Annealing Part 2 – Temper Annealing, Cycling, and Final Properties

Thanks to Ryan Rodriguez and Matt Peter for becoming Knife Steel Nerds Patreon supporters!

I was interviewed by Shawn Houston on his Youtube channel about the micrographs article. See the video here.

Background

Make sure you read the Annealing Part 1 article, or this information won’t make as much sense.

Temper Annealing

A modification of subcritical annealing that is sometimes called “temper annealing” requires a martensitic mirostructure instead of pearlite. Instead of air cooling or furnace cooling from high temperature, the steel is quenched to form hard martensite. Martensite has carbon evenly distributed throughout. During normal tempering, small carbides are formed and the martensite steadily drops in carbon content. The higher the tempering temperature, the lower the carbon in the martensite, the larger the carbides are, and the lower the hardness is. This process occurs much more rapidly than spheroidization of pearlite. In part 1 I showed micrographs of 1040 steel with subcritical annealing of pearlite that took 200 hours to fully spheroidize:

Image from [1]

Here is the same 1040 steel from above heated to 1290°F for 21 hours starting from martensite:

Image from [1]

The carbides are also smaller than in the typical subcritical anneal because the overall time required is shorter. The carbides that spheroidize first have a long period at high temperature for “Ostwald ripening” which was covered in Part 1.

To understand the mechanism by which tempering works, you can read my article on tempering. The tempering of martensite is relatively intuitive to anyone who heat treats knives: higher tempering temperatures means lower hardness. Here is a tempering curve for a simple carbon steel similar to 1095:

Image from [2]

Just like normal tempering, the subcritical temper anneal can be performed in 2-4 hours, relatively short when compared with subcritical annealing of pearlite.

Temper Annealing of High Alloy Steels

In studies of T1, M1, M2, and M42 high speed steels [3][4], optimized temper annealing was found to lead to finer grain size in the final heat treated product because the carbide distribution could be controlled to lead to a fine austenite grain size. The grain size comparisons were with a conventional “Transformation anneal” which is also called DET as explained in Part 1.

Image from [3]

T1 high speed steel with different grain sizes (a) 12 microns, (b) 8 micron, (c) 6.8 microns, (d) 5.7 microns. Image from [4]

The grain size of the austenite of the final heat treated steel is improved by providing more “nucleation sites” for austenite through more fine carbides. Below I have compared the number of austenite “nuclei” (red circles) that form in a steel with coarse carbides compared with how many form in the presence of many fine carbides:

Coarse carbides (black) in conventionally annealed steel leads to few austenite nuclei (red)

Fine carbides formed in temper annealing (fine black lines) leads to more austenite nuclei

When there is a higher density of austenite nuclei, the resulting grain size is finer. You can imagine those individual austenite nuclei growing until they impinge on each other; less growth is required to do that when there are more initial grains. The carbide size was also reduced through the temper anneal. A combination of finer carbides and finer grains leads to better toughness. Below is “bend fracture strength,” a type of toughness measurement, for M2 steel with temper annealing compared with transformation annealing:

Data adapted from [3]

They looked at a range of variables for using the temper anneal, including the austenitizing temperature, the tempering temperature, and the hold time for both of those steps. A sufficiently high austenitizing temperature is required to form martensite upon cooling prior to the tempering step. Yet higher temperatures leads to either grain growth or excessive dissolution of carbide. An optimum temperature for T1, M2, and M42 was found to be 900°C (1650°F) which is similar to the normal recommended temperature for transformation annealing in the datasheets, about 30°C more than the datasheet recommends for those steels.

Data adapted from [3]

They also found that there is an optimal tempering temperature. While a lower tempering temperature leads to finer carbides, they can be so small that they dissolve prior to being available to serve as an austenite nucleation site. So relatively high tempering temperatures were found to be optimal for leading to the finest possible grain size. Similar mechanisms are in play when it comes to time at tempering time. Too short leads to carbides that are too fine and readily dissolve. Too long and the carbides coarsen.

Data adapted from [3]

The final grain size also depends on the austenitizing temperature that is chosen, because more temperature means faster grain growth. However, the temper anneal led to finer grains across the range of austenitizing temperatures chosen.

Data adapted from [3]

The average carbide size of the heat treated steel was also found to be decreased from 11 microns in the transformation annealed steel to 5.5 microns in the optimized temper anneal.

