The Inner Workings of Early-Type Galaxies: Cores, Nuclei and Supermassive Black Holes (Including a Critical Comparison of Nuker and core-Sersic/Sersic models)

Abstract

Recent years have seen dramatic progress in the study of the core and nuclear properties of galaxies. The structure of the cores has been shown to vary methodically with global and nuclear properties, as cores respond to the mechanisms by which galaxies form/evolve. The dynamical centers of galaxies have been found capable of hosting two seemingly disparate objects: supermassive black holes (SBHs) and compact stellar nuclei. In a drastic departure from previous beliefs, it has been discovered that both structures are common: galaxies lacking SBHs and/or stellar nuclei are the exception, rather than the norm. This review explores the connection between cores, SBHs and stellar nuclei in early-type galaxies, as revealed by the ACS Virgo Cluster Survey. We find that, from parsec to kiloparsec scales, the surface brightness profiles of early-type galaxies are well described by a Sersic model -- a parametrization that is shown to be preferable to the so called `Nuker' model. Relative to the inward extrapolation of the best fitting Sersic model, the profiles within the inner few tens of parsecs transition from luminosity `deficits', in the brightest (M_B < ~-20) galaxies, to luminosity `excesses' in the faintest. The compact stellar nuclei responsible for such luminosity excesses comprise approximatively 0.2% of the total galactic mass. Remarkably, this is the same fraction believed to reside in the SBHs, which are predominantly found in the brightest galaxies showing luminosity deficits.

Full text

Page 1

arXiv:astro-ph/0612139v1 6 Dec 2006
Black Holes: from Stars to Galaxies – across the Range of Masses
Proceedings IAU Symposium No. 238, 2006
V. Karas & G. Matt, eds.
c ? 2006 International Astronomical Union
DOI: 00.0000/X000000000000000X
The Inner Workings of Early-Type Galaxies:
Cores, Nuclei and Supermassive Black Holes
Laura Ferrarese1and Patrick Cˆ ot´ e1
1National Research Council of Canada, Herzberg Institute of Astrophysics, 5071 West Saanich
Road, Victoria, BC, V9E 2E7, Canada
email: laura.ferrarese@nrc-cnrc.gc.ca; patrick.cote@nrc-cnrc.gc.ca
Abstract. Recent years have seen dramatic progress in the study of the core and nuclear
properties of galaxies. The structure of the cores has been shown to vary methodically with
global and nuclear properties, as cores respond to the mechanisms by which galaxies form/evolve.
The dynamical centers of galaxies have been found capable of hosting two seemingly disparate
objects: supermassive black holes (SBHs) and compact stellar nuclei. In a drastic departure
from previous beliefs, it has been discovered that both structures are common: galaxies lacking
SBHs and/or stellar nuclei are the exception, rather than the norm. This review explores the
connection between cores, SBHs and stellar nuclei in early-type galaxies, as revealed by the ACS
Virgo Cluster Survey.
Keywords. galaxies: elliptical and lenticular, cD, galaxies: dwarf, galaxies: fundamental pa-
rameters, galaxies: kinematics and dynamics, galaxies: photometry, galaxies: structure, galaxies:
nuclei, galaxies: bulges
1. Introduction
Cores, a term we will use loosely to describe the central few hundred parsec region of
a galaxy, represent an integral part in our understanding of the global galactic structure,
for very good reasons. Cores act as recording devices of a galaxy history. Dynamical
timescales are shorter here than elsewhere in the galaxy; the morphology, dynamics and
history of star formation and chemical enrichment of the cores are a sensitive tracer of the
gas, dust and dense stellar systems, either intrinsic or accreted through merging events,
that are drawn to the bottom of the potential well throughout cosmic times. Furthermore,
core and global properties are linked through a number of scaling relations. In particular,
those involving supermassive black holes (SBHs) – which are almost always associated
with galactic cores – underscore the importance of nuclear feedback in galaxy evolution
(Ferrarese & Merritt 2000; Gebhardt et al. 2000; Graham et al. 2001; Ferrarese 2002;
Haring & Rix 2004).
The study of galactic cores received a tremendous push forward with the deployment
of the Hubble Space Telescope. HST images brought into focus a plethora of structural
features, including nuclear stellar disks, bars, “evacuated” regions (possibly scoured out
by the evolution of SBH binaries), and an entire spectrum of dust features - from small
irregular patches to large, organized dust disks. In early-type galaxies, cores were found
to fall in two distinct classes: those with a shallow surface brightness profile, and those
whose surface brightness kept increasing, in roughly a power-law fashion, to the inner-
most radius accessible given the resolution limit of the instrument (Ferrarese et al. 1994;
Lauer et al. 1995,2005; Ravindranath et al. 1996; Rest et al. 2001). Galaxies falling into
the first class have become known (somewhat unfortunately) as “core” galaxies, galax-
ies falling into the second class as “power-law”. The division between the two classes
1

