Monday, January 20, 2014

The 'Temper of Iron'

"He [Fiore Dei Liberi] also longed to learn the temper of iron, of the nature of each weapon and of its aptness for defense and offense." -- Fiore Dei Liberi, Getty MS as translated by Tom Leoni.

In an earlier post we discussed some of the basics of phase diagrams, using the iron-carbon (i.e. steel/cast iron) system as an example. While some talk of heating, heat treatment and quenching came up in the discussion, this is such an interesting and important area I wanted to give it its own space. Besides, Fiore apparently thought it was important too.

Let's first start with the use of hot working in blacksmithing: I am willing to bet that pretty much everyone reading this blog has seen a blacksmithing scene in a movie that goes something like this:
[Dramatic music plays over a dark scene, a Blacksmith, covered in soot with biceps like sacks of melons, forearms like hams, and a beard ZZ Top would be proud of stands before a glowing forge.]
[Blacksmith pumps bellows as flames jump out of the forge]
[Blacksmith grabs glowing hunk of metal from the forge, and begins hammering it madly]
[Blacksmith looks at the piece, and dips it into a vat of water. The water sizzles as steam dramatically rises from the water]
[Blacksmith then plunges the dark metal back into the forge and repeats the process]
[Fade to black, Fade music]

Let's start with why the above process is ridiculous (and not surprisingly, is not done by blacksmiths)

Hot Working Metal: The Real Deal

In metal work, there are essentially two types of work performed on wrought (i.e. not cast) pieces: cold and hot. The difference is that cold work (not to be confused with work hardening, which sometimes goes by a similar name) is performed on pieces near the ambient temperature, while hot work is done at elevated temperatures. If we take a look at a plot of the change in yield strength and elastic modulus at different temperatures, we can see why hot work gets used:
Typical yield strength and Young's (elastic) modulus reduction with temperature for a carbon steel.
Notice how the yield strength and elastic modulus dramatically decreases between about 400 °C and 900 °C (the dull red to orange glow range), and continue to decrease at a slower rate above about 900 °C. In practical terms, this translates to a material that is much easier to deform, both elastically and plastically. The deformation a blacksmith is interested in causing is plastic: they want to permanently change the shape of a piece. If you recall my earlier posts on strengths of materials and materials science: elastic deformation results from the stretching of a lattice, and therefore relaxes once the force is removed. On the other hand, plastic deformation occurs primarily through the permanent displacement of atoms via dislocations. Even at 900 °C, the energy barrier to 'undo' dislocation motion is far larger than the average kinetic energy available to overcome the barrier. In other words, plastic deformation at this high temperature is still plastic deformation and will remain after the sample is cooled.

For reference, from the book Backyard Blacksmithing by Laurelei Sims, the recommended temperature for forge operations varies between about 980 °C and 1150 °C (bright orange to yellow), while bending can be done at 760 °C (red). Note that these are well after that marked decrease in yield strength and Young's modulus, maximizing the efficacy of moving material without getting too close to the temperature where liquid starts to show up.

In short: hot working metal is to make it easier and faster to work metal. That's it.

But: heat and controlled cooling do get used to modify certain properties of a material, as we'll discuss next, but this is a very different process than heating and hammering. By cooling and re-heating a piece so you can hot work it, all you do is increase the time to get back to a good working temperature. Why?

If we heat a piece and plastically deform it, at some point the piece leaves the temperature range where we can easily work it. At that point, you can get a bigger hammer (seriously: you could go get a multi-ton machine and do the whole thing cold, if you really wanted), or you can warm it back up to where you can easily work it with hand tools. By hot working, you also gain the ability to do some additional things that may not work at lower temperatures without cracking or tearing the material. Any decrease below the minimum working temperature just translates to an increased time to reach the working temperature again. The reason can be seen by looking at phase diagrams and transformation diagrams: if you increase temperature sufficiently after a quench, you effectively undo any microstructural changes you made along the way (for example: turning martensite back into austenite).

So our movie blacksmith isn't doing anything useful, they are just being really inefficient. This is why blacksmiths generally just chuck the piece back into the forge after it cools past the working point.

Now let's get to the good stuff: heat treating steel. But first, let's revisit some of our discussion on cooling from an earlier post:

Isothermal vs Continuous Cooling

As it ends up, the cooling rates, hold times, and temperatures required for isothermal transformations aren't terribly practical for many applications. It is more common to encounter continuous cooling schemes, such as those described in our earlier discussion of martensite formation, where a sample is cooled continuously to room temperature. And as we'll get into later, there are processes which combine aspects of both.

