Use water for plain medium-carbon shaft steel with no added alloy, in a simple shape of medium thickness that must be hard all the way through. Use oil for alloy steel (chromium, molybdenum, nickel) and for parts with a hole, thread, sharp corner or a thin wall next to a thick one. Water cools harder: it hardens deeper, but it also cracks and warps more. So the rule is to use the mildest bath that still gives the part the hardness it needs.
What does the quenching bath do to the steel?
The bath pulls heat out of the part fast, and that speed is what hardens the steel. Think of the blacksmith who takes glowing iron out of the fire and plunges it into the bucket, with that hiss. Quenching is that motion: heat the steel until it glows and cool it fast so that it hardens.
In the furnace, the iron atoms rearrange and open room for carbon to dissolve between them. This hot arrangement is called austenite.
If the steel cools fast, the carbon stays trapped between the iron atoms and martensite is born, the hard structure that quenching is after. If it cools slowly, the carbon has time to regroup, and the steel goes soft again. The step-by-step, with a drawing, is in the guide to oil quenching.
The liquid in the tank that removes this heat is the quenchant. In industry, the four most used are water, brine, polymer in water and oil.
After quenching comes tempering, a gentle reheating that relieves the stress and brittleness that martensite leaves in the part.
Choosing the wrong bath is expensive, because the part reaches the tank after hours of machining (cutting and drilling on the machine). A part that comes out soft can still go back to the furnace, at a cost and with a risk of warping. A cracked part becomes scrap.
For those who need the number: before the bath, carbon steel and low-alloy steel are usually heated to between 800 and 900 °C, depending on the steel.
How severe is each quenchant?
The force with which a quenchant removes heat from the part is called severity. With a still bath, the order runs from brine, the most severe, to water and, last, oil, the mildest; polymer in water sits between water and oil, depending on the concentration.
Brine is water with sodium chloride, common salt, dissolved in it. Polymer in water contains a type of plastic that dissolves in the water and thickens it.
With still baths, water removes heat 3 to 4 times harder than an ordinary oil. The difference comes from the three phases in which the part loses heat; anyone who has seen a drop of water dance on a hot stove plate, without boiling away, knows the first one.
Vapor blanket. The liquid that touches the glowing part turns to vapor and forms a blanket around it. The blanket insulates, and the heat leaves slowly.
Boiling. The blanket breaks, the liquid touches the metal and boils in bubbles. Each bubble carries heat away, and this is the fastest phase.
Convection. The surface is already cooler than the liquid's boiling temperature, the bubbles stop, and the heat leaves only with the liquid that circulates around the part. It is a slow phase.
The second variable weighs as much as the liquid: tank agitation. An agitated bath breaks the vapor blanket earlier. In convection, it carries the heat away faster, like hot soup that cools faster when you stir it.
So an oil with violent agitation comes close to still water. The severity ranking only holds when you compare baths with the same agitation.
For those who need the number: the classic measure of severity is the Grossmann H factor, in which still water is close to 1. It serves to compare quenchants.
| Bath agitation | Oil (H) | Water (H) | Brine (H) |
|---|---|---|---|
| None | 0.25 to 0.30 | 0.9 to 1.0 | 2.0 |
| Light | 0.30 to 0.35 | 1.0 to 1.1 | 2.0 to 2.2 |
| Moderate | 0.35 to 0.40 | 1.2 to 1.3 | no data |
| Good | close to 0.4 | 1.4 to 1.5 | no data |
| Strong | 0.5 to 0.8 | 1.6 to 2.0 | no data |
| Violent | 0.8 to 1.1 | 4.0 | 5.0 |
Source: Grossmann table (ASM, 1964), reproduced in Thermal Processing, May 2023. The cooling-curve test, further below, counts for more than the table.
Figure 1. The strength of each bath by the Grossmann H factor, still and with violent agitation: oil with violent agitation comes close to still water. Values for mineral oil without additive; polymer varies with the concentration.
