Quasar Redshifts from S-CAM Observations: Direct
Colour Determination of ~12 Gyr-Old Photons
Jos H. J. de Bruijne1, A. P. Reynolds1,
M. A. C. Perryman1,2, A. Peacock1, F. Favata1,
N. Rando1, D. Martin1, P. Verhoeve1,
1: Research and Scientific Support Department of the European
Space Agency. 2: Sterrewacht Leiden. 3: Hamburger Sternwarte
CCDs have revolutionised
astronomy in the last quarter of the 20th century, yet measuring energy
distributions of celestial objects still requires the indirect methods
of filter photometry or dispersive spectroscopy. The development of
superconducting tunnel junction (STJ) detectors (Perryman
et al., 1993; Peacock et al., 1996,
has opened up the possibility of measuring individual optical photon
energies directly. The first time-and spectrally-resolved observations
of cataclysmic variables and pulsars using these techniques have been
reported (e.g., Perryman et al., 1999,
et al., 2002), and the first direct measurements of the redshifts
of quasars using an imaging detector with intrinsic energy resolution were
published early this year (de
Bruijne et al., 2002; cf. http://astro.esa.int/SA-general//Research/Detectors_and_optics/).
We observed 11 Lyman-limit quasars, selected from the literature
in the range z = 2.2 – 4.1, using S-Cam2 (Rando
et al., 2000) on the William Herschel Telescope in October 2000.
S-Cam2 is a 6×6 imaging array of 25×25mm2(0.6×0.6
arcsec2) tantalum junctions, providing individual photon arrival
times to ~5μs, a resolving power of ℜ~8 at 500nm, and a high sensitivity
from the atmospheric cutoff to ~720nm (this cutoff is currently set by
long-wavelength filters which reduce thermal noise photons). All targets
show strong Ly-α and CIV emission lines which, at redshifts ~2 – 4, fall
in our wavelength response range.
Information on each detected photon consists of arrival time,
coordinates of the junction, and an energy channel E in the range 0–255.
Laboratory measurements have confirmed that all junctions have a highly
linear energy response, so that an incident photon of energy Ep
is assigned to an energy channel E ~ 42.5· Ep[eV] –
Bruijne et al., 2002).
We determined quasar redshifts z by fitting the calibrated observed
energy distributions with a single template HST quasar spectrum, i.e.,
by minimising the function χ2(z) (de
Bruijne et al., 2002). Examples of observed and modelled spectra are
shown in Figure 1. The overall shape of these spectra, in particular the
falloff at low energy channels (long wavelengths), is due to the combined
response of the instrument and telescope. In practice, the Ly-α line and
the associated break at shorter wavelengths contribute most to the redshift
Figure 1. Results for QSO 0127+059, 0148–097,
and 0642+449. Left: observed (black) and modelled (grey) energy channel
distributions (arbitrary units). Our model is based on a single template
HST quasar spectrum. Insets indicate the Poisson noise. Numbers above the
top left panel show the mapping between energy channel and wavelength. Right:
the corresponding dependence of χ2 on z. Vertical dashed lines
indicate the literature redshifts; the dotted line for QSO 0127+059 indicates
z=3.04 (see text). [ JPEG | TIFF ]
Figure 2 compares the best-fit redshifts with the literature values.
QSO 0127+059 is our single prominent outlier. It was discovered in a
thin prism survey, classified as a possible quasar, and tentatively assigned
a redshift of z ≈ 2.30 with a questionable line identification. Our fit
provides a good representation of the data (Figure 1), yet the derived
redshift, z = 2.976, differs significantly from the literature value. We
therefore obtained a spectrum of this object with the Siding Spring Observatory
2.3-m telescope, from which a redshift z = 3.04 was deduced, in excellent
agreement with our estimate!
Figure 2. Observed versus literature redshifts.
Numbers refer to the objects (de
Bruijne et al., 2002). Symbol sizes correspond to χ2; smaller
symbols indicate a poorer fit. QSO 0127+059 has an incorrect literature redshift
of 2.30; follow-up spectroscopy has yielded z=3.04, moving the point to
the position shown in grey. The dashed line shows the 1:1 correlation. [
JPEG | TIFF ]
As all fits have reduced χ2>> 1, none of them is formally
acceptable. The general consistency between the models and the observations,
combined with the pronounced, deep and narrow, minima in all χ2(z)
plots, nonetheless indicates that our model fits the data well. Small systematic
errors related to, e.g., template mismatch, are, although largely hidden
due to the limited detector resolution, the key to this paradox (de
Bruijne et al., 2002).
Pronounced χ2(z) minima are already present in
our data truncated a posteriori to observation times as small
as, e.g., 10 – 20s for QSO 0000–263 (z = 4.1; V = 17.5mag), where ~350
source photons s–1 were recorded (Figure 3). We therefore
conclude that efficient low-resolution spectroscopy of faint extragalactic
sources is possible with STJ devices, enabling the determination of redshift.
Extraction of detailed physical information from the spectra presented
here is limited by the modest resolving power of S-Cam2 (ℜ~8). A significant
improvement in energy resolution is, however, foreseen in the future (e.g.,
et al., 2000), promising enhanced physical diagnostic capabilities.
It has, for example, been shown that an STJ detector with Â~20 would
allow the determination of galaxy morphological type and perhaps emission
and absorption line ratios (Jakobsen,
& Brunner, 2000).
Figure 3. Top: χ2(z) using the first
x=1,...,21 seconds of data of QSO 0000–263 (z = 4.1). The dashed and solid
lines indicate the literature and best-fit redshifts, respectively. Bottom:
as top panels, but showing the observed QSO spectra (black) and best-fit
models (grey). Panels have differing vertical scales. [ JPEG | TIFF ]
STJ instrument development within ESA is currently also aimed at producing
larger format arrays, to facilitate sky subtraction and possibly allow
for multi-object spectroscopy, and at extending the wavelength response
further to the red. The latter objective, which is consistent with the
fundamental device response characteristics, would open up a larger accessible
We acknowledge the ING staff for their excellent support during
the S-Cam observing campaigns.