An example of continuous quenching is pulling a heated rod out of a forge and placing it into either the open air or a liquid bath (both typical methods for blacksmiths, depending on the goal). An important thing to remember is that continuous cooling does not necessarily mean a constant cooling rate.

Heat Transfer During Quenching

Now let's look at what happens when we quench a sample. The 'Captain Obvious Strikes Again' answer is that it cools down. But how does that go down?

The rate of cooling will depend on a number of factors. These factors are related to the three main forms of heat transfer:

Conduction: Conduction is the transfer of of thermal energy via atomic vibration.

Think about it this way: if one atom in a lattice were 'heated up' (forced to vibrate about its equilibrium position by increasing its kinetic energy), it would cause others around it to begin to vibrate. These vibrations would then cause their neighbors to vibrate and so on. Bam: conduction. As you might imagine, the closer the system is to an ideal lattice structure, the more readily these vibrations can be transmitted. This is why solids tend to have much better thermal conductivities than liquids or gasses: solids have a structure which can readily transmit the vibrations (which can be thought of as setting up waves in the atomic structure known as phonons).
Schematic of conduction in a crystalline material.

Convection: Convection is the transfer of thermal energy via bulk motion.

Imagine a loose collection of molecules as in a liquid or gas. If one atom were 'heated up' (increasing its kinetic energy), it would then transfer some portion of its energy when it smacks into others (not literal contact but the mental image is good enough) and they to others they contact and so on. On a larger scale, you can see effects like buoyancy taking place due to coordinated motion. Because liquids and gasses are able to readily undergo bulk motion, convection tends to occur much more readily than in solids, where diffusive motion events are much more rare.

Two types of convection exist: natural or free convection, which is what I describe above and forced convection which takes place when there is also an externally applied bulk fluid motion, such as a flowing liquid being used as a coolant. Fundamentally, the difference is the amount of energy being transferred and how exactly it is transferred. There is a final form called advection, that refers to transfer via bulk motion of the fluid alone (i.e., not including the diffusive transport seen in free and forced convection).
Schematic of natural convection caused by applied heat. Image courtesy of wikimedia commons user Oni Lukos.

Radiation: Radiation is the transfer of energy through the emission or absorption of electromagnetic waves.

Heat transfer via radiation is usually discussed in terms of infrared (IR) radiation, though anything that affects lattice or molecular vibrations or atomic motion works: microwaves exciting rotational modes, for example, or even the energy loss due to radiating visible light. Radiation is the only form of heat transfer that can occur through a vacuum. Also, it's really radiation that kicks off conduction between bodies in contact, unless there is sufficient contact and energy for lattice or molecular vibrations to directly transfer (either good thermal contact or a fluid layer). In the case where you have very poor thermal contact (say a two pane window with a vacuum space, being heated on one side), you only get radiation as the mode of heat transfer, which represents only a portion of the thermal energy of the heated pane of glass. This is the main reason this sort of window is used.

The classic example of heat transfer via radiation at work is the sun warming your skin. Your skin absorbs the IR radiation emitted by the sun (a rather long journey) and the heat you feel is the increase in vibration of the molecules in your skin it causes. This effect is actually used in IR spectroscopy to identify chemical species.

Continuous Quenching in a Fluid Bath

Let's look at the most common form of continuous quenching: placing the sample into a fluid bath at a known temperature. A fluid can refer to a gas or a liquid, such as air or water. But which of the above mechanisms of heat transfer is relevant if we want to know what dominates the cooling of the sample?

To find out, we can do a comparison of the heat fluxes, or heat transferred per unit area, associated with each mechanism. For simplicity, let's assume that the volume of our sample is small compared to the quenching bath so that we can assume the quenching fluid doesn't change temperature significantly. In practice, this situation is the desirable one, is generally easy to set up and sometimes involves not just a large bath but also a mechanism to stir the fluid.

