Hello Naif,

I opened your file and have two comments:

1. Note that in the layout in the paper, the slit and sensor are symmetrically spaced about the symmetry axis of the lens and grating. In your design the slit is centered, crowding the rays into half the lens. This can be implemented using coordinate breaks. 
2. More importantly, you have not implemented the lens as part of a double pass system. The lens appears only once in the Lens Data Editor. Since this is a sequential design, the rays encounter surfaces in the order in which they appear in the LDE. In your design, once rays reflect from the grating, they do not see a lens between the grating and the image plane. If you look closely at the layout, you will see that rays leaving the slit are refracted by the lens, but rays reflected by the grating pass through the lens as though it were not there. In fact, it isn’t.

One way to correct the 2nd issue is to make the system double pass. In such a system the lens will occur twice in the LDE, with the second occurrence positioned to exactly coincide with the first in global coordinates. There is a knowledge base article on that method here.  Another way is to use non-sequential mode to model the system.

Hello David,

Many thanks for the notes. That is what I was missing, indeed.

It looks great now.