Failure analysis of die case (strange surface layer?) 6

[coreman73]

I was wondering if I might be able to get some input from you guys regarding a failure I'm currently working on. I am investigating a die case that has fractured into two pieces. I've attached a .pdf of what I have so far. I've included some short background information as well that will explain a few things (too long to type up here).

Just to summarize my findings to this point:
1. steel chemistry meets spec for H13 hot work tool steel
2. bulk hardness is approximately 49.3 HRC, which is slightly above spec of 46-48 HRC
3. microhardness showed a decarburized layer of roughly 0.003" along ONLY the threaded area of inside diameter
4. failure mode is fatigue with propagation occurring directly through the bottom land of final thread on inside diameter
5. unknown surface layer (looks like some sort of scale/oxide) that has hardness of ~80 HRB found ONLY along the entire threaded area (all other surfaces are ~47-49 HRC)
6. microstructure and grain size are as expected - no issues here
7. no surface damage or deformation anywhere in the vicinity of fracture site (threads included)
8. due to ratchet/beach marks on fracture surface, this defect was already present prior to the die assembly being reworked (explained more in attachment)

I realize this is a lot of information but I would really appreciate some input as I'm a bit stumped right now. The only root cause I can come up with is that failure was initiated due to this strange surface layer along threads, which served as stress riser with the final thread being the weakest site hence failure initiating and propagating here. What could this unknown surface layer be and where might it have come from (manufacturing defect)??

Thank you in advance.

 

[coreman73]

I was looking for a simlar type of surface layer appearance in my ASM failure analysis book and it looks like it could be sulfidation. However, it looks like this type of corrosion can only occur in a high temperature environment. I was told this die case was used for cold forming.

Also, I had thought this layer was decarburized since it showed such low hardness. However, after etching, this layer simply got extremely dark. So it looks like it is NOT decarburization, which would make sense if it was sulfidation.

Is this what the layer might be? I've asked for confirmation on whether this die case was really used for cold forming or if in fact it was for hot.

 

[metengr]

coreman73;

unknown surface layer (looks like some sort of scale/oxide) that has hardness of ~80 HRB found ONLY along the entire threaded area (all other surfaces are ~47-49 HRC)

I would strongly suugest you have this surface evaluated using SEM/EDS. If necessary, send it to an outside met lab.

 

[coreman73]

Metengr,

I agree with that advice and will arrange to have it sent out for SEM/EDS as I don't have that capability here.

In the meantime, I would really appreciate any ideas of what it might be based on the information I provided. I know it's speculation but I don't believe there's any harm in going through some possibilities while waiting for EDS results.

As for sulfidation, I found out that the die was in fact used for cold forming but that temperatures typically reach as high as 400 degrees C while continuously running. Some sources I found mention that sulfidation could occur as low as 250 degrees C in some environments.

 

[metengr]

coreman73;
The surface contaminant that is resuting in grain boundary oxides could be from residue during forming and not removed prior to heat treatment, poor atmosphere control during heat treatment, ....

After conducting many failure investigations over the years don't try to come to conclusions too quickly and miss the obvious. This is all I am saying...

 

[CoryPad]

Was the hole drilled and tapped prior to heat treatment? Was the rest of the part turned or ground or polished or coated after heat treatment? I think the unknown layer looks like intergranular oxidation that occurred during heat treatment. This can reduce fatigue life.

 

[unclesyd]

The composition of the unknown layer will be most interesting. At first it looks like some heavenly barbarized material. In general this presents a metallurgical notch for the crack to propagate from.
Cracking in first thread leads one to believe that the thread was overload or a had relatively loose fit which allowed relative movement.
We had trouble with H-13 parts that were heat treated using a salt bath and later being cleaned in a salt bath. The layer formed in this case was more general but was much softer that the parent metal.
We also had trouble with H-13 due cracking from nascent Hydrogen from the process.
Another quick test would be to re-temper a section at the temperature stated in the literature and see what happens to the hardness.
As posted the composition of the layer will be interesting, but I think getting the fracture surface in an SEM will tell you a lot. i think you should be on hand if you don't have access to one and be the pilot.
There is one possibility that you had a quench crack that let go when the load was applied. It is not unusual for tool steel dies to let go on the first hit. We had several failure of this type where the part broke on the shelf. All were attributed to an aggressive heat treatment.

I sure hate it when a tool steel fails by fatigue as normally the fatigue markers are not very discernible and it takes a little work to make them visible.

Is this a wrought product?

