I always wondered why quench cracks were intergranular. That’s one of the questions that I did not notice wasn’t answered when I was in school. But some years back I realized that if two grains are adjacent, and during cooling one is getting smaller due to normal thermal contraction and the adjacent one is getting bigger at the moment of transformation to martensite, then there could be a lot of stress at the grain boundary.
I spoke with several people at the time, and nobody knew. But nobody that I spoke with at the time was willing to say that they thought it was a totally stupid idea. But at the MS&T Conference in Pittsburgh this year, the idea of delayed quench cracks due to hydrogen embrittlement as a possible result of hydrogen diffusing into the material from an endothermic-gas protective atmosphere was brought up. This was pretty shocking to me, and I will have to think about it more and do more research.
A closing thought is that often parts may have discontinuities from the day they were cast or the raw stock was rolled. When I see a heavy two-toned oxide lining a crack that was found after heat treating, I believe that it pre-existed the quenching. The crack surface is usually nondescript. The oxidation process has obliterated any true crack features. I don’t think that temper scale would be associated with this type of heavy multi-composition oxidation.
When I wrote a chapter on heat-treating failure analysis for an ASM book (which was multiply peer-reviewed), I made this argument, and it ended up in the book. But last week at the conference, I met someone who was as skeptical of this theory as I was about the hydrogen causing intergranular quench cracks. I think that both of us are reasonably competent, so time will tell. And more information is needed to be able to diagnose the “cause” of quench cracks with higher certainty. Stay tuned!