b) No. Any supported format that contain GTF information can be employed as input file for this kind of analysis, such as .mwfn, .gms, .molden. See Section 2.5 of Multiwfn manual for description.

c) Right. However, if your code (such as ORCA) can produce molden file, you can use subfunction 2 of main function 100 to convert the molden file to fch file, then if use the fch file as input file, you can normally let Multiwfn utilize cubegen to calculate ESP.

d) I cannot give you definitive answer without having your ORCA input file and .wfn/wfx file. You can upload them to a netdisk, I will check.

It is important to notice that if you employ ORCA to carry out MP2 calculation, by default the exported wfn/wfx file only contains HF orbitals, evidently the analysis accuracy will be quite poor. Without special reasons, I strongly suggest using normal DFT functional instead, such as B3LYP, PBE0, wB97X-D3... If you really want to study ESP at MP2 level, you have to ask ORCA to generate natural orbitals at MP2 level, additional cost will be needed.

For an ORCA user, if Gaussian is available and the system to be studied is not small, my general recommendation of ESP analysis is using ORCA to perform optimization at DFT level (not double-hybrid functional), then convert the resulting gbw file to molden file (see beginning of Chapter 4 of Multiwfn manual on how to do), then convert molden file to fch file (as mentioned above), then use the fch file as input file to perform quantitative surface analysis of ESP.

Thanking You.