There is a really nice question about quenching swords in dragon blood:

Quenching swords in dragon blood; why?

That made me curious about a more realistic example: Mercury

Could this substance be used to quench swords ending up with some improved properties in comparison with conventional quenching liquids?

Toxic fumes shall not be a problem here, either there are some protective measures in place, or the work is done by cheap slave-goblins or whatever, so it does not matter.

The boiling point of Mercury is 357 °C, the thermal conductivity is 8.3 W/mK, In a more medieval setting water would be used which has a boiling point of 100 °C and thermal conductivity of 0.597 W/mK, so judging from the answers of the linked question Mercury would give a better temperature control. I don't know what kind of oils are used for quenching nowadays but I am curious about this comparison as well.

My initial guess about problems would be the formation of soft amalgamates on the surface, though these could be polished off if the reaction stays at the surface. The swords can be forged out of steel or some other metals like bronze if you think it opens up possibilities for interesting reactions.

  • 5
    $\begingroup$ I'm not a metallurgist or metalworker, but as I understand it, the point is to cool it quickly to get a hard edge, but if you cool it too quickly you get too many small crystals, the blade will be brittle and shatter when you try annealing it $\endgroup$ – nzaman Mar 8 '19 at 14:01
  • 42
    $\begingroup$ "or the work is done by cheap slave-goblins" you are CEO material. $\endgroup$ – Renan Mar 8 '19 at 14:01
  • 10
    $\begingroup$ @Renan ah yes, that is just what a Chaotic Evil Overlord would do! $\endgroup$ – VLAZ Mar 8 '19 at 14:06
  • 4
    $\begingroup$ something like this? disneyparkhistory.files.wordpress.com/2014/09/freddie.jpg $\endgroup$ – Guran Mar 8 '19 at 14:10
  • 4
    $\begingroup$ This is not a very good question. Asking if real world elements and real world techniques go well together has a simple answer if one cannot find examples of it being used in real life: No, otherwise someone would already be doing it. And if there are examples of it being used online then it is not a worldbuilding question either way. $\endgroup$ – Mindwin Mar 8 '19 at 17:22

Mercury is heavy. Specifically, it has a density of about 13.5 g/cm3 - as opposed to steel, which varies but generally hovers around 8 g/cm3, less than two-thirds as much. (Compare also to the density of water, which is 1 g/cm3 by definition.) In order to quench your sword in mercury, you need to displace more than one and a half times its own weight in mercury - and you need to put a corresponding amount of pressure on the blade.

However, quenching is done while the metal is still hot and partially malleable. Shoving it into a pool of mercury is going to put large and unexpected stresses on the blade right as it's cooling, which is generally a Bad Thing. It would be frightfully easy to twist or fracture the blade and ruin it.

  • 3
    $\begingroup$ @Eth I'm not sure offhand, but I'd still be worried. The mercury would still be putting a lot of weight on the steel, just from a different direction. $\endgroup$ – Cadence Mar 8 '19 at 15:32
  • 6
    $\begingroup$ @And How do you even make a smeltery for Tungsten? $\endgroup$ – T. Sar Mar 8 '19 at 18:17
  • 8
    $\begingroup$ Um, displacing 1.5 times its own weight gives it the experience of 50% antigravity. It will have less stress in many senses tha it has just hanging around in the air, or being suspended in water (-50% is smaller in absolute magnitude than 88%). Are you talking about something else? Like viscosity? $\endgroup$ – Yakk Mar 8 '19 at 18:47
  • 2
    $\begingroup$ I think the "stress" put on the metal here is being given too much weight. When forming a sword, it's not spaghetti-noodle soft. The metal is strong enough that it must be hammered or otherwise beaten into shape. Putting less downward pressure on the blade to quench it than upward pressure normally required even to lift it off the anvil isn't going to be a problem. $\endgroup$ – JPhi1618 Mar 8 '19 at 20:57
  • 7
    $\begingroup$ The point of the people talking about the "antigravity" effect of buoyancy on the blade is: There is more stress being applied to the blade when it's picked up than there would be when holding the blade under the mercury. There would be less stress on the blade while it's cooling than if the blade were being held under water or mineral oils, not more as this answer states. $\endgroup$ – Ghedipunk Mar 8 '19 at 21:26

If the point of quenching were to cool it as fast as humanly possible, we'd use liquid nitrogen. Quenching is the process of cooling it at the appropriate speed — the faster you quench, the sharper the blade and the more brittle the blade. The slower you quench, the softer the blade but the less likely it is to break. That's why katanas have a hamon — the edge was quenched in water quickly while the back of the blade is coated in clay and cools slower. This gives the katana a relatively sharp edge with a more flexible, stronger backing without using significant extra steel (iron/steel was rare in Japan, and had to be conserved, as opposed to Europe where the backs of single-edged blades were just thicker for the same reasons).