It is important to note that at least one of these studies used temper annealing as a “re-anneal” meaning the steel had already seen a conventional anneal from the mill. So for stock removal makers there may be performance improvements possible by performing a temper anneal treatment. This process is somewhat similar to the “prequench” treatment that was described in this article. But the tempering step may lead to further improvements.

Cycling

Another method of annealing is to cycle above and below the critical temperature which can also spheroidize pre-existing pearlite. A 1060 steel (0.6% C, 0.3% Si, 0.5% Mn) was heated in a furnace at 810°C (1490°F) for 6 minutes followed by forced air cooling to room temperature. This was repeated up to 8-cycles and the microstructure and properties were measured. The pearlite can be seen to be spheroidizing through further cycles, after one cycle (top), five cycles (middle), and eight cycles (bottom):

Images from [5]

The cementite “lamellae” in the pearlite partially dissolve and spheroidize on heating to austenite, and then those carbides continue to grow on cooling by the Divorced Eutectoid Transformation. This continues until sufficient cycles are completed to fully spheroidize the prior pearlite.

A similar study [6][7][8][9] was performed on 1080 steel (0.79% C, 0.73% Mn, 0.33% Si) austenitizing at 775°C (1425°F) for 6 minutes followed by either cooling in still air, forced air, or quenched in brine. Forced air, and especially quenching, led to more fragmentation of the pearlite so that it spheroidized more rapidly with further cycles. The quenched specimens were pearlite-free after only the second cycle. Four cycles was not enough to lead to full spheroidization with either forced air or still air:

Data adapted from [8][9]

However, while cooling in still air led to a gradual reduction in hardness, further cycles with forced air cooling led to an increase in hardness. Quenching, of course, led to very high hardness due to martensite formation. Therefore, using either forced air or quenching would require a subcritical temper anneal to reduce hardness.

Data adapted from [9]

Fracturing the samples you can see a change in behavior because of the shift from pearlitic to spheroidized microstructure. When the steel is pearlitic there is a wavy “cleavage” fracture surface indicating cracks that grew through pearlite. With further cycles there is a shift to a dimpled surface which is the result of fracture around spheroidized carbide. There are also “craters” in the fracture which are formed around pearlitic-type cementite. With more cycles there is a decrease in the number and size of craters. The change in the cementite structure led to improved ductility in both conditions.

1-cycle (a), 2 cycles (b), 3 cycles (c), and 4 cycles (d). All for still air cooled samples. From [9]

Image from [9]

Another change that occurs from the cycling of the steel is grain refinement. This process is similar to multiple austenitizing and quenching cycles described in this article. With each successive cycle new austenite grains are formed, and the new grains are small if the steel is heated sufficiently rapidly to a relatively low temperature and not held for too much time (to prevent grain growth). The steel is then cooled down to “lock in” the new grain size with newly formed pearlite or carbides. The size of the pearlite, carbides, and grain boundaries is reduced and the density is increased. When reheated, new grains form again, only now there are more nucleation sites (grain boundaries, pearlite, and carbides) for austenite to form so that the new grain size is even finer. This works up to a certain number of cycles until a steady state is reached. Grain growth can occur in some cases with too many cycles. The 1060 steel grain size was found to be refined from 147 microns to only 7 microns with eight cycles.

1060 steel as-received (a), after 1 cycle (b), 3 cycles (c), 5 cycles (d), and 8 cycles (e). The pearlite (black) is replaced with cementite and the grain size was reduced with further cycles. Image from [5]

In a study on a low carbon 0.16% carbon steel [10], cycling was performed by heating to 910°C (1670°F) for 6 minutes and quenched. It was found that only one cycle was necessary to refine the grain size and that further cycles led to little change:

Image from [10]

Heat Treatment Response and Hardness

The process of austenite formation from an annealed microstructure is described in this article. Below are micrographs showing the process of austenite formation in 1080 steel. First in (a) there is a spheroidized carbide microstructure with ferrite. This is annealed steel. The steel was held at 1350°F for increasing amounts of time, so you can see the austenite (dark) replacing the ferrite; the ferrite is absent after 300 seconds in (e). After a 900 second hold (f), almost all of the carbide has been dissolved as well.

