Thanks to Josh Warren and Alex K for becoming Knife Steel Nerds Patreon supporters!
Edge retention testing is continuing, we have now tested 18 different steels, and four of those steels with multiple heat treatments. Come to Patreon if you want updates on testing as it is completed.
Powder Metallurgy Knife Steels
I wrote about how powder metallurgy steel production works and the history of it in this article. Rather than pouring liquid steel into a mold and letting it slowly cool, the liquid is passed through a gas spray which rapidly solidifies small particles, creating a steel powder. The powder is then placed into a canister for “hot isostatic pressing” (HIP) where the steel is heated to high temperature under pressure to create a solid ingot. The ingot has a very small level of porosity, which is eliminated in the forging and rolling process. The rapid cooling to produce powder rather than slow cooling to form an ingot leads to less segregation of alloy in the steel and a finer microstructure. Below shows a comparison between CruWear (conventional) and CPM CruWear (powder metallurgy). The carbides (white particles) are finer and more uniform in the powder metallurgy version.
Cruwear (conventional)
CPM CruWear (powder metallurgy)
Impurities and Inclusions
Some of the major claims about differences in powder metallurgy production are related to impurities and inclusions, so we need to briefly introduce what those are before explaining further. The most common types of inclusions in steel are sulfides (compounds of sulfur) and oxides (compounds of oxygen). Sulfides are usually present in the form of manganese sulfide, as manganese is added intentionally to avoid iron sulfides. Iron sulfides melt at relatively low temperature, leading to liquid sulfides at forging temperatures, which makes hot rolling steel likely to fail. Oxides can be in many forms including FeO, MnO, Al2O3, SiO2, and others. There are various methods used in an attempt to limit the oxide and sulfide content of steel though it is impossible to reduce it to zero. I won’t be able to describe all of those methods in this article, however. In some cases higher sulfur levels are intentionally added because a high content of MnS leads to greater ease in machining at the cost of some toughness. Powder metallurgy is reported to lead to less detrimental effects of MnS, and the poor machinability of high wear resistance PM steels means that sometimes the tradeoff is worth it.
Another impurity in steel apart from O and S is phosphorous (P), though it is not typically in the form of an inclusion so it is somewhat separate from the other two elements. Phosphorous, however, tends to segregate to grain boundaries and to reduce toughness in steel.
3rd Generation Powder Metallurgy Technology
Bohler-Uddeholm (now voestalpine High Performance Metals) have advertised their shared powder metallurgy technology as being a “3rd generation” version of powder metallurgy with an even finer powder size and fewer inclusions in the steel [1]. The powder size is about half of the size of the “1st gen” process which is claimed to lead to a decrease in carbide size even relative to the original powder metallurgy process performed by Crucible.
The difference in technology has to do with the “tundish” that the liquid steel is poured into prior to atomization. In the original process, the steel is poured into the tundish several times to make a whole batch of powder. On top of the liquid steel is where a layer of “slag” is formed of various oxides and other impurities, and we would like to avoid any slag getting into the final solidified product. However, during the re-pours it is possible for the slag to be pushed down into the liquid steel and through the atomization nozzle resulting in excess oxides in the final product.
Image from [2]
Erasteel, maker of powder metallurgy ASP-series of steels and RWL34, produced a new powder metallurgy facility in the early 1990s which was developed to have a larger tundish. A large tundish alone does not solve all of the problems because the steel can solidify while it sits in the tundish waiting to be atomized into powder. So they also added graphite electrodes that keep the steel heated. A layer of conductive slag on the liquid steel surface allows the electrodes to heat the steel and offers some protection from the atmosphere. They also use argon gas stirring to maintain a consistent temperature. Erasteel’s process is called “2nd generation” by Bohler-Uddeholm. Erasteel around 2006 began advertising a further reduction in inclusion content through a process called “Dvalin” but very unhelpfully they never say what that process is, just that it exists and makes the steel better. Below shows their values for inclusions with 1st gen “PM”, 2nd gen “ESH” and whatever Dvalin is:
Image from [3]
When Bohler-Uddeholm built their plant around 1999 they used a very similar process but instead of argon gas stirring they used electromagnetic stirring which is somewhat superior to argon stirring. The B-U process also directly encapsulates the powder for the HIP process to avoid effects of handling powder, since sand and other impurities can end up in the powder, and the high surface area powder is prone to oxidation. The finer powder size results from a reduced nozzle size for a slower atomization process which provides fine powder without requiring sieving. The heating of the tundish with electrodes allows the longer process to take place for the finer powder.
