3.3.6. cD Galaxies. II. The Extended Envelope

The halos of cD galaxies are enormously larger than those of ordinary galaxies, whether tidally distended or not. Figure 15 shows the brightness profile of a typical example, NGC 6166, whose large size compared to M31 has been dramatized in Figure 1 of Morgan and Lesh (1965). If NGC 6166 were substituted for the Galaxy, it would reach at least 70% of the way to M31 (distance = 0.65 Mpc, de Vaucouleurs 1978). This is by no means the largest cD galaxy known; the cD in A1413 has a limiting radius of 2 Mpc (Oemler 1976). Thus these halos are more nearly cluster-size than galaxy-size. Recent work reveals further similarities between cD halo and cluster properties, and suggests that these halos, unlike those of ellipticals, are formed by cluster processes.

Figure 15. Brightness profile of the cD galaxy NGC 6166 (Oemler 1976). The straight line is the adopted r1/4-law fit. This has a smaller range of validity here than in other cDs because of the multiple nuclei, which result in a large "core radius". Also, the cD halo is unusually bright. This halo begins at r1/4 ~ 2.8 (r 60" 56 kpc if H0 = 50 km s-1 Mpc-1). The outermost measured point is at a radius of 460 kpc. For comparison, the major-axis radius of M31 at 25.8 B mag arcsec-2 is 20 kpc (de Vaucouleurs 1958), and the limiting radius of M87 is ~ 150 kpc (Oemler 1976).

Oemler (1973) has shown that the brightness profile of the cD in A2670 is proportional to that of the cluster as a whole, but fainter by a factor of 1.9. If the cD and cluster profiles continue to be parallel beyond the outermost measured point, then the halo contains 35% of the light of the cluster. This is not enough to explain the "missing mass", but it is dynamically significant. Similarly, the diffuse background light in the Coma cluster is distributed like the galaxies, and contributes about one-fourth of the light of the cluster ( 19 ± 4%, Melnick, White and Hoessel 1977; ~ 31%, Thuan and Kormendy 1977).

The amount of halo light is strongly correlated with cluster richness (Oemler 1976; Thuan and Romanishin 1981). As shown in Figure 16, the total luminosities L₁ of the brightest galaxies in rich clusters increase with increasing cluster luminosity as L₁ L_cluster^1.25. If the luminosity of the main part of the galaxy (Section 3.3.4) is subtracted, the halo luminosity L_halo is found to satisfy L_halo L_cluster^2.2. The halo luminosity goes to zero at L_cluster ~ 1.5 - 2.0 × 10¹² L, where M₁ = - 2.5 log L₁ + constant = - 23 (H₀ = 50 km s^-1 Mpc^-1). The poor clusters containing cD-like galaxies (Morgan et al. 1975; Albert et al. 1977) generally have L_cluster 2.5 × 10¹² L, which is approximately the luminosity at which cD halos first appear. And, in fact, none of these "cDs" are found to depart significantly from r^1/4 laws (Thuan and Romanishin 1981). These galaxies appear to be nothing more than very bright ellipticals. Lacking halos, they have a shallower L₁ - L_cluster relation than cDs in rich clusters, but the two relationships are continuous (Fig. 16). Both Oemler (1976) and Thuan and Romanishin (1981) emphasize the intimate connection between halo formation and cluster processes. The data are qualitatively consistent with a picture (e.g., Richstone 1976) in which tidal stripping by galaxy encounters and by the cluster potential (White 1982) produce an intergalactic sea of stars which is identified as the cD halo. To some extent, the elliptical galaxy at the center of the cD is a "lucky bystander" (Oemler 1976) which happens to be at the bottom of the same potential well; the halo really belongs to the cluster. However this elliptical is the ultimate recipient of galaxies which spiral inward due to dynamical friction in the halo (both visible and dark). Victims on their way to being cannibalized leave (both visible and dark) parts of their halos behind, adding to the cD envelope. It is not an accident that the brightest elliptical in the cluster tends to be at the center, because dynamical friction and merging are most rapid for the most massive galaxies. The above is of course an over-simplified sketch of one possible way of making cDs; a detailed review is given by White (1982). Within this picture it is not surprising that poor clusters lack cD halos: there is not enough material present to make them. For example, the brightest halo in Figure 16 has M_V -25.8 or L ~ 10^12.2 L. This is as bright as the whole cluster luminosity at which cD halos first appear. Thuan and Romanishin (1981) also point out that the collision time in poor clusters is longer than a Hubble time.

