ULTRADEEP OPTICAL SPECTROSCOPY WITH PMAS using the NodandShuffle Technique

1、1ultra-deep optical spectroscopy with pmasusing the nod-and-shuffle techniqueabstract:pmas, the potsdam multi-aperture spectrophotometer is a new integral field spectrograph in the optical, optimized for good transmission and high image quality from 350nm to 1m. we present our plan to implement a cc

2、d charge-shuffle mode to allow for beam switching with a very high degree of sky subtraction accuracy for faint object 3d spectroscopy.key words:integral field spectroscopy faint object spectroscopy beam switching1.introductionthe exploration of the high-redshift universe is certainly among the most

3、 exciting topics of modern astrophysics, both on the grounds of advances with numerical simulations in theoretical cosmology, as well as observa-tionally with the light collecting power of 8-10m class telescopes. while hst has provided us with images of galaxies up to a redshift of z5, ground-based

4、imaging and spectroscopy even with the largest available telescopes is embarrassingly difficult and expensive in terms of observing time.the intrinsic faintness of high-redshift galaxies is only one of the reasons for the problem. ground-based observations are confronted with the fact that these obj

5、ects are much fainter than the surface brightness of the sky, typically 1% of the sky or less. therefore, the detection limit for low and medium resolution spectroscopy is dominated by the limited accuracy in subtracting the bright and variable sky background. as an instructive example, fig.1 shows

6、images and a spectrum of a galaxy at redshift 6.56 (pictures taken from a press release of the ifa, univ. of hawaii), obtained as part of a total of 24 hours of multi-wavelength broad- and narrowband imaging at the keck- and subaru telescopes, and with 4 hours worth of integration using keck and the

7、 lris spectrograph (hu et al. 2002). figure-1: ly emission at z=6.56, detected with an onband/offband imaging technique. the narrow-band filter transmission curve (insert) has been strategically chosen to fall in a gap between two strong oh night sky emission bands ( note that the sky lines have bee

8、n scaled down by a factor of 100 ). the strong residuals redward of the redshifted ly line (near 9200 ) are typical for the problem of low/medium resolution spectroscopy in the red. the 10 resolution lris spectrum on the top (taken from another source) demonstrates that more than half of the availab

9、le wavelength range is contaminated by the disturbingly bright oh emission bands (pictures from ifa press release by hu et al. 2002, and i. lehmann, aip).2.beam switching techniquesthe spectacular images of the hubble deep fields, obtained with integration times on the order of 100 hours, may seem t

10、o suggest that it is merely a matter of allocating sufficiently long observing time at 8-10m class telescopes to also obtain good quality spectra of the most distant galaxies of the universe. that this is not so due to the presence of the bright night sky has been known for some time and is explaine

11、d e.g. in cuillandre et al. 1994. chiefly, the limited accuracy of a conventional slit spectrograph in subtrac-ting the sky contribution from a spectrum containing (object + sky) flux results in systematic errors which propagate multiplicatively. contrary to photon shot noise and others sources of r

12、andom error the resulting signal-to-titel:z6_picture.epserstellt von:fig2dev version 3.2 patchlevel 1vorschau:diese eps-grafik wurde nicht gespeichertmit einer enthaltenen vorschau.kommentar:diese eps-grafik wird an einenpostscript-drucker gedruckt, aber nichtan andere druckertypen.error! no text of

13、 specified style in document.3noise ratio cannot be reduced by increasing the observing time. instead, the s/n is saturating at some limiting value, no matter how large the telescope is, or how long we expose our spectra. the effect is particularily worrisome near bright night sky lines, dominating

14、the sky spectrum at wavelengths above 6000 (for spectrophotometric data, see massey & foltz 2000).cuillandre et al. 1994 proposed a method to overcome this limitation, called “va-et-vient”, which is essentially an adaptation of the familiar beam-switching technique, known from ir astronomy, to ccds.

