AGN Continua and SEDs

About 25 years ago Baldwin (1977) discovered an anti-correlation between the equivalent width in CIV1549, W(CIV), and the continuum luminosity, L(1450), measured at 1450 A (``Baldwin Effect'', hereafter BEff). The BEff was subsequently confirmed in CIV1549, as well as other BELs such as Lya and OVI1034 (for a recent review, see Osmer & Shields 1999). The BEff is an important diagnostic tool to study the AGN structure and, perhaps, metal abundances (Korista et al.1998). The relation between the continuum luminosity and the relative emission line strengths and ratios can be used to study the evolution and physics of the quasar phenomenon (Baldwin 1999). In particular, this correlation can be used to test model predictions for the dependence of the shape of the continuum spectral energy distribution as a function of luminosity. It was suggested by Mushotzky & Ferland (1984) that the BEff can be explained by an anti-correlation of the ionization parameter, U, and the continuum luminosity, assuming that in addition, the covering factor is decreasing with increasing continuum strength. The BEff might be caused by a fundamental correlation between the continuum luminosity, L_c, and the shape of the ionizing EUV - soft X-ray continuum (Binette et al.1989; Zheng & Malkan 1993; Zheng et al.1997; Korista et al.1998). Netzer (1985,1987) and Netzer et al.(1992) suggested accretion disk models as an attractive scenario to explain the observed continuum and emission line correlations. Wandel (1999a,b) included the aspect of the black hole mass evolution by accretion in the accretion disk model. These models are attractive because the UV - X-ray spectral softening has been well documented by observations (Tananbaum et al.1986; Wilkes et al.1994; Green et al.1995).

We have compiled a large sample of rest-frame optical and ultraviolet spectra for type 1 AGNs. A majority of the spectra were obtained by several groups for different studies over the last 20 years using ground-based instruments as well as IUE and HST (Bahcall et al.1993; Baldwin et al.1989; Chaffee et al.1991; Corbin & Boroson 1996; Kinney et al.1991; Lanzetta et al.1993; Laor et al.1995; Sargent et al.1988,1989; Schneider at al.1991a,b; Steidel 1990; Steidel & Sargent 1991; Steidel (priv.comm.); Storrie-Lombardi et al.1996; Weymann et al.1991,1998; Wills et al.1995; Zheng et al.1997). Furthermore, we observed a large number of the quasars in our sample at redshifts z>3 (Constantin et al.2002; Dietrich et al.1999,2002a,2003a,b) Dietrich & Wilhelm-Erkens 2000). For this study we exclude Broad-Absorption Line quasars (BALQSOs). The sample we investigate consists of 744 type 1 AGNs.

Fig.1 - Redshift distribution of the AGN sample as a function of intrinsic luminosity at 1450 A. The open symbols represent radio-loud quasars and the filled symbols radio-quiet quasars. The dashed horizontal (vertical) lines indicate the luminosity (redshift) ranges which were used to calculate composite spectra. At the left side (top) of the figure is the number of the individual spectra contributing to each composite spectrum. For the luminosity range 43 < log L(1450) < 44 composite spectra were calculated for the redshift intervals marked by the vertical dashed lines.

We computed composite spectra in different intervals of luminosity and redshift as indicated in the Figure above. The normalized spectra show very obviously a strong relation between the equivalent width, W, and the continuum luminosity represented by L(1450), i.e., the Baldwin Effect, for most of the emission lines (Figure below). The BEff can be seen easily in strong lines such as CIV1549, OVI1034, Lya1216, and CIII]1909. For the first time, we report about the BEff for NIV]1486, OIII]1663, and NIII\1750. In spite of the large scatter, the broad Fe emission features in the ultraviolet show some indications of a BEff. Close inspection of the NV1240 emission line indicates that this line does not exhibit a BEff like other high ionization lines. The careful deblending of the Lya - NV1240 emission line complex reveals that NV1240 remains nearly constant in W over ~6 orders of magnitude in continuum luminosity.

Fig.2 - Normalized composite spectra are shown for the luminosity bins (Delta log L(1450) = 0.5 dex) as displayed in Fig.1. The horizontal dashed lines indicate the continuum-level for the individual normalized composite spectra which are vertically shifted for better display.

We selected a narrow luminosity range to search for a dependence of emission line properties with cosmic time. To cover a redshift range as large as possible for an almost constant luminosity, we selected a luminosity range of 43 < log L(1450) < 44 (Fig.1). Figure 3 shows the composite spectra for each redshift bin. These composite spectra are normalized by the power law continuum which we derived from the multi-component fits. The equivalent widths of the lines remain constant within ~30 %. The stronger emission lines, however, like Lya, CIV1549, and CIII]1909, show marginal trends for a relation of W and redshift (Dietrich et al.2002). W stays nearly constant within less than ~20 % for redshifts less than z = 2 to 3, but then for z > 3 there are marginal indications for a slight increase of W at constant luminosity.

Notice, that the slope of the BEff becomes steeper for increasing chi_ion (Zheng et al.1995; Espey et al.1995; Espey & Andreadis 1999; Dietrich et al.2002). In Figure below the slopes beta of the log W vs. log L(1450) relations are plotted as a function of chi_ion for the entire luminosity range (filled symbols) and for log L(1450) > 41 (open symbols) to show the relationship of the BEff slopes beta to chi_ion.

Fig.3 - The slope beta of the Baldwin Effect as a function of the ionization energy chi_ion needed to create the specific ions. The filled symbols represent the slopes based on the entire luminosity range. The slopes of the BEff for higher luminosities (log L(1450) > 41) are plotted as open symbols.

The Baldwin Effect, its steepening towards higher luminosities, and the correlation of the slope beta with chi_ion can all be well explained in the context of a luminosity dependent spectral energy distribution of the ionizing continuum. Assuming that the SED can be described as a combination of a powerlaw continuum and a thermal UV bump, the ionizing continuum becomes softer for increasing luminosity as the UV bump is shifted to longer wavelengths. This behaviour can be explained with accretion disk models as suggested by Netzer et al.(1992) and Wandel (1999a,b). The lack of a BEff for the high-ionization feature NV1240 suggests that the chemical composition of the gas is an additional parameter that can strongly influence the equivalent width of this and possibly other lines.

Publications

This page is under construction. Last update: April 30, 2005.

This page created by Matthias Dietrich
Questions or comments should be sent to: dietrich@astronomy.ohio-state.edu

	Return to Astronomy Department Home Page
	For comments and suggestions contact the department at webmaster@astronomy.ohio-state.edu About this site (department disclaimer, logo and recommended web browser)