Figure 16. Correlations with total cluster luminosity Lc or LV, cl of the absolute magnitude of the brightest member. The panel at left is from Oemler (1976), that at right from Thuan and Romanishin (1981). The upper filled circles and plus signs in both panels represent the total luminosity of the first-ranked galaxy. Open circles (left) and the lower filled circles and plus signs (right) are the "reduced magnitudes", Mred = - 2.5 log(I0 a2) + constant, of the part of the galaxy which is described by a Hubble law. These yield total magnitudes for the E-galaxy part (crosses, left-hand panel) via an empirical calibration for ellipticals (Oemler 1976). There is still a dependence on Lc, but it is weaker than for magnitudes which include the halo. Also shown in the left panel (chained dots) is the difference between the total and E-galaxy luminosity, i.e., the halo luminosity. This increases steeply with increasing Lc. The cD halo first appears at a cluster luminosity Lc ~ 1012.2 L; this is approximately the luminosity of the brightest "poor clusters" at right.

Further evidence for or against the above picture can be supplied by measurements of color gradients. Available data are sparse and generally inconclusive. e.g., Gallagher, Faber and Burstein (1980) saw a weak blueward gradient at large radii in NGC 6166, but the measurements (at µ 25 V mag arcsec^-2) barely reach the start of the halo (see Figure 15). Also, Mattila (1977) measured a patch at 27 G mag arcsec^-2 in the background light of the Coma cluster. He found B - V 0.54 ± 0.19, which is bluer than the light of E and S0 galaxies. Since many ellipticals become bluer at larger radii (Strom and Strom 1978a, b, c, 1979b), this observation may indicate that the halo consists of stars stripped from the outer parts of galaxies. However, the errors are large. Basically, color measurements are another tool that awaits conclusive exploitation.

The most compelling evidence that cD halos are stripped stars moving in the cluster potential is provided by measurements of velocity dispersions . An early study (Faber, Burstein and Dressler 1977) derived = 470 ± 250 km s^-1 at 43 kpc radius in the cD in A401. This is equal to the nuclear value of 480 ± 120 km s^-1, and clearly smaller than the cluster dispersion of 1390 km s^-1. However, the errors are large, and compatible with a contribution as large as 70% from a halo having the cluster dispersion. More conclusive results were obtained by Dressler (1979), who measured the dispersion as a function of radius in the cD galaxy in A2029 (Figure 17). The inner halo shows a dramatic rise of velocity dispersion with radius. These measurements are difficult, and need to be confirmed. Nevertheless, they imply that cD halos are dynamically very different from elliptical galaxies, which have constant or decreasing dispersion profiles (section 4.2.5). Dressler was able to fit the brightness profile, the dispersion data, and the cluster's profile of galaxy densities with an indicative model. This consisted of a normal elliptical (M/L = 10), a luminous halo of tidal debris having M/L = 35 and a dark halo (M/L > 500) which binds the cluster. Dressler also discussed other evidence for the origin of cDs through the accumulation of tidal debris. Finally, a similar rise of with radius has recently been seen in IC 2082 by Carter et al. (1981).

Figure 17. Velocity dispersion as a function of radius in the cD galaxy in the cluster A2029 (Dressler 1979). The measurements were made with a SIT spectrograph and the Palomar 5 m telescope. Open and closed symbols refer to points on opposite sides of the center. The solid curve describes a King (1966) model with constant M/L and a core radius of 10 kpc.