So, the goal with quenching is to get the right balance of ability to hold an edge (fast cooling) and ability to take a blow without shattering (slow cooling). Mercury is likely to cause too many headaches to provide any significant benefit.

Also, I'm not a metallurgist, but I know that most of the mercury-metal reactions I've heard of don't stay on the surface, but the mercury travels deeper into the metal (see: mercury-aluminum, mercury-gold, etc.) so polishing off any problems might not help.

  • 16
    $\begingroup$ "Mercury is likely to cause too many headaches" +1 $\endgroup$ – user39548 Mar 8 '19 at 19:12
  • 1
    $\begingroup$ I saw in a documentary (for whatever that's worth), that Japanese swordsmiths each had their "trademark"/"signature" hamon, formed with specific clay mixtures and patterns; the hamon could be used to differenciate a genuine weapon from an otherwise near-perfect reproduction. Thought that was pretty cool =) $\endgroup$ – Mathieu Guindon Mar 8 '19 at 19:16
  • 4
    $\begingroup$ It might be good to note that this is explicitly why oil is used for most quenches nowadays - you can heat it warmer than water so there's a lower temperature differential and slower cooling than with water. Without fantastic temperature control of the blade, water quenches can seriously damage a blade by introducing a lot of stress and can cause excessive hardness (brittle). Oil leads to a slightly less hard blade, which is good because even poorly quenched steel is pretty damn hard - it needs that little flex to be durable. $\endgroup$ – Delioth Mar 8 '19 at 22:25
  • 2
    $\begingroup$ I don't think liquid nitrogen would cool a blade faster than water due to Leidenfrost effect. $\endgroup$ – Tomáš Zato - Reinstate Monica Mar 10 '19 at 11:57

The point of quenching is to use an heat sink to rapidly subtract heat from a hot piece of metal, so that a certain phase transition happens. When applied to steel, quenching is used to freeze an otherwise unstable crystallographic phase, and the freezing is due to the rapid cooling.

The boiling point of mercury is, as you note, higher than water. This implies that the freezing would be slower or not happen at all. A slower freezing might be wanted for metallurgic reasons (lower stress on the structure, better properties for the particular usage) and this is the reason why sometimes specific oils are used, but no quenching at all simply defies the purpose of quenching.

About the formation of amalgama, I think you are underestimating the extent of the damage: the quenched layer is the superficial one and the effect of quenching fades away while going more in depth, thus scraping away the surface would remove also the hardened layer. Plus, removing a layer from a stressed material will likely result in induced cracks. For this very reason quenching of worked metal is done after all material removing steps in the production process.

  • 4
    $\begingroup$ Mercury also has a much lower specific heat capacity than water so will certainly not cool the steel as quickly. Water has an unusually high specific heat capacity, which is one of the reasons it's so useful as a material :) hyperphysics.phy-astr.gsu.edu/hbase/Tables/sphtt.html $\endgroup$ – Ynneadwraith Mar 8 '19 at 15:13
  • 4
    $\begingroup$ The boiling point isn't really that important. While the latent heat involved in the transition of the water to vapour means that some more heat is transferred, it still depends on the heat capacity of the water to make this happen. It's a slight advantage in fact because if all the water turned to vapour there'd be no water left and just steam which is less good a conductor, but since we know real-world metal-workers can quench without this happening it's a pretty moot advantage compared to the more pressing matters of conductivity and density. $\endgroup$ – Jon Hanna Mar 8 '19 at 16:15
  • $\begingroup$ "For this very reason quenching of worked metal is done as last technological step." James says not quite in the chat. $\endgroup$ – user39548 Mar 8 '19 at 19:12
  • $\begingroup$ Quenching definitely isn't the last step. After quenching you have to, at least, temper the steel to cut some of the brittleness and relieve some of the massive internal stress that quenching causes, and sharpen the blade. You actually want to put most of the blade on an already-quenched piece of steel, since a thinner piece is more likely to warp during quenching. The quench also definitely isn't just surface-layer, it goes some distance in. $\endgroup$ – Delioth Mar 8 '19 at 22:29
  • $\begingroup$ @Delioth, after quenching one can do an annealing to release some of the stress, but that doesn't involve removal of any material. I made it clear in my edit $\endgroup$ – L.Dutch - Reinstate Monica Mar 9 '19 at 3:40

A couple of notes: mercury generally doesn't form amalgams with iron, see for example https://chemistry.stackexchange.com/q/28916/54697 where a reference is given for low-temperature amalgams. My recollection is that it was sold in units of steel 76-lb flasks so one might think that the mercury isn't going to mess up the steel very much. Now if meteoroic iron were used with its relatively high nickel content the situation might be different; I don't know.

