Thanks to Kevin Colwell, Prentice Goodwin, HallMade Knives, and Saint Apocryphal for becoming Knife Steel Nerds Patreon supporters! And Fixall for increasing his contribution amount.
YouTube Video
There is a video version of the following content:
Background of Austempering and Bainite
I previously wrote about austempering and bainite in this article: Bainite vs Martensite – The Secret to Ultimate Toughness? And of course there is a chapter about it in my book Knife Engineering. In normal heat treating of steel we heat the steel up to high temperature and then quench rapidly to room temperature to form martensite. And then temper at a temperature such as 300-500°F to improve toughness at the cost of a little hardness. With bainite we do a single step heat treatment where after heating to high temperature we quench to an intermediate temperature like 400-600°F and hold to form a phase called bainite. Bainite is sort of like tempered martensite; it is difficult to differentiate between the two even with electron microscopy. So this process, austempering, is somewhat like forming pre-tempered martensite.
However, there are important differences, and greatly improved toughness has been reported in some studies, such as the following on 1095 cited in the ASM Handbooks:
With high carbon martensite there is a switch in the type of martensite that forms, from “lath” martensite up to about 0.6% carbon to “plate” martensite at 1% carbon. And in between there are increasing amounts of plate martensite. Plate martensite has poor toughness and can even have microcracks within it which makes it undesirable.
A steel that has 1% carbon doesn’t necessarily have all of that carbon “in solution” in austenite because we can leave some carbide in the steel. The higher the temperature we austenitize, the more carbide dissolves and the higher carbon our austenite is. Learn about austenitizing in this article. It is the carbon content of the austenite that controls the formation of plate martensite during quenching. However, with low alloy steels with greater than about 0.85% carbon it is difficult to heat treat them without having at least 0.6% carbon in solution. You can read more about the toughness of low alloy knife steels in this article. With bainite, however, there is no “plate bainite” so with high carbon steels we often see the biggest jump in toughness. This was observed in previous research [1] where they found that the toughness difference between bainite and martensite increases with higher carbon content.
Another reason for the toughness difference with bainite is “tempered martensite embrittlement.” Bainite is typically lower in hardness than martensite tempered at 300-400°F. To match the hardness of bainite we have to temper in the range of 500°F or higher. This puts the steel in the embrittlement range where toughness is reduced from forming large carbide particles. So to compare bainite and martensite 1:1 at the same hardness it often requires tempering martensite in an unfavorable temperature range.
Data from [2]
1095, O1, and 52100
O1 and 1095 both have shown relatively low toughness in my previous testing, likely because of plate martensite. 52100, however, despite having a similar carbon content shows much greater toughness. The 1.5% chromium addition reduces the carbon in solution for a given austenitizing temperature and makes it easier to avoid plate martensite. Read more in this article about 52100. This makes 52100 a good comparison point because we can avoid plate martensite to see if that is the cause of improved toughness in bainite. 1095 is a simple carbon steel with 0.95% carbon and small amounts of Mn and Si. Steels like it have been around basically ever since the invention of steel. O1 was the first “oil hardening” steel developed in 1905 which you can read about here. Some of the steel I used was donated by Alpha Knife Supply and New Jersey Steel Baron.
1095 quenched from 1475°F
O1 quenched from 1475°F
52100 quenched from 1500°F
Fortunately for us there are TTT (Time-Temperature-Transformation) diagrams available on these three steels so that we can have some idea about how to austemper them for bainite.
There are several important aspects of these diagrams we will be using. One is the martensite start temperature (Ms) which is the temperature at which martensite begins to form. If we want full bainite we have to austemper above that temperature. If we austemper below Ms we can still form bainite but there will be some amount of martensite that is also formed. This isn’t necessarily bad, but we have to keep it in mind. The other important aspect of these diagrams is the start and finish points for bainite formation. For example, if we look at O1, austempering at 500°F the bainite formation begins at about 10 minutes and finishes at about 2 hours. So to have full bainite we need to austemper for at least 2 hours. You can hold it longer to ensure that the transformation is complete if you want without negatively affecting the steel. Continuing to hold at the austempering temperature after the transformation is complete will just mean more tempering of the bainite, but the hardness drop from tempering is relatively minor in bainite. If we don’t austemper long enough there will be some martensite that is formed in the remaining austenite during cooling to room temperature after the austemper. Adding a tempering step after austempering ensures there is no “untempered” martensite. Another aspect of the diagrams is that hardness measurements after austempering is listed to the right, such as 60 Rc for austempering O1 at 400°F, or 56 Rc for austempering at 500°F.
