What length scale mismatch specifically prevents DFTB+MBD from capturing mode I-to-mode II crack transitions?

The provided preprints don't directly address mode I-to-mode II crack transitions or the specific length scale mismatches that prevent DFTB+MBD from capturing them.

The closest document, [AG-2024.04-1576], describes the DFTB+MBD framework itself—combining Density Functional Tight Binding with many-body dispersion for van der Waals interactions—and mentions "limitations of simplified modeling" in materials mechanics, but the abstract does not specify crack propagation modes or the length scales at which the method fails.

To answer your question rigorously, I'd need preprints that explicitly study crack tip physics, the competition between Mode I (tensile opening) and Mode II (in-plane shear) fracture, and where DFTB+MBD's treatment of covalent bond breaking or stress transfer breaks down relative to atomistic detail or continuum elasticity.