Page 2

2 L. Ferrarese & P. Cˆ ot´ e
was found to correlate neatly with galaxy luminosity, with core galaxies being exclu-
sively bright giant ellipticals, while fainter galaxies are (with few exceptions) classified
as power-laws. The stark separation between the two classes has been attributed to dif-
fering formation/evolutionary histories. Power-law galaxies have been claimed to be the
result of dissipation during galaxy formation, with some authors further claiming that
all power-law galaxies host stellar disks; while core-galaxies are believed to be the result
of the merging of fainter (power-law) galaxies, and of their central SBHs.
Dynamical detections of SBHs exist in approximatively three dozen galaxies (see Fer-
rarese & Ford 2005 for a review); indeed, balancing the SBH mass function from the QSO
epoch to the present day requires virtually all local galaxies brighter than a few 0.1L∗to
host a SBH (e.g. Shankar et al. 2004; Marconi et al. 2004). Recent observations, however,
have made it clear that SBHs are not the only objects to enjoy a priviledged position at
a galaxy’s dynamical center. Stellar nuclei, or nuclear star clusters, have recently been
detected in a large fraction (70% to 90%) of galaxies of all Hubble types and luminosities
(B¨ oker et al. 2002; Lotz et al. 2004; Grant et al. 2005). Follow-up spectroscopy of stellar
nuclei in spiral galaxies (Walcher et al. 2005,2006; Rossa et al. 2006) has shown them to
be massive, dense objects akin to compact star clusters. Luminosity-weighted ages range
from 10 Myr to 10 Gyr, younger than the age of the galactic disk, and with the younger
clusters found preferentially in the later type spirals.
This review explores the connection between cores, nuclei and supermassive black holes
in light of recent results from the ACS Virgo Cluster Survey (ACSVCS), the largest HST
imaging survey designed specifically to provide an unbiased characterization of the core
structure of early-type galaxies.
2. The ACS Virgo Cluster Survey
The ACSVCS (Cˆ ot´ e et al. 2004) consists of HST imaging for 100 members of the Virgo
Cluster, supplemented by imaging and spectroscopy from WFPC2, Chandra, Spitzer,
Keck and KPNO. The program galaxies span a range of ≈ 460 in B-band luminosity and
have early-type morphologies: E, S0, dE, dE,N or dS0. All images were taken with the
Advanced Camera for Surveys (ACS; Ford et al. 1998) using a filter combination roughly
equivalent to the g and z bands in the SDSS photometric system. The images cover a ≈
200′′× 200′′field with ≈ 0′′.1 resolution (≈8pc at the distance of Virgo, 16.5 Mpc).
This review summarizes results from the subset of ACSVCS papers which deal with the
morphology, isophotal parameters and surface brightness profiles for early-type galaxies
(Ferrarese et al. 2006a), their central nuclei (Cˆ ot´ e et al. 2006) and scaling relations for
nuclei and SBHs (Ferrarese et al. 2006b). Other ACSVCS papers have discussed the data
reduction pipeline (Jord´ an et al. 2004a), the connection between low-mass X-ray binaries
and globular clusters (Jord´ an et al. 2004b), the measurement and calibration of surface
brightness fluctuation distances (Mei et al. 2005ab), the connection between globular
clusters and ultra-compact dwarf galaxies (Ha¸ segan et al. 2005), the luminosity function,
color distributions and half-light radii of globular clusters (Jord´ an et al. 2006ab; Peng
et al. 2006a), and diffuse star clusters (Peng et al. 2006b).
3. The Core Structure of Early-Type Galaxies
Over the three-decade radial range between a few tens of parsecs and several kiloparsecs
(i.e. to the largest radii covered by the ACSVCS images), the surface brightness profiles
of the ACSVCS early-type galaxies are well described by a simple S´ ersic model (S´ ersic
1968) with index n increasing steadily with galaxy luminosity. Notable, and systematic,