The heat fluxes for the mechanisms in our scenario are as follows:
\[q_{cond} = -k\frac{dT}{dx} \approx -k \frac{(T_{s} - T_{int})}{L_c}\]
Where \(k\) is the thermal conductivity for the body, \(T\) is the temperature (as a function of position) and \(x\) is position within the conducting body. As an approximation, we can consider the derivative to be the gradient of the temperature from the interior (\(T_{int}\)) to the surface (\(T_s\)), over a characteristic distance \(L_c\).

\[q_{conv} = h(T_s-T_{\infty})\]
Where \(h\) is the convective heat transfer coefficient, \(T_s\) is the surface temperature of the sample being placed in the quenching fluid and \(T_{\infty}\) is the bulk temperature of the quenching fluid.

\[q_{rad} = \epsilon \sigma_{SB} (T_s^4 - T_{\infty}^4)\]
Where \(\epsilon\) is the emissivity of the sample, \(\sigma_{SB}\) is the Stefan-Boltzmann constant, \(T_s\) is the sample surface temperature and \(T_{\infty}\) is the bulk temperature of the quenching fluid.

Main Mechanisms in Fluid Bath Quenching

The easiest way to get a quick and dirty idea of which one(s) of the above mechanisms are important is to do an order of magnitude analysis on a typical part (since it could change based on the particulars of the system). If we assume we have a 5 cm diameter plain carbon steel round rod that was heated to a uniform 1000 °C, being quenched in water at 20 °C (room temperature), we end up with the following:

Conduction through the bar and into the water at the surface:
Plugging in some relevant values into the conduction equation, we get:
\[q_{cond}^{max} = -(30\;\mathrm{W/m*K}) * \frac{(293\;\mathrm{K} - 1273\;\mathrm{K})}{0.025\;\mathrm{m}} \approx 1.2 \mathrm{x}10^{6}\;\mathrm{W/m^2}\]
Where I've assumed the surface temperature equals the quench bath temperature to get a high estimate of the gradient, and the radius is the characteristic length. It is worth noting that in general the thermal conductivity varies with temperature, and for plain carbon steel it increases nearly linearly from 30 W/m*K at 1000 °C to a value of about 76 W/m*K at 20 °C.

As the cooling process continues, we'd also expect a change in the dimensions of our rod, since the bar would have expanded due to the thermal expansion during heating, and will shrink as we cool it. But because the coefficient of thermal expansion is on the order of \(10^{-6}\) 1/K, the expansion of the radius would be only be on the order of 10 \(\mu\mathrm{m}\) at 1000 °C: a negligible amount for our purposes.

Convection from the surface of the bar to the quench water:
Compared to the thermal conductivity, the convection coefficient is a bit more complicated as it encapsulates a number of effects. In general, its value is determined via experiment or computer simulations, but it can be determined analytically for simple geometries and flow fields.

For natural convection in still water, the convection coefficient value is on the order of 20-100 \(\mathrm{W/m^2*K}\), while for air it is about 10-100 times smaller and oils tend to be in the middle. Brine solutions (~3+ wt% NaCl in water) or 'caustic soda' (1+wt% NaOH in water- an old method) have even higher coefficients than plain water. The specific value depends on things like the geometry and orientation of the piece being quenched, as well as the details of the fluid and flow field. If we look at the maximum heat flux, assuming \(h=100\;\mathrm{W/m^2*K}\) and \(T_s\) = 1000 °C, we end up with:
\[q_{conv}^{max} = (100\;\mathrm{W/m^2*K}) (1273\;\mathrm{K}-293\;\mathrm{K}) \approx 10^{5} \mathrm{W/m^2}\]

But there's a hitch here: In our case, until the rod surface temperature drops below 100 °C, we're also causing a localized phase change in the water (boiling and vaporizing it). The phase change results in a change in the heat flux as we increase the temperature over the saturation temperature (i.e. boiling point), the temperature at which the thermal energy is maximized before boiling begins (~100 °C for water at atmospheric pressure). This change in heat flux is represented below:
Boiling curve for water at atmospheric pressure. Image courtesy of wikimedia commons.

This plot is called a boiling curve and describes the total heat flux (\(q_{s}''\)) from a surface as a function of the surface temperature (\(T_s\) relative to the saturation temperature (\(T_{sat}\)) of the fluid to which it is exposed (think: the bottom of a pot of water). Our 1000 °C bar (\(\Delta T_s\) = 900 °C) begins above the Leidenfrost point, where a vapor layer is produced between the surface and the bulk of the fluid, resulting in relatively poor heat transfer. Eventually our bar will cool below the nucleate boiling region into the free convection regime where the calculation above will be more reasonable. Based on the boiling curve, we'd expect to see heat fluxes between about \(10^3-10^6 \;\mathrm{W/m^2}\).