There are a couple complications of simply using the above diagrams, however. One is that the composition of steel can vary some from batch to batch or manufacturer to manufacturer. For example, the 1095 TTT diagram above is for 0.89% carbon 1095, which is actually outside of the accepted AISI range of carbon for 1095. If instead we had a higher carbon 1095 such as 1% carbon, the Ms temperature may be lower from more carbon in solution and the time for transformation is likely also shifted. Below is a TTT diagram for 1.14% carbon W1 which would also be just outside of the 1095 specification but on the high side.
Another complication is the austenitizing temperature as mentioned previously. A higher austenitizing temperature means more carbide is dissolved, higher carbon is in solution, and the Ms temperature is reduced. In some cases there is also grain growth with high austenitizing temperatures which increases “hardenability” which makes bainite formation take longer. As an extreme example below is the TTT diagram for 52100 austenitized at the very high temperature of 1950°F. The Ms temperature was reduced from ~475°F from a 1555°F austenitize to 300°F from 1950°F. And the time for transformation was increased from 15 minutes at 500°F to 15 hours by austenitizing so high.
I wanted to compare austempering at 400°F from 1550°F (with 52100) where there was some martensite formed prior to bainite, and also an austemper at 400°F which was 100% bainite. That requires we austenitize higher to reduce Ms, without austenitizing so high we are getting significant grain growth or other issues. Using various sources I estimated the required austenitizing temperature as 1625°F.
The Heat Treatments
I did a range of heat treatments by austempering at 400 and 500°F. I wanted to do as many heat treatments as possible from the relatively low temperature of 400°F because knives need high hardness. There are not a lot of studies on austempering in this range when comparing toughness and other properties. The two different 400°F austempers on 52100 were done for the reasons explained in the previous section. These heat treatments were performed with an EvenHeat salt pot furnace using low temperature salts.
O1 Steel
So now we can compare the toughness that we got from these heat treatments with previous quench and temper treatments. Let’s start with O1; below are the previous Q&T treatments that I have done with the steel, some of these were previously reported in this article.
You can see that with previous heat treatments using a lower austenitizing temperature consistently led to improved toughness at comparable hardness, with 1425-400 leading to the best properties. Likely higher hardness could be obtained by combining the 1425°F austenitize with reduced tempering such as 300-350°F. This is because of the reduction in plate martensite as explained previously. The 1380°F austenitize led to very low hardness with little improvement in toughness, however. The low hardness is probably from not fully austenitizing (some ferrite remained). I had austenitized for 30 minutes to try to overcome this, but at low temperatures it takes longer for the steel to heat through and longer for the austenite to form. But even if 60 minutes would do it, the 1380°F is probably dancing too close to the line. So now that we have our baseline established, let’s compare with the austempered conditions:
You can see that we successfully increased the toughness, even significantly above the 1425°F austenitize. This likely means that even the 1425°F austenitize led to some plate martensite. Another factor is the amount of carbide dissolved during austenitizing. The higher austenitize (1475°F) used when austempering means that we dissolved more carbide which may have also helped to improve toughness. The 58.4 Rc achieved with the 400°F austemper was somewhat lower than the hardness reported on the TTT (60 Rc), but the 500°F hardness was similar: 56.6 vs 56 Rc. Based on these hardness results I certainly wouldn’t austemper any higher than 400°F. Even 58.4 Rc is on the low end for most knives.
1095 Steel
With 1095 we also saw an improvement in toughness by austempering, but the toughness was much lower than with O1. In this case we had hardness somewhat higher than was reported on the TTT diagram for the 0.89% carbon steel but lower than for the 1.14% carbon W1, as we might expect.
I’m not sure why the toughness is so limited with this 1095 even with austempering. Perhaps a steel with better “cleanliness” or prior processing would lead to improved toughness. I can’t say without comparing different 1095 steels from different sources.
52100 Steel
With 52100 we had done a range of different heat treatment and toughness experiments which were reported in this article. There are a lot of different comparisons there but I have simplified the results to the best of the 52100 quench and temper (Q&T) results and overlaid the austempering results.
The toughness improvement with 52100 is much smaller than with the previous two steels. This seems to confirm the hypothesis that it is plate martensite that results in the difference in toughness between martensite and bainite. With the 1625-400 heat treatment the toughness is very similar to the Q&T conditions. Perhaps we got some grain growth with 1625°F limiting the toughness of the bainite. The 1550-400 and 1550-500 austemper conditions had reduced hardness but improved toughness vs the Q&T and 1625-400 conditions. These 1550°F heat treatments look very much like what would happen when tempering 52100 at 450-550°F if tempered martensite embrittlement was not a reality. So again this shows that bainite is best for intermediate hardness levels where with martensite it would require tempering in the embrittlement range.
