Structure and Formation of Elliptical Galaxies
John Kormendy
University of Texas at Austin;
MPE & University Observatory, Munich
Accomplices:
David Fisher
Ralf Bender
Mark Cornell
Summary
•
What I learned about galaxy formation from Bernard (Gräftåvallen, Sweden, 1992)
•
Sérsic-profile Es & astrophysically diagnostic departures from Sérsic functions
•
Brightness profiles of Virgo cluster elliptical galaxies – all of them!
•
The dichotomy between elliptical and spheroidal galaxies
•
The dichotomy into 2 kinds of ellipticals includes
- Sérsic index n
- central cores and extra light
 conclusions about E galaxy formation
- parameters correlations
}
Projected Brightness Profiles of E Galaxies
Are Sérsic (1968) Functions
de Vaucouleurs (1948) r1/4 law:
Sérsic (1968) generalization:
n = 1  exponential (many disks)
n = 4  de Vaucouleurs law
The Sérsic (1968) Function
Caon et al. (1993) showed that residuals
of Sérsic fits are systematically smaller
than residuals of r1/4-law fits.
They suggested that the shape
parameter n is physically meaningful
because it correlates
with effective radius.
Diagnostic Departures From Sérsic Profiles. I. Cores
Near the center, the profiles of many Es break below the inward
extrapolation of the outer Sérsic profile into a nonisothermal “core”
(Kormendy 1977; King 1978; Lauer 1985; Kormendy 1985).
Kormendy 1985
Lauer 1985
We adopt this definition
of cores (cf. recently,
e. g., Trujillo et al. 2004,
AJ, 127, 1917).
HST Era: “Cuspy cores” are shallow power laws.
Nuker function (Kormendy et nuk. 1994; Lauer et nuk. 1995; Byun et nuk. 1996),
,
is a (yet another) empirical, analytic
fitting function with
 = slope of inner power law;
 = slope of outer power law;
 = sharpness of break between them.
This function fits many Es and bulges
over the central 10.
Byun et nuk. (1996)
Core “fundamental plane” correlations
define what it means to be an elliptical galaxy.
globular
clusters
E ≠ Sph ≈ S,Im
Kormendy (1985, 1987)
Origin of Cores = Black Hole Scouring?
Galaxy mergers preserve the highest central density in the progenitors.
But higher-luminosity galaxies have fainter µoV.
Problem: How to prevent mergers from destroying the core FP relations
(Kormendy 1993; Faber et nuk. 1997).
Possible solution: make cores via scouring by supermassive black hole binaries:
- Merge two galaxies that contain black holes
- They form a black hole binary
- The binary orbit decays as the black holes fling stars to larger radii
- Result: depleted mass ~ mass of the binary
(Ebisuzaki et al. 1991, Faber et nuk. 1997, Milosavljević et al. 2002, Merritt 2006)
The E Dichotomy: There are two kinds of elliptical galaxies
(Bender 1988; Bender et al. 1989; Kormendy et nuk. 1994; Kormendy & Bender
1996; Gebhardt et nuk. 1996; Tremblay & Merritt 1996; Faber et nuk. 1997)
Normal and low luminosity Es
– rotate rapidly,
– are nearly isotropic oblate spheroids,
– are substantially flattened (E3.5),
– are coreless
– have disky-distorted isophotes.
Giant ellipticals
– are essentially non-rotating,
– are anisotropic and triaxial,
– are less flattened (E2.5),
– have cuspy cores,
– have boxy-distorted isophotes.
Kormendy & Bender (1996)
Dichotomy: Cores vs. No Cores
NGC 4621 (Lauer)
NGC 720 (Lauer)
Dichotomy: Cores vs. No Cores
Gebhardt et nuk. (1996)


Cores are real breaks
in the volume density profile
(Kormendy 1999).
Disky ellipticals are rotationally flattened, like bulges

DEFIS
NGC 821: disky, a4>0
NGC 2300: boxy, a4<0
• Disky Es are rotationally flattened, like bulges. They continue the
Hubble sequence beyond S0s toward lower disk-to-bulge ratios.
• Boxy Es show dynamical signatures of triaxiality (minor-axis rotation).
Bender 1987, 1988, 1990, Bender et al.1989, Nieto & Bender 1989, Kormendy & Bender 1996
Faber et nuk. (1997);
Kormendy (1999);
also Nieto, Bender,
& Surma (1991)
Kormendy (1987)
M32
Dichotomy:
Cores vs. No Cores
Core galaxies are boxy & slow rotators;
power-law galaxies are disky & fast rotators.