Page 3

Cores, Nuclei and Supermassive Black Holes3
deviations from a S´ ersic model are however registered in the innermost regions. For eight
of the 10 brightest galaxies (MB? −20.3) the measured inner profiles (typically within
0′′.5 to 2′′.5, corresponding to 40 to 200pc) are shallower than expected based on an inward
extrapolation of the S´ ersic model constrained by the region beyond. For these galaxies,
the surface brightness profile is best fitted by joining the outer S´ ersic profile to an inner,
shallower, power-law component (such composite models are referred to as “core-S´ ersic”
Graham et al. 2003; Trujillo et al. 2004),
The opposite is seen in fainter galaxies, ≈ 80% of which show a clear upturn, or
inflection, in the surface brightness profile within (typically) the innermost few tens
of parsec region (see Figure 1 of P. Cˆ ot´ e, these proceedings). The upturn signals the
presence a stellar nucleus that is most likely structurally distinct from the main body
of the underlying galaxy. When a nucleus is present, the inner surface brightness is, by
definition, larger than the inward extrapolation of the outer S´ ersic model.
The picture that has emerged from the ACSVCS is therefore one in which, in moving
down the luminosity function from giant to dwarf early-type galaxies, the innermost 100-
parsec region undergoes a systematic and smooth transition from light (mass) “deficit”
(relative to the overall best fitting S´ ersic model) to light “excess”. Although the subset of
ACSVCS “core-S´ ersic” galaxies coincides with the galaxies that were classified as “cores”
in previous investigations, there are critical differences between our study and the ones
that preceded it. Compared to previous work, the ACSVCS has emphasized the role of
stellar nuclei; the fact that the frequency, luminosities and sizes of the ACSVCS nuclei
are in remarkable agreement with those measured (using different techniques and as-
sumptions) by recent independent surveys in both early and late type galaxies, supports
the robustness of the ACSVCS analysis. Recognizing the nuclei as separate components
has allowed us to revisit the issue of the division of early-type galaxies into “core” and
“power-law” types. Such division was based on the fact that the distribution of the loga-
rithmic slopes, γ = −dlogI/dlogr, of the inner surface brightness profile had been found
to show various degrees of bimodality. Such bimodality is absent in the γ distribution
of the ACSVCS galaxies. In agreement with previous studies, in galaxies brighter than
MB≈ −20.3, γ is indeed found to decrease with galaxy luminosity, while the opposite
trend is seen for fainter galaxies, however, the transition is smooth, rather than abrupt. In
a further departure from previous studies, we find that the low-γ end of the distribution
(corresponding to the galaxies with the shallowest surface brightness profiles) is occupied
mostly by the faintest dwarf stystems, rather than by the brightest giant ellipticals. We
note here that the absence of a bimodal behaviour in γ does not automatically invalidate
a picture in which brighter galaxies evolve mainly through merging while fainter systems
are largely left untouched. Indeed, such picture does not necessarily explain the perceived
stark separation of galaxies in “core” and “power-law” types for which it was formulated.
The extent to which structural parameters are compromised by merging of galaxies (and
their supermassive black holes) depends on the the masses of the progenitors (e.g., Bour-
naud et al. 2005; Milosavljevic & Merritt 2001); given a continuous distribution for the
latter, combined with a galaxy luminosity function heavily biased towards low-mass sys-
tems, allows for the possibility of a smooth transition between progenitors and merger
products.
4. Compact Stellar Nuclei in the ACSVCS
At the outset of the ACSVCS, it was known that at least ≈25% of the program galaxies
contained nuclei, based on ground-based classifications from the VCC (Binggeli et al.
1985). Stellar nuclei in the ACSVCS images were identified by a variety of indicators,