Images from [11]

I first covered the relative speed of austenite formation and carbide dissolution during austenitizing in this article. The coarser the cementite/carbide, the slower austenite formation and carbide dissolution is because carbon has to diffuse farther. Martensite actually transforms to austenite the fastest because it is essentially carbon-enriched ferrite which has approximately the same carbon content as the new austenite. Therefore little or no carbon diffusion has to take place. Very fine pearlite is a little bit slower, followed by coarse pearlite. The slowest is coarse spheroidized carbide. You can imagine the difference in diffusion distance by looking at these schematic images:

Martensite – evenly distributed carbon

Pearlite – carbon is located in the cementite bands, or “lamellae”

Spheroidized carbide – carbon is located in the carbides (black circles)

Because of the shorter diffusion distance, pearlite leads to shorter required hold times at the austenitizing temperature for full hardness:

Data adapted from [12]

The slower hardening response of the spheroidized microstructure is not necessarily a bad thing. A longer hold time can be used to ensure the full piece has reached temperature, for example. And the pearlite sample which has transformed to austenite more rapidly has more time for grain growth if given the same hold time. However, it is important to know the response to heat treatment that the prior processing has provided. For example, if some knives are made with stock removal using the as-received annealed microstructure, and some are forged and heat treated from a pearlitic microstructure, the steel will respond differently and will require a different austenitizing time-temperature combination for the same hardness and properties.

Knife Steel Nerds Studies

A finer microstructure in the annealed steel leads to finer carbides in the final heat treated steel for better toughness. Therefore, annealing procedures leading to a finer microstructure are likely to lead to the best combination of properties. We looked at different prior microstructures with toughness studies of CruForgeV and 52100.

CruForgeV Study

In the CruForgeV toughness study we looked at three annealing conditions:

1. Subcritical anneal – 1250°F for 2.5 hours after normalizing
2. Temper Subcritical anneal – 1450°F for 10 minutes, quenched in oil, 1450°F for 10 minutes, quenched in oil, 1250°F for 2.5 hours
3. DET – 1460°F for 30 minutes, furnace cool at 670°F/h to 1260°F followed by air cooling

The DET anneal was used for a broader range of heat treatments including different forging, austenitizing, and tempering temperatures. All of the annealing types were found to offer a similar hardness-toughness balance. The temper anneal gave slightly higher toughness in the longitudinal direction, but was not superior in the transverse direction, which makes me think it was simply experimental scatter that led to superior behavior of the temper anneal condition. None of these three anneals lead to a very coarse structure, so it may simply be that each was sufficiently fine for a good heat treatment response and toughness.

52100 Study

We also compared a few different prior microstructures in our toughness study of 52100. One condition was as-received 52100 which has a presumably conventionally DET annealed spheroidized structure (labeled “spheroidized”). One heat treatment used a pearlitic microstructure which was normalized at 1650°F for 10 minutes followed by an air cool to magnetic, then 1550°F for 10 minutes followed by an air cool, and then 1450°F for 10 minutes followed by an air cool. The rest of the samples were given the same DET anneal as the CruForgeV study which was determined by Verhoeven for a very fine spheroidized structure in 52100. The Verhoeven-DET samples (labeled “DET”) had a higher hardness and a higher hardness-toughness balance than the other two conditions:

The pearlite condition led to similar hardness as the spheroidized condition, the samples were harder because they were tempered lower. Showing tempering vs hardness there is no apparent difference:

We weren’t sure if the Verhoeven-DET samples led to a better hardening response or if it was because the spheroidized and pearlite samples used a shorter austenitizing time (10 vs 15 minutes). Knifemaker Shawn Houston heat treated some 52100 where two samples were heat treated as-received, either from 1475 or 1525°F. Another condition was normalized and given a Verhoeven-DET anneal prior to the heat treatment from 1525°F. All were held for 15 minutes at the austenitizing temperature.

The Verhoeven-DET condition led to higher hardness than using the as-received microstructure. Therefore, the DET anneal with the relatively short austenitizing time (30 minutes), and relatively rapid cooling rate (670°F/hour) led to a fine spheroidized microstructure that gave a very good hardening response and toughness.

