Host Galaxy Morphologies in the GOODS-South HST ACS Images B. D. Simmons, C. M. Urry (Yale U.), S. Ravindranath (STScI), F. Bauer (Penn State U.), C. Conselice (Caltech), E. Chatzichristou (Yale U.), N.A. Grogin (JHU), A. M. Koekemoer, E. Schreier (STScI), and the GOODS Team There is growing evidence that the formation of galaxies and the growth of central black holes must be closely related. This “grand unification” of AGN and galaxies can be tested through studies of the AGN host galaxies. Using the deep multi-color HST ACS imaging data, we describe the morphologies of AGN host galaxies in the GOODS (Great Observatories Origins Deep Survey) southern field, which coincides with the Chandra Deep Field South. We compare these to the morphologies of normal galaxies at the same (photometric) redshifts, also taken from the GOODS ACS data. This work is supported by NASA. Results Introduction Of the 346 X-ray sources in the CDF-S catalog, 215 were matched to within 100 pixels (5’) of an optical source from the ACS catalog. 146 sources remained after imposing restrictions such as a magnitude cut of mi < 25.0 and stellarity index (from Sextractor) less than 0.8. Of these, 68 sources were well-fit by GALFIT by the SGxy case, with detected Gaussian point sources. These sources have redshifts ranging from 0.05 < z < 3.0 and Sérsic indices from 0.57 = n = 7.55. GOODS is a multi-wavelength survey aimed at uniting deep observations from space (using HST, SIRTF, Chandra, and XMM-Newton) with the most powerful ground-based observations, to survey the distant universe to the faintest flux limits across the broadest range of wavelengths. GOODS will survey a total of roughly 320 square arcminutes in two fields centered on the Hubble Deep Field North and the Chandra Deep Field South. The space-based observations will be complemented by ground-based imaging and spectroscopy, including an extensive commitment of ESO and NOAO observing time. One of the major goals of GOODS is to study Active Galactic Nuclei (AGN), including a possible population of heavilyobscured AGN at z ~ 2. Figure 1 shows LX vs. LOPT for each of the 68 well-fitted sources. This figure serves as a sanity check on our fitting method; as expected, higher LX values correlate with higher LOPT values. Also, higher LOPT sources are seen at higher redshifts. The present paper describes the morphologies of AGN host galaxies in the southern GOODS field, based on deep optical imaging with the Hubble Space Telescope Advanced Camera for Surveys and deep X-ray observations with the Chandra X-Ray Observatory. Figure 1. Method Figure 2. Figure 1: as expected, optical and Xray luminosity are correlated. Downward arrows indicate that the position shown is an upper limit to the X-ray flux. Spectra of X-ray sources in the CDF-South are not yet publicly available, so we do not have firm AGN identifications; instead, we consider all X-ray sources detected with Chandra. After matching X-ray sources to galaxies in the ACS catalog using a simple algorithm with no astrometric corrections, we use the two-dimensional fitting algorithm GALFIT (Peng et al. 2002, details described in poster by Ravindranath et al.) to fit Sérsic profiles and point sources to the host galaxy. We then use photometric redshifts to determine the luminosity of the point source (if one was detected). We then divide the sample into objects with luminosities characteristic of AGN (LOPT > 3 x 1042 erg/s) and those with smaller luminosities, as well as those with hardness ratios greater than and less than zero, to look for systematic differences. We also examine the dataset for systematic changes in morphology with redshift, spectral type (from the SED used to determine the photometric redshifts), hardness ratio, and LX/LOPT. Figure 2: histograms of Sérsic index for each spectral type. Due to the small total number of sources in each curve, it is difficult to determine whether there are any systematic differences in n between spectral types. Figure 3: Sérsic index vs. Xray hardness ratio. No correlation between n and h is found. Hardness ratios of +1 and -1 correspond to sources which were not detected in the soft and hard bands, respectively. Note that most of the sources were best fit with n > 2; dominating disks are relatively rare in this sample. Photometric Redshifts The determination of photometric redshifts is discussed in Mobasher et al. (2003, in preparation) and will not be discussed in detail for this paper. A multi-wavelength catalog was created by matching ACS objects with objects in the R-band selected FORS1 catalog. Photometric redshifts were determined by fitting a template Spectral Energy Distribution (SED) to each galaxy. Galaxies were divided into five types based on the best-fit SED used for photometric redshift determination: E, Sbc, Scd, Im, and starburst (SB). It should be noted that only a small fraction of the photometric redshift catalog is considered “comfortably” reliable (i.e., greater than 97% confidence; Mobasher 2002, private communication). Luminosity Determination Figure 3. Using the photometric redshifts, the luminosity distance DL to each source was determined assuming ΩM = 0.3, ΩΛ = 0.7, H0 = 65 km/s/Mpc. The fitted magnitudes of each Gaussian point source was converted to a flux using ACS calibration of AB magnitudes. With the use of K-corrections for the i-band for each spectral type (Fukugita et al. 1995, Annis 2000), the optical luminosity of each galaxy’s central point source was determined from the flux (in cgs). X-ray fluxes are available from the CDF-S catalog, and these were converted to X-ray luminosities assuming a spectral index of 1 (no K-correction). Figure 2 shows superimposed histograms of Sérsic index for each spectral type. The spectral type is determined from the best-fit SED template used to find the photometric redshifts. From these it is difficult to determine whether a systematic trend with spectral type exists due to the small number of samples. It is important