Water might not be as effective a coolant as one might think due to the Leidenfrost effect. I know that I have stuck my fingers into liquid nitrogen for a couple of seconds without harmful effects, but doing the same with liquid propane may be more risky if not immediately injurious. Lower boiling temperature doesn't necessarily imply faster cooling. Don't try either experiment!

Looking at Bergman, Lavine, Incropera, and Dewitt, Fundamentals of Heat and Mass Transfer, sixth edition, John Wiley and Sons, New York 2011, p. 581, we see that $$\overline{Nu}_D=\frac{\bar hD}k$$ Where $\overline{Nu}_D$ is the average Nusselt number for a cylinder of diameter $D$, $\bar h$ is the average heat transfer coefficient, and $k$ is the thermal conductivity of the fluid. Then a correlation for $\overline{Nu}_D$ is given as $$\overline{Nu}_D=\left\{0.60+\frac{0.387Ra_D^{1/6}}{\left[1+ \left(0.559/Pr\right)^{9/16}\right]^{8/27}}\right\}^2$$ For example, at $400\,°C$ table A.5 gives a Prandtl number of $Pr=152$ for engine oil and $Pr=0.0163$ for mercury. The Rayleigh number is given on p. 573 as $$Ra_D=\frac{g\beta\left(T_s-T_{\infty}\right)D^3}{\nu\alpha}$$ Where the gravitational acceleration $g=9.81\,m/s^2$ on earth, the volume thermal expansion coefficient for oil is $\beta=0.70\times10^{-3}/K$ while for mercury it's $\beta=0.181\times10^{-3}/K$, and maybe the difference between the surface temperature and the free stream temperature is about $T_s-T_{\infty}=500\,K$. We are hoping that the sword will be reasonably approximated by a cylinder of diameter $D=5\,cm$ for the purposes of convective heat flow analysis, the kinematic viscosity of oil is $\nu=10.6\times10^{-6}m^2/s$ and for mercury $\nu=0.0882\times10^{-6}m^2/s$ and the thermal diffusivity of oil is $\alpha=0.695\times10^{-7}m^2/s$ and of mercury is $\alpha=54.05\times10^{-7}m^2/s$.

Plugging in all these numbers for oil we get $Ra_D=5.82\times10^8$ so $\overline{Nu}_D=136$ while for mercury $Ra_D=2.32\times10^8$ and $\overline{Nu}_D=32.8$. But it's the average heat transfer coefficient that is proportional to the rate of heat flow and for oil the thermal conductivity $k=134\times10^{-3}\frac W{m\cdot K}$ so $\bar h=364\frac{W}{m^2K}$ while for mercury $k=9800\times10^{-3}\frac W{m\cdot K}$ so $\bar h=6430\frac W{m^2K}$.

Metals can use their conduction band electrons to conduct heat as well as electricity so mercury looks like it might be $20\times$ as effective as oil at transferring heat by convection. Now, whether this is a good thing or a bad thing is a question for the blacksmith to answer.

  • 1
    $\begingroup$ For a sword, I'd use a value below .5 cm. Why? The relevance of a cylinder in the equation is about the distance of a given amount of material from the surface (where the heat exchange with the outside happens). Virtually all of the material of something as large as a classic broadsword is within a fraction of cm from that surface. Since NuD scales as D in the equation above, we'd expect all of those values to shrink by a factor of 10. $\endgroup$ – Derrell Durrett Mar 11 '19 at 17:48

Not for the blades.

Water has a very good feature - it boils at 100 C, quickly taking away a large amount of heat which would not be possible with liquids that are boiling at a higher point.

Quenching in a high point boiling liquid like oil is definitely a thing, but this leads to lower hardness - not something we would want for a sword blade.

However, mercury can definitely be used in other metallurgical processes directly related to quenching - soaking and tempering, which require temperatures higher than 100 C.

