Next: The German ROSAT XUV Data Center and a ROSAT XUV Pointed Phase Source Catalogue
Previous: FPTOOL: A GUI for the Analysis of Fabry-Perot Data in IRAF
Table of Contents --- Search --- PS reprint


Astronomical Data Analysis Software and Systems V
ASP Conference Series, Vol. 101, 1996
George H. Jacoby and Jeannette Barnes, eds.

A Reexamination of the `UV Leak' of the High Resolution Imager on ROSAT

J. D. Silverman

Smithsonian Astrophysical Observatory, Cambridge, MA 02138

K. Kearns

Wesleyan University, Middletown, CT 06459

D. E. Harris

Smithsonian Astrophysical Observatory, Cambridge, MA 02138

Abstract:

The `UV Leak' is apparent in observations of Vega which produce an HRI count rate of 0.1 cts/sec whereas a much lower rate is found by the PSPC. This discrepancy led Arp (1995) to interpret extended features adjacent to NGC 1097 (on a `soft' map) as evidence for extragalactic UV emission. We show that the majority of Arp's features are caused by a detector characteristic which produces a slightly higher noise level for PHA 1--3 in a region where the gain is lower, i.e., a higher fraction of background photons produce counts in the lower PHA channels for detector regions with lower gain.

1. Introduction

In response to a report (Arp 1995) which concluded that features on an X-ray map were caused by celestial UV emission rather than X-rays, we have initiated a re-evaluation of the so-called `UV-leak' in the ROSAT HRI (David et al. 1995). We believe Arp's results can be attributed to HRI gain effects although there remain unanswered questions about the `UV Leak'.

2. Effects of the Gain Map

Based on the `UV leak' of Vega, Arp suggested that the high count rate in PHA 1--3 of spatial regions surrounding NGC 1097 was a result of UV emission rather than real X-rays. From two observations of NGC 1097 separated by six months, it is apparent that the count rate increase in specific regions of the detector is caused by the spatial variation of the gain (mean PHA) across the detector. Therefore, it is necessary to first evaluate the complete spectral distribution of the source or spatial feature. Looking only at the first three PHA channels (as in the case of NGC 1097) could result in a misleading analysis of the observations.

Figure 1 shows an image of the prelaunch gain map of the ROSAT HRI for the element CU.

  
Figure 1: The ground based spatial gain map for the spectral line of copper (0.93 keV). This image has been blocked by a factor of 64 thereby producing a scale of 32 arcsec/pix. This image includes the greyscale of the spatial variation of the gain overlayed with the contour lines of the mean PHA channel. The contour lines begin with a PHA of 4.0 in the central dark region an extend outward to a PHA of 7.0 on the periphery.
Figure 1: PS 125 Kb

PHA distributions in the darker areas will be shifted to lower channels.

The observations of NGC 1097 provide an important means of distinguishing between the effects of the gain map on a detection and the `UV Leak'. Arp has analyzed an HRI detection of NGC 1097 observed in Jan. 1994.

  
Figure 2: Two observations of NGC 1097 separated by six months. A smoothed image of a `soft map' for the observation in Jan (a) and June (b) are shown.
Figure 2: (left) PS 272 Kb, Figure 2: (right) PS 528 Kb

By looking at the features on a soft map (PHA 1--3, Figure 2a), he suggested that there appears to be emission from spatial features adjacent to the region of NGC 1097 (top right circle). Since these features are not detected in the more sensitive PSPC, Arp attributed the features to emission from UV celestial sources.

In a second observation of NGC 1097 taken six months later, in which the detector is rotated about 180 from the first observation, features which Arp interpreted as UV emission are no longer apparent in that region of the sky. The higher count level on the `soft map' has now shifted to the opposite side of the field (Figure 2b) as expected because of the rotation of the detector.

We have compared the spectral distributions for the two opposite regions from the image of NGC 1097 for both observations.

  
Figure 3: Spectral distribution for the marked regions of the two observations of NGC 1097. The solid line corresponds to the top right region while the dashed line corresponds to the lower left region. These results are consistent with the rotation of the gain map by 180.
Figure 3: (left) PS 124 Kb, Figure 3: (right) PS 124 Kb

In Figure 3a, the top right region with the features described by Arp falls within the low gain area of the detector and shows an increase in the count rate for PHA 1--3. Then in Figure 3b, we see that the increase in low PHA channel counts now occurs on the other half (lower left corner) of the field.

3. The `UV Leak'

Throughout the mission, there has been evidence that sources with strong UV emission cause higher count rates in the lower pulse height amplitude channels (PHA) of the HRI than predicted. The observed count rates for Vega (0.1 cts/sec) and Beta Car have been shown to significantly exceed the expected values for the HRI. Based on the spectral distribution from IUE measurements, Zombeck (1992) has predicted an HRI count rate of 4.7e-4 cts/sec for Vega. For the PSPC, however, the observed count rate of Vega agrees with the filter transmission measurement and Vega's excess UV emission. From preflight measurements, the peak UV transmission of the UV filter is within the 1500--1700 Angstrom band. Therefore, we might have expected the `UV Leak' photons to produce a PHA distribution which would be indistinguishable from that of the UV calibration lamp (1800 Angstroms). Since the mean PHA of Vega is less than that of the UV cal lamp (David et al.), we are not confident that Vega's observed high count rate can actually be caused by 1500--1700 Angstrom emission.

4. Conclusion

The `UV Leak' of the ROSAT HRI has motivated an analysis of the contributing factors that affect the spectral distribution of a source. We have shown that the gain map directly affects the PHA distribution of collected photons. Even though Vega and Beta Car have high count rates in the lower PHA channels, it is incorrect to assume that the excess counts in PHA 1--3 are always attributed to UV radiation because of the extremely limited spectral resolution of the HRI. Therefore, it is necessary to determine where on the detector the events occurred and how its count rate is affected by the variation of the gain.

Two questions still remain. Why is the predicted count rate of Vega different from the observed measurements? And why does the spectral distribution of Vega differ from the UV calibration lamp?

Acknowledgments:

This work was supported by NASA contract NAS5-30934.

References:

Arp, H. 1995, A&A, 294, L45

David, L. D., Harnden Jr., F. R., Kearns, K. E., & Zombeck, M. V. 1995, The ROSAT High Resolution Imager (HRI)

Zombeck M. 1992, internal memo


Next: The German ROSAT XUV Data Center and a ROSAT XUV Pointed Phase Source Catalogue
Previous: FPTOOL: A GUI for the Analysis of Fabry-Perot Data in IRAF
Table of Contents --- Search --- PS reprint
Wed Jul 3 08:07:12 MST 1996