Page 4

4 L. Ferrarese & P. Cˆ ot´ e
including direct inspection of the ACS frames, color changes in the g − z color images,
and sudden upturns in the surface brightness profiles. Based on these criteria, 60 to 80%
of ACSVCS galaxies host stellar nuclei (with the precise fraction depending on galaxy
magnitude), in line with the fraction reported in both spiral and elliptical galaxies based
on recent high-resolution surveys (Carollo, Stiavelli & Mack 1998; Matthews et al 1999;
B¨ oker et al. 2002, 2004; Balcells et al. 2003; Lotz et al. 2004; Graham & Guzman 2003;
Grant et al. 2005), but a factor ∼ 3 higher than expected based on the VCC.
Our analysis shows that surface brightness selection biases in the VCC data are largely
responsible for the difference: in galaxies with central g-band surface brightnesses lower
than ≈ 20.5 mag arcsec−2, the agreement between the ACSVCS and VCC is nearly
perfect, while above 19.5 mag arcsec−2, virtually all nuclei were missed by the ground-
based survey. Selection effects might, of course, still be at work in the ACSVCS sample,
implying that our estimate for the frequency of nucleation, fn ≈ 60 − 80%, is almost
certainly a lower limit on the true frequency. As will be discussed shortly, the luminosity
and half-light radii of stellar nuclei correlate strongly with the magnitude of the host
galaxy; it is therefore possible, for each galaxy classified as non-nucleated, to determine
whether a nucleus, if present, could have gone undetected. Based on these tests, with very
few exceptions, the only galaxies for which the existence of a nucleus can be confidently
excluded are those brighter than MB ≈ −20.3 mag. These are the same galaxies with
central light “deficits” for which the surface brightness profiles are well represented by
“core-S´ ersic” rather than S´ ersic models (§3)
4.1. Scaling Relations for Stellar Nuclei
For 51 galaxies in the ACSVCS the sharp upturn in the surface brightness within ≈ 1′′
is conspicuous enough that a measurement of the nucleus’ photometric and structural
parameters is possible. These parameters are recovered by adding a central King model
(King 1966) to the underlying S´ ersic component when fitting the surface brightness
profile.
The luminosity function of nuclei follows a Gaussian distribution with dispersion in the
range 1.5−1.8 mag and peak absolute g−band magnitude ≈ −10.7 mag, a factor ≈ 25×
brighter than the peak of the globular cluster luminosity function. With a half-dozen
exceptions, nuclei in the ACSVCS galaxies are clearly spatially resolved (thereby ruling
out an AGN origin), with individual sizes ranging from 62 pc down to the resolution
limit of 2 pc, and a median half-light radius of ?rh? = 4.2 pc. Unlike globular clusters,
for which size is largely independent of magnitude, nuclear sizes are found to scale with
luminosity according to the relation rh∝ L0.50±0.03(Figure 1, left panel).
One of the most credited models posits that the formation of nuclei proceeds trough the
coalescence of globular clusters drawn to the bottom of the potential well by dynamical
friction (e.g. Tremaine et al. 1975). While the size-magnitude relation observed for the
ACSVCS nuclei is consistent with the prediction of such model (Bekki et al. 2004), a more
complex picture is put forth by the observations that nuclei, again unlike globular clusters,
display a color-magnitude relation (Figure 1, right panel). Monte Carlo simulations show
that mergers of globular clusters through dynamical friction are unable to explain the
observed color-magnitude relation; indeed the existence of this relation suggests that the
chemical enrichment of nuclei is governed by local or internal factors, along the lines of
the various “gas accretion” models (e.g. Mihos & Hernquist 1996). Note that the nuclei’s
color-magnitude relation is better defined for galaxies fainter than MB ≈ −17.6 mag,
while the nuclei belonging to brighter galaxies frequently show very red colors, (g −z) ∼
1.5. If confirmed (measurements are more uncertain for these nuclei, due to the high

Page 5

Cores, Nuclei and Supermassive Black Holes5
Figure 1. (Left Panel) The size-magnitude relation, in the g−band, for the 51 ACSVCS nuclei
for which structural parameters could be measured (solid circles) and the sample of globular
clusters from Jord´ an et al. (2006) (points). Typical errorbars for the nuclear sizes are shown
in the right hand side of the panel. The arrow shows the ”universal” half-light radius of 0′′.033
(≈ 2.7 pc) for globular clusters in Virgo (Jord´ an et al. 2005), while the dashed line shows a
conservative estimate for the resolution limit of the ACS images. The solid diagonal line shows
the best fitting relation for the nuclei (rh ∝ L0.5), while the dotted line shows the prediction
of the globular cluster merging model of Bekki et al. (2004). (Right Panel) Color-magnitude
diagram for the ACSVCS nuclei.The size of the symbol for the nuclei is proportional to the
magnitude of the host galaxy, shown in the legend. The dashed line shows the the best fit
relation for the nuclei of galaxies fainter than BT = 13.5 mag.
underlying galaxy surface brightness), this observation might suggest that these nuclei
may constitute a separate type of objects following a different formation route.
A third model, namely nuclear formation through two-body relaxation around a black
hole, is inconsistent with the observation that nuclei are spatially resolved in most of
the ACSVCS galaxies. Nuclei formed through this mechanism are predicted to extend to
approximately 1/5 of the SBH sphere of influence (e.g. Merritt & Szell 2005), and would
therefore be spatially unresolved by the ACS in all of the ACSVCS galaxies, clearly
contradicting our observations.
Finally, we note that the luminosity function and size distribution of the ACSVCS
nuclei shows remarkable agreement with those of the “nuclear star clusters” detected in
spiral galaxies (B¨ oker et al. 2002,2004). This points to a formation mechanism for the
nuclei that is largely independent on both intrinsic and extrinsic factors, such as host
magnitude and Hubble type, and immediate environment.