Radiation from the surface of the rod into the water:
The heat flux due to radiation from the surface into the water can be estimated as follows:
\[q_{rad}^{max} = 0.2 (5.6703 \mathrm{x}10^{-8}\;\mathrm{W/(m^2 K^4)}\;((1273\;\mathrm{K})^4 - (293\;\mathrm{K})^4)\approx 3\mathrm{x}10^{4} \mathrm{W/m^2}\]
Where I've assumed that the emissivity doesn't change with temperature, though in general it does: for mild steels it varies from about 0.1 to 0.3 from room temperature to the liquid state.

Note how this heat flux is approximately a tenth of the convective value and a hundredth the value of the conduction through the bar. Also, because of the power of the temperature, the difference term \((T^4-T_{\infty}^4)\) decreases rapidly as the bar surface temperature approaches the quench bath temperature. So this effect is not negligible at very high temperatures (several hundred  degrees or more), but it is small compared to convection in the liquid and conduction through the bar. This actually changes a bit after boiling begins near the surface of our bar because of an effect called the Leidenfrost Effect, where the rapid boiling of the water near the surface of the bar produces a vapor layer where convection is suppressed and radiation from the hot surface can become significant compared to convection.

Conduction through the water from the surface of the bar:
In liquid water at 20 °C, the thermal conductivity is only about \(6\mathrm{x}10^{-4} \mathrm{W/m*K}\), while the thermal conductivity of the gas phase at the boiling point is about an order of magnitude smaller. Both of them are over ten thousand times smaller than the thermal conductivity in the metal bar. In order for the conduction from the surface through the quench water to be on the same order as the convective heat flux, the characteristic distance \(L_c\) would have to be about 1 \(\mu m\) from the surface. Sufficed to say, this would require an unreasonably small quenching bath and thus conduction through the quenching fluid is not terribly relevant compared to convection for most of the cooling process.

As a short aside, this low thermal conductivity of water is why it can be used as a thermal insulator if convection can be suppressed: in a wetsuit, for example. The exchange of water from within the wetsuit to outside the wetsuit is slow, so you are able to warm up the water inside the suit through convection (cooling yourself a bit in the process) but once the water temperature is the same as your body temperature, convection becomes ineffective. Radiation is poor because your temperature is too close to the temperature of the water inside the suit and the material of the suit. The only remaining major heat transfer mechanism is conduction: and water is quite bad at it. Air is as well, which is why jackets and drysuits work in cold weather, assuming they are loose enough to maintain a decent air layer inside.

Summary of mechanisms in quench bath:
Except in the first moments when the bar is plunged into the quench water, convection from the surface of the bar and heat conduction within the bar are the dominant effects. The convective cooling at the surface will drive the temperature gradient within the bar (and therefore the conductive heat flux), thus causing the quenching fluid to control the rate of cooling of the bar. For cases where the dimensions of the bar (or the thermal conductivity) are such that the conductive heat flux within the bar approaches or exceeds the convective heat flux at the surface, the net effect is that the bar will tend to change temperature uniformly. Thus, we can get a qualitative estimate of how quickly a given fluid will cool a piece by comparing convective heat transfer coefficients, say relative to air. So holding all other things equal, the speed of quench (sometimes called the quench velocity) based on a still fluid would be something like:
\[\mathrm{Brine} > \mathrm{Water} > \mathrm{Oil} > \mathrm{Air}\]

Tempering and Annealing

Now that we have some idea of how quenching works, let's get back to talking about how heating and cooling affect the structure and properties of a piece. You've probably seen the color change that occurs when you heat a piece of clean steel, producing a rather brilliant array of colors:

Surface color change in a heated steel. Image courtesy of Wikimedia commons.
The coloration you see is quite different than the glowing that occurs as you heat the steel: the above effect is a result of the reflectivity of the thin oxide layer that forms during heating, as opposed to radiation of photons by the surface. This effect is called thin-film interference, and the color changes as the oxide layer thickens. The rate at which the oxide layer grows is related to the temperature of the surface, so for a given relatively short period of time, the color designates the temperature at which the heat treatment occurred. But if the piece is left at a given temperature for a very long time (say hours), the oxide layer can grow thick enough to no longer produce a distinct color. Because of the coloration effect, blacksmiths have long used the color to determine the heat treatment temperature:

Temper color standards for steel. Image courtesy of Wikimedia commons.

