arXiv:astro-ph/9305025v1 19 May 1993
LICK SLIT SPECTRA OF THIRTY-EIGHT
OBJECTIVE PRISM QSO CANDIDATES
AND LOW METALLICITY HALO STARS1
DAVID TYTLER,2 , 3XIAO-MING FAN,3, 4
VESA T. JUNKKARINEN,3and ROSS D. COHEN3
To appear in the Astronomical Journal, August, 1993
1Based on observations obtained at Lick Observatory, University of California.
2Dept. Physics, University of California, San Diego (TYTLER@cass155.ucsd.edu).
3Center for Astrophysics and Space Sciences, 0111, University of California, San Diego, La Jolla, CA
4Department of Astronomy, Columbia University, New York.
We present Lick Observatory slit spectra of 38 objects which were claimed to have pronounced ultraviolet
excess and emission lines. Zhan & Chen selected these objects by eye from a UK Schmidt telescope IIIaJ
objective prism plate of a field at 0h0.0◦(l ≃ 98◦, b ≃ −60◦). We concentrated on mJ ≃ 18 –19 objects
which Zhan & Chen thought were most likely to be QSOs at redshift zem≥ 2.8.
Most of our spectra have FWHM spectral resolutions of about 4˚ A, and relatively high S/N of about
10 – 50, although some have FWHM ≃ 15˚ A or lower S/N. We find eleven QSOs, four galaxies at z ≃ 0.1,
twenty-two stars and one unidentified object with a low S/N spectrum.
The ZC lists are found to contain many QSOs at low z but few at high z, as would be expected. Of
eleven objects which ZC suggested were QSOs with zprism≤ 2.8, eight (73%) are QSOs. But only three of
twenty-five candidates with zprism≥ 2.8 are QSOs, and only two (8%) of these are at z ≥ 2.8. Unfortunately
the ZC prism redshifts are often incorrect: only five of the eleven QSOs are at redshifts similar to zprism.
Six of the QSOs show absorption systems, including Q0000+027A with a relatively strong associated
C IV absorption system, and Q0008+008 (V≃ 18.9) with a damped Lyα system with an H I column density
The stars include a wide variety of spectral types. There is one new DA4 white dwarf at 170 pc, one
sdB at 14 kpc, and three M stars. The rest are of types F, G and K. We have measured the equivalent
widths of the Ca II K line, the G-band and the Balmer lines in ten stars with the best spectra, and we derive
metallicities. Seven of them are in the range −2.5 ≤ [Fe/H] ≤ −1.7, while the others are less metal poor. If
the stars are dwarfs, then they are at distances of 1 to 7 kpc, but if they are giants, typical distances will be
about 10 kpc.
Subject Headings: quasars: general – galaxies: distances and redshifts – stars: fundamental parameters –
white dwarfs – surveys
Zhan & Chen (1987a,b, 1989a,b, hereafter ZC1, ZC2, ZC3 and ZC4) presented lists of several hundred
QSO candidates, which they selected by eye from a single UK Schmidt telescope IIIaJ objective prism plate.
The candidates were chosen because they had emission lines and UV excess in the range 3200–5400˚ A, and
each was assigned a reliability index, Q = Q1+Q2, where Q1was 1, 2 or 3 for increasing strength of emission
lines, and Q2was similarly valued for increasing strength of UV excess.
We have obtained slit spectra of 38 of these QSO candidates with the Lick Observatory 3 m telescope,
on three separate occasions, during unrelated projects. Two of the objects have Q = 6, thirty-four have
Q = 5 and the remaining two, with Q = 4, are the least likely to be QSOs. Since Q ≥ 4 for all thirty-eight
objects, all should have both emission lines and UV excess.
Table 1 is a journal of our observations, with the instrumental setup (§2 below), wavelength range and
integration time. We also give the reference to the Zhan & Chen paper which contains the object coordinates,
magnitude and finding chart. Note that these charts have East to the right, and that the English translation
of ZC3 lacks charts, while for the other papers the charts are often better in the translation.
Our spectra were taken in support of three different observing programs, at three different times, and
with three different instrumental setups, although in all cases we used a Cassegrain spectrograph with the
Shane 3-m telescope at Lick Observatory.
2.1 Setup A: Ten z ≃ 3 Objects
These targets were selected as bright QSO candidates with zprism≃ 3. They were selected and observed
by VTJ and RDC.
The UV Schmidt camera was used on the Cassegrain spectrograph (Miller & Stone 1987) with a 300
g/mm grating blazed at 4230˚ A in first order. A thinned TI 800×800 CCD with 15 µ pixels, and 7 e−
readout noise was used, giving 3.9˚ A per pixel across 3100˚ A. All observations were made on October 4,
1988, when the sky was clear and the seeing was about 1 arcsecond. For the ten program objects we used a
wide 2.88 arcsecond slit, giving a FWHM resolution of 3.5 pixels, which is 14˚ A, but for the flux standard
star we used a 7.9 arcsecond slit. The slit was not rotated to the parallactic angle, but it was rotated to
PA = 248◦to simultaneously record 0003+011A & B, and to PA = 92◦for 0011−002A & B. Hour angles
ranged from 2 hours 30 minutes East to 3 hours 10 min West. All exposures were 600 seconds, and the
spectra were reduced in the usual way.