to note that for these fits, spectral type does not necessarily agree with morphology as determined by the Sérsic index (i.e., spectral types Scd do not always have Sérsic profiles best fit by a disk), although in some cases this could be due to large uncertainties in the Sérsic index. Figure 3 shows Sérsic index n vs. hardness ratio from the CDF-S X-ray catalog. There does not appear to be any systematic change in morphology with hardness ratio, though the uncertainties are fairly large. Figure 4 shows Sérsic index vs. z for hard and soft X-ray sources. Figure 5 displays Sérsic index vs. z for objects with AGN luminosities, LOPT = 3 x 1042 erg/s, and for objects with luminosities LOPT < 3 x 1042 erg/s. Note that objects in the latter category could be obscured AGN. Given the mean error in Sérsic index and redshift, it is not possible to tell at this time whether there are systematic differences between hard and soft sources, or sources with luminosities above and below LOPT ˜ 3 x 1042 erg/s. It could be true that objects with LOPT < 3 x 1042 erg/s, higher redshift objects tend to have values of n around 4, but this could also be a small-number effect. Figure 6 shows CAS parameters 1/concentration vs. asymmetry for X-ray matches in the GOODS CDF-S field. Blue points have SEDs which fit late-type galaxies, and green points have SEDs which fit early-type galaxies. The brown starred points are X-ray sources. Figure 7 shows the ratio LX/LOPT for both hard and soft X-ray luminosities. It seems possible from this figure that objects with a higher LX/LOPT tend to have Sérsic indices near n = 4 (a deVaucouleur profile), but again the small number of sources and the relatively large uncertainty in the Sérsic index makes this conclusion soft at best. Figure 6: CAS parameters 1/concentration vs. asymmetry for X-ray matches in the GOODS CDF-S field. Blue points have SEDs which fit late-type galaxies, and green points have SEDs which fit early-type galaxies. The brown starred points are X-ray sources. Morphological Fitting We determine initial guess parameters for GALFIT from ACS the i-band Sextractor catalog. We then fit the following profiles to each galaxy: (1) deVaucouleur only [dV], (2) deVaucouleur + Gaussian (point source approximation) [dVG], (3) Sérsic + Gaussian [SG], and (4) Sérsic + Gaussian + Exponential Disk [SGE]. In the case of dVG, SG, and SGE fits, we also performed independent fits allowing the (x, y) position of the Gaussian point source to vary. Further additional fits in which all components were allowed to vary in (x, y) were performed for the SG and SGE fits. Figure 6. Figure 4. Figures 4 and 5: Sérsic index vs. redshift for hard/soft sources and sources with luminosities greater or less than LOPT ˜ 3 x 1042 erg/s. Average uncertainties for n and z are shown with the triangular point in the upper-right of each panel. We found that the best fit was generally determined with the Sérsic + Gaussian fits in which only the Gaussian point source was allowed to vary its (x, y) coordinates. Our results which follow are for this fit (SGxy). Figure 3. In the nearby universe, these three parameters correlate with the scale of a galaxy, such as mass and bulge to disk ratio (concentration), star formation (clumpiness), and major galaxy mergers (asymmetry) (Conselice 2003, submitted). The idea behind the system is that understanding past and present star formation and merging activity, the evolutionary history of the galaxy population can be traced. Conclusions 1. As best shown by figure 3, most of the sources well-fit by GALFIT are bulge-type systems with n > 2. This does not mean that there are no disks in these systems, but disks do not dominate most of the galaxies we have fit. This is supported by the fact that relatively few of the sources were well-fit by the SGE fits. 2. The optical morphology of the X-ray matches varies, with n ranging between 0.57 = n = 7.55. 3. We detect no trend in Sérsic index n with redshift z. 4. We do not detect any systematic difference in morphology detected with LX or LX/LOPT. In the latter case, those sources with high LX/LOPT tend to have n near the deVaucouleur value of n = 4, though the sample is too small to determine this with any certainty. When spectroscopic redshifts are available, we will revisit the question of systematic differences in morphology with X-ray and optical properties as well as redshift. CAS Parameters In the CAS (concentration, asymmetry, clumpiness) morphology system all nearby major galaxy types fall in well defined corners of a volume created out to these parameters. Early types are those with a high light concentration, low asymmetry and low clumpiness values. Early type spirals have a lower light concentration, higher asymmetries and high clumpiness values. Later type disks on average have even lower concentrations and slightly higher asymmetries and clumpiness values. Mergers are galaxies which have high asymmetry values. Figure 5. 5. Future work includes a detailed comparison between AGN host galaxies and other field galaxies (see poster by Ravindranath et al.). Also, given the relatively large number of galaxies that were not well-fit by any of the fitting profiles, we will explore the possibility of manually fitting each galaxy in the sample of X-ray matches, to reduce the morphological fitting errors. Figure 7: Sérsic index vs. LX/LOPT for soft and hard X-ray luminosity. Average uncertainty in n is shown in the top right of each panel. Sources with higher LX/LOPT seem to preferentially inhabit the area near n ˜ 4, though this will be clarified in future studies. Figure 7. References Annis, J. 2000. “The SDSS Filter Set K-Corrections,” http://home.fnal.gov/~annis/astrophys/kcorr/kcorr.html Fukugita, M., Shimasaku, K., Ichikawa, T. 1995, PASP, 107, 945. Lucas, R.A., et al. 2003, this meeting. Mobasher, B. et al., 2003 (in preparation) Peng, C. Y., Ho, L. C., Impey, C. D., Rix, H. 2002, AJ, 124, 266. Ravindranath, S., et al. 2003, this meeting.