If water is stronger and hardens more, we still need to understand why it also cracks more.
Why does water crack more than oil?
Because, at the moment the steel starts to harden, water is still cooling the part in the fast phase, and oil, in general, is already cooling slowly. Think of a boiled potato dropped in cold water: the skin cools in seconds, and the middle stays hot for minutes. A steel part also cools first at the surface.
Two facts complete the picture. First, martensite starts forming at a temperature of its own for each steel, the Ms (from martensite start), at around 300 °C in many common steels.
Second, on turning into martensite the steel "swells" slightly, because martensite takes up a little more room than austenite.
Now put the three phases together. Water boils at 100 °C, so the fast phase lasts until the surface gets near 100 °C, well below Ms. Oil boils at a much higher temperature, between about 230 and 480 °C depending on the oil: the bubbles stop when the surface cools to that range, and in general the slow phase has already begun by the time the surface reaches Ms.
Figure 2. In water, boiling lasts until about 100 °C and covers the Ms range. Oil boils between about 230 and 480 °C, depending on the formulation, and the fast phase ends when the surface cools to that range. Qualitative form: the ranges vary with the formulation, the agitation and the steel.
In water, this sets up a chain of two steps:
The surface turns to martensite while the core is much hotter, which leads to a hard outer layer around a core that has yet to transform.
When the core transforms, it swells and pushes against the already rigid surface, which leads to tension (a force that tends to pull the material apart) at the surface and to cracking at a sharp corner, a hole or a change of thickness.
More carbon makes martensite more brittle and raises this risk.
In oil, when the surface reaches Ms, the core is closer to the surface temperature, and the stress drops. Oil also cracks parts if it is too severe for them, but with a much wider margin.
Figure 3. Surface and core of a part leaving 850 °C, in still water and in still oil. The red bar is the difference between the two when the surface reaches 300 °C, a typical Ms for a medium-carbon steel. Time is not to scale and is the same in both graphs. Qualitative form.
How this shows up in your plant: cracks that appear minutes or hours after quenching, before tempering; a long shaft that is warped; and, in water, soft spots (regions without hardness) in holes and cavities, where the vapor blanket stays trapped.
Water above about 25 °C, or contaminated with oil, soap or dirt, lengthens the vapor blanket and makes the hardness vary from part to part.
For those who need the number: Ms is usually between about 200 and 400 °C and falls with more carbon and more alloy. Quenching oil boils, depending on the formulation, between about 230 and 480 °C; convection begins when the surface cools to that range (Herring and Balme, Gear Solutions, 2007).
So why does anyone still quench in water?
When is oil not enough to harden the steel?
When the steel has little alloy or the part is thick: in those cases, oil does not cool fast enough to harden the part as deep as it needs.
Each steel has a minimum cooling speed to become martensite, the critical cooling rate. Below it, the carbon escapes and the steel stays soft; the reason is in the guide to oil quenching.
Carbon decides how hard the martensite gets.
The ease with which steel hardens down to the core is called hardenability, and what raises it is alloying elements (chromium, molybdenum, nickel): with them the steel hardens even with gentler cooling. Boron, in tiny amounts, does the same in some low-alloy steels used in bolts, without removing the section-size limit.
Manganese, which every carbon steel already contains, helps a little but is not enough for oil to harden a medium-size part.
Thickness also counts: a big potato takes much longer than a small one to cool in the middle. The thickness of the part at the point that matters is called the section, and in a thick section the core may not reach the critical rate and stay soft.
Hence the rule: steel with little alloy, or a thick section, calls for a stronger bath; alloy steel hardens in oil.
Steel names help to read this. Numbers such as 1045 and 4140 are SAE designations, from the American society of automotive engineers. The first two digits give the family (10 is carbon steel with no added alloy; 41, steel with chromium and molybdenum). The last two give the carbon in hundredths of a percent: 1045 has about 0.45%.