The 1550-400 condition is somewhat interesting because we should have some tempered martensite present in that one, about 50% according to the TTT. However, the toughness appears similar to the 1550-500 condition with pure bainite if it was at higher hardness. So there does not appear to be an issue with austempering below Ms if higher hardness is desirable. It might be interesting to try low temperature austempering treatments at 350°F, and with other steels like O1. However, the data available for sub-Ms austempering is very limited and would require more experimentation.
Comparisons with Other Q&T Steels
It probably makes sense to look at the toughness of these steels in the broader picture with other steels that were quenched and tempered. Below I have charts of toughness for other low alloy steels as well as high alloy non-stainless and stainless steels.
The 30-35 ft-lbs of the austempered 52100 is excellent, though steels like L6, 5160, and 8670 achieved better toughness at similar or higher hardness. Those steels all have lower carbon to avoid plate martensite and also have little or no carbide. So 52100 likely has better wear resistance from its carbide. Expanding our steel comparisons to high alloy we find 420HC, AEB-L, Caldie, 3V, A8 Mod, and Z-Tuff also achieving toughness in the 35+ ft-llbs range, and many of those at over 60 Rc. Those high alloy steels also have the advantage of harder chromium or vanadium carbides for similar or better wear resistance. There are many steels with more carbide and lower toughness, of course. 30+ ft-lbs or even 25+ ft-lbs is probably more than sufficient for most any knife. Only very large blades or swords would you likely need more. The high alloy steels are usually easier to austenitize where they don’t have plate martensite because the higher alloy changes the austenitizing behavior. Sort of like the chromium addition to 52100 but even more extreme.
Austempered O1 being in the 20-25 ft-lbs range is much more impressive than with previous Q&T heat treatments. That makes it more competitive with other steel choices, though at the relatively low hardness of 56.6-58.4 Rc from austempering. 1095 was still relatively disappointing with its ~13 ft-lbs. Again I’m not sure what is keeping the toughness low in that steel.
Should You Austemper?
I’m a bit divided on that question because it is possible to have improved toughness with bainite. However, the potential hardness is also somewhat reduced when compared with Q&T. I tend to prefer a bit higher hardness because of the benefits to strength and edge retention. With knife edges it is often deformation that occurs before chipping, even at 60+ Rc. There is also some required investment to be setup to work with low temperature salts, as opposed to using typical quenching oils. Utilizing steels that avoid plate martensite like 52100, 1084, and 80CrV2 can be a cheaper and simpler method to achieving an excellent combination of hardness and toughness.
Conclusions and Summary
Austempering for bainite can lead to improved toughness because brittle plate martensite and tempered martensite embrittlemt is avoided. When using austenitizing/steel combinations that avoid plate martensite the potential improvements are much smaller. In those cases bainite can still have specific use cases for lower hardness steel (56-58 Rc) that is tougher by avoiding plate martensite. I am divided on whether austempering is something I would generally recommend, as choosing steels that avoid plate martensite may be a simpler way to achieve a good combination of hardness and toughness, and I think somewhat higher hardness (60 Rc+) is generally more desirable in a knife.
[1] Niccols, Edwin H. Literature Review: Impact Toughness of Bainite vs. Martensite. No. WVT-TR-76012. WATERVLIET ARSENAL NY BENET WEAPONS LAB, 1976.
[2] http://www.crucible.com/eselector/prodbyapp/tooldie/champloy.html
Hi Larrin
Very interesting and informative article. It looks like a lot of different variables can affect the outcome.
There would be potential for retained austenite especially with the higher hardening temperatures you used.. and was curious if the austempering process can effectively eliminate retained austenite and get a full conversion to bainite.
The data I found in that area showed extremely low values for retained austenite in low alloy tool steels. It could be that that changes with sub-Ms austempering but that is not a greatly studied area. I agree that if we are getting significant RA that might make the austempering less appealing. With high Si contents we would definitely be stabilizing some RA.
