Or in other words: how do cracks inside metal survive given the phenomenon of cold welding? If metal is overstretched, one or more fine cracks result along imperfections in the metal lattice. Why don't they disappear again if the stress force is removed or even changes the direction, f.e. if a force is applied that enables the 2 adjacent surfaces to touch again or even to be pressed together as in case with oscillating forces. From length standard calibration gauge blocks it is known that they can be cold welded together if 2 of them are put one above the other. That way both polished surfaces can be linked with each other's lattice. It's not a matter of atmospheric pressure as with bathroom hooks. Normal contacts between 2 metal parts as in screw threads, flanges and many other technical connections probably do always have enough separating layers of oxides, grease, dust etc. even if pressed together with high forces to avoid cold welding.
A crack is not surrounded by even surfaces but those fresh surfaces should perfectly match if no tiny bits did break away, and if the crack is completely inside the metal, no sulphuric or oxide layers can settle, what is the case with every metal surface that gets in contact with air after very short time.