As before, you're better off with a chemist, or better yet a metallurgist, but I'll do the best I can.
The steel has different "regions," think of it as edges of different crystal structures. Depending on how the steel is alloyed and how it's made, the regions will be different, as in ferrite, cementite, and martensite -- different configurations of steel.
The boundaries of these regions will exist as microscopic imperfections. Along with the type of steel itself, these boundaries govern the overall properties of steel. Hence a low-quality steel can have irregular crystals, while a very high-quality steel would be a single crystal or amorphous, and stronger, higher heat conduction, less vibrational damping, etc. as a result.
Caustic chemicals like our sulfur compounds will preferentially attack these region boundaries -- they're the weakest bound and easiest points of attack. The sulfur reacts with the steel, forming new compounds and exposing additional areas for future reactions. This is the "intergranular" part.
The new sulfur compounds and the steel can form a "eutectic mixture," that is to say, a combination of elements that has a melting temperature lower than any of its components. The example I use in my whitepaper is of ice and salt. Ice melts at 0oC, salt melts at 802oC, but the mixture melts at a lower temperature than either -- perhaps -20oC.
Something similar happens in the steel. Where the sulfur-iron compounds and ordinary iron compounds are well-mixed, a eutectic forms, and this can melt away, leaving holes and curled edges, etc. But -- and this is the key point -- neither the steel nor the sulfur compound, by themselves, melts. These survive. This is what Dr. Biederman is reporting on.
And with that, I am way out of my field, and heading still further out. Anyone who knows more about this stuff, please feel free to correct me about any of the above.