Image from [4]
Carpenter Steel
When Bohler-Uddeholm is comparing with the “1st generation” they are usually referencing Crucible Steel. However, another manufacturer of tool steel is Carpenter, known for grades like XHP, B75P, and 204P. There is an email response from a Carpenter employee on the Spyderco forum [5] that provides some information on their powder metallurgy process. In that email he said that Carpenter uses “reticulated refractory filters” to achieve a finer powder size than Bohler-Uddeholm. I don’t know what reticulated refractory filters are, unfortunately. And while he says that they have the ability to produce powder under vacuum for very low oxygen content, he also states that the knife steels are produced in air like Crucible, Carpenter, and Bohler-Uddeholm. Erasteel also has the facilities for producing powder under vacuum for specialty metals like titanium, but because of the higher cost and smaller batches this does not seem to have ever been used for knife steels that I have seen.
Carbide Size
I have done a series of comparisons between powder metallurgy steels from Crucible, Carpenter, Bohler, and Uddeholm to look at impurities, inclusions, and carbide size. I have previously written about carbide size differences in this article on micrographs. In comparisons between various products there does not seem to be much difference in carbide size between different companies. For example, Uddeholm Vanadis 4 Extra and Crucible CPM-4V are basically identical steels and the carbide size is also roughly identical:
Crucible CPM-4V
Uddeholm Vanadis 4 Extra
And in looking at Bohler M390 and Crucible 20CV, which also have identical composition, the carbide size is very similar:
Crucible CPM-20CV
Bohler M390
Effect of Composition and Carbide Type
You may notice that the carbide size is more different between M390 and Vanadis 4 Extra than between 4V and Vanadis 4 Extra. In other words, the steel composition is more important for carbide size than the production process. One major difference between M390 and Vanadis 4 Extra is that M390 has a large volume of chromium carbides while Vanadis 4 Extra is exclusively vanadium carbide after heat treatment. After atomization, the two carbide types (vanadium and chromium) are similar in size, but during the HIP and forging processes the chromium carbides coarsen more rapidly than vanadium carbides. Therefore it is the resistance to coarsening of the carbides that primarily controls the final size of the carbides. For a given carbide type, the greater the volume of the carbides the larger they are as well. You can read about different carbide types in this article. Chromium carbides (M7C3 and M23C6) and molybdenum/tungsten carbides (M6C) coarsen the fastest, and vanadium carbides (MC) coarsen much more slowly. Nitrides [6] and niobium carbides [7] coarsen even slower than vanadium carbides.
Uddeholm has redeveloped several of their steels over the past couple decades to reduce or eliminate the Cr and Mo/W carbides so that the overall carbide size is reduced. One example is Vanadis 4 Extra which was a redesigned Vanadis 4 with reduced chromium content to eliminate chromium carbides. Vanadis 8 is a replacement of Vanadis 6 and Vanadis 10, again with reduced chromium content. And Vancron Superclean is a replacement of Vancron 40 with reduced Mo to eliminate Mo carbides. Below shows micrographs of Vanadis 4 Extra and Vanadis 4 where you can see the carbide size difference. The chromium carbides are grey and the black carbides are MC (vanadium); the chromium carbides are larger than than the vanadium carbides.
