Corrosion Resistance Testing of Stainless Knife Steels

No new Patreon supporters this week! I think this is the first article where I didn’t have any new people to thank at the top. Thanks to all my current Knife Steel Nerds Patreon supporters.

Corrosion Resistance of Steel

At its most basic, chromium in steel leads to the formation of a chromium oxide layer on the surface which protects the steel from rusting. Molybdenum and nitrogen are known to help with “pitting resistance” and can support the chromium oxide layer. Chromium, molybdenum, and nitrogen can form carbides or nitrides which leads to less of the alloying element available for contributing to corrosion resistance. So the overall composition and the heat treatment are significant parts of corrosion resistance not just the “bulk” composition. For example, D2 has about 12% Cr which would be enough to be stainless but with its high carbon content there is insufficient chromium “in solution” to be stainless. Read more about D2 corrosion resistance here.

Previous Corrosion Resistance Article

I have previously written other articles about the corrosion resistance of steel, including this article where I talked about what gives steel its corrosion resistance and used JMatPro calculations and reported experiments to rate each steel. The amount of chromium and molybdenum in solution at the recommended hardening temperature for each steel was estimated with the software. I used some reported experiments on knife steel corrosion resistance to develop a simple equation to determine the relative contribution of Cr and Mo and then ranked each steel accordingly. However, there were a couple limitations to that previous article which I hope to improve here:

1. The corrosion resistance experiments reported were either a mass loss after a set amount of time in acid or a voltage measurement of “pitting resistance” in salt water. Neither was a measure of rust formation itself. It may be that one of these tests correlates well with rust formation but I don’t know which of them it would be. Rust would be the main issue people are concerned about with knives so that is the basis on which we want to compare corrosion resistance.
2. I used JMatPro at the time rather than ThermoCalc because JMatPro is faster to use and I was not sure which had the better estimate of alloy in solution. I found cases where one was closer to experimentally reported values than the other so it seemed overall to basically be a tie. However, I have recently learned that ThermoCalc previously updated their database based on tool steels with high Cr and V content that conflicted with JMatPro and earlier versions of ThermoCalc [1]. Old databases showed steels like Vanadis 6 and the original Vanadis 4 to have no chromium carbide in them at typical hardening temperatures while in reality they have a significant amount of it. This means that stainless steels with high vanadium content like S30V, M390, Elmax, S35VN, S90V, etc. would have estimated Cr in solution which is too high. However, ThermoCalc is now better at calculating Cr in solution for these steels. ThermoCalc seems to regularly update their database while looking at several versions of JMatPro the database seems never to change. Therefore, I have more confidence in ThermoCalc in general. The differences between the two software packages varies somewhat, in general they are not radically different, but in some cases the differences are significant.

New Experiment

For each of the corrosion resistance experiments that I performed I started with ~1/8″ starting stock and cut out 1 x 1.5 inch pieces. Each was wrapped in heat treating foil and austenitized in my Evenheat furnace for 15-30 minutes depending on the steel. The pieces were quenched between aluminum plates and given a cryo treatment in liquid ntirogen, then tempered twice for two hours at 400°F. The austenitizing temperature selected for each steel is shown below along with the ThermoCalc-estimated Cr, Mo, and N in solution. The exception is CPM-154 which uses the Cr solution measured experimentally in this article. Each piece was then ground and finished to 400 grit. With corrosion testing of materials the steel is often mirror polished because scratches can reduce corrosion resistance. That helps with making it a test of the material only rather than the finish. However, I decided to use a more practical finish that would better represent the state of consumer knives.

You can see that Cr in solution ranges from 9.4 to 14.4%, Mo from 0 to 2.7%, and N from 0 to 0.31%. These steels were selected to have these wide ranges so that the effect of the different alloy elements could be evaluated.

Water Spray Test

I first started by testing S35VN, AEB-L, CPM-154, 14C28N, LC200N, and XHP. I sprayed each with distilled water every 8 hours and continued the test for 4 days. Only XHP saw any rust, and it had significant rusting after only 10-12 hours. I was concerned about potential contamination of steel dust so I re-finished the sample and repeated the experiment but it rusted after about 10 hours the second time as well. In my previous article about XHP I speculated that XHP had insufficient Cr in solution to be stainless and this experiment appears to have confirmed that speculation. Interestingly, CPM-154 with a similar Cr in solution did not have rust, which is an indication that the high Mo is making an improvement to corrosion resistance apart from pitting resistance in saltwater solutions only, which is where Mo has the biggest effect.