There are four main processes used in heat treatment of steels:
  • Tempering
  • Annealing
  • Precipitation Hardening
  • Surface Treatments
Let's dig into these a bit to understand the difference between them and why we may want to use them.


Tempering is the process of heating a piece to a specified temperature (below the eutectoid, typically between 150 °C and 700 °C), holding for a period of time then allowing it to rapidly cool. This process is usually only performed on pieces that contain a greater portion of martensite than desired (i.e., those that were rapidly cooled after hot working). The result is typically an improvement in ductility and toughness through a structural transformation of a portion of the martensite due to the diffusion of carbon while at the elevated temperature.

The phase transformation that occurs results in tempered martensite, consisting of small uniformly dispersed particles of cementite in a ferrite matrix. Tempered martensite resembles a spheroiditic microstructure but the cementite particles are typically much smaller and can't be seen with optical microscopy. The small cementite particles tend to act as barriers to dislocation motion, improving strength and hardness over pearlite or spheroidite but generally at a loss compared to pure martensite. But the ferrite matrix is ductile and tough, compared to martensite, thus resulting in an improvement in the ductility and toughness with a relatively small loss of hardness and strength. The size of the cementite particles relative to the ferrite matrix determines the balance of these properties: increasing the particle size (decreasing the overall phase boundary area) tends to decrease the strength and hardness while improving ductility and toughness. The size of the cementite particles in tempered martensite is determined by the temperature and soak time used in the tempering operation. Typically tempering is performed as a constant temperature heat treatment, with the temperature determining the rate of softening (and decrease in hardness) as a result of changes in diffusion. The soak time is then based on the desired change in property.

However, not all steels respond identically to tempering. For some, the tempering process can result in a reduction of toughness, called temper embrittlement. It tends to occur when the steel is tempered above 575 °C and slowly cooled to room temperature or when the steel is tempered between 375 °C and 575 °C and rapidly cooled. This has been found to occur in steels containing significant amounts of manganese, nickel, or chromium as alloying elements and low concentrations of antimony, phosphorus, arsenic or tin impurities. The combination of these impurities and alloying elements results in a significant increase in the ductile-to-brittle  transition temperature with room temperature being in the brittle region. Temper embrittlement can be avoided by avoiding the combination of alloy elements and impurities or by modifying the tempering process temperatures and cooling rates.

Tempering doesn't always occur on purpose: it can occur in cutting tools due to high friction while machining or near the heat-affected zone of a weld joint, for example. For cutting tools, this tempering is generally avoided by modifying cutting rates and applying a coolant because the property changes are not desirable. In welding, process changes (including changing the method of welding entirely) are used as well as taking this behavior into account in the weld joint design.

Austempering and Martempering:

There are two other processes with 'tempering' in the name, though they are not really tempering processes in the traditional sense: austempering and martempering.

In austempering, the piece is rapidly cooled to a temperature above the martensite formation temperature, then held at that temperature until the piece is entirely transformed into bainite (or stabilized austenite in cast irons), then cooled to room temperature. This process is actually a quenching technique that at least partially alleviates the need for a subsequent tempering operation, thus saving time and money. This process first saw widespread adoption during WWII, in the manufacture of rifles as an improvement over conventional quench-and-temper operations.

In martempering, another quenching method, the piece is cooled to the martensite transformation, but is cooled slowly through the martensite transformation. The main reason for doing this is to prevent cracking and distortion of the part. It results in a martensitic structure just as rapid cooling would, and does not eliminate any subsequent tempering steps.


Annealing is a more general category of heat treatments, typically used to relieve internal stresses, increase softness, ductility and toughness, or to produce a desired microstructure. Annealing processes can be broken down into the following stages:
  • Heating to annealing temperature (this varies with desired effect and composition)
  • Soak time
  • Slow cooling to room temperature
In other words, annealing is the combination of an isothermal transformation and continuous cooling process. Let's look at some common annealing processes.

Process Annealing:
Process annealing is used to 'undo' some of the hardening and loss of ductility that results from work hardening a piece. One major reason to do this is to improve the ability to further deform the material without cracking. For example: to ease the ability to shape a piece as it is being cold worked (which results in work hardening as discussed before).