15、 like in many other measurement problems regardless of any specific technical domain, small signals affected by temporarily varying bias or gain can be determined more accurately by rapidly switching between the signal of interest and a reference (chopping). the problem for ccds in adopting this sch

16、eme, however, has been the prohibitively long readout-time required for low read-noise operation of the detector. “va-et-vient” makes use of the fact that in frame transfer mode charge can be clocked in two directions, thus allowing to consecutively obtain (sky + object) and (sky) exposures on the s

17、ame frame without having to read out each sample immediately. this scheme requires to provide on the ccd for one active zone (exposure) plus two masked areas (storage), and to parallel-clock 2/3 of the resulting charge image back and forth, synchronized with the telescope which has to alternate betw

18、een the object and a sky pointing, accordingly. the important point is that both (object + sky) and (sky reference) are always observed quasi-simultaneously through the identical optical train, with identical slit/aberration/distortion/flexure/ properties, and with identical ccd pixels, which exhibi

19、t essentially the same flatfield and fringing behaviour.a more recent paper by glazebrook and bland-hawthorn (2001) describes the method which they dubbed “nod-shuffle-spectroscopy”, and a practical implementation with the ldss spectrograph at the aat. based on a wealth of test data, the authors cla

20、im that, in principle, nod-shuffle should be able to reach a limiting flux level of 10-4 of the sky brightness - provided one can spend enough observing time.3.pmaspmas, the potsdam multi-aperture spectrophotometer, is a new integral field spectrograph, which was designed as a flexible travelling in

21、strument, and which has seen first light at the calar alto 3.5m telescope in may 2001 (roth et al. 2002b). the instrument has been developed at the astrophysical institute potsdam (roth et al.1997, 1998a, 1998b, 2000, fechner et al. 2000).figure-2: pmas during the process of mounting to the calar al

22、to 3.5m telescope casse-grain flange for first light on may 28, 2001.a major motivation for this development has been the goal to use the advantages of integral field spectroscopy (ifs) for the difficult observation of individual sources in crowded fields, e.g. resolved stars and nebulae in nearby g

23、alaxies, analogous to the development of crowded field photometry after the advent of ccds in astronomy. we have also considered the benefits and limitations of ifs for faint object spectroscopy (roth et al. 2000) which can be summarized as follows:absence of slit losses (there is no slit), image sl

24、icer effect for extended sourcessky background estimate can be derived more reliably from an annulus or an extended region around the target rather than from interpolation along a slituncritical telescope pointing for faint targets due to extended 2-dimensional fov differential atmospheric refractio

25、n and associated slit losses are irrelevant for ifserror! no text of specified style in document.5figure-3: single 2k4k ccd frame of pmas fiber spectrograph internal calibration exposure (continuum+ spectral line lamp). the magnified insert shows two groups of 16 well-separated spectra, correspondin

26、g to two rows of 16 spatial elements of the lens array. figure-4: the pmas fiber spectrograph supports a large spectroscopic field of view: 4k4k with 15m pixels, corresponding to 3200 coverage at 3 resolution (600 l/mm grating).4.nod- and shuffle with pmas with funding from the verbundforschung of t

27、he german bmbf, and in collaboration with lutz wisotzki and joachim wambsganss, university of potsdam, we have started in april 2002 the “ultros” project which is intended to implement a variant of the nod-and-shuffle technique in pmas. figure-5: pmas charge-shuffle mode experimentally implemented i

28、n the lab (see text). the idea is to use the strips of free space between neighbouring spectra as storage areas for the shuffling process rather than masked storage space at the edges of the chip. fig.5 demonstrates how this mode has been implemented in the lab: while the left frame shows a normal c

29、alibration frame, the right frame has twice the number of spectra generated by error! no text of specified style in document.7clocking charge 7 pixels (half the average vertical distance between spectra) back and forth, synchronized with the shutter.acknowledgementsthe pmas project has received fina

30、ncial support from the german bmbf under verbundforschung grants 05 3pa414/1 and 05 al9ba1/9. the ultros project is funded under grant 05ae2baa/4. tb acknow-ledges support from dfg grant ha1850/10. we wish to thank our colleagues of the eso odt and their former team leader jim beletic for allowing u

31、s to copy the eso design vlt detector head, the ace ccd controller, developed by roland reiss, and generous support and advice throughout, without which the project would not have become a success. thanks also to ingo lehmann for providing us with the lris frame (fig.1).references1 cuillandre, j.c.,

32、 fort, b., picat, j.p., soucail, g., altieri, b., beigbeder, f., dupin, j.p., pourthie, t., ratier, g. 1994, a&a 281, 6032 fechner, t., wolter, d., roth, m.m. 2000, in proc. optical detectors for astronomy ii, kluwer dordrecht, eds. james w. beletic, paola amico, p.1533 glazebrook, k., bland-hawthorn, j. 2001, pasp 113, 1974 hu, e.m., cowie, l.l., mcmahon, r.g., capak, p., iwamuro, f., kne