XHP 10 hours after being sprayed with water

Salt Water Spray Test

I did not continue to test XHP because it did so poorly with water. I next tested the same steels (apart from XHP) with a 3.5% saltwater solution, which is a similar salt content to seawater. However, differentiating between the different steels was somewhat difficult. LC200N did the best, 14C28N was second best, and AEB-L was last but differentiating between S35VN and CPM-154 was difficult. Significant rusting was seen after only 12 hours apart from LC200N which saw no corrosion. So I decided to reduce the salt to 1% and perform the experiment with a few more steels (now the full matrix shown above) to better differentiate between the different grades. Every 8 hours I cleaned the samples, photographed them, and then sprayed them with the 1% saltwater spray again. Here are the specimens, after 24 hours on the left and 48 hours on the right:

LC200N

S110V

204P

14C28N

S35VN

CPM-154

S90V

AEB-L

Comparisons Between Steels

The Mo-free AEB-L had relatively poor corrosion resistance when compared with these other steels. 14C28N had much better corrosion resistance than AEB-L. 14C28N was designed by Sandvik to have improved corrosion resistance vs their 13C26 which is identical to AEB-L. The 14C28N was designed to have much more Cr in solution than AEB-L which is why its corrosion resistance is so much better. S110V was also found to have improved corrosion resistance when compared to its predecessor, S90V. No corrosion was observed with LC200N due to its high Cr in solution in combination with Mo and N. 204P and presumably also M390/20CV has very good corrosion resistance due to its high Cr in solution in combination with Mo and W. CPM-154 has decent corrosion resistance due to the high Mo in solution despite its relatively low Cr in solution.

Relative Effects of Cr and Mo

Similar to the method used by Carpenter in development of XHP [2], I assigned a rating from 1 to 10 of the surface of each specimen, where 10 means no visible rust, 9 means small light spots, 8 means easily visible rust spots, 7 means the rust spots have grown to be “medium-small,” 6 means 5-10% covered, 5 means 10-20% covered, 4 means 20-40% covered, 3 means 40-60% covered, 2 means 60-80% covered, and 1 means 80%+ covered with rust. In some cases I gave a 0.5 addition if I had difficulty assigning one of the values. I then plotted the result vs chromium in solution and got the following result:

There was a strong effect of both chromium and molybdenum on corrosion resistance. Whether there is an effect of nitrogen is somewhat less clear. The 14C28N steel seems to be closer to the 0.6-1.0% Mo steels than AEB-L and that may be from the ntirogen addition. But LC200N has such high Cr in solution so it is hard to separate out the nitrogen effect leaving 14C28N as our only datapoint. CPM-154 is on the same trend line as S110V and S35VN despite its significantly higher Mo in solution (~2.7%), which may mean either the Mo effect is not as strong with low Cr in solution or perhaps that the effect of Mo is saturated at some amount greater than 1%. This makes sense as the chromium oxide layer is most important, and the Mo in solution supports the existing Cr-oxide layer. I calculated a regression for the effect of Cr and Mo in solution for the 24 hour data which was Rating = -13.6 + 1.528*Cr + 2.419*Mo, which provides a strong correlation:

Dividing the Mo coefficient by Cr (so that Cr is equal to 1) gives 1.6 for the 24 hour and 48 hour data. The 1.6 coefficient for Mo is much lower than the PREN equation, which is Cr + 3.3*Mo + 16*N. The 1.6 value for Mo is relatively similar to the coefficient I calculated to in the previous article. The PREN equation was derived for austenitic stainless steels that have very high Cr in solution (16%+) so the chromium oxide layer was relatively strong in all of them. So it makes sense that the behavior would be somewhat different for these knife steels where 15% Cr in solution is a very high number. Also PREN is from pitting resistance specifically, not necessarily for rust formation alone.

Tungsten is reported to have half of the contribution to molybdenum so it would have a coefficient of 0.8. And if we are generous and assume that the contribution of nitrogen is significant to 14C28N, that would give a coefficient of approximately 6 to nitrogen. So our final equation for predicting the effect of each alloying element in solution is Cr + 1.6*Mo + 0.8*W + 6*N.