Carbon steel has three carbon ranges, and each calls for a different path:
Low carbon, such as 1020: barely hardens if quenched without preparation, and usually gains carbon at the surface beforehand, in a furnace treatment called carburizing.
Medium carbon, such as 1045 and 1050: the typical case for water.
High carbon, above about 0.6%: forms harder, more brittle martensite, cracks easily in water and, in a thin part, usually goes to oil; in a larger part, it calls for alloy steel.
Highly alloyed steels, such as tool steels and martensitic stainless steel, are quenched in oil, air, gas or salt, depending on the steel.
Figure 4. In the same section, 1045 in oil barely hardens the surface, in water it hardens deeper, and 4140 hardens in oil almost to the center. Qualitative drawing.
For those who need the number: hardenability is measured by the Jominy test (ASTM A255), with a standard bar cooled by a water jet at one end. Usual carbon ranges: low, up to about 0.3%; medium, from about 0.3 to 0.6%; high, above about 0.6%.
With that, we can answer the most common question.
Is 1045 steel quenched in water or in oil?
It depends on the section and the shape of the part. 1045 has little alloy, so thickness decides almost everything. As an order of magnitude, not as a limit:
Thin section: that of a pin, a bolt or a blade, which cools all the way through almost at once.
Medium section: that of a shaft or a gear as thick as a tool handle.
Thick section: that of a shaft or block as thick as an arm or more, where the core is slow to lose heat.
The exact boundary for each steel comes from its own Jominy test.
In each case:
Thin part: oil may be enough, and it is the way to go when the shape cannot take the shock, accepting lower hardness in the core.
Medium section that needs hardness in the core: water or brine, because oil does not cool fast enough for this steel.
Thick section: not even water hardens the core of 1045.
For the thick section, there are two ways out. The first is to switch to an alloy steel, such as 4140, which hardens in oil. In a very thick section, not even 4140 reaches the center, and a steel with more alloy comes in, such as 4340, with nickel, chromium and molybdenum.
The second applies when a hard surface is enough: a 1045 part can be quenched only at the surface by induction, in which an electromagnetic field heats only the outer layer, which is cooled right afterwards.
And if the 1045 part has a hole, a thread or a sharp corner and cracks in water? Then a middle ground between water and ordinary oil comes in, described in the next section and tested beforehand on production parts. In the opposite direction, with alloy steel, switching from oil to water rarely brings hardness that makes up for the cracking risk.
Maximum hardness comes from the carbon; the depth it reaches depends on the bath, the alloy and the section.
For those who need the number: hardness at each depth changes with the steel lot and the agitation. It comes from the Jominy curve of the steel you bought, crossed with the part diameter and the H factor of the bath.
Where do brine and polymer come in?
They come in when neither plain water nor ordinary oil will do. Brine gives more uniform cooling than water. Polymer and accelerated oil, a faster oil than the ordinary kind, sit in the middle ground between water and oil.
Brine. The salt makes the vapor blanket break earlier, which makes the cooling more uniform and reduces soft spots. Its gain is in the hot part of the cooling, where water still forms vapor. In the Ms range, it cools like water, so the cracking risk stays the same.
On the other hand, brine corrodes the part and the equipment and calls for washing right after quenching.
Polymer in water. On the hot part, the polymer forms a film around it and slows the heat leaving. The concentration adjusts the severity between that of water and that of oil, and the bath does not catch fire.
Because severity depends on the concentration, polymer calls for routine control in the tank. The instrument used is the refractometer, which measures how much light bends when it passes through the liquid.
Accelerated oil. It contains additives, chemicals in small doses, that break the vapor blanket earlier. It cools fast in the hot part and, in convection, generally stays mild like an oil. It serves steel that is hard to harden, in a part of complex shape: it aims to harden almost like water without cracking like it, the dilemma of Figure 3.