2.2 Setup B: Eleven Intermediate z Objects with Close Neighbours
These eleven QSO candidates were observed on August 24 or 27, 1990 by DT and FXM to search for
Mg II absorption systems which might show large scale (≃ 100 h−1Mpc) correlations in three dimensions.
This program was motivated by the finding of Tytler et al. (1987) that a few QSOs each had more Mg II
systems than were expected if they were all intervening, a two sigma result which has since been refuted by
much larger samples which do not show any sign of such correlations (Sargent et al. 1988; Steidel & Sargent
1992, Tytler, Sandoval & Fan 1993 §2.4). The QSO candidates which we observed were chosen because
they had one or more neighbours within about 1◦. Here we present our observations of only the ZC QSO
candidates, two of which were also observed by VTJ and RDC with setup A. Other QSOs observed in this
program will be discussed elsewhere.
Spectra were obtained with the Cassegrain spectrograph using a 600 g/mm grism, blazed at 4840˚ A in
first order (Miller and Stone 1987). We used a thinned TI 800×800 CCD with 15 µ pixels, and 7 e−readout
noise, giving 3.43˚ A per pixel from 4312 to 7059˚ A. The wavelength range was chosen to maximize the chance
of detecting redshifted Mg II absorption line systems, and it unfortunately misses blue wavelengths which
are most useful for stellar spectral classification. A 2.09 arcsec slit was used giving a FWHM resolution of
2.5 pixels, which is 8.6˚ A. The slit was not rotated to the parallactic angle, but the spectra were reduced in
the usual way.
2.3 Setup C: Nineteen High z Objects
Nineteen QSOs were observed by DT in November 1992 with the superb new Kast double spectrograph,
which records blue and red spectra simultaneously. We used a dichroic with a nominal wavelength of 5500˚ A.
Light with wavelengths to the blue of this were dispersed with a 600 g/mm grism blazed at 4310˚ A, while
the red light was dispersed by a 600 g/mm reflection grating blazed at 7500˚ A. A thinned Reticon 1200×400
CCD was used in each of the blue and red cameras. In the blue we recorded from 3320 to 5485˚ A with
1.81˚ A per pixel, and a two pixel FWHM of 3.6˚ A, and in the red from 5530 to 8270˚ A with 2.34˚ A per pixel
and a two pixel FWHM of 4.7˚ A. A 1.5 arcsec slit was used, and was rotated to the appropriate parallactic
angle. The sky was clear, the seeing about 1.5 arcseconds, and the spectra were reduced in the usual manner.
We first discuss various problems with the spectra, then slit magnitude and color estimates.
3.1 The Spectra
The spectra shown in Figure 1 are grouped by setup (A, then B, then C) to make spectral features easier
to identify. The flux is fν, in units of micro-Jansky, and all wavelengths given in this paper are vacuum,
but they are not heliocentric. Wavelength scales should be accurate to about one pixel or better. A one
sigma error trace is shown beneath the spectra from setups B and C. Peaks in the error correspond to sky
emission lines. Note that poorly subtracted sky emission lines can appear as emission and/or absorption
in the spectra. In setup B there is frequently a bogus absorption at the extreme blue end of the spectrum
(4311˚ A), a bogus emission feature near 4325˚ A, and a second bogus absorption near 4370˚ A. The former
two arise from poor flux calibration, while the third is bad sky subtraction.
For setups A we did not attempt to correct for atmospheric absorption, hence the B band (6867˚ A) is
visible. For setups B and C we did use early type star spectra to attempt to remove the B band, the A band
(7600˚ A), and OH absorption at 7160–7340˚ A but with varying success.
The spectra have been corrected for atmospheric extinction, but not for interstellar extinction (b ≃
−60◦). Table 2 is a summary of our results.
3.2 Slit Magnitudes
Magnitudes listed by ZC were obtained from image sizes on a direct Schmidt plate, using the King et
al. (1981) calibration of the dependence of BJ magnitude on image size. ZC5 noted that these magnitudes
may be too bright by 0.5 – 1.0 mag. because they found most objects at ≃ 18.5, a whole magnitude brighter
than the peak of the otherwise similar survey by Savage et al. (1984). However H¨ ortnagl, Kimeswenger
& Weinberger (1992) have shown that King et al. measured larger image diameters at a given mj, which
suggests that the ZC magnitudes may actually be too faint, rather than too bright, for mj ≥ 18. We can
not determine which is correct because we do not know how ZC measured image sizes.
