel goes non-magnetic) some degrees above decalescence. If you overheat, large grains suddenly form. BUT - if you get back to the lower transformation temp (Ar3 - which if I understand right corresponds with recalescence) without hitting the high temp then you will lock in the smaller grain size.

Another thing to fix in my old brain: At some higher temperature (unique for each steel) grain size suddenly increases with little regard for soak time.

I probably don't need to think about recrystallization temp unless I get into cold working.

I can find a number of sites where the composition of various steels are listed (Niagara Specialty, Kevin's site, ZKnives...) and for 8 common knife steels Kevin has specific heat-treat advice (THANK YOU KEVIN!) - but other than Kevin's notes - specific heat treat info seems hard to find.

I'm assuming that the generic steel phase diagram is only valid for simple steels - and the addition of Chromium, Molybdenum, Silica, etc. makes the diagram invalid for more complex steels.

Is there any central source for various steels' phase change temps?

Any central source for the grain growth "danger" temps for various steels?

I've seen the book "Steel Metallurgy for the Non-Metallurgist" touted. I've been holding off on spending the $100 - but if it has this information in it (specific temps for specific steels) I'll pull the trigger today rather than leaving it on my wish list.

I have certainly heard that thermal cycling reduces grain size. But then again I've heard that annealing reduces grain size and normalizing does not (I now gather this ain't necessarily so). I was confused about soak times, temps, and grain growth. I am *very* thankful for the time you both have taken to untangle my thinking.

I suppose part of my "intuitive" trouble was my room-temperature experience noting the growth of salt or sugar crystals - where crystal growth is totally dependent on time.

The picture you paint of steel grain growth is completely different - dependent on critical temperature ranges with very little consideration of soak time.

I'm indebted to Kevin for posting heat treat temps for 1084, 5160, and half a dozen others. Hopefully there is some central listing of these phase change temps for some of the other knife-maker steels - - - and a knowing the grain growth "danger" temp for the various steels would be a comfort!

Many thanks!

~ Michael
 
Last edited:

Michael Kemp

Well-Known Member
I went back and re-read Kevin's post on Normalizing. This is making more sense thanks to the above notes from Kevin and me2.

I find myself still wishing for a central source for these temperatures to supplement the heat treat info on Kevin's website. And if there are ways to detect those temperatures in a regular knife-maker forge where you do not have the precise temp control of a digital kiln.

After the full normalization cycle you have all of your grains a nice and even size but now you may want to focus on making that size smaller, for this you can follow the initial normalization with some other heats that do not exactly fit the true definition of normalizing but will work hand in hand with it.

The first cycle can be followed with subsequent heats involving decreasing maximum temperatures before cooling. The next could fall in line with the appropriate hardening temperature for that steel to leave carbide untouched but reduce grain size. This in turn could be followed by an even lower heat to initiate yet another even finer grain set with no chance the grain enlargement; it is worth noting however that eventually there is a point of diminishing returns as the grain coarsening temperature drops in conjunction with size.
 

me2

Well-Known Member
The temperatures where grain size will really grow is typically up in the forging range, 1650+ depending on alloy. I've seen some steels heated to 2200+ that don't really show any grain growth out of control, but they are highly alloyed and have to have that temperature to harden properly.

The salt and rock candy growth is pretty much the same thing. The biggest difference is that those are grown in a liquid (if this is the same thing I'm thinking of). It's much easier to move the crystal molecules in liquid than it is to move carbon and other atoms in a solid. They just can't move that far that fast.

Try searching for Verhoeven's book on Metallurgy for People Who Forge and Heat Treat Steel. I have a copy I can email you if you can't find it.

The grain growth above the growth temperature isn't instant, but it's pretty fast. Steels with lots of alloying and carbides slow it down. Here's a video model in 2 D

http://www.youtube.com/watch?v=J_2FdkRqmCA

Faster heating and cooling results in smaller grain size. Say heating in an oven vs heating in a liquid lead bath (not recommended). Rapid heating provides more, smaller grains. Rapid cooling gives a smaller structure when going through the transformation temperatures. These 2 things build on each other, and also amplify with each thermal cycle.
 

vapor1898

New Member
To normalize I heat 5160 to 1600f - soak for 10 minutes......i`m a newbie...what do you mean by soak
thx
edit-never mind, found out 2-21-2013
 
Last edited:
Top