Original Vanadis 4 with larger chromium carbides (grey)
Vanadis 4 Extra with a more uniform carbide size lacking in chromium carbides
There are many older powder metallurgy steels which were designed with this fact in mind, however. 10V was one of the earliest powder metallurgy steels, being patented in the 1970s, which is lacking in chromium carbides after heat treatment. The difficulty in eliminating chromium carbides in stainless steels is part of the reason why the non-stainless tool steels like 3V, 4V, Vanadis 4 Extra, 10V, etc. have better toughness for a given level of wear resistance.
Impurities and Inclusions in Powder Metallurgy Steels
As described earlier in this article, impurities phosphorous (P), sulfur (S), and oxygen (O), are the primary elements we look at in seeing how “clean” a steel is. We want the content of each of those elements to be as low as possible. So I looked at the content of each of those elements in three identical steels, Bohler M390, Crucible 20CV, and Carpenter 204P, along with Uddeholm Elmax which is in a similar class to those steels. Here are the compositions of each steel as provided by the manufacturer:
We measured S and O with LECO combustion testing and other elements including phosphorous with optical emission spectroscopy (OES). Below I have shown the measured P, S, and O:
The P and S of each steel is very similar, in fact Uddeholm Elmax has the highest levels of both of those elements. There is always going to be some range of resulting impurity contents and this is only one measurement. But the similar values likely mean that there are similar levels in general between the different manufacturers. There are various processing methods for reducing P and S content, but whatever methods each company is using the result appears to be pretty similar. So when Bohler-Uddeholm is claiming superior cleanliness and reduced inclusions they are apparently not referring to phosphorous and sulfur.
There are differences in oxygen, however. The lowest values are for Uddeholm Elmax and Bohler M390, Crucible has roughly double the oxygen content of Bohler-Uddeholm, and Carpenter 204P has about 50% more than Crucible. These values are about in line with expectation, as Bohler-Uddeholm attempts to restrict their oxygen content to below 0.01% [8], and values for oxygen in Crucible patents also hover around the 0.015% value [9]. However, I had not seen oxygen measurements for Carpenter PM steels and this value was much higher than I expected; I thought it would be closer to Crucible.
Micrographs
I also took micrographs for each of these four steels. When steel is polished (without etching) and imaged with microscopy, the inclusions show up as black features. This allows the observation of the size and distribution of the inclusions in each of the steels. Showing representative images is somewhat difficult because there is a statistical distribution of inclusions. For example, see this data on oxide inclusions from Bohler-Uddeholm intended to demonstrate the superiority of their product:
Image from [1]
They counted inclusions that were larger than 11 microns, and even in the “bad” example of Crucible steel they still measured less than one inclusion of such a size per square centimeter, which is a relatively large area in microstructure terms. So I am showing two images from each steel to get a slightly better feel of the inclusions, at the lowest magnification the microscope had:
Crucible 20CV
Carpenter 204P
Bohler M390
Uddeholm Elmax
Interestingly I did not see any MnS elongated in the rolling direction, perhaps the powder metallurgy process makes them small enough that they are not as noticeable as in conventional steels. The overall “volume” of inclusions appears very similar, probably because the sulfide content is similar between them. There appears to be a greater density of inclusions in the Bohler-Uddeholm steels, however, which makes sense given the previous diagram where they emphasized the number of relatively large oxides. None of the steels looks especially “clean” when observed in this way, however. It may be that the occasional larger inclusions show up more frequently in the Crucible and Carpenter steels though I did not look at such a large area as what B-U reportedly observed. The size of the inclusions may be significant, because the larger the inclusion the easier crack initiation is at the oxide. When the oxides are larger than the carbides then the easiest point for crack initiation is the oxide rather than the carbides.
Other Elements
I also made a comparison between a few of the “extra” elements in the steel which are not listed in the datasheet. There are also unintentional “impurity” levels of many elements, and some may be higher because of the use of recycled steel, such as nickel which is usually present around 0.2% in recycled steel.
