High resolution (R ~ 8,000) work on faint stars/distant galaxies with large telescopes is difficult for a variety of reasons. The most prominent is the mismatch of the image scale in the focal plane of the telescope with the width of the entrance slit in ths spectrograph. Widening the slit does not help as your spectral resolution decreases at the same time.
An IFU allows you to collect light over an extended area on the sky and also obtain spectral information for each spatial element in that area. This means you can produce maps of various emission lines, absorption lines, continuum and two-dimensional velocity maps. To do this you need to reshape the light from an extended area in the sky onto a long thin slit suitable for feeding into a spectrograph.
There is an HTML-ised paper written for the "Optical Telescopes of Today and Tomorrow" symposium SPIE 2871 (1997). This paper covers all the equations concerned, complete with diagrams. If you have a printer, there is a postscript (376K) version of this paper available.
A brief description of SPIRAL
The diagram on the left shows the physical set-up of SPIRAL on the telescope.
There is an optical table bolted onto the Cassegrain focus of the telescope on
which the fore-optics and lens array are mounted. The fore-optics re-image the
scale of the focal plane so that each lenslet corresponds to 0.5 arcsec on the
sky.
The picture on the right shows the aperture in the primary mirror at the top,
with the fore-optics just below. Further down the optical table is the lens
array unit, and the fibres pass through the black conduit into a strain-relief
box in the bottom left and out of the Cassegrain cage.
The conduit now passes through the base of the Cassegrain cage which contains
the lens array, and drops down to the floor of the telescope dome some twelve
metres below. The spectrograph can now be seen with it's light-proof covers
off. As the telescope swings around at night, we checked to our satisfaction
that the conduit did not catch on any surrounding catwalks.
The fibres pass into the spectrograph through the fibre slit where they
are rearranged into a 1-D line. Due to the design of the spectrograph, the
dewar with the CCD inside is situated next to the slit - it is a large red
cylinder which is behind the grey box of electronics.
Here is a detailed picture of the fibre slit (seen on the right) coming into
the spectrograph. The light is reflected off a prism barely visible in the
corner of the lens closest to the red dewar. The light passes down to the
bottom left of the picture towards the grating, then back through the same
lenses into the CCD.
Here is the whole spectrograph with no light-proof covers on. The fibre slit is
hidden behind the red dewar and the grating is now easily visible. Another
advantage of having small optic fibres as a 'slit' is that there is almost
perfect matching of fibre diameter to pixel size on the CCD. This allows
construction of a Littrow configuration spectrograph, where camera and the
collimator are the same set of lenses, allowing the grating to be used most
efficiently.
First light for SPIRAL
The comissioning run for SPIRAL was in the middle of February 1997 at the AAT.
We suceeded in observing for one and a half nights and we lost one and a half
nights to bad weather.
The aim of the run was to measure the throughput of SPIRAL, and to see if we could detect Lithium absorbtion in a bright star, in preparation for examining candidate brown dwarfs. We also wanted to demonstrate the integral field capability and to this end we looked at the cores of nearby Seyfert II galaxies (the Circinus galaxy and NGC 1068) around 7500Å.
On the second night we obtained the Lithium absorption line (as seen on the
right here) from a bright star.
On the left is a flux intensity map of the continuum in the middle of the
Circinus galaxy, an active galaxy. The centre can be seen in the lower left
corner, together with some structure surrounding the core. On the right is the
same region of sky, but this is a velocity map of the Hydrogen alpha
line. The range is from -130 km/s to +140 km/s with blue signalling gas moving
towards us. With an R of 10,000, there is an error of ±15 km/s.
Performance and signal to noise calculations
The efficiency of SPIRAL is shown in the table below as compared to some other
spectrographs. The overall efficiency from top of atmosphere to the detector
(including QE) is 11.8% for 6750 Angstroms with the telescope
pointing at the zenith.
| SPIRAL | IDS | ISIS | FOS | RGO | ||
|---|---|---|---|---|---|---|
| Atmosphere | 0.88 | x | x | x | x | |
| Telescope | 0.61 | x | x | x | x | |
| Spectrograph slit (SPIRAL fibre optics + fore-optics) | 0.68 | 0.36 | 0.41 | 0.40 | x | x |
| Spectrograph optics | 0.52 | |||||
| Detector | 0.62 | x | x | x | x | |
| Measured throughput | 11.8% | x | x | 12.5% | 6.0% | |
| NOTE: Other spectrograph efficiencies were taken with a wide slit, usually 2 arcsec or more. SPIRAL throughput was with an effective slit width of 0.5 arcsec. | ||||||
Below are 5 sigma limiting fluxes for an emission line in a high-redshift starburst galaxy, and for pure continuum
| Exposure time | 0.5 hr | 1.0 hr | 3.0 hr |
|---|---|---|---|
| Continuum in R band | 18.0 | 18.8 | 19.9 |
| Emission line flux | 2.05x10-17 | 18.8x10-17 | 19.9x10-18 |
| With 270R grating which gives R ~ 3600 (arc line width of 85 km/s) For a S/N of 5 in the continuum in the R band and an emission line with S/N of 5 that falls on one spectrally resolved element. A SLOW readout speed on the TEK, and a 6 day moon on the AAT. | |||
| Exposure time | 0.5 hr | 1.0 hr | 3.0 hr |
|---|---|---|---|
| Continuum in R band | 18.0 | 18.8 | 19.9 |
| Emission line flux | 1.98x10-17 | 1.01x10-17 | 3.64x10-18 |
| With 1200R grating which gives R ~ 10000 (arc line width of 30 km/s) For a S/N of 5 in the continuum in the R band and an emission line with a S/N of 5 that falls within one spectrally resolved element. A SLOW readout speed on the TEK, and a 6 day moon on the AAT. | |||
I am currently writing IFU display software for SPIRAL and COHSI. The software
can work with any hexagonal array geometry - the lens pattern is input via a
simple numeric text file, and data to be displayed can be either by a text
file, or more usefully, an IRAF multispec file. The IFU software is written entirely in SPP, the core language for IRAF. The two main reasons for this were: (i) IRAF is one of the most popular astronomical packages and (ii) the cross-platform capability of IRAF SPP means that the source code can be compiled on any IRAF platform with the minimum of fuss.
Jim Lewis of the RGO and Alfonso Aragon-Salamanca (IoA) have helped and guided me through the programming of SPP and the astronomical requirements for the software. As of 21/7/97 the software can now display data files and can examine multispec files. In the next month I hope to start modifying 'dofibres' to work with SPIRAL and COHSI data frames.
URL http://www.ast.cam.ac.uk/~optics/spiral/spiral.htm - Revised: 20 Jul 97
We are part of the Institute of Astronomy ,
which is part of the University of Cambridge