I’ve done some searching on the internet and haven’t come up with much on this subject. It could be that austenite doesn’t require as much driving force to convert to bainite as it does to martensite..and that might account for the low values of RA you found in the data. The necessary strain energy to get full martensite would be lost as the quench is interrupted above the ms start point.. so maybe It’s possible that at or below the ms point a guy might have some high levels of RA to deal with. Very interesting subject, there’s a lot I don’t know here. If you ever get this figured out, I would be interested to know what you find. Thanks for the article and the comments.
CATRA? Would be interesting to see the difference.
One thing at a time. Hopefully we will get to it. My guess is no difference; we will see.
Interest has risen. Can’t wait. Just multiquenched today a thick D3 chopper. I’ll match it with one more like it and multiquench it but the last quench will be slowed to bainite. Then, chopping oak will reveal the difference.
Thank you for the article, I’ve read articles on bainite before and rever really grasped how it fit into the real world heat treatments.
One suggestion, I think combining the two O1 graphs into one with colored points would be clearer than the two similar ones.
It certainly does seem like the process would be ideal for swords. Was this ever actually done or was the sort of precision long term temperature control required (one can imagine blundering into the technique without actually understanding the science) more or less impossible in historic times.
I’m not aware of any historical swords with a majority-bainite microstructure, I think it was mostly a modern discovery. I’m not an expert on ancient swords, however.
While Im certain they didnt know the microstructural reasoning, gunsmiths have long quenched mainsprings in molten lead and left them submerged for extended periods. This pre colonial era practice probably came into popularity because it produced bainitic steel springs that where incredibly strong and reliable. How far back does the practice go? I have no idea, but prior to gunsmithing, crossbow prods must have been produced by some means that didnt result in inevitable catasrophic failure and injury…
Very interesting. Can you share the source?
I dont know that I have one off the top of my head aside from an old film made from the colonial Williamsburg gun shop. Here is a link to the video on youtube https://m.youtube.com/watch?v=X_O1-chxAdk
It features Wallace Gustler, and details the traditional techniques used in the era for long rifle production. I have had other sources for this information on molten lead quenching, a few traditional gun smithing books, but as I said, none come to mind at the moment.
I dont know if it is gone over in this specific video, but, to be clear, It is a combination of molten lead and bismuth at a 2 to 1 ratio. This combination of metals has a melting temp somewhere around 400 degrees and I believe is mentioned in J.D. Verhoven’s book. This method of using lead and bismuth as a quench and temp hold media is not to be confused with another common practice of quenching in oil to achieve martensite and then putting the spring in straight lead while just barely molten to achieve a blue, or spring temper. Different process, still works, but the tempered martensite spring is not as resilient as the austempered bainite version of the same size and shape. Once again I apologize for a lack of source material at this time. All my old gunsmith books are collecting dust in storage.
Larrin, any chance we will see S7 or CPM 1V on the toughness chart?
I have a very thick piece of S7 so that could come at some point. I have not successfully obtained a piece of 1V.
Lath martensite fraction in hypereutectoid steel seems to be a function of grain size, if https://doi.org/10.1016/j.msea.2004.07.002 is to be believed.
Seeing the bainitic 52100 outgunned by other low-alloy steels was interesting. These steels are formulated for toughness so it was expected. Bainitic L6 in particular seems popular for swords, I wonder it would compare to the bainitic 52100.
Yes, that paper makes the interesting finding that a smaller grain size can reduce the amount of plate martensite for a high carbon steel. But plate martensite is still a function of carbon in solution, not grain size. The plate martensite was not eliminated in that paper, for example, with the grain size they were able to achieve.
The more I learn about 1095 the more I am baffled by how it is recommended to beginner knifemakers. Not only is it a bad steel for beginners, it’s a bad steel in general. The O-# steels aren’t much better
So does 1095 not require a soak time? Just heat to 1475 and quench?
Thank you,
Roger
I prefer to use a hold time with simple steels when heat treating in a furnace. Reducing temperature is easier to control for with the furnace.
The more I read about in on this site the more baffled I am that it’s used in most of the carbon steel knives you find around. Especially when they go for a temper of only 58-60 on average. Mediocre toughness for the hardness.. You’d be better off getting a random leaf spring that’s probably 5160 and forge austenitize from pearlite, throw it in the freezer, and tempering in a toaster oven.
Interesting results on the bainite below Ms. This seems to be somewhat related to what I’ve seen someone doing, by quenching and slow cooling with 400f oil over many hours, maybe getting a partially bainitic structure and self tempered martensite. they do cryo to remove retained austenite and some kind of secondary temper again at 300 ish. They have some interesting results, though grain refinement makes up a fair portion of additional toughness, I assume. Would be interesting to dig deeper into it, with more science.