Fibers and Voorwerpjes
While preparing for more observations of the Galaxy Zoo giant AGN clouds (Voorwerpjes), this is a good time to introduce more complete ways of obtaining astronomical spectra. Traditionally, we’ve put a long slit in front of spectrographs, so we can measure everything along that line without worries about overlapping spectra of different objects or pieces of sky. In some cases, as with the optical fibers used by the Sloan Digital Sky Survey, we get the light summed within a circular aperture on the sky (with Sloan, from hundreds of different objects at each pointing of the telescope). But many of the things we want to understand are large and oddly shaped, so these approaches limit us to a very partial view (or to making many observations to cover everything of interest). Enter the Integral-Field Unit (IFU), which is any kind of device that lets us get the spectrum of every point in some region of the sky. They often use fiber optics to rearrange light from the object, so each small region of it comes out at a different place on what would otherwise be the spectrograph slit. After that it all becomes a software problem.
IFUs are becoming more common on large telescopes. We’ve gotten excellent data on some Voorwerpje systems with the unit on the 8-meter Gemini North telescope. Here’s a sample of raw data on UGC 11185. Each horizontal streak is the spectrum of an area 0.2 arcseconds square. The sampling, sensitivity, and image quality are superb, revealing multiple clouds of gas moving within a total span of almost 1000 km/s.
On the other hand, if we want to use its whole wavelength range, the Gemini device covers only 3.5×5 arcseconds of sky at once. I’m headed to the 3.5m WIYN telescope on Kitt Peak to use a complementary device called Hexpak, newly commissioned by instrument designer Matt Bershady of the University of Wisconsin (who I’ve been emailing about this since I learned of the project three years ago). This fiber bundle plugs into the multipurpose spectrograph kept in a climate-controlled room below the telescope, and combines small and densely-packed fibers in the middle (for things like galactic nuclei, small and bright with lots of structure) and large fibers near the edges (collecting a lot of signal from large diffuse surrounding material – sound familiar?). Matt and his team were able to get a short exposure through thin clouds of UGC 11185 as a feasibility test – here’s a piece of that raw data frame, showing the small central fiber and the larger surrounding ones (which show brighter night-sky airglow lines as well as more object signal; the bright [O III] lines and H-beta are near the middle, with wavelength increasing to the right for each spectrum). I hope to get a lot more data like this shortly.
Elsewhere, the European Southern Observatory has commissioned an enormous IFU, and the Sloan team has rebuilt their fiber bundles so that each one now makes multiple IFUs which can be placed on many galaxies at a time – this part of the Sloan survey extension is known as MANGA. Then there is the Spanish-led CALIFA project for hundreds of galaxies, which has publicly released data for their first two subsets. Then there are SAURON (whose data ca be tamed in software by GANDALF) and the upgrade of SCORPIO-2 and more… Swimming in data as we sift for knowledge, I am reminded of this anonymous computer error message in haiku form:
Out of memory.
We wish to hold the whole sky
but we never will.
Excellent, so there is 10X or 100X the amount of data for the same length of exposure time?
Indeed. In this case, about 20x faster than using the long-slit spectrograph setup first employed to confirm the clouds. (There is a cost in extra time it takes to process the data, but that’s not expensive telescope time under a dark sky).
Very cool Bill!
Are there gaps in the integrated spectra? In both images you posted, there seem to be stripes, with gaps in between. To get full coverage, do you have to ‘dither’?
The stripes are the 1-dimensional spectra through each aperture, which then need to be reassembled into “data cubes” knowing the two-dimensional location of each one. For Gemini, the spatial coverage is essentially continuous (with hexagonal apertures collected by tiny lenses in front of each fiber). For Hexpak, where the circular fiber ends are surrounded by the cladding of each fiber, a 3- or 4-point dither pattern (almost) fills the gaps.