Thanks to my Patreon supporters I have a salt pot furnace and was able to get some high temperature salts to test out austenitizing in the salt pot for this article. Patreon dollars also went towards metallography for this study. If you want to support further knife steel research become a Patreon supporter today!
YouTube Video
There is also a video for the following information:
Salt Pots and What I Used
“Salt bath” or “salt pot” heat treating refers to the use of molten salts for heating steel. Specific salts are used for different temperature ranges, so that they melt under the necessary temperature without certain bad things happening by overheating them. For these experiments I used “Neutral Salt B” by Hubbard-Hall which I purchased from High Temp Tools: https://hightemptools.com/products/high-temperature-salt They list the working range as 1300-1700°F but also says 1700 can only be used for short periods of time. I have also used low temperature salts for austempering which is a totally different use of salt pots; you can read those experiments here. Of course I used a different low temperature salt for those austempering experiments, which you can also buy from High Temp Tools. If you are using molten salts to heat treat stainless or high speed steels you will also need different salts as that would need higher temperatures. These salts are certainly available; salt pot heat treating of high speed steel is actually pretty common, but I don’t know where to buy small amounts of them. If you know a good source let us all know in the comments.
The furnace I used was the EvenHeat Salt Bath 709 which accomodates a 9″ long pot (plus extra that sticks out), though they are also available in larger sizes. There is a control thermocouple within the furnace as well as within the pot itself. So when operating the furnace you heat to the target temperature using the furnace thermocouple first and wait for the salt to melt. Then you insert the second thermocouple into the molten salt and tell the controller to switch modes to the second thermocouple.
Why Use Salt Pots?
Salt pots have a few advantages. One is that you don’t need to protect the steel from oxygen. The steel is present within the molten salt so there is no atmosphere at all. No need for coatings or foil. Just stick the steel into the molten salt and pull it out to quench. Another is the rapid heat transfer. Liquid is much better at transferring heat than air. This is why sticking your hand into boiling water at 212°F (100°C) is much, much scarier than sticking your hand in an oven at the same temperature, or even 500°F (260°C) for that matter.
Salt Pots and Safety
Probably the number one thing you hear about operating salt pots is how dangerous they are. This is true; if you drop significant amounts of water into the salt bath it could explode and you would be in significant danger. You have to take it seriously, wear protective equipment such as a face shield and long sleeves, and avoid letting any water contact the salt. However, I have also heard some knifemakers take this to extremes, saying things like not to even leave any sharpie markings on the steel and this is exaggerating the level of danger. Don’t take it lightly, but if you follow proper procedures and take safety precautions you can use these furnaces without issues.
Time to Heat Steel in Molten Salt vs Air
As I mentioned, one of the big benefits of salt pots is the shorter time required for heating the steel. To test this I used 1/8 x 1-1/2 x 2-1/4 inch (3.3 x 40 x 55 mm) coupons of 1084 and used 1 minute increments (separate coupons for the different hold times). This is total time in the furnace, not soak time at temperature. Of course, larger coupons would take longer to heat up. I then did the same experiments by using my normal furnace where the coupon was heated in air:
Unfortunately I was a bit late on the 2 minute coupon and it was actually 2 minutes and 20 seconds, so we can’t see exactly how long it took to heat up the steel with the salt pot. But certainly by 3 minutes the steel had reached the target temperature. There was then a small increase in hardness up to 4-5 minutes in the salt as more carbide dissolved. With a 3 minute hold in the conventional furnace the hardness didn’t change at all, meaning it still hadn’t formed any austenite (somewhere under 1335°F/723°C or so). It wasn’t until 5-6 minutes that the steel reached the target temperature. So you can save a few minutes of heat treating time when using salt instead of air. How much this matters will depend on the individual, of course.
The required time to dissolve carbides also depends on the temperature. Below shows published micrographs of 1080 steel [1] with different hold times at only 730°C (1346°F) where it took 5 minutes just to transform the steel to austenite (g), and then another ten minutes to dissolve most of the carbide (h). You can also see how the transformation happened, as the white ferrite was replaced by dark martensite (austenite at high temperature before quenching) with increasing hold time. Then more carbide (white particles) were dissolved with further hold time. You can read more about how this happens in this article on austenitizing.