Recommended Annealing Procedures

Any of the annealing procedures can work, whether subcritical annealing, temper annealing, cycling, or DET/transformation annealing. My favorites are probably the temper anneal for high alloy steel and the shortened DET anneal developed by Verhoeven for low alloy steels. The temper anneal in high alloy steels provides an array of fine carbides with sufficient stability to survive to the high austenitizing temperatures required in hardening which leads to a finer grain size and carbide size. The DET anneal developed by Verhoeven combines a relatively short low temperature austenitize and intermediate cooling rate to provide very fine spheroidized carbides which provides the good machinability of spheroidized carbides but with better hardening response and toughness due to finer carbides.

High Alloy Steels

Most high alloy steels (A2, D2, M2, M4, 3V, stainless steels, etc.) have similar annealing temperatures. A relatively generic temper anneal would be:

1650°F for 1 hour, quench through air cooling or plate quench, temper at 1450°F for 12 hours

Low Alloy Steels without PID-Controlled Furnace

For low alloy steels heat treating with simple equipment, heat the steel to a little above nonmagnetic (not as high as for quenching for full hardness) and then place the steel in vermiculite for slow cooling. That will provide an anneal that is similar to the shortened DET anneal described by Verhoeven. That annealing treatment can be performed after cycling the steel for grain refinement. So the total process would be 1) high temperature (~1600-1700°F) normalize, air cool, 2) low temperature austenitize (just above nonmagnetic) and air cool 1-3 times for grain refinement, 3) anneal by heating to just above nonmagnetic and place in vermiculite.

Low Alloy Steels with Furnace

For low alloy steels that will be annealed in a furnace, heat to 1385°F for simple carbon steels, higher for alloy steels, hold for 30 minutes, then cool at 670°F/h to 1200°F and air cool. Higher carbon steels can use a little higher temperature, see the chart below for the approximate temperature range for simple carbon steels. For steels with significant alloy additions like 52100 you need to know the approximate critical temperature, which is described here. Low alloy steels with relatively high hardnenability like L6 may need somewhat slower cooling rates, learn about hardenability here.

---

[1] Chandler, Harry, ed. Heat treater’s guide: practices and procedures for irons and steels. ASM international, 1994.

[2] Roberts, G A, and Robert A. Cary. Tool Steels. Beachwood, Ohio: American Society for Metals, 1980.

[3] Neumeyer, T. A., and A. Kasak. “Grain size of high-speed tool steels.” Metallurgical Transactions 3, no. 8 (1972): 2281-2287.

[4] Bhattacharyya, D., A. Hajra, A. Basu, and S. Jana. “The effect of grain size on the wear characteristics of high speed steel tools.” Wear 42, no. 1 (1977): 63-69.

[5] Saha, Atanu, Dipak Kumar Mondal, and Joydeep Maity. “Effect of cyclic heat treatment on microstructure and mechanical properties of 0.6 wt% carbon steel.” Materials Science and Engineering: A 527, no. 16-17 (2010): 4001-4007.

[6] Mishra, Alok, and Joydeep Maity. “Structure–property correlation of AISI 1080 steel subjected to cyclic quenching treatment.” Materials Science and Engineering: A 646 (2015): 169-181.

[7] Mishra, Alok, Atanu Saha, and Joydeep Maity. “Microstructure Evolution in AISI 1080 Eutectoid Steel Under Cyclic Quenching Treatment.” Metallography, Microstructure, and Analysis 4, no. 5 (2015): 355-370.

[8] Maji, Soma, Amir Raza Subhani, Bijay Kumar Show, and Joydeep Maity. “Effect of cooling rate on microstructure and mechanical properties of eutectoid steel under cyclic heat treatment.” Journal of Materials Engineering and Performance26, no. 7 (2017): 3058-3070.

[9] Mishra, Shakti, Alok Mishra, Bijay Kumar Show, and Joydeep Maity. “Simultaneous enhancement of ductility and strength in AISI 1080 steel through a typical cyclic heat treatment.” Materials Science and Engineering: A 688 (2017): 262-271.

[10] Saha, Atanu, Dipak Kumar Mondal, Koushik Biswas, and Joydeep Maity. “Microstructural modifications and changes in mechanical properties during cyclic heat treatment of 0.16% carbon steel.” Materials Science and Engineering: A 534 (2012): 465-475.

[11] Samuels, Leonard Ernest. Light microscopy of carbon steels. Asm International, 1999.

[12] Stickels, C. A. “Carbide refining heat treatments for 52100 bearing steel.” Metallurgical Transactions 5, no. 4 (1974): 865-874.