Forging Knives

Heat Treatment

Updated 4/24/08

Bend or Break?

This is one of the central paradoxes of steel. We want steel that's hard and strong and the harder/stronger it becomes the more likely it is to break. Another way of asking the question is how fragile do you want your knife to be? Better edge-holding ability equals greater fragility. Tool steels are created with this paradox in mind. In the graph below, we can see that F2 is more fragile than S1 and that's because of composition.

Once a particular steel has been selected, heat treatment is the process by which it's bend/break point is set. For example, F2 can be made softer and therefore tougher. S1 can be hardened more, losing some toughness and gaining wear resistance. Is there an optimal bend/break point for each steel?

Every knife steel has an optimal hardness and one of the easiest tests for this can be made while sharpening. Soft steels create a burr when sharpened--during the process of sharpening the steel at the edge of the blade BENDS and forms a burr. If the steel were progressively hardened, at some point burr formation would cease. This is right at the point where the steel BREAKS rather than bends. This BEND/BREAK point is the optimal hardness for any given knife steel. A burr means too soft, no burr means too hard. It's right at the hint of a burr that things are just right--hardness/strength is optimal. Apply the Three Bears Rule (too hard, too soft and just right) to gauge optimal hardness.


There you are, tongs and a piece of red hot steel in one hand and scratching your head with the other. What to do? Time's running out!

OK, put the steel down and let's plan this out. We're going to use a brine quench. It's easier, cheaper, safer (the shop won't burn down) and more interesting.

Mix one cup of salt to each gallon of water to end up with something close to a 7% solution. Store the brine in a plastic container if you want. Use a metal bucket for quenching. The saltwater should be over 70F and up to 140F during the quench. If you like quenching warmer then get a Bucket Heater, it'll make raising quenchant temperatures much easier. It can also be used with oil if you want. Have at least 2-3 gallons of brine to quench a knife--more doesn't hurt. Whatever brine temperature you end up using, check it before you quench.


When does the clock start ticking? The clock starts when steel temperature falls below 1333F. There's no need to quench at or near the Peak temperature. So when to quench? Between the Peak and 1333F, it's your call. There's no reason to subject the blade to any more thermal shock than necessary, so quench at the lowest temperature you can and still be above 1333F.


Place the blade into the muffle and when it gets hot enough, watch/check for Critical, using a magnet. Now the question is how much higher to heat the steel? Time and temperature are both at play here. A slower temperature climb over a longer soak period is about the same as a faster climb and shorter soak. Use both the 1400F and 1500F Tempilstiks, 1400F is right at Critical. Let the temperature rise and if you hit 1500F then withdraw the steel. You'll have the temperature bounded, lower and upper, with Tempilstiks. Shoot for 50-100 degrees above Critical and a soak of 1-3 minutes.

Withdraw the blade from the muffle and as it cools, watch the color change. Know what Critical looks like and quench right as you pass Critical going down. If you start seeing 'shadows' move across the steel you've hit the recalescence point and the steel's too cool to quench. Put it back into the muffle, go back up to peak, soak a bit and try again. This routine can be practiced with 1018 and although it won't harden you can get the procedure down.

One important point to understand: When the steel has reached Critical during the initial heating phase, the magnet test will work. However, the magnet doesn't work coming off Peak during the cooling phase. Further, the magnet won't work until the steel has cooled below Ar1 and is heated again. Steel needs to be 'reset', as it were.

Once the quench begins, it's off to the races, the clock has started. We need to get past the nose and down to maybe 900F in a second or so. That's pretty easy, cool until the color disappears from the steel. When the color is gone the steel is cooler than about 900F. At this point we have time, 15 minutes, to drop the temperature to 130F. In particular, we want things to happen slowly during the Martinsite transformation, about 450F to 130F, that's when the blade is getting hard. It doesn't happen during the quench, although that's where the drama is, it happens during the transformation.

Quench for less than a second--plunge and withdraw. If there is any color, immediately plunge and withdraw again. Be sure and get to 900F or cooler in under two seconds. The time period for plunging will depend to a large extent on knife size and thickness. What you want, when you withdraw from the initial plunge, is the color gone out of the steel, but that it's still hot enough to quickly evaporate water from the surface. These are the two temperature tests to bracket the blade's heat--no color and almost instant sizzle.

Now you have plenty of time. Take you time. You'll need something like a wad of wet burlap that you can touch with the back of the blade to cool it. Slowly, draw the heat out of the blade. Touch and sizzle, touch and sizzle. The sizzle should get longer and less violent until at around a minute or two there not a whole lot of sizzle left. The blade should be somewhere around 450F. You can test this temperature by watching tempering colors, you'll want straw yellow. Get to 450F within two minutes. In many cases, you can just let the blade air cool to 450F.

Rest now and let the blade air cool in air. You want to make it to about 130F within 15 minutes. Keep an eye on things, it wouldn't hurt to have an egg timer running at this point. Make the 130F/15 minute-mark and you've brought the blade in for a smooth thermal landing, perfect! When you can hold the blade in your bare hand--warm, but not hot to the point of discomfort, then begin the temper/draw phase of heat treatment. Discomfort is wondering when you can set the blade down, when you feel you could hold it indefinitely the temperature is about right. Don't let the blade drop below 125F. With thin sections, the step from 450F to 130F will often happen in less than 15 minutes so it doesn't hurt to insulate the blade and check it every few minutes. Extending the time for Martensite transformation to a full 15 minutes will be repaid many times over. The whole transformational magic happens during this period and it happens best when it happens slowly.

