It looks like the call that sets up field variables is where we define the index (step), then we add particle data. There is a pending merge request (MR1413) that should fix the issue, but it needs testing.
Currently MFIX-Exa skips MultiFab (field) data when the fluid is not solved. MR1413 always sets up volfrac, the geometric volume fraction MultiFab, so the idex can be defined.