Summary of diploma thesis, Hamburg University 1996

X-ray spectral slope of AGN in the ROSAT All Sky Survey

Volker Beckmann

1. Abstract

The spectral properties of AGN in the X-ray energy band are still controversial. Many investigations on various subclasses of AGN have been made, obtaining different spectral slopes. To get rid of the problems due to selection effects we present results using a method based upon all AGN candidates found in the Rosat All Sky Survey identified on objective prism plates of the Hamburg Quasar Survey. Applying the log N - log S relation for RASS AGN and using galactic absorbtion we obtain a mean spectral photon index of Gamma = 2.53 at low hydrogen column density and Gamma = 2.18 for high absorption. Our results support the model of concave X-ray spectra and soft X-ray excess.

2. Data Base

This work is based on data, observed by the ROSAT satellite during the ROSAT All Sky Survey (Voges et al. 1996), containing about 80.000 objects in the PSPC energy band (0.07 - 2.4 keV) with a detection likelihood >= 10, from which the 18 811 brightest sources were compiled in the ROSAT Bright Source Catalogue (RASS-BSC, Voges et al. 1999). Basis for our Hamburg/RASS Catalogue of optical identifications (HRC, Bade et al. 1997) in the northern hemisphere are the objective and direct plates of the Hamburger Quasar Survey (HQS, Hagen et al. 1995) taken with the Schmidt telescope on Calar Alto/Spain. The flux limit is usually B = 18.5mag but depends on the quality of the plate. The direct plates have a brightness limit of B = 20mag. For the evaluation and determination of the most plausible optical counterpart to a RASS-object all information gathered is put together. The result of the identification shows the following distribution: 16% of the RASS-BSC objects could not be identified, 41% are AGN, 6% galaxies and clusters of galaxies, 34% stars and 3% seem to be empty fields, where no optical counterpart to the X-ray source is found (Bade et al. 1997).
This work is based upon the identification on 70 HQS fields, covering an area of about 1.900 squaredegrees on the northern hemisphere, where 843 AGN-candidates have been found.

3. Determination of the X-ray spectral slope

The method used to ascertain the spectral index depending on the X-ray position and the countrate as given by the RASS-BSC, has been described by Maccacaro et al. (1988). Assuming that the objects are randomly distributed in an Euclidean void and that the absolute magnitudes of the objects are randomly distributed as well, we have:
log(N / No) = -1.5 * log(S / So)
One can replace N by the AGN density by sky area. Because the flux S depends on absorption, a hydrogen column density is assigned to every element of sky area (we use the values derived by Stark et al. 1992). Assuming a single power law plus absorption the X-ray fluxes are formed by
fx = CF * countrate, with the conversion factor CF being a function of the spectral index and the hydrogen column density (Tananbaum et al. 1979). Plotting the density of AGN per NH-Intervall against galactic column density the interesting parameters can be determined by Chi Square minimization. This method benefits from the independence on object selection, but needs precisely determined AGN-densities by area in relation to the galactic column density.

4. Determination of AGN-densities by area

Assuming that the identification of X-ray objects is regular on the choosen HQS fields, the spectral index could be calculated as described above. But due to different quality of our objective prism plates, we had to correct the densities first. If the identification is based upon photographic plates of poor quality, the AGN-density is underestimated in comparison to a deeper prism plate.
Regarding on this, we had to determine flux limits for the photographic plates first. Working on the assumption, that direct plates of the HQS are usually one or two magnitudes deeper than spectral plates (Hagen et al. 1995) we determined the plate limit of the objective prism plates from comparison to the corresponding direct plate. We fixed the flux limit at the value, where half of the objects from the direct plate can be found on the prism plate.
For 336 HQS fields, where identification of RASS-BSC objects has been made up to now, we were able to calibrate only 70 fields, because not enough standard stars were available for calibrating all fields. The flux limits of the photographic plates are now used to correct the measured AGN-densities (Beckmann 1996).

5. Spectral slope of RASS-AGN

Using 3305 AGN-candidates and their countrates, we could not find a steep course for the log N - log S relation as expected due to evolution effects (Hewett & Foltz 1994). Plotting log N(>countrate) against log (countrate) the gradient of this curve is identical to the gradient of log N - log S relation. Using linear regression taking the errors in both values into account, we obtain log N / log S = -1.39 +- 0.07.
The corrected AGN densities by sky area have been plotted against galactic hydrogen column densities (Stark et al. 1992). Afterwards we determined the photon index which reproduced this distribution best.
At low absorption NH = 0.5-5 * 10^20 1/cm² we obtain a mean value of the photon index of Gamma = 2.53 +- 0.42 and at high hydrogen column density NH = 1-10 * 10^20 1/cm² Gamma = 2.18 +- 0.32.
The dependence of the spectral slope to the absorption was expected, because objects at high column densities have to have a flatter spectra to be detectible, meaning they have enough flux in the hard X-ray band to surmount the absorption.
We have two explanations for different spectral indices measured at high and low hydrogen column density:
- different AGN have different Gamma, but the spectral slope of each can be described accurately by a simple power law plus absorption. At high absorption only those AGN with flat spectra are detectible, at low absorption we measure both steep spectra as well as flat spectra, providing a mean value. One would expect even many AGN with steeper spectra Gamma > 2.8 in this picture.
- AGN have concave spectra, being described not only by one photon index. In that case we would measure hard tail of spectra at high absorption and mean value of both, steep and flat tail, at the low end of absorption. The steep part of the spectra then would be steeper than the measured Gamma = 2.53, but at higher energies and high column densities the value of Gamma = 2.18 would represent the real spectral slope, if there is not another significant flattening towards the harder X-ray band E > 2.4 keV.
We support the latter model, which is in correspondence to other investigations of this topic (e.g. Schartel et al. 1996, Walter & Fink 1993). Applying the model of concave spectra, we obtain basically information about the spectral slope of AGN: The mean value for the spectral photon index of all AGN types is Gamma = 2.2 at E = 2 keV and Gamma > 2.5 at the soft end of the ROSAT energy band. One should expect that generally radio quiet AGN will have steeper, radio loud will have flatter spectral slopes than the values given here. If there is any intrinsic absorption, and based upon various investigations we do not think this significant, the spectral slope would be steeper when computed with the method used here.


Bade, N., Engels, D., Voges, W., Beckmann, V., et al. 1998, A&AS, 127, 145
Beckmann, V. 1996, Diploma thesis, Hamburg University
Hagen, H.-J., Groote, D., Engels, D., Reimers, D. 1995, A&AS, 111, 195
Hewett, P. C., & Foltz, C. B. 1994, PASP, 106, 113
Maccacaro, T., Gioia, I. M., Wolter, A., et al. 1988, ApJ, 326, 680
Schartel, N., Green, P.J., Anderson, S. F., Hewett, P. C. 1996, MNRAS, 283, 1015
Stark, A. A., Gammie, C. F., Wilson, R. W., et al. 1992, ApJS, 79, 77
Tananbaum, A., Avni, Y., Branduardi, G., Elvis, M., et al. 1979, AJ, 234, L9
Voges, W., Boller, Th., Dennerl, K., et al. 1996, MPE-Report 263, 637
Voges, W., Aschenbach, B., Boller, Th., et al. 1999, A&AS, 349, 389
Walter, R., & Fink, H. H. 1993, A&A, 274, 105

Last change: March 2004 by