Diagnostic Departures From Sérsic Profiles. II.
“Extra Light” Near the Center
Kormendy (1999) found that some low-luminosity ellipticals have a central
excess of light above the inward extrapolation of a Sérsic fit to the outer profile.
Interpretation:
The excess light closely resembles
the extra central components
made in dissipative merger simulations:
Mihos & Hernquist (1994) :
TREESPH code with
1. stars, gas, DM;
2. ∑SFR  ∑GAS1.5 ;
3. bulge/total = 0.
Dense Central Components Made In Dissipative Mergers
made in starburst
Springel (2000) simulates mergers of disks
+ gas (20 % of the disk mass) that cools
+ supernova feedback tuned to make
star formation ∑SFR  ∑GAS1.4.
Dissipation results in a merger remnant with a
dense central stellar component.
Springel (like Mihos & Hernquist) thought that
this result disagrees with observations. He
suggested that different star formation
physics* subdues the central starburst.
*E. g., AGN energy feedback
Also: Cox et al. (2005), …
Interpretation of Outer Brightness Profiles in the Context of
Hierarchical Clustering and Galaxy Mergers
Schweizer (1981)
Agreement of merger remnants (observed or
simulated) with r1/4 laws has been used as a
diagnostic that mergers make normal ellipticals.
HST ACS
caveats!
Simulation: J. Barnes
Mergers Can Make Sérsic Profiles
Van Albada (1982) simulated dissipationless collapses with lumpy initial
conditions. Collapse+merger violence was measured with 2T/|W|.
T=Kinetic Energy; W=Potential Energy; Equilibrium  2T/|W| = 1.
He tried to make good r1/4 laws but instead made Sérsic profiles.
More violent collapses give larger values of n.
n>4
n<4
n>4
Dense Central Concentrations Made By Dissipative Mergers
Our interpretations of deviations from Sérsic fits:
• Cores are indicative of “dry mergers” – ones with little dissipation and
with binary black scouring after the last major merger.
• “Excess” central light is a signature of “wet mergers” – of
dissipative starbursts during the last major mergers.
Similar conclusions: Faber 1995, NNG2. Further tests: new photometry
Surface Photometry of Virgo Cluster Ellipticals
We measured surface brightness profiles for all elliptical
galaxies in the Virgo cluster. We combined HST data with
new ground-based, wide-field measurements and with
published data for each galaxy.
Sources of Published Profiles:
• Lauer et nuk. 1995
• Peletier et al. 1995
• Caon et al. 1990
• Bender et al. 2005
• de Vaucouleurs (various)
• Davies et al. 1985
• Kormendy (various CFHT)
• Lauer 1985
• Dressler 1987
Sources of Measured Profiles:
• McDonald 0.8 m: 1.4˝ pixel-1 over 46´
• CFHT 12K (Kormendy & Wainscoat): 0.21˝ pixel-1 over 42´ x 28´
• One AO image from CFHT: 0.035˝ pixel-1
• HST ACS Virgo cluster survey (Côté et al.): 0.049˝ pixel-1 over 3.3´
• SDSS
Composite Profiles
We combine multiple data sets for each galaxy to provide large dynamic range :
- reduced systematic errors (e. g., sky subtraction)
- more accurate values of µe , re, n (as shown by tight parameter correlations)
- more reliable detection of diagnostic departures from Sérsic fits
n = 6.45
major axis
minor axis
Caon et al. 1994
This Study
Virgo Cluster
Distance = 17 Mpc
1 arcsec = 82 pc
27 elliptical galaxies = 1/3 of Virgo luminosity
Omit S0s: Too little leverage on bulge parameters
Kinematically
decoupled
center
Kinematically
decoupled
center
Kinematically
decoupled
center
Kinematically
decoupled
center
Kinematically
decoupled
center
Central Light Concentration and Profile Shape
In the Virgo cluster + NGC 4434 + NGC 4261,
- 15 of 15 MB > -19.93 Es have extra light
(not power-law profiles)
- 14 of 16 galaxies with extra light have n  4
(exceptions: NGC 4621, NGC 4459);
- All core galaxies have n > 4.
Other studies find some aspects of these results (Lauer et nuk. 1995;
Rest et al. 2001; Ravindranath et al. 2001; …; Ferrarese et al. 2006),
… but not all of them. Also, they did not
notice the dependence on n, or
notice that the extra light is ubiquitous, or
connect it with dissipative mergers.