By speed, quenching oils form three groups:
Conventional oil: the mildest, for steel that hardens easily.
Medium-speed oil: the middle ground, for steel of intermediate hardenability or a part of medium section.
Accelerated oil: the fastest of the three, for steel that is slow to harden or a thicker section.
Oil is fuel, and hot oil in the tank can catch fire. Water falling into hot oil turns to steam suddenly, foams and splashes oil. So the oil in the tank must stay dry, and its water is measured by the Karl Fischer method, a chemical reaction that counts the water.
Where oil cannot be kept on site, polymer takes its place, because it is not flammable and the concentration adjusts the severity.
For those who need the number: brine typically works near 10% sodium chloride, and from there up the vapor blanket practically disappears (MacKenzie, Thermal Processing, July 15, 2024). The most common polymers, the polyalkylene glycols (PAG), come out of solution at the hot surface and form the film.
With the four quenchants on the table, it is time to choose.
Which quenchant should you use for each steel?
Use the mildest quenchant that still hardens the part as deep as the design asks. The less alloy the steel has, the more severe the quenchant must be; the more complex the shape, the greater the risk of that severe quenchant.
| If the part is | Use | Why |
|---|---|---|
| Medium-carbon steel (1045, 1050) in a simple shape (shaft, pin, block) and medium section, with hardness in the core | Water or brine, between 15 and 25 °C, with agitation | With little alloy, only a severe quenchant takes the hardness inward; the simple shape can take the stress |
| Carbon steel in a thin part, including high carbon | Oil, checking the hardness on a test lot | The thin part hardens in oil; high carbon cracks easily in water |
| Carbon steel of complex shape (hole, thread, sharp corner) that cracks or warps in water, or steel that is hard to harden | Accelerated oil, or polymer at the concentration chosen by the laboratory test | Accelerated oil is fast in the hot part and mild in the Ms range; polymer sits between water and oil, depending on the concentration. Both crack and warp less than water |
| Alloy steel (chromium, molybdenum, nickel), including in a precision part | Conventional oil, or medium-speed oil when the section is medium; accelerated only when the thicker section does not harden in the other two | The alloy hardens with gentle cooling; cooling faster than needed only adds stress |
| A forged part (shaped while hot, by hammer or press) that goes straight to the bath, a part heated by induction, or a tank where oil is a fire risk | Polymer in water, at the concentration chosen by the laboratory test | It does not catch fire, and the concentration adjusts the severity |
Polymer is mild compared with water, but it usually cools faster than oil in the slow phase, when the surface passes through the Ms range. That is why alloy steel that is sensitive to cracking stays in conventional oil.
A part that warps or cracks a lot, with thin and thick walls together, may call for martempering: a hot oil bath in which surface and core equalize their temperature before cooling further. It is suitable for steel with enough alloy to harden in that slower bath. The Technical Team evaluates this case.
Low-carbon steel stays out of the table, because it gains carbon at the surface, in carburizing, before quenching. After that, the carburized part usually goes to oil, often accelerated oil; the Kelpen Oil Technical Team evaluates the case.
Figure 5. Decision tree for the quenchant: first the fire risk, then the alloy, the carbon, the section and the shape of the part.
We at Kelpen Oil formulate quenching oils in two lines: the Micro Temp line, with options for steels that are hard to harden, and the Austenity line, with conventional and accelerated quenching options for ferrous metals (those based on iron, such as steel). We also formulate a non-flammable polymer-in-water quenching fluid.
The quenching products are in heat treatment oils. For medium speed, martempering or any doubt about which product to use, talk to the Kelpen Oil Technical Team, which recommends by the steel, the section and the shape of the part.
And what about those who already quench in water and want to move away from it?
Can you swap water for oil or polymer without losing hardness?
Sometimes you can. The swap works when the steel has the hardenability to harden in a milder quenchant; in medium-section carbon steel, ordinary oil can leave the core soft.