Micrographs from [1]
Time to Heat vs Time to Soak
1084 is a simple steel with only carbon, manganese, and silicon added to it. The main element that needs to diffuse for transforming the steel to austenite is carbon. Carbon is a very small element and therefore it diffuses very fast. However, when other alloying elements are added it can lead to significantly longer times required for austenitizing. I used 52100 with the factory-annealed condition to illustrate this. 52100 has 1.5% chromium which significantly slows down the time for carbides to dissolve, even at a higher temperature of 1525°F (830°C):
While the 1084 was nearly 66 Rc after only 3 minutes, 52100 with the same hold time at 50°F higher temperature was only 59 Rc. This isn’t because the steel didn’t heat up as fast but because the carbides in the steel did not dissolve as quickly from the chromium addition. The hardness didn’t “level off” until 7+ minutes in the molten salt, and was still climbing after 11 minutes. This is partly why I recommend having a sufficiently long soak time to ensure that you reach a steady state before quenching. Sometimes knifemakers in my comments will ask why I am soaking for so long (10-30 minutes) and say that they only use 5 minutes or even shorter. That results in poor consistency because the hardness can change significantly with a small change in soak time. This also means that the heat treatment becomes very sensitive to the size of the piece as well. Maintaining the same time but with different sized knives would yield very different results.
Another thing to note is that this was with the factory-annealed condition, with relatively coarse spheroidized (round) carbides. The austenitizing time could be shorter with a different starting microstructure. I wrote about this difference and how to use it with forge heat treating here: How to Thermal Cycle Knife Steel. Below shows the difference between pearlite and spheroidized carbides for austenitizing 52100:
Data from [2]
Rapid Austenitizing for Better Performance?
Rapid heating and short soak times are also used in some cases to achieve a finer grain size, and a finer grain size can result in higher toughness. I tested this by heat treating 1084 coupons with a total of 3 minutes in the salt at 1475°F. A true “soak time” includes only the time actually at temperature, so the true soak time is something like 1-2 minutes. I also tried a “triple quench” with another coupon by repeating this procedure two more times. The coupons were quenched in Parks 50 and tempered at 400°F twice. I compared with another coupon that was given a more conventional soak for ten minutes in my normal EvenHeat furnace in air with the same 1475°F, Parks 50 quench, and 400°F tempering.
The conventional heat treatment with the ten minute soak resulted in the same toughness but slightly higher hardness. So despite the potential for finer grain size this did not result in improved toughness. This is relatively similar to a result I saw with 8670 heat treatments where using a 5 minute soak resulted in lower hardness and toughness than a 10 or 15 minute soak:
Prior experiments with 8670 steel in a conventional furnace
Temperature matters much more than soak time for grain size, and in general you don’t need to be afraid of grain growth with a longer soak. Here is a study on 52100 [3] that illustrates this:
Data adapted from [3]
The big surprise was the triple quench which resulted in much lower toughness than the other two heat treatments. The ~11 ft-lbs of the triple quench makes the steel much closer to 1095 and O1, significantly higher carbon steels (~0.95% carbon), which were around 8 ft-lbs at the same hardness:
The older 1084 datapoints on this chart are slightly lower than these new values. That 1084 was from a different steel manufacturer.
The fracture grain of the single and triple quenched 1084 was very fine, so I knew the drop in toughness was not from significant grain growth. So I decided to look at the microstructure of the two conditions:
Single Quench 1084 (3 minutes in 1475°F salt)
“Triple Quenched” 1084 (3 minutes in 1475°F and quenched, repeated)
The main thing I noticed is that there was much more carbide in the single quench version. This makes some sense given that it had 1/3 the total soak time. However, the ten minute in air condition had a similar soak time and did not have a significant drop in toughness. It appears that performing the austenitize and soak repeatedly led to more rapid carbide dissolution than a single soak. Heat treating from a martensitic condition (after quenching the first time) is known to lead to more rapid austenite formation, so I believe this is what happened with the multiple quench. With more carbide dissolved, this put more carbon in solution which also reduces toughness for a given hardness. This is the same reason why O1 and 1095 have lower toughness, because they had high carbon in solution with the same 1475°F ten minute soak. You can read more about this (and some things that can be done to improve toughness with high carbon steels), in this article about austempering.
