The positions of the five orders that can fall on the detector.
The grism disperses the light of a celestial source into a series of spectra, just like a prism, but with a difference. The optics are more complicated than a simple prism in order to optimize the quality of the light in the first order. As a result the zeroth order is extended toward the UV, and the minus-first order is very fuzzy. Two more orders can fall on the detector, the second and third order. In the XMM-OM grisms, and the UVOT V grism they overlap the first order, while in the UVOT UV grism they may curve away from the first order. The sensitivity of the second and third order are low and lower, but especially the second order must be accounted for in very blue objects, like White Dwarfs and GRBs.
Order overlap in the V grism and curvature in the UV grism
The V grism and UV grism show different types of order overlap.
In the V grism the second and third order fall right on top or in
line of the first order. For all but the most blue objects
the contribution of the second and third order to the first order light
can safely be neglected. However, for very blue objects there may be
some second order light above about 4000A in the V grism first
order. The magnitude of this effect still has to be
determined.
In the UV grism the spectra are curved. This is probably because one
of the elements in the grism optics is slightly tilted. This has the
fortunate result to offset the second and third orders from the first
order over part of the detector so that their contributions can be
separated from the first order. When they overlap, the magnitude of
contamination can be estimated, if not corrected for, in principle.
The curvature of the UV grism spectra was not in the original design. Indeed, it is not present in the XMM/OM UV spectra, which has the same instrument and filter design. The characterisation of the second order was done by tracing the position of the orders. In the part of the detector where the overlap is negligible, the positions were derived by interpolating the solutions on the detector relative to the anchor point. Once the relative order positions are known for a given spectrum, the spectrum can be extracted and spectral features in the second order can be used to determine the dispersion of the second order and its offset to the first order. In practice, those were then scaled to the predictions of the unmodified Zemax optical model to provide a consistent approach to our grism calibration.
Examples of the first and second order spectrum are given in the Example Spectra section.