The measure that decides is the cooling curve: the record of the temperature of a standard test part in the bath over time. That test part is called a probe, and the test is explained in the guide to oil quenching.
Two numbers from the curve decide the swap. The maximum cooling rate shows whether the candidate fluid still hardens the steel. The rate at 300 °C, the cold part of Figure 3, shows the risk of cracking and warping.
The safe path has three steps:
Compare the curve of the candidate fluid with that of the current bath.
Quench a test lot with parts from your own production.
Check the hardness at the surface and in the core and look for cracks before changing the production, because the real part cools differently from the probe.
The Kelpen Oil laboratory in Arujá (SP) measures the cooling curve of oil by ASTM D6200 and that of polymer by ASTM D6482, on the IVF Smart Quench equipment. For other companies, it is a paid service, including with a fluid from another maker.
| Quenchant | What to measure | What it reveals | Standard or control |
|---|---|---|---|
| Oil | Cooling curve | Maximum rate, the temperature at which it occurs and the rate at 300 °C: whether the candidate oil hardens without cracking | ASTM D6200 |
| Polymer | Cooling curve with agitation and concentration | The same reading at the concentration and agitation of use | ASTM D6482; concentration by refractometer |
| Water and brine | Bath temperature, salt in the brine and cleanliness | A bath that is hot or dirty with oil, soap or grime lengthens the vapor blanket and gives soft spots | Process control |
The other tests are in the guide to laboratory tests, standards and equipment.
For those who need the number: ASTM D6200 is equivalent to ISO 9950 and uses a nickel-alloy probe 12.5 mm in diameter. The test records the maximum rate, the temperature at which it occurs, the rate at 300 °C and the times to 600, 400 and 200 °C.
Frequently asked questions
What is the ideal water temperature for quenching? Between 15 and 25 °C, with agitation and clean water; hotter or dirtier, the hardness varies from part to part.
Why does a part quenched in water crack hours later? Because martensite leaves the part under stress, and that stress can open a crack over time if the part is not tempered soon. Temper as soon as the part has cooled to near room temperature, not before and not long after. Before that, part of the steel has not yet turned to martensite, and it forms later, brittle, without having been relieved.
Does water quenching make steel harder than oil? In medium-section carbon steel, often yes, because oil can leave it soft. The ceiling comes from the carbon: in a thin part or an alloy steel that already hardens in oil, water adds little hardness and a lot of cracking risk.
Can I quench in burnt oil or engine oil? In production, it is not recommended: the severity of that oil is unknown. The additives in engine oil (detergents and others) change how it cools from one batch to the next, so the result does not repeat. Used oil can also contain water and fuel, and fuel lowers the temperature at which the oil vapor ignites, which raises the fire risk in the tank.
Glossary
Quenchant: the liquid in the tank that removes heat from the part (water, brine, polymer in water or oil).
Severity: the force with which the quenchant removes heat from the part; the classic number is the Grossmann H factor, with still water close to 1.
Martensite: the hard structure that forms when steel cools fast and the carbon stays trapped between the iron atoms.
Ms: the temperature at which martensite starts to form; it varies with the carbon and the alloy.
Hardenability: the ease with which steel hardens down to the core of the part.
Section: the thickness of the part at the point that matters; the thicker it is, the slower the core cools.
Brine: water with sodium chloride dissolved in it, the most severe quenchant among the cold liquids.
Talk to the Technical Team
The Kelpen Oil Technical Team evaluates the steel, the section and the shape of the part, the hardness asked for at the surface and in the core, the volume and agitation of the tank and the safety requirements of the site, and confirms the right quenchant and product. If the bath is already in use, the Arujá (SP) laboratory performs the analysis of the oil in use, with the cooling curve, as a paid service. Before changing baths, write down the steel, the thickest section of the part and the hardness the drawing asks for: those are the three data points that decide.