With O1 I was able to improve toughness by austenitizing at a lower temperature of 1425°F. It is quite likely that I could improve the toughness of the triple quenched 1084 by using a lower austenitizing temperature as well. However, I think this result is still a good warning to those that think that any “special heat treatment” they perform could potentially help and certainly wouldn’t hurt. With any specialized heat treatment, the potential improvement is usually small but the potential for being detrimental is always there, and probably more likely.
O1 toughness was improved by using a lower austenitizing temperature (1425 vs 1475°F)
We did a triple quench experiment with CruForgeV in a conventional furnace back in 2018. In this case we did each soak for 10 minutes in the furnace, the first two at 1450°F and the final at 1500°F. With this experiment we also saw an increase in hardness and a drop in toughness by triple quenching, though the drop in toughness was not as extreme as the salt pot triple quench of 1084. The smaller drop is likely due to 1) the chromium addition to CruForgeV so its carbides dissolve more slowly and it is not as sensitive to overheating, and 2) the lower temperature used in the first two quenches.
Single vs Triple quench of CruForgeV
Prior Triple Quench Salt Pot Experiments by Dr. John Verhoeven
There is an experiment reported by Dr. John Verhoeven in his book Steel Metallurgy for the Non-Metallurgist [4] that he wrote for knifemakers and bladesmiths. He tried it with 1045, 1086, and 5150 low alloy steels. First he heated the steels to 1650°F for 15 minutes and quenched in oil. This was the starting condition where he measured the grain size. Then he heated each steel to 1450°F for 4 minutes in a salt pot and quenched in oil, repeating the procedure two more times (triple quench). He found a decrease in grain size relative to the initial condition after the 1650°F and quench (larger ASTM number means finer grain size):
In this case his results make sense given that the 1650°F austenitize would definitely result in a larger grain size. The repeated cycles at 1450°F were beneficial in refining the grain size, though the result from one or two cycles was not tested. So I think it makes sense for this particular experiment to show a finer grain size from a rapid salt pot “triple quench” while my toughness experiments did not show an improvement. It is also possible that my triple quench actually did improve the grain size but the higher carbon in solution was a more important factor. Measuring very fine grain sizes is difficult.
Do I Recommend Salt Pots for Austenitizing?
I heat treat a wide range of steels with a wide range of austenitizing temperatures so having to use multiple salts and pots doesn’t really work for me. Cleaning out a pot to switch to a different type of salt is definitely not an exciting prospect. So I would probably only recommend salt bath furnaces for those that are using a narrow range of steel types and benefit from shorter soak times and the clean surface coming out, both of which lead to saving them time. But there are several knifemakers out there that swear by salt pots so there might be some benefits I am not thinking of.
Summary and Conclusions
Salt bath furnaces have much faster heating rates than conventional furnaces and also do not require any protection of the steel furnace (there is no oxygen present). Toughness experiments did not reveal an improvement to properties despite the potential for grain refinement with rapid heating and short soak times.
[1] Samuels, Leonard Ernest. Light microscopy of carbon steels. ASM International, 1999.
[2] Stickels, C. A. “Carbide refining heat treatments for 52100 bearing steel.” Metallurgical Transactions 5, no. 4 (1974): 865-874.
[3] Khzouz, Erik Ryan. “Grain growth kinetics in steels.” (2011).
[4] Verhoeven, John D. Steel metallurgy for the non-metallurgist. ASM International, 2007.
Is this triple quench an actual quench in oil three times, or is it heating and then cooling in air or forced air, etc?
You are aware of my bad habits with short soaks and such, but I have had no luck with anything that anyone would call a triple quench. The grain bloats.
If there is something to be gained with quick heats, it’s to do a sort of weak quench after normalizing, then grain cycles (which seem to do more from that point, and then the final quench is much more stable) and then push it on a final quench.