To try to reduce this uncertainty, we have estimated magnitudes from our slit spectra. These magnitudes
are highly uncertain because we used narrow slits. A broad band flux F is defined as
where T(λ) is the band transmission (Kitchin, C.R. 1984), f(λ) is the flux per unit wavelength, and λmin
is ideally −∞, but in practice was the minimum wavelength of the spectrum.
an estimate of the flux in the dichroic filter gap, which was only 20˚ A. We obtained magnitudes from
U = −2.5log10FU+ KU, and similarly for B and V, where the constants KU, KB and KV were obtained
from the standard star spectra.
Our slit magnitudes are listed in Table 3. They have automatically been corrected for atmospheric
extinction by the usual flux calibration process, which converts from recorded photoelectrons to flux above
the atmosphere, as a function of wavelength. Galactic reddening E(B−V) values from Burstein & Heiles
For setup C we added
(1982) are 0.01, 0.02, or 0.03 for the targets. The colors (B−V)0listed in Table 3 have been corrected for
these reddening values, but we have not corrected the individual magnitudes because their zero point errors
are much larger than the corrections of AV≃ 0.03 − 0.09.
We have checked our magnitudes in five ways. First, we calculated a magnitude from each of our thirteen
standard star spectra. We obtained standard deviations of s = 0.17 magnitudes for the V magnitudes and
s = 0.21 for the B magnitudes, which we regard as lower bounds on the external errors of our other magnitude
Second, two objects were observed twice, and in both cases the magnitudes were 0.7 – 1.0 fainter in
setup A, because a wider slit was used for the flux standard than for the program objects, although the
(B−V) colors differ by much less (0.02 and 0.14).
Third, when we compare our slit (B−V) colors with those estimated from the strength of the Balmer
lines in §3.4 below, we find excellent agreement which suggests that our slit (B−V) colors from setup C (the
only ones to have the silt aligned to the parallactic angle) have a 1σ error of under 0.04 mag.
Fourth, in Figure 2 we show the difference between our slit B magnitudes and the ZC image size BJ
magnitudes. Ours are on average 1.4 magnitudes fainter. ZC5 noted that their magnitudes were probably
too bright, but they guessed by only 0.5 – 1.0 magnitudes, which suggests that some of our magnitudes may
be too faint. We do not see any systematic differences between our three setups.
Fifth, four of the QSOs (0004-005B, 0006+020B, 0006+025 and 0010-002B) have been found indepen-
dently by Foltz et al. (1989). Their BJmagnitudes, which we list in §6, are brighter than ours by 0.3, 0.54,
1.19 and 1.57 magnitudes respectively, where the first three are from setup B, and the last one is from setup
These tests suggest that our colors, but not necessarily the individual magnitudes, from setup C are
good. Both the magnitudes and colors from setups A and B setups are also suspect because the slit was
narrow (1.5 to 2.88 arcsec), it was not rotated to the parallactic angle, and the TV camera guides on the
red, so we expect that the B and especially the U magnitudes will be systematically too faint, and the V
magnitudes should be the least bad. In addition there are several reasons why we expect our magnitudes
for the program objects to be systematically too faint. The standard stars were observed for much shorter
times than the faint targets, they should be better centered on the slit, and better focused, and for setup A
a wider slit was used for the standard stars than for the program objects.
Papers by Gunn & Stryker (1983) and Jacoby, Hunter & Christian (1984) were consulted to obtain
rough stellar classifications. Berg et al. (1992) present a simple classification scheme which they used on
their 6˚ A FWHM optical spectra of QSO candidates. Beers et al. (1992a) present digital spectra covering
3700 – 4500˚ A at 0.7 to 1.2˚ A FWHM for various hot halo stars (A, DA, sdO, sdB, Horizontal Branch =
HB), together with classification criteria, while Beers, Preston & Shectman (1992b, hereafter BPS2) present
spectra of cooler halo F and G stars of various metallicities.
Greenstein (1980) discusses the difficulty of distinguishing white dwarfs (WDs) from hot halo (subdwarfs
sdO or sdB, hot HB) stars. The separation is hardest for the hottest stars because the physical differences
in temperature and gravity are also small, so that DAwk (weak Balmer lines) and sharp lined DAs can be
confused with sdBs. But the distinction is easy below 12,000 K because the WDs then have stronger Balmer
lines. Greenstein (1980) shows that any star with W(Hγ) ≥ 15˚ A must be a DA, but as W(Hγ) drops from
15 to 5˚ A, one has either DA stars of increasing T, or stars from the sequence HBA (HB type A), HBB, sdB
and sdO, which is one of both increasing T and gravity. For the hot stars we searched for but did not find
any He I 4026, 4388, and 4471, or He II 4686.
Our spectral classifications are given in Table 2. For the stars we list the spectral type implied by the
spectral features, and then, in parentheses, that which would correspond to the (B−V) if the star had solar
metallicity. Most of the stars actually have significantly lower metallicities, so their (B−V) colors are from
0.1 to 0.3 magnitudes bluer (e.g. Beers et al. 1990; hereafter BPSK) than those of solar abundance stars
with the same MV. This deblanketing effect accounts for why the spectra type deduced from the colors are
hotter than those from the spectral features. Notes on the classification of individual objects are given in §6