NIRSpec assignment #2

NGST ASWG NIR spectrograph subcommittee

A strawman set of spectrometer modules will be drawn up by Peter, Bob, and Marcia. This will include how the performance of these modules depends on R, pixel size on the sky, detector properties, and other factors like scattered light.

Following the discussion during the telecon on Monday 11 Oct. regarding the priorities expressed in the Crampton/Lilly study report, the following three spectrograph options are being investigated with the first two being top-equal in priority:

0. Low R (100+) - need ultimate sensitivity (at the expense of spatial multiplexing if necessary). The actual value of R we choose depends on how well we expect the detectors to behave.
1. Mid R (~1000) - tuned to high sensitivity emission line studies (astrophysics). Moderate multiplexing needed if objects are preselected. If the detectors are quiet, this does a pretty good job of continuum and absorption lines as well. Some spatial multiplexing on individual objects would be useful.
2. Kinematic R (3000-5000) - need spatial multiplexing at high resolution. The ground will do a pretty good job for Ha for z<2.6 and for [OII] for z<5 and a bit. Do we need to do this for wavelengths longer than K? As the detectors get better (after the NGST ones have been bolted onto the rocket), and telescopes get bigger and AO gets better... the ground will get relatively better and better.

What are this issues which bear on the choices we have to make? I'll divide these into 1^st and 2^nd order things.

First order

• The match to the DRM and other science we consider important.
• The anticipated groundbased performance.
• The anticipated detector performance (and its evolution after the NGST devices are procured).
• The capabilities described in the instrument study reports.

Second order

• The dependence of the spectrograph 'fidelity' (ie, straylight, cleanliness of the LSF, single source throughput, etc) on the type and degree of the optical mulitplexing.
• The dependence of ultimate performance on detector 'nasties' (ie, particle hits, stability, cosmetics...)

In making the choice between the instrument concepts, the nature of the front-end multiplexor will, in addition to the degree of multiplexing, determine the level of what is likely to be the dominant source of scattered light in the instrument. The effect of this level must be included in the spectrograph figure of merit.

Consequently, I have been spending a few idle moments this week trying to figure out how to quantify this in a meaningful way. The MacKenty team have addressed this for the MMA using diffraction theory to show that the skylight (zodi - which generally dominates over object light) is much lower than the 'in aperture' zodi. I have tried to do this in a generic way by looking at a pixel in the detector array and comparing the photon rate from an open aperture looking at sky (+ detector dark) with the rate from leakage from all other apertures (open and closed). I'm not sure I want to publish the numbers yet (some checking to do) - but you need to ensure that the leakage/diffraction from an individual closed aperture is at a VERY low level (~~0.1% of the sky on the aperture hitting the whole detector array). This is in a best case situation. If there are any bright objects in the field (on closed apertures) the criterion is clearly more stringent.

My conclusion from this is that we must evaluate the cost of multiplexing in terms of the limiting performance on individual objects. To my mind, the first priority is single object sensitivity with multiplexing coming as a bonus. If we know the price tag for multiplexing (performance, cost, reliability, complexity of operation) and the density of 'interesting' objects, we can make a decision. Without this knowledge, we are shooting in the dark.