 

[coreman73]

Metengr,
I hadn't thought about the possibility that it could be grain boundary oxidation as a possible result of forming. I just found it very strange that the ONLY surface areas that showed this was isolated to the threads. I do understand what you're saying though. I may have been a little hasty in come to a conclusion but I guess that's what happens when there's a deadline to be met and lots of people are waiting for who to blame.

CoryPad,
I have asked the vendor for details regarding forming and am waiting to hear back. Unfortuately, it was made in Taiwan and so getting these details isn't the easiest/quickest process. It does make sense that this layer was likely picked up during heat treatment though.

unclesyd,
As suggested by metengr, I will submit this sample for EDS/SEM and find out the composition of this layer. I am very interested in this as well. I do want to specify about the fracture location though. It was actually through the final thread (deep inside the die case). The were no signs of use/wear or deformation on ANY threads. I suppose it could have been possible that there was a quench crack but this part seems to have been in service awhile due to the quite clear and extensive fatigue markings. I'm not sure if it's a wrought product or not.

 

[TVP]

coreman73,

Die cases such as these are typically turned and ground after the quench and temper operation to ensure excellent straightness, concentricity, etc. I agree with CoryPad that the surface of the threads appear to have EXTENSIVE intergranular oxidation (IGO) that occurred during heat treating. The rest of the case has been machined so that the oxidized surface is no longer remaining. It is extremely poor practice to produce that level of IGO on a part, and even worse to leave it on any surface that is heavily loaded, and the threads of a die case certainly qualify as that. The reduction in fatigue life will probably be on the order of 10x for that condition.

 

[coreman73]

TVP,

Thank you for the response. It seems that this layer is definitely oxidation (intergranular). I just heard back from the vendor in Taiwan. Their response is "The threads on the inside diameter of die case are machined. Then we do the heat treatment. After that we will check the threads by gage. If the thread is undersized, we will machine it again with correct size. There is no surface coating applied on the die case. We found that the threads without re-machining are getting oxidized easier. Please let us know anything we have to do to improve the thread quality."

So it seems that they're aware of the problem but are still allowing them to be produced in such way. I am debating whether or not to send this out for EDS of the surface layer now that even the vendor mentions oxidation.

 

[TVP]

c73,

If there are any financial or legal implications to your failure analysis, then I suggest having the EDS analysis performed. By this I mean if you are going to force the questionable supplier to pay for your lost production time, or to make some type of legal claim regarding the injured person, etc. If you just need to document the failure in your report, and your boss is going to complain about spending another $250-500 on outside testing services when your company has already been been burdened by this questionable supplier, then perhaps you want to omit this. Take a look at your ASM reference to see if there is a good photo of intergranular oxidation, it is likely in reference to gear heat treating, since the carburizing gas atmospheres in traditional furnaces are often oxidizing in nature. That should be sufficient as a reference, in addition to the expert opinions of a bunch of Internet yahoos that have weighed in on the matter.

 

[unclesyd]

If you can swing it I think you should get the analysis. You may not be able to swing for the EDS but a couple of micrograph's of selected areas would provide you with the information to backup your finding. I would be particular interested in the area from your marked initiation point and the anomaly shown in figure 2-A. The number of striations and surface morphology would be interesting.

It stills bothers me that the failure location was strange as with any thread or material anomalies the threads will normally shell out at the entrance to the threads. If I/m looking at the micrograph's correctly the fatigue crack appears to take off at an angle, which to me would indicated a several overload and this coupled with the metallurgical notch initiated the crack. Though we haven't had a lot of fatigue failures with high strength bolting, H-11 or H-13, all were generally pure radial in direction from the root of the thread.

Please kept us informed on the progress and resolution of your failure.

bcc:
Getting to examine this failure in an SEM would be a very good exercise for your failure analysis career.

 

[coreman73]

Well, I decided to go ahead and get the EDS analysis of this layer. I have attached the results. I'm honestly not very familiar with this type of test so what conclusions could be drawn from these results as it relates to the possibility of intergranular oxidation? What should we be seeing if anything?

TVP,
Hey, I appreciate all the experienced input from all you Internet yahoos very much.

unclesyd,
I would love to get some experience on an SEM but it's just not possible with my current position. I do agree that it's a must have area though and am trying to change that.

 

[coreman73]

I did notice the Sulfur content is very high at 0.272 wt%. When I measured the bulk chemical composition of this die case, I only measured Sulfur to be 0.0157 wt%. Could this actually be that my original guess at Sulfidation might be correct afterall?

