@Michael Cheng  - I looked at the link provided by @Julien.Gateau  and it appears to me that modeling that part would require a 2D (rotationally symmetric even?) RCWA grating DLL. Do we currently have such a thing? Or do we have any option to model a diffraction grating with a user defined spacing?

I’ve not heard of this type of grating before but looking at the website posted I think that the Optically Fabricated Hologram would handle this. The great power of the OFH is that the grating is formed by the interference between two other optical systems, and so the fringes do not need to be parallel or equidistant.

Good tip, @Mark.Nicholson !

For @Julien.Gateau  here is an article on OFH: https://support.zemax.com/hc/en-us/articles/1500005489901-How-to-use-the-Optically-Fabricated-Hologram-to-correct-for-aberrations

I agree with Mark. This is probably manufactured by two beam interference. The webpage also mentioned this is a holographic blaze grating.

I think you could start from a simple hologram and then change to a more complicated OFH. I have attached an example for your reference.

The simple holograms are all some variant of a point source and so will produce straight line fringes. The described hologram has varying line spacing and orientation, and so will need the OFH.

The next question is: does the vendor provide the information for you to construct the two fabrication systems?

Oh, great stuff. Thank you Jeff.

Thanks a lot to all for your answers!
 

@Jeff.Wilde Thank you for your references and Zemax simulation overview. I am currently having a look at the Hologram 1 Surface and  reproducing what you did.

Sorry for my lack of knowledge, but which merit function is used to optimize the position of the diffracted centre wavelength at the midpoint of the linear detector?

@Alissa Wilczynski  @Michael Cheng@Mark.Nicholson  I will have a look to the Optically Fabricated Hologram, the article sounds also interesting.

Concerning the information they shared, they only gave to me a pdf giving distances and angles vs wavelength (see attached document).

From the latter, I can certainly verify if the Zemax simulation and the coordinates they have are fitting to each other.

Maybe I can also use these values to add some constraints in the optimization loop?

To center the diffracted spectrum on the linear detector array, you can use the REAY operand to direct the chief ray for the center wavelength to the y=0 coordinate position on the detector (which in my model is the center of the detector in its local coordinate system).  

Also, make sure you select ‘ignore lateral color’ in the optimization wizard. For a spectrometer, lateral color is the intended behavior, and not an aberration.

I wouldn’t worry about he Optically Fabricated Hologram if the simpler hologram surfaces do what you need. 

Yes, it’s definitely important to have the MF wizard ignore lateral color. 

One more quick note.  It’s easy to see the spatial frequency chirp across the hologram by using the “Hologram Fringes” utility which is included as one of the ZOS-API User Analyses programs:

For the demo model shown above, we can see the hologram fringe frequency varies from 262 to 284 cycles/mm.

For more details, see: Analyzing hologram construction fringes with a ZOS-API User Analysis

I managed to repeat your simulation and almost obtained similar results. I am now working on the implementation in our optical setup.

The operand REAY is indeed quite useful! Good to know

Having a series of angles vs wavelength from their file, I also used the operand RAEN, which also seems to help configure the flat field grating.

Quite an interesting approach. I would like to thank you all again for your help in this optical design process.