E profiles are bimodal (Gebhardt et nuk. 1996;
Lauer et nuk. 2006): either they have cores,
or they have “extra light”
(Kormendy et al. 2006).
Gebhardt et nuk. (1996)
Kormendy et al. 2006
Lauer et nuk. 2006
Extra light is often a kinematically decoupled center:
r = 5
Halliday et al. (2001) :
NGC 4458 has a
“clear signature of a
KDC within r < 5 arcsec.”
r = 5
Also in:
NGC 4551
NGC 4478
NGC 4387
This result supports our interpretation that the central extra light is a distinct
component that originates in an accretion event.
Extra light is often a kinematically decoupled center:
Sauron V, fields
Kormendy 1984
Emsellem + (2004)
NGC 5813: kinematically decoupled center at
r ≤ 6 – 9 arcsec (Efstathiou et al. 1982,
Kormendy 1984).
We find extra light at r ≤ 8 arcsec.
This result supports our interpretation that the central extra light is a distinct
component that originates in an accretion event.
The E Dichotomy: There are two kinds of elliptical galaxies
Normal and low luminosity Es
– rotate rapidly,
– are nearly isotropic oblate spheroids,
– are substantially flattened (E3.5),
– are coreless & have extra light near
the center above the inward
extrapolation of the outer Sérsic profile,
– have disky-distorted isophotes.
Giant ellipticals
– are essentially non-rotating,
– are anisotropic and triaxial,
– are less flattened (E2.5),
– have cuspy cores,
– have boxy-distorted isophotes.
Kormendy & Bender (1996)
Profile Shape (i. e., Sérsic n) Participates in the E Dichotomy
Low-luminosity ellipticals have extra light and n  4.
Core galaxies have n > 4.
The E Dichotomy: There are two kinds of elliptical galaxies
Normal and low luminosity Es
– rotate rapidly,
– are nearly isotropic oblate spheroids,
– are substantially flattened (E3.5),
– are coreless & have extra light near
the center above the inward
extrapolation of the outer Sérsic profile,
– have disky-distorted isophotes,
– have n  4.
Giant ellipticals
– are essentially non-rotating,
– are anisotropic and triaxial,
– are less flattened (E2.5),
– have cuspy cores,
– have boxy-distorted isophotes,
– have n > 4.
–the last mergers were gentle and
last mergers
were violent
Galaxy formation: we interpret thethe
above
as evidence
that:and
dissipative. BH binary scouring was
dissipationless; they were
overwhelmed by dissipative starburst.
followed by BH binary scouring.
[/Fe] is consistent with prolonged
merger history.
[/Fe]  star formation finished
in first billion years; subsequent
dissipationless mergers are OK.
The E Dichotomy: There are two kinds of elliptical galaxies
Normal and low luminosity Es
– rotate rapidly,
– are nearly isotropic oblate spheroids,
– are substantially flattened (E3.5),
– are coreless & have extra light near
the center above the inward
extrapolation of the outer Sérsic profile,
– have disky-distorted isophotes,
– have n  4.
Giant ellipticals
– are essentially non-rotating,
– are anisotropic and triaxial,
– are less flattened (E2.5),
– have cuspy cores,
– have boxy-distorted isophotes,
– have n > 4.
Galaxy formation: we interpret the above as evidence that:
–the last mergers were gentle and
dissipative. BH binary scouring was
overwhelmed by dissipative starburst.
the last mergers were violent and
dissipationless; they were
followed by BH binary scouring.
[/Fe] is consistent with prolonged
merger history.
[/Fe]  star formation finished
in first billion years; subsequent
dissipationless mergers are OK.
Formation of the Two Kinds of Elliptical Galaxies
Normal and low luminosity Es
seem entirely consistent with
standard galaxy formation in a
hierarchically clustering universe.
Giant ellipticals
and maybe also the biggest BHs
seem to have formed in a
suspiciously anti-hierarchical
manner.
Are we missing something from
our galaxy formation picture?
Galaxy formation: we interpret the above as evidence that:
–the last mergers were gentle and
dissipative. BH binary scouring was
overwhelmed by dissipative starburst.
the last mergers were violent and
dissipationless; they were
followed by BH binary scouring.
[/Fe] is consistent with prolonged
merger history.
[/Fe]  star formation finished
in first billion years; subsequent
dissipationless mergers are OK.
Formation of the Two Kinds of Elliptical Galaxies
Giant ellipticals
and maybe also the biggest BHs
seem to have formed in a
suspiciously anti-hierarchical
manner.
Are we missing something from
our galaxy formation picture?