 

[Maui]

You have a fatigue failure on your hands, and you have done a good job of documenting it. Based on the information that you have supplied and the attached images, this would appear to be a wrought H13 product that was heat treated and tempered in a furnace with an oxygen containing atmosphere. The threads, as you were informed by the supplier, where tapped in the part prior to heat treatment. This is clear due to the presence of oxide on the thread surfaces. The balance of the component surfaces likely received some form of surface removal such as a final grinding step to clean up the part to the required dimensions and tolerances after heat treatment was completed. Unfortunately, your supplier may not possess the capability to clean up the thread surfaces after heat treatment by removing this oxide layer. And even if they did it would likely oversize the hole and bring the part out of tolerance.

The hardness that you measured coupled with the appearance of the tempered microstructure would indicate that the part may not have been thoroughly tempered. There is a way to verify this. I suggest that you contact your supplier and find out what austenitizing temperature and tempering temperatures were used, as well as the number of temper cycles that were performed – and be persistent. The first person you talk to may claim to know nothing about it, and if this is the case ask to speak to someone else who might. Someone there likely has that information, and it’s just a matter of finding out their name and contact information. Please post any information you are able to locate on how this part was heat treated. As suggested by a previous member, you could perform an additional temper cycle at the specified tempering temperature. By examining the resulting re-tempered microstructure, you should be able to discern any difference from the microstructure you received. And by measuring the hardness you should be able to determine if executing an additional temper reduces the hardness so that it falls within your required range. If this is the case, then it suggests that the part was not adequately tempered. This would make the component more brittle, and therefore more susceptible to premature failure by crack propagation.

The initiation site for the fatigue crack shows that the cracks were initially growing at an angle to the direction of the applied stress, and then transitioned to growing perpendicular to the stress axis. This is indicative of a transition from Stage I (initiation) to Stage II (propagation) fatigue crack growth. Although this transition is often difficult to discern in steel alloys, the fatigue cracks generally initiate and coalesce from slip-plane fracture – they initially grow on the crystallographic planes where the fluctuating shear stresses are the greatest. And these are usually oriented roughly 45 degrees to the direction of the applied stress. After a relatively small amount of growth the fatigue cracks transition to growing perpendicular to the direction of the applied stress. For those who are interested, this is discussed to some degree in Chapter 16 “The Mechanical Durability of Metals and Alloys” in

http://www.asminternational.org/portal/site/

http://www/AsmStore/ProductDetails/?vgnextoid=eb0b76d47ff2f210VgnVCM100000621e010aRCRD

One way to eliminate the problem in the threads is to tap them after heat treatment has been completed. But this is often difficult to do in hardened parts, and your supplier may be unwilling or unable to do this. An alternative is to apply no-carb paint to the threads prior to heat treatment. This greatly reduces the formation of oxide on the surfaces of the parts. It also burns off during heat treatment, so there is little or no clean-up required after heat treatment is completed. Let us know what you find.

Maui

 

[TVP]

c73,

Getting to your specific question about sulfidation, we need a little more information. First, how did you obtain the bulk S composition, Leco method? Second, can you find out the exact type of EDS system that was used, and specifically ask how low in atomic number they can detect? Third, what is the sample orientation in the SEM image (is that a mounted specimen or free surface)? And fourth, can you find out whether a lubricant was used on the threads during assembly of the die, and if so, was it MoS₂?

I still think that this is IGO, perhaps with some junk on the surface from extrusion oil, assembly lubricant, etc. As you remarked upon earlier, sulfidation is usually associated with significantly higher temperatures. The 400 C temperature is usually an instantaneous surface temperature at the tool-workpiece interface, with the temperature much lower than that in the die case where the threaded section is.

 

[coreman73]

TVP,

All I know for now is that the independent lab does EDS with their SEM. The sample orientation is a mounted specimen I sent them. I will try and find out the other information and then respond ASAP.

In the meantime, and maybe this is a stupid question, but if this layer is mostly oxide then shouldn't the EDS results show that O was detected?

 

[coreman73]

TVP,
I just got a response that there was no lubrication used during the assembly of the die. It was mentioned that everything was assembled "dry".

Maui,
Just wanted to say thank you for the excellent response. Lots to think about and very educational. I appreciate it very much.

 

[CoryPad]

coreman73,

Low atomic number elements like carbon and oxygen are not detectable by every EDS device, which is why TVP asked the question about the exact EDS system.

 

[coreman73]

CoryPad,

Ok that makes sense. I learn lots from you guys.

I contacted the independent lab why I didn't see any oxygen in the results and he basically indicated what you just said. He said there was definitely O present but the accuracy couldn't really be proven. He gave me new results with O included and it looks like there is quite a bit. I've attached these results.

So, I'm not sure where this leaves me. Is there enough overall proof available now to back up the root cause 'most likely' being intergranular oxidation?