Rating other Steels

I next used ThermoCalc to estimate the Cr, Mo, W, and N in solution for each steel and plugged those into the equation above. I assumed that the contribution of Mo, W, and N saturates at an overall contribution of 2.6%, corresponding to 1.6% Mo (1.6*1.6 = 2.6). Each of those elements supports the chromium oxide layer rather than replacing it so I grouped them together with this contribution. Using that equation I then normalized to a 10-point scale to provide a rating for each of the steels below. I don’t think these ratings are perfect but I think it is a reasonable approximation and is an improvement over the table in the previously published article. I have a solid line which is where I think the rough border is between “stainless” and “non-stainless” steel. Apart from XHP, ZDP-189 is also below the line.

I have not listed alternate names for each steel. For example, LC200N is also called Cronidur 30 and Z-Finit. M390 is also called 20CV and 204P. S90V is also called 420CV or 20CP. I do not differentiate between CPM and conventional steels with identical composition like 154CM and CPM-154. In some cases PM technology may make the alloy more evenly distributed for better corrosion resistance but this is not factored into a ThermoCalc estimate.

Effect of Heat Treatment

The heat treatment of the steel matters. In the table above I used the recommended “austenitizing,” or hardening, temperature for each steel. If a single recommend temperature was not given then I took the average of the recommended range. With higher austenitizing temperatures more carbide is dissolved putting more Cr, Mo, etc. in solution. You can read more about austenitizing here: Part 1, Part 2, Part 3. There are limits to how high the temperature can be, whether because hardness drops from excessive retained austenite or because the steel melts. For example, no austenitizing temperature is sufficient to make D2 stainless.

The tempering temperature also affects corrosion resistance, as high temperature tempering (>750°F) leads to precipitation of chromium carbides and therefore a reduction in chromium in solution. Some steels are relatively commonly tempered in that high temperature range and therefore the results of the testing would be somewhat different. Non-stainless steels and high speed steels like M2, M4, Maxamet, Vanadis 4 Extra, 10V, 3V, CruWear, etc. are often heat treated with a high temperature temper. And certain stainless steels like CPM-154 and BG-42 are relatively commonly tempered at a high temperature. In general I recommend using a low tempering temperature instead for better corrosion resistance and in some cases a low temperature temper gives better toughness as well.

Effect of Surface Finish and Contamination

As noted previously, a rough finish leads to faster corrosion, in general the finer the finish the better the resistance to corrosion. Contamination of steel particles, particularly those of a non-stainless steel, also can create sites for corrosion. Therefore, when two knives are compared to each other it is not necessarily a pure comparison of the steel type because there is a contribution from heat treatment and surface finish, assuming no contamination.

Corrosion of Edges

Corrosion is not only a cosmetic issue, it can also affect edge retention as shown in this study.

Future Studies

I would like to use the less extreme water spray test to compare more non-stainless steels to find the effects of a couple variables. One thing to test is whether the high Mo and W contents in high speed steels are sufficient to improve corrosion resistance when using low tempering temperatures or if the low Cr content means that those steels behave similarly to low Mo steels like A2. This would also help to test whether the contribution of Mo/W/N does indeed saturate at some value as assumed above. I would also like to see if a patina makes any difference with corrosion in water.

Conclusion and Summary

Higher chromium in solution leads to better corrosion resistance, as expected. ThermoCalc was found to be better than JMatPro for predicting Cr in solution. Molybdenum was found to improve corrosion resistance whether in water or saltwater. The contribution of Mo appears to saturate at some amount of Mo around 1.6%, and this is likely true of the contribution of W and N as well. XHP was found to have poor corrosion resistance and I don’t consider it to be a “stainless” steel. ZDP-189 also likely has poor corrosion resistance. Using a regression derived for the contribution of each element in solution, a rating system was devised for a range of knife steels. This rating system should be improved when compared with the previous attempt at ranking knife steels in terms of corrosion resistance. The rating system may be updated using information from future corrosion resistance experiments.

---

[1] Bratberg, Johan, and Karin Frisk. “An experimental and theoretical analysis of the phase equilibria in the Fe-Cr-VC system.” Metallurgical and Materials Transactions A 35, no. 12 (2004): 3649-3663.

[2] Novotny, Paul M., Thomas J. McCaffrey, and Raymond M. Hemphill. “Corrosion resistant, martensitic steel alloy.” U.S. Patent 5,370,750, issued December 6, 1994.