One interesting difference is nitrogen, where it appears that Bohler and Carpenter are making an intentional addition of about 0.2% while Crucible with 0.07% is probably an incidental level from the atmosphere and the nitrogen gas used in atomization. A nitrogen measurement of M390 from 1995 reported 0.11% in one heat and 0.15% in another [10]. So I wonder if the nitrogen level has been intentionally raised since then or if there is simply wide variation in nitrogen levels with M390. The 0.11% in Elmax may also be the result of a small intentional addition. The small increase in nitrogen with M390 and 204P helps some with achieving higher hardness, it may also lead to a small increase in corrosion resistance and wear resistance.
Another interesting difference is the higher Co content in the Bohler-Uddeholm steels, and the relatively high tungsten content in Elmax which is not an intentional addition according to the datasheet. Typically recycled steels have less than 0.1% tungsten. This higher Co and W content may mean than Bohler-Uddeholm is using some percentage of recycled high speed steels that Carpenter and Crucible are avoiding. High speed steels are more likely to be alloyed with Co and W. However, the amount of those elements probably has little effect on the final properties.
Resulting Properties
Does any of this matter then? The impurity and inclusion content is not going to affect edge retention, after all. However, as mentioned above, the inclusion content may affect the resulting toughness. We have done a couple comparisons between the toughness of similar steels from different manufacturers, such as Bohler M390 and Crucible 20CV. In that case the toughness was essentially identical, the small difference is probably just due to experimental variability. Each was given the same heat treatment of 2140°F austenitize, plate quench, cryo, and 365°F temper. The small difference in hardness may be the result of the nitrogen difference or possibly just variation between heat treatments or composition between heats of steel.
We also have some comparisons between Uddeholm Vanadis 4 Extra and Crucible 4V but have only condition that is identical between the two, because we had some specimens that ended up being undersized and could not be machined for testing. We tested both longitudinal and transverse specimens because transverse specimens are typically lower in toughness and may reveal superiority in carbide size, inclusions, or processing. For transverse specimens that were heat treated with a 1975°F austenitize, plate quench, cryo, and 400°F temper, the 4V had significantly better toughness (12 vs 9 ft-lbs), though at slightly lower hardness, and the toughness was only slightly less than the longitudinal Vanadis 4 Extra (13 ft-lbs). On an unrelated note, the 1000°F temper of 4V led to lower toughness than the 400°F temper, which was also found with CPM CruWear/Z-Wear.
I think the superior toughness of 4V is due to the minor differences in composition leading to lower “carbon in solution” rather than superior processing technology by either company. However, this comparison illustrates that B-U steels do not necessarily have superior toughness.
Therefore in our testing we did not find better toughness of Bohler-Uddeholm steels than Crucible steels. This may be because of the very low number of larger oxide inclusions, even in the Crucible steel. When there is only roughly 1 medium or large oxide in a square centimeter, that oxide is unlikely to be in the volume of steel being tested for toughness. Or it could be that the carbides are the limiting factor for toughness rather than the oxide inclusions simply because there is such a significant volume of carbides. The highest toughness powder metallurgy steel we have tested was Z-Tuff/CD1 which is produced by Carpenter. Carpenter had the highest oxygen we tested, at least in the case of 204P. Crucible-produced 3V also had excellent toughness. Both of those steels gain their high toughness through low carbide content rather than differences in oxides.
Summary and Conclusions
Erasteel and Bohler-Uddeholm have newer powder metallurgy production facilities that allow a reduction in oxygen content of steel. Bohler-Uddeholm and Erasteel also reportedly produce a finer powder size but this does not result in a significant change in carbide size because the limitation in carbide size is coarsening during processing rather than the size of the structure in the powder. The composition of the steel has a much greater affect on carbide size than the powder metallurgy manufacturing technology. The oxygen content was confirmed to be reduced in Bohler-Uddeholm steels relative to Crucible and Carpenter, but a resulting reduction in toughness was not seen in testing. Therefore I think that steels can be purchased from any of these companies without worrying about missing out on “3rd generation” technology advantages.
