Nuclear Instruments and Methods in Physics Research A 348 (1994) 188-191
North-Holland
NUq[.;LI- AH
INSTRUMENTS
& METHODS
IN PHYSICS
RESEARCH
Seclton A
Tabulated responses of X-ray CCDs to optical astronomical sources
Alan Owens *, Kieran J. McCarthy
Department of Physics and Astronomy, Leicester University, Leicester LE1 7RtI, UK
Received 16 March 1994
We calculate the response of both front-illuminated and back-illuminated X-ray CCDs to various astronomical optical inputs. The results
are given in tabular form and are useful when calculating the expected optical yields encountered during an observation or in designing
optical filters for soft X-ray telescopes.
1. Introduction
In recent years, charged coupled devices (CCDs) have
emerged as the preferred detectors on all new X-ray
astronomy missions [1-4]. This is largely because they
offer a combination of superior spectroscopic and spatial
resolution [5] coupled with small size and all the inherent
advantages of semiconductor technology. However, unlike
traditional X-ray detectors, such as proportional counters,
the response of deep-depletion CCDs does not conve-
niently cut off in the XUV or soft X-ray band, but extends
through the optical region down to infrared wavelengths
(~ 10000 A). This is illustrated in Fig. 1 in which we plot
the broad-band quantum efficiency of a typical front-il-
luminated X-ray CCD from the optical to the X-ray band.
We can see almost 30% of the total response lies in optical
region. Since most astrophysical objects emit many more
optical photons than X-rays (by a factor of 10 for M type
stars and 106 for A type stars), the signal from the
detection plane can easily be swamped by the visible light
component. Even worse, for telescopes employing grazing
incidence optics, the light from such objects is focused at
the same spot as the X-rays, giving rise to excessive dark
current. This in turn results in measurable base line shifts,
broadening, decreased throughout and spectral distortion.
In most cases, filters with high optical to X-ray extinction
ratios are required. Unfortunately, highly attenuating filters
at visible wavelengths also attenuate soft X-rays and a
trade-off between optical contamination and EUV/XUV
attenuation must be made. A good understanding of the
expected optical yields is therefore essential to this study.
In a previous paper [6], we described in detail the rational
behind filter design for X-ray telescopes. In this paper, we
evaluate the optical yields for a variety of astronomical
* Corresponding author.
objects. Three types of EEV device were considered in the
present analysis; a standard "TV" CCD and two large
format devices based on the P88000 series [7] - one being
a deep-depletion front-illuminated CCD and the other a
deep-depletion rear-illuminated CCD. The thicknesses of
the depletion regions are ~ 3, 30 and 30 p~m, respectively.
2. Response to optical sources
The linear absorption coefficients of silicon are shown
in Fig. 2. From the graph we see that the optical depths for
visible photons and keV X-rays are comparable - explain-
ing the large sensitivity of X-ray CCDs at optical wave-
lengths. Optical photons interact in the first few microns of
the depletion region releasing single photoelectrons. The
light induced charge causes a shift in baseline over that
introduced by readout noise. This, in itself, would not be
such a problem were it not that the latter is steady-state
while the former is source dependent. For an incident
monoenergetic X-ray beam of energy, E, the correspond-
ing peak energy is shifted by a factor,
~E/E = (E + wr + OJNe)/(E + ogr), (1)
where w is the energy required to liberate an electron-hole
pair (3.68 eV in Si), r is the readout noise in units of
equivalent number of electrons (typically 3-5 electrons)
and N e is the average number of electrons produced by the
optical flux. The additional charge resulting from the
presence of optical photons also results in a broadening of
the detector full-width-at-half-maximum (FWHM) energy
resolution.
AE = 2.355~/o-g 2 + ~2 + o02, (2)
where ~2, o.gZ and 0-2 are the variances (in units of eV 2)
of the noise components due to carrier generation statistics,
0168-9002/94/$07.00 © 1994 - Elsevier Science B.V. All rights reserved
SSDI 0168-9002(94)00497-U