X-ray Leak of Deep Survey
May 28, 1996
To: Science Team
From: John Vallerga
Summary: The short wavelength response of the Deep Survey
has only been measured at 44A and 68A. The released
interpolation between these points is inaccurate and
the published extrapolation below 44A is very inaccurate.
Both overestimate the magnitude of the x-ray leak.
However, EUV photometry with the DS on highly absorbed objects
is reliable if the source spectrum convolved with ISM attenuation
and the DS effective area curves peak above 68A. This is true
for most objects with column densities log Nh < 20 such as
Her X-1, the pulsar J0437-4715, and the Coma and Virgo clusters.
Because of the increasing use of the Deep Survey
instrument to measure extragalactic objects as well as
galactic x-ray sources, Roger Malina requested that I look into
the calibration of the response of the Deep Survey lexan/boron quadrant
to wavelengths shorter that 68A. Unfortunately, measured
data points are scarce for both the full instrument response
as well as the individual subcomponents (mirror, filter and detector).
The current "official" effective area file for the DS is
However, it only goes down to 51A. Jerry Edelstein published an
extrapolation down to 11A using models of filter and mirror and detector
QE. It is published in ApJ Vol 454:454-446, 1995 November 20. Both
models show strong fall-off in response as the wavelength get shorter,
mostly because of the poorer mirror reflectivity.
What I wish to add to the discussion is the uncertainties in these
extrapolations. We only have measurements at 44A (Carbon K alpha)
and 68A (Boron K alpha). So the question becomes how to connect the
dots between 44 and 68 and then how to extrapolate shortward.
The mirror reflectivities are based on the optical constants of
gold and are pretty reliable as they have been measured well over this
range. The detector photocathode QE show much structure in this regime
and are known for being inconsistent from one deposition to the
next. The detector QE is also a strong function of input angle
as the photoelectrons that result from the penetrating xrays
have trouble escaping from the surface.
Probably the biggest error in our interpolations/extrapolations are
the shapes of the filter transmission. The curves DO NOT model the
filter edges properly. The boron edge at 66A and the carbon edge
at 43.7A are not strongly evident in the effective area curves.
The boron edge, for example, should show a factor of 50 change
in effective area. These errors are the result of our reliance
on our own measurements and an inattention to modeling the out of
band response of lexan. The results of these errors, however, is an
over estimate of the x-ray leak, especially on the short side of the
edges. Evidence for this comes in the detection of the bright
xray source Sco X-1 at 0.26 cps. Based on a model spectrum
of Damian Christian's from analyzing the SSS results from Einstein,
we expected to see 2.2 cps using the Edelstein eff. area curve,
mostly shortward of 44A. If you use only the eff. area above
44A, the prediction is 0.18 cps
The good news is that most if not all science published using the
DS detections are not affected by the errors in the DS effective
area. The key here is the wavelength of the detected photons. Even
with a column density of log Nh=20, the weighted peak of the response
to a flat input spectrum is around 80A with response shortward of 68
an order of magnitude less. The column densities to the pulsar
J0437-4715, Her X-1, Coma cluster and Virgo cluster are low enough
such that the response increases faster than the flux decreases
and we are detecting EUV photons. Researchers need worry
only when the source spectrum attenuated by the ISM and folded
through the DS effective area peaks shortward of 68A.
Recommendation: If a researcher has a detection in the
DS which he/she believes is due to short wavelength radiation,
then better filter and detector models must be developed
for the DS instrument, only constrained by the measurement