The octahedral void exposes S in FeS2 on the surface and made it vulnerable to a free Li2. One Fe−S ionic bond is broken with the formation of one Li2−S bond. This one broken bond out of six with a big Li jammed in the lattice causes the lattice structure to crack, and data from X-ray absorption suggests that Li2FeS2 is amorphous. At room temperature, on recharge the lattice does not reform and a different sets of redox reactions applies which results in the precipitation of Fe and S. Both are undesirable in the operation of the battery.
It's the hole's fault, we need a bigger hole.
A mix of smaller "late" transitional metal might allow Fe to open up. Cobalt Co is a suitable candidate.
And CoS2 (beta form) can serve as a backing lattice to keep FeS2 structural integrity. But Co is further down the reactivity series, it is likely that Co is reduced by Li before Fe in which case we would use FeS2 as the backing lattice and use CoS2 to receive Li2. CoS2 will insulate FeS2 from Li2 while FeS2 provides structural integrity.
Another big "late" transitional metal is Mn with crystal radius of 0.81×10−10m. Mn is more reactive than Fe. In place of Fe as the backing lattice and dopant for a reactive CoS2 layer might open the lattice (CoS2 doped with Mn lattice) further and make S more accessible to Li2.