Tempered Martensite

[GregPerry]

According to literature the following steel:

BS 970 817M40 Condition 'T'

should have a final microstructure consisting of tempered marensite.

We have had a fatigue failure on a solid shaft and testing revealed the following:

Chemical properties all within specification
Mechanical Properties all within specification
Bulk hardness within specification

The mataturgist report indicated "...presence of residual bands enriched with the pressence of ferrite..."

My question is: Should this material be 100% Tempered Martensite or is there a minimum specification for % Tempered Martensite and other phases?

Thanks
Greg

 

[rustbuster1]

Sounds like a red herring to me.
Fatigue properties relate so strongly to the mechanical properties that I suggest this is where to pay all your attention. After all, there may be many more of these shafts that have not broken and have the same microstructure. You need to do a detailed analysis of the INITIATION point(s) of the fatigue crack(s) first. Did a machining mark or something else initiate the crack??

I suspect the presence of some ferritic phase is a non-issue. Tempered martensite means what it says - the martensite should be tempered to a less sharp-needle like appearance - but this is subjective and the resulting mechanical roperties are far more important.

 

[CoryPad]

I disagree with rustbuster1. Fatigue performance is dependent upon microstructure more than static performance is. A banded structure indicates two different structures - the ferrite-enriched bands are likely to have lower mechanical properties. In a tension test specimen, the low strength bands are averaged with the high strength regions (since the tested volume is relatively large compared with the volume of a single band), and the results may meet specification requirements. Conversely, during fatigue loading, the material may develop persistent slip bands or preferential plastic flow in the banded regions, resulting in poor fatigue performance.

It is common to specify a minimum martensite content - e.g. "Parts shall be austenitized and quenched to form 90% minimum martensite phase in the part core."

 

[Carburize]

The real question is whether or not any of the microstructural banding played any role in the fatigue process and in particular the initiation. Alloy steels frequently show some segregation particularly as nickel levels increase in the alloy but in a commercial engineered product it is usual for stress concentrations, machine marks, surface damage etc to overwhelm the microstructual aspects, as Rustbuster1 stated,

 

[metman]

rustbuster and carburize make some good points especially if you read the 40 posts to the thread on the Aeronoatical forum here with question for mental challenge about surface defects. So yes where did the fatigue crack initiate. It would appear that the analyst did not specifically answer this question for Greg.

Greg, I suggest you asttempt to have your failure analyst answer that question. Do you have a copy of the part that has been in service or a new part. You can do some macro analysis with a magnifying glass to see if the part has machining or other marks (stress risers) especailly to be looked for at changes in section size and or at known high stress areas.

But I would like to persue the other question posed by GregPerry because even though the real culprit in this instance may very well be surface defects, there are ample examples of fatigue failure from banding. So back to Greg's question which is:

My question is: Should this material be 100% Tempered Martensite or is there a minimum specification for % Tempered Martensite and other phases?

CoreyPad gave a very reasonable answer such as it is. Which is:

It is common to specify a minimum martensite content - e.g. "Parts shall be austenitized and quenched to form 90% minimum martensite phase in the part core."

Whether or not this banding could be the primary culprit if surface defects are ruled out or secondary if surface defects are not ruled out depends on part geometry, stress distribution and how far from the surface the banding resides.

My question which shows my rustiness: Is it possible to remove the banding by homogenization during proper austenitizing? Corey's answer seems to imply this even though as carburize says, "Alloy steels frequently show some segregation." Segregation is not a nice word in my book at least not in fatigue applications as Corey so aptly summarized. If banding cannot be removed by proper austenization, must one normalize then austenitze and quench or what?

 

[TVP]

BS 970 Grade 817M40 is similar to SAE grade 4340 in the U.S. This alloy is capable of quenching to nearly 100% martensite in normal/practical sections used for engineered components. The presence of residual ferrite indicates a problem during the thermomechanical processing of this bar.

Now, like the others have already mentioned, did this have an effect on fatigue initiation? The metallurgical report should have conclusively said that the fatigue origin was in the area of residual ferrite if that were true.

Is it possible that this residual ferrite banding is due to an inadequate austenitization? Yes. Can this problem be reversed? Yes. As I mentioned previously, this alloy is capable of quenching to near 100% martensite, so as long as the initial structure is not overly segregated, and proper austenitization is performed, then residual ferrite should not be a problem.

 

[CoryPad]

Carburize,

I don't think it is true that "it is usual for stress concentrations, machine marks, surface damage etc to overwhelm the microstructual aspects". I think this mentality stems from stress-based fatigue analysis and infinite life approximations for steels. This is not the most accurate fatigue analysis method. Every part has something that will initiate fatigue (e.g. a nanometer scale disbonded inclusion or one grain with fully tangled dislocations), so the real trick is how to avoid propagating the crack. The strain-based and crack growth analysis methods are much better at quantifying part performance, and illustrate how microstructure (and crack propagation through it) is extremely important.

Regarding metman's final paragraph:

It is possible to decrease banding. More segregation is removed the higher the temperature used (which can cause unwanted austenite grain growth), so a separate homogenization step, then austenitization, then quenching and tempering would result in maximum properties. Of course, controlling initial solidification and hot working to avoid segregation is the most preferred option. It is true that alloy steels frequently show segregation, and it also is true that segregated steels have worse fatigue performance. Since nickel, chromium, and molybdenum have relatively slow diffusivity, if an initially segregated ingot/rod/plate/etc. (e.g., due to inadequate cast microstructure breakdown) is austenitized, quenched, and tempered using normal parameters, then alloy-deficient areas will form ferrite. If this occurs, and it lowers material properties enough, then a new high-temperature normalizing treatment, followed by standard A+Q+T should be performed.