I can generally get obscene hardness even with 52100 without visible grain growth (69/70 out of the quench), though I’m quenching in brine – no LN at the tail, but the brine is not warmed. If it’s in that ballpark and the final post temper hardness is 64/65ish after a 375-400 double temper, it loses its toughness in chisel, which is an asset to a point. A 60 hardness 52100 chisel would hold a foil as it fails and is a terrible woodworking tool, but 65 post HT hardness is hard tempered and acts like W1 or 1095, and both of them fall short a little in terms of making a good chisel.
Every picture I have of a reasonable temp triple quench, though, shows a little visible snapped grain growth under the scope and the tool acts like it. Even the slightest visible growth at high magnification that’s also associated with full heats makes for an undesirable result.
The salt pots would be attractive for stainless for me, but not if you can’t find the salt to use them on stainless, or the toxicity of the high temp salts would make for a no-go situation for amateurs.
Separately, the mention of the earlier quench – but done at a low temperature after normalizing, just as magnetism is returning ….without that, 52100 and short heats is a no go.
26c3 is far more amenable to quick heats and a fast quench – it takes a lot less discretion/skill.
A triple quench is heating and then quenching in oil. There have been many previous studies on multiple quenching for grain refinement. Grain growth of 1095 or W1 with such a treatment would mean the temperature is not well controlled during the multiple austenitizing treatments.
I find your results confusing – but I see the same thing even without triple quenching if getting compact grain but unexpectedly high hardness. Yes on the implication that grain is growing with each heat – though I will admit I have never revisited the idea of actually quenching three times vs. running grain cycles heating only until steel is transitioning. Now that you have a salt pot, it would be interesting to see variations, especially on something like 52100 – descending quenches vs. descending heats but air cooling and then the same heat temp prior to quench.
I have trouble knowing for sure what the value of quenching something is instead of letting it partially cool unless martensitic structure is seen as a way to make each successive quench more effective shrinking grain than starting from a non-martensitic structure.
And my apologies – I missed your comment reading this morning before a meeting that perhaps the 1475 temp is a little high for multiple quenches for 1084. Still, the results are surprising.
The salt pot, though, could open up possibilities to test a lot of cases without spending nearly as much time as waiting for a furnace.
Because the martensite leads to more nucleation sites for austenite than pearlite.
Thanks for the explanation. I still use a pair of forges, but over the years, my results have been greatly improved just by reading stuff on your site to confirm things I see and feel or suggest that maybe I need to look further because the cause of something isn’t what I think it is.
At some point, not sure if it’s too eggheady, but there’s a big divide between the information on heat treatment here and then what’s actually done industrially. For example, there is a maker of woodworking tools that uses XHP, presumably ordering melts from Carpenter – I don’t know – but industrial process could probably produce a very narrow hardness range. Tested samples that I’ve come across can vary by 2 1/2 points. Why does that occur? For woodworking tools, what you describe as edge stability on a thin knife is pretty essential and hardness has a pretty significant effect.
Put differently, what you find with salt pots and thermal cycles is wildly interesting to someone like me or someone making knives but trying to get them right, but how much does it affect day to day in stuff that’s made in the upper production or lower boutique kind of market? Something like “what they do at Peters” in a furnace that holds up to 5000 pounds in a batch vs. the person in a garage using a forge, or more importantly for most of your audience, a programmable electric furnace. What do those commercial setups do – do they thermally cycle things in big batches or is the process generally faster with some things that provide marginal improvement removed.
I had a thought on the toughness with respect to triple quenching. higher carbon steels like 1095 form plate martinsite, and in other articles you’ve mentioned that the plates can impinge on each other forming micro cracks. if heating to austenitizing temps does not repair this, wouldn’t that mean that triple quenching induces three sets of those micro cracks?
The micro cracks would be repaired from re-austenitizing. But it would be a fun study for some fancy microscopy to make sure they are gone.
Excellent work Larrin! Keep it coming.
30 years ago i was forging blades from O1. I found that triple normalizing or triple quenching made them last four times as long on cutting rope. I didn’t test for toughness though.