Process annealing in steels is carried out at a temperature below the eutectoid temperature, with a hold period and a slow cooling process. During the soak time, the piece is held at the annealing temperature for sufficient time to relieve stresses through lattice relaxation and allowing recrystallization to occur. However, because of the Hall-Petch effect, the heat treatment is usually only carried out for long enough to form a relatively fine microstructure.

Stress Relief:
A stress relief heat treatment is essentially meant to counteract the residual stresses and stress gradients that can form during various processes. For example: uneven cooling rates in a piece, grinding or machining, or excessive localized deformation. These heat treatment processes generally consist of heating a piece to a given temperature and holding for long enough to ensure the entire piece is at a uniform temperature before finally slowly cooling the piece to maintain a nearly uniform temperature during cooling. In essence, this allows relaxation of the microstructure to occur at the elevated temperature and 'locks in' the relaxed structure during cooling, minimizing the residual stresses.

In some processes, such as rolling, extrusion or stamping, the large plastic deformations can result in large irregularly shaped and sized grains. Normalization is an annealing process that seeks to produce a more uniform distribution of smaller grains, thereby producing a tougher and more ductile material. This process involves heating the piece to above the eutectoid for a sufficient time to allow the structure to completely transform to austenite (a process called austenitizing) then quenching the piece in room temperature air. It generally results in a fine pearlitic microstructure.

Full Annealing:
This process is usually performed in low and medium carbon steels that will be processed by machining or excessive plastic deformation. The process begins with austenititizing like the normalization process above, but the cooling is done slowly under controlled conditions in the heat treatment furnace. The resulting microstructure is generally coarse pearlite that is more ductile than the normalized part and the microstructure tends to be more uniform through the part due to the slow cooling process. The main drawback to this process is the long period of time it takes to cool the part: on the order of hours.

For higher carbon steels, a microstructure consisting of coarse pearlite may still be too difficult to machine or deform. To deal with this issue, these alloys may be heat treated to develop a spheroiditic microstructure, which tends to be softer and more ductile than both coarse and fine pearlite. As discussed previously, spheroidite can be produced by heating a piece to just below the eutectoid temperature and holding for an extended period of time (12 hours to a day, possibly more) and cooling to room temperature.

Tempering vs. Annealing

Occasionally you'll hear (or read) something to the effect that "Tempering is hardening and annealing is softening" (actual quote from an online forum).

However, this isn't at all true, as described above: both processes are meant to improve improve ductility (among other things), and both processes reduce hardness. I can only assume folks who say this confuse the process of forming martensite with tempering, but their only similarity is that they both involve rapid cooling.

We discussed in a previous post how heating a carbon steel to above the eutectoid point then quickly cooling to room temperature can result in the formation of martensite, a hard and strong microstructure. But what other processes can increase the hardness and strength of a steel?

Precipitation Hardening (Aging)

A method of enhancing the strength and hardness of some alloys is to form very small uniformly distributed particles within the original material. These particles are called precipitates and the process by which the alloy is modified is called precipitation hardening or age hardening. The first name comes from the fact that the formation of precipitates is the mechanism by which the hardening occurs, whereas the second comes from the fact that the material property changes occur over a period of time.  Two common designations of age hardening types are: natural aging (performed at room temperature, or close to it) and artificial aging (carried out at elevated temperatures). This process is commonly used in aluminum alloys and in some iron-carbon alloys.

The formation of precipitates during the hardening process requires two things:
  • A significant solubility of one component within the matrix, typically on the order of several weight percent.
  • A significant decrease in solubility limit of that component as temperature  decreases.
But another aspect is required for the precipitates to have the desired hardening effect: the formation of the precipitates must also result in a local distortion of the crystal structure near the precipitates. This lattice distortion is responsible for the impeding of dislocation motion (i.e. making plastic deformation more difficult) that we measure as a hardening effect.

Precipitation hardening of a sample is performed in two steps:
  1. Solution heat treating
  2. Precipitation heat treating
In the first, the sample is heated to a sufficiently high temperature that all of the solute phase has dissolved into the matrix, this solid solution is then rapidly cooled to a low temperature (often room temperature) to prevent any diffusion or coalescence of the solute. The result is a non-equilibrium state with a distributed solute, and the resulting material is usually soft and weak.