[1] Tidesten, Magnus, Odd Sandberg, and Lennart Jönson. “PM Tool Materials: An Optimised PM Produced Cold Work Tool Steel.” In European Congress and Exhibition on Powder Metallurgy. European PM Conference Proceedings, vol. 3, p. 1. The European Powder Metallurgy Association, 2004.
[2] I got this image from Tool Steels: Properties and Performance by Rafael Mesquita. In the book he cites the image as coming from a Bohler document. However, the image does not appear in that particular document so I’m not sure where it came from.
[3] Raising the game in the demanding world of PM high speed steel, Metal Powder Report, vol. 61, no. 1, pp. 16-19, 2006.
[4] https://www.bohlerperu.com/app/uploads/sites/92/2018/08/ST035DE_Microclean.pdf
[5] https://forum.spyderco.com/viewtopic.php?t=49823&start=40
[6] Lindwall, Greta. “Multicomponent diffusional reactions in tool steels: Experiment and Theory.” PhD diss., KTH Royal Institute of Technology, 2012.
[7] Kajinic, Alojz, and Andrzej L. Wojcieszynski. “Cold-work tool steel article.” U.S. Patent 7,615,123, issued November 10, 2009.
[8] Liebfahrt, Werner, and Roland Rabitsch. “cold work steel alloy for the manufacture of parts by powder metallurgy.” U.S. Patent 6,773,482, issued August 10, 2004.
[9] Pinnow, Kenneth E., and William Stasko. “Wear resistant, powder metallurgy cold work tool steel articles having high impact toughness and a method for producing the same.” U.S. Patent 5,830,287, issued November 3, 1998.
[10] Pinnow, Kenneth E., William Stasko, and John Hauser. “Corrosion resistant, high vanadium, powder metallurgy tool steel articles with improved metal to metal wear resistance and a method for producing the same.” U.S. Patent 5,936,169, issued August 10, 1999.
Very interesting insight into pm steels… one question, the appearance of 20cv and m390 microgaphs… It looks like there are a larger amount of smaller carbides as well as the ones that are the maximum size. It almost looks like they obscure the grain boundaries. Is this real or imagined?
I’m not sure I know exactly what you’re asking. There is a lot of carbide in 20CV/M390, yes. And the carbides have a distribution in size from small to large. And carbides are more likely to be found at grain boundaries, both because boundaries are a likely place for them to form and the carbides that are already there “pin” the boundaries. Does that answer your question?
Yes, I just thought it looked like the m390 had a larger portion of small carbides than 20cv and they were more aligned to the boundaries… do you have programs to do kinda “spatial analysis” or “network analysis” of the relations between different microstructures or should I say carbides to martensite to grain boundaries?
So yes you answered the question to an extent
To do statistical analysis on microstructure features it would take a lot of images. I don’t know if anything so specific could be surmised from these two images. But thanks for looking with me.
Hi Larrin
Good article, very useful information. It answered some questions I’ve always had about buying from these companies.
Clay
Your discussion of trace tungsten content contradicts the values in the table:
“the relatively high tungsten content in Elmax which is not an intentional addition according to the datasheet. Typically recycled steels have less than 0.1% tungsten. This higher Co and W content may mean than Bohler-Uddeholm is using some percentage of recycled high speed steels that Carpenter and Crucible are avoiding.”
Elmax had the *lowest* tungsten content according to the table, at 0.20%. Crucible had the highest level, at 0.49%. Carpenter was also high at 0.41%. So it can’t be the case that Carpenter and Crucible are avoiding recycled high-speed steels if those steels are supposed to be high in tungsten, since their steels here are also high in tungsten (high for “trace” levels – roughly 0.5% is surprisingly high to me, given how expensive tungsten is…)
Hi Joe. 20CV/M390/204P have an intentional tungsten addition according to each steel’s datasheet while Elmax does not. Other Crucible steels have much lower tungsten.