Diamond as an IR window is better than ZnS IR window. This because diamond is mechanically strong and relatively chemically inert.
Even it can be made to coat the ZnS IR window, so that one can use combined window. For more information please see the link:Article Diamond coatings for IR window applications
As Dr. Zekry said, diamond is relatively chemically inert and mechanically stronger. However, diamond has strong bands in the middle of the IR spectrum, from c. 2700-1800 cm-1, as you can see in this single-beam spectrum:
This leads to reduced throughput and higher noise in this region. Ergo, while it's generally not a problem for qualitative analysis, if you plan to do any quantitation where sensitivity could be an issue, then if your sample is compatible with zinc selenide, then that may be the best way. Take care, though, ZnSe scratches easily and the scratches will lead to diminished throughput and also poorer contact with the sample.
Diamond has a much broader transparency than ZnSe. ZnSe starts to absorb below 3µm whereas diamond goes all the way to ~220nm. The only thing to watch out for in diamond is the two phonon adsorption around 5µm. However, if you don't care about below 3µm ZnSe could be better as it doesn't have ther 4-5µm wobble of diamonds' 2 phonon adsorption.
One major problem with ZnSe is is it pretty toxic and not that stable. Despite what someone said above you cannot coat it with diamond (DLC perhaps), that would create hydrogen selenide gas which is the most toxic of all selenide compounds. Don't even think about it...