The second stage consists of heating to an intermediate temperature where diffusion rates are significant, and holding at that temperature for a given amount of time. The time, called the aging time, is based on the desired amount of precipitation of the solute from the non-equilibrium solution. This amount of precipitation, and therefore the aging time, is based on the desired strength or hardness. Once the desired amount of time has passed, the sample is cooled to room temperature. There are actually some non-ferrous alloys that can undergo precipitation hardening at room temperature over moderate and even short periods of time: they are generally quenched below room temperature and stored at low temperatures.

Maraging Steel

One phrase that folks who've done a lot of modern fencing may recognize is maraging steel: there are several olympic fencing weapons manufacturers who use this prominently in their marketing. So what is this stuff, and why is it used? (For the record, I've no idea how you're supposed to pronounce it: I suppose 'Mar-aging', for reasons that follow)

Maraging comes from 'Martensitic' and 'Aging': and the material is best known for being ductile while also having a high toughness and strength (relative to other steels). However, they tend to not hold an edge well and therefore is not used in swords that are meant to be sharpened. The reason for this is that these steels are actually very low carbon steels whose principle alloying element is nickel (15-25 wt%), and also commonly include cobalt and molybdenum. The strengthening mechanism of these steels is through the formation of intermetallic compounds from the solution as opposed to the formation of carbides in carbon steels.

Maraging steels are generally produced by first heating the sample to 820 °C to form an austenitic microstructure, then quickly quenching the sample in air to room temperature. This results in a soft, dislocation-heavy martensitic structure. Next, the material is precipitation hardened over several hours to produce dispersed intermetallic precipitates. However, if the treatment goes on too long, the martensite can decompose and many of the desirable properties of the material will be lost.

Some are probably wondering now: are the maraging steel fencing blades actually different than their carbon-steel counterparts? Well, the short answer is this: Yes, there is a difference but if both blades are FIE homologated, the main practical difference will be in useable life and abuse tolerance. The maraging steel blades may tend to have a longer useable lifetime and greater tolerance for abuse (e.g. takes more to take a permanent set). However, this all depends on the particular blade, usage and maintenance. All blades must be properly maintained, inspected, and replaced when appropriate and no material can make up for lack of proper care.

Also, I have no idea why the product of maraging isn't called 'maraged steel'. I can only assume it is due to the fact it was coined by engineers, and as a group we are not known for excellent grammatical skill.

Surface Treatments

In addition to performing a hardening operation on an entire piece, there are treatments which only affect a small region near the surface of a part. These are known as surface treatments or surface hardening. Let's look at two common methods:

Case Hardening

Case hardening is a surface treatment that is intended to improve the surface hardness and fatigue life of steels. It is accomplished by introducing carbon (carburization), nitrogen (nitriding) or boron (boriding) into the surface of a part via diffusion, producing a surface layer (the case, hence the name) whose hardness is greater than the underlying metal. This layer is usually on the order of 1mm deep. The local structural changes (residual stresses) can actually improve the fatigue life of the entire part.

The process itself usually consists of exposing the part to be hardened to a carbon, nitrogen or boron-rich compound and heating it to increase the diffusion rate. The concentration gradient of carbon, for example, between the compound and the piece acts as a driving force for the diffusion into the surface.


Another method for improving fatigue life and hardness of a part is called shot-peening. In essence, a part is blasted with hard beads that act as tiny ball-peen hammers: they each plastically deform the surface of the part, introducing residual compressive stresses in the surface. One reason this process is used is to counteract the stress concentrations (small notches and scratches) introduced during manufacturing. It also tends to produce a matte finish that can be appealing in some contexts.

In Closing: We've hardly scratched the surface

As with the alloy and phase diagram discussion, this is hardly an exhaustive discussion. But hopefully you have learned a bit of the rather complex world of which Fiore spoke when he mentioned the 'temper of iron'.

Next up, we'll get back to some classical mechanics!


  1. Hiya! I know this is kinda off topic but I'd figured I'd ask. Would you be interested in exchanging links or maybe guest authoring a blog post or vice-versa? My blog covers a lot of the same subjects as yours and I think we could greatly benefit from each other. If you might be interested feel free to shoot me an e-mail. I look forward to hearing from you! Terrific blog by the way!
    CNC Machining edmonton

    1. Thanks! Unfortunately, I just don't have the time I used to so this blog is sort of dead.

      But I'm glad people still find it and like it!

  2. Highly beneficial information particularly the last part. I take care of this kind of information much. I was seeking this specific information for a long time. Thanks and good luck.
    Safety Nets Manufacturer