Immediately place the blade into a kitchen oven preheated to 450F (see tempering chart below). Using an oven thermometer wouldn't hurt. Close the oven door and wait two hours. Retrieve and let cool in air to room temperature. The resultant hardness of O1 steel should be around HRC 60. That's it, that's the heat treatment.

First, we hardened by bringing the temperature down from 1475F to 130F in a controlled manner, then we softened a little by tempering in the kitchen oven. See the tempering graph below.

It should be noted that 130F water and 130F steel are two very different things. Putting your hand in water that hot will create discomfort if not injury.

Make any adjustments to this procedure as you see fit and/or as results warrant.


Directions from above in graphic form for O1 tool steel.

1. 1350F to 900F in one second
2. 900F to 450F in a minute or two, starting the Martinsite transformation (yellow)
3. 450F to 130F, completing the transformation, in air, in about 15 minutes

The rule is simple. You can do anything you want with temperature so long as you don't get trapped above the Nose or bump into the S Curve. Design and follow a cooling path that ends with Martensite and keeps thermal stresses low.

This heat treatment applies to all steels in this tutorial. The only difference is that for steels faster than O1 (F2, 1095, 1084, W1) we need to get to 900F in under a second. After that, the thermal landing is about the same.

Read the TTT graph for your particular steel and plot the cooling path accordingly. Plan your moves before making them.

CCT Diagram for O1

Composition as it applies to this diagram:
C ---- Si ---- Mn ---- Cr ---- V ---- W
0.95 -- 0.25 -- 1.20 -- 0.48 -- 0.13 -- 0.55

Martensite is about 4% less dense than Austenite, thus during the A to M transformation steel expands. 4% is a huge amount of stress. This is why we're taking plenty of time with the Martensite transformation. Martensite is hard and strong and a cooling path which avoids Ferrite, converts most of the Austenite to Martensite. By and large, a successful cooling path is measured by the resultant hardness. Taking 15 minutes from 450F to 130F gives plenty of time for the Martensite transformation to take place in a smooth and trouble-free manner.


Tempering graph for O1

It looks like the optimal hardness is about HRC56, right where the lines cross.
But that's an illusion, as the Impact Scale is in arbitrarily sized units. I've always suspected that
Bo Randall misread this graph and that's why Randall knives are hardened they way they are.

For a look at how the Randall people forge a knife.


In particular and as it applies to the Three Bear Rule from above, it should be pointed out that final blade hardness is set by the drawing/tempering temperature. This is the place to adjust things in order to get the exact burr you want--hence the exact bend/break point you want. For example, you can temper at 400F or lower and then Burr Test the blade. If you want it softer then temper again at a slightly higher temperature. Keep going until the Burr Test is to your satisfaction.


In reality ...

The heat-treating directions above sound reasonable and precise, but how do they work out in the real world? Blade temperatures change fast! Both in heating and cooling. Temperature readings (from the various methods) are valid for microseconds. In the time it takes to say the temperature, it has changed to a considerable extent. In particular, sections as small as we're using don't have enough thermal mass to slow the process down and even it out.

The steel along the edge should be the hottest, the blade's spine can and should be cooler. And there's no need to heat the handle at all. Key in on the half inch along the edge.

Attaining a suitable peak temperature (and soak) isn't that hard as you can control the torch with one hand and the blade with the other. In time, you can do this by eye alone. Have a reference clock running throughout heat-treating. It's surprising how much time distortion occurs in the heat of the moment!

The quench is problematic. With sections this thin the quench needs to be fast. In and out. One way to extend the time is to heat the brine to 140F. That way a rapid quench and withdrawal will leave the blade temperature somewhere around 900F.

The second problem is the cooling phase from 900F to 450F. How to gauge temperatures? 900F is when blade color is gone, but you will come to know it by how fast the brine evaporates when the blade is withdrawn from quenching. That sound will tell you most everything you need to know. It should be fairly instantaneous. Now for 450F. There are probably some good ways of judging this temperature but a Tempilstik is as good as any. Get the 425F just to be on the safe side.

Temperature methods: Tempilstiks, magnet, steel color, rate of evaporation. Use them all and recalibrate your senses from time to time.

In the end, the proof is in the pudding. If you're creating blades that do what you want, you must be doing it right.

Other than that? Practice.


How is brine different than an oil quench? Not much really. Oil has to get the blade past the nose quickly (the same as brine) and then create a smooth landing. We've done the same thing with brine--even better. And we have greater control of the process than just dunking the blade into an oil quenchant and hoping for the best. No smoke, no flames, no fuss. An oil quench is over in seconds rather than minutes and although the Martensite transformation takes a little longer with oil than water, it's still a fraction of the time available with this method.

Something knife forgers run up against is that most of the steel and heat treating information is geared for industrial sizes and processes. Most spec sheets aren't talking about knife blade thicknesses, they're usually taking about inches thick. Most industrial heat treatment isn't done one small item at a time. They want a quench where unskilled labor, using a crane, can dunk 1000 pieces at once and have some assurance the hardness will come out about right. What makes sense for industrial heat treatment, might not for the craft of forging knives one at a time.

It's often said that heat treatment is the soul of a knife. Get involved, the treatment outlined above is radically different than attempting to find the right quenchant to do the whole job for you, automatically. This method puts you in the driver's seat.

Should a paring knife and a Bowie be treated differently? Of course. And what's needed is your good judgement and skill during the heat treating process. Have some degree of pride in knowing the crystalline state of what you're holding with the tongs. Take pleasure that you control the rates of change.


During the Martinsite transformation, say the Good Words, speak to the knife, put some power, beauty and intention into that matrix. Otherwise, why forge knives at all?