  • 3
    $\begingroup$ Just adding a blacksmithing perspective - we normally use oil with a high flash point for quenching regular high-carbon swords/tools, because using water would make them be too hard, and be likely to shatter. It is a good point that mercury may be useful for those other processes though, assuming minimal chemical reactions/adhesion between the two. $\endgroup$ – HammerN'Songs Mar 8 '19 at 19:31
  • $\begingroup$ @HammerN'Songs thanks for your expertise - looks I've got a wrong idea from Wikipedia article. $\endgroup$ – Alexander Mar 8 '19 at 19:40

In addition to the density and specific heat issues raised in other answers, steel is susceptible to liquid metal embrittlement. If exposed to a liquid metal (say, mercury) and at the same time a tensile stress (say, from quenching) catastrophic fracture will occur.

If you are interested in a steel treatment that is toxic to workers, look into cyanide salts. They are both a great way to carburize steel and extremely toxic. (Carburizing adds carbon just to the surface, creating a part with a harder wear-resistant surface while maintaining a softer crack-resistant core.)


Lots of answers that get part of the chemistry and metallurgy right, but that also either miss key features or are wrong in some conclusions (some facts are right, but the reasons for those facts lead to wrong conclusions). This won’t be exhaustive since I haven’t worked with mercury much, and it’s not in my home library, but this will fill in many of the gaps in other answers.

Quenching in oil versus water leads to lowered hardness (as other answers indicate), but contrary to some answers (and in alignment with others) this is neither good nor bad specifically.

Hardness of steel will alter how the metal interacts with other substances: if you hit a softer object you will cut it (in general), but increased hardness increases the risk of both chipping and breaking.

Other answers indicate that Japanese steel used differential hardening to get a hard edge and flexible backing, which while true, was because of the poor quality of iron ore and metallurgy throughout the Tokugawa period, which wasn’t significantly improved until the Meiji Restoration in the late 1800s, by which time sword development had largely ceased. While very good steels can be obtained from Japan today, it’s not because of their inherently good ores, but despite them. The opposite was true in Spain and Sweden, which had naturally occurring high quality ores that lead to superior steels with the same processing. This is a large part of why arms and armor development was so different between Western Europe and Japan. With modern steels differential hardening as seen in Japan adds negligible benefits.

One comment suggested that water was less effective for hardening because of the rate of absorption. This is incorrect: waters primary drawback is due to hydrogen embrittlement caused by the absorption of elemental hydrogen into the steel matrix, which weakens it in every regard. For maximum hardness you would want to use a high thermal conductivity liquid to quickly freeze the matrix into place forming a higher magnetite to austenite ratio.

Depending on desired features, a “best” method would be edge hardening with a liquid that doesn’t contain hydrogen and with a high thermal conductivity (pure magnetite) paired with a spine of softer spring steel (much more austenite). Assuming that mercury (or dragons blood) doesn’t form an alloy (metallurgists, please chime in if you know).

Also keep in mind that there is no “best” steel, everything is a trade off. For example, if I had to make my own sword from scratch, there’s no way I’d try any kind of laminated or pattern welded blade, but just get a nice Swedish or Spanish ore and forge a monosteel blade and be happy with it as is. The extra steps would be better, but not worth the cost. If you’re interested, different oils have different thermal conductivity, and will make different hardness blades, but remember that the real reason to use oil instead of water is to avoid hydrogen embrittlement, not slow cooling, as you can always address that through annealing (which is required under all circumstances anyway).



Mercury offers no advantage over oils and has many moe drawbacks. the vapor point of mercury is lower than many quenching oils, so sou can't get it any hotter than oil. Cooling faster offer no benefit if you are still cooling to the same temprature, in fact it makes it worse. It leads to stress fractures, even water cools too fast for better quality steels, and waters thermal conductivity is nearly an order of magnitude less.

The toxicity cannot be hand waved either quenching is a technical process it can't be done by unskilled labor and it will kill your skilled labor very quickly. There is no medieval protection from mercury vapor.

Lastly getting enough mercury to actually quench a blade would cost more than a small army, mercury ores contain very small amounts of mercury, and are not easy to process.


I have made several swords and repaired them many times. I have tried several methods of quenching but nothing beats just throwing it in a fast flowing stream. The only comparable sword was a rough clip sword I hastily made out of scrap mild steel from a building site using a mold carved in a block of slate and quenched in the sea. I usually case harden my swords in a mix of charcoal, hay and animal bone. The finished result was as good as any EN45 sword I have.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.