Popular explanation of “anti-hierarchical” star formation in giant Es:
AGN energy feedback quenched star formation after < 1 Gyr.
Problem: How do you guarantee that an AGN was switched on
every time a gas-rich galaxy got accreted?
Implications of Extra Light in Low Luminosity Es
We believe that cores are produced by binary black hole scouring.
Low-luminosity Es like M32 and NGC 3377 contain black holes.
They are believed to have formed via major mergers.
But they contain extra, not missing, light near the center.
Why did binary black hole scouring fail?
Suggestion:
The dissipative starburst that produced the extra light
also swamped black hole scouring.
But:
AGN feedback easily prevents star formation  extra light.
Implication:
Maybe AGN feedback is a strong function of black hole mass
(weak feedback from low-mass black holes).
Core “fundamental plane” correlations
define what it means to be an elliptical galaxy.
globular
clusters
E ≠ Sph ≈ S,Im
Kormendy (1985, 1987)
Schematic core “fundamental plane” correlations
(1) in the HST era and (2) with larger samples.
Kormendy et al. 2006
Binggeli 1994
Non-Parametric Version From
Our Virgo Sample (Top 2 Panels)
Here r10% is the radius that
Contains 10% of the light of
the galaxy and 10% is the
surface brightness at r10%.
Bottom panel: Sérsic n
versus absolute magnitude.
Note:
Our Sph galaxies () are biased
in favor of those that are
most like ellipticals.
From Sérsic fits to major-axis profiles:
Edge-on FP has small scatter
(Saglia et al. 1992;
Jørgensen et al. 1996)
From integrating 2-D profiles:
We confirm that
E and Sph galaxies are physically unrelated.
Their formation processes were almost certainly
very different. Best guess:
Es were formed by (wet or dry) major mergers.
Sphs are defunct S+Im galaxies transformed by
internal processes like
supernova-driven baryon ejection (Dekel & Silk 1986)
and by environmental processes
including reionization in the early Universe,
galaxy harrassment, ram-pressure gas stripping, etc.
Summary
Observational results:
Spheroidal galaxies are not faint ellipticals.
They are structurally similar to late-type galaxies.
New aspects of the E - E dichotomy:
Core-anisotropic-boxy
Es have Sérsic n > 4
Coreless-“isotropic”-disky Es have Sérsic n  4
and extra (not missing) light near the center.
Interpretation:
Core-anisotropic-boxy Es formed in gentle, wet mergers;
a dissipative starburst made the extra light;
dissipative starburst swamped BH scouring.
Coreless-“isotropic”-disky Es formed in violent, dry mergers;
binary black hole scouring formed cores.
AGN energy feedback
may have been strong enough in hi-L galaxies to quench star formation
but was weak in wet merger events that made low-L ellipticals.
Two Disagreements with Ferrarese et al. 2006:
(1) “The widely adopted separation of early-type galaxies between ‘core’
and ‘power-law’ types … prompted by the claim of a clearly bimodal
distribution of [inner profile slope] values … is untenable based on the
present study.”
(2) “Once core galaxies are removed, dwarf and bright ellipticals display a
continuum in their morphological parameters, contradicting some
previous beliefs that the two belong to structurally distinct classes.”
Why? The slope bimodality is clearcut.
Compare derived parameters  accuracy given by composite profiles
helps us significantly but is not the main reason.
Sample selection is important:
– We omit S0s and spirals: too little leverage on bulge parameters after
bulge-disk profile decomposition.
– We distinguish E and Sph galaxies. Important: Sphs are not faint Es!
Ferrarese et al. deny that the E-E dichotomy exists based on inner profile
slope and then rediscover it based on a break in profile slope between
large radii (steep Sérsic function) and small radii (shallow power law).
But many people always defined cores based on profile breaks.
Also core vs. no core correlates with many other physical parameters.
Next Steps In Progress
1. Expand this study to all ~ 230 nearby ellipticals that have
HST data using V-band imaging from McDonald Observatory
and from other sources.
2. Study dynamics with (e. g.) V/ -  diagram
3. Estimate star formation timescales with [/Fe] using HET.
4. Compare with n-body merger simulations.
NGC 4621
A Puzzle and a Problem For Hierarchical Mergers
1. What is the origin of the dichotomy into two kinds of elliptical galaxies?
2. In galaxy formation by hierarchical clustering,
A. How do you make the biggest galaxies form their stars so quickly?
B. How do you stop them from forming stars afterward?
The [/Fe] chronometer:
 elements are produced by SN II;
Fe are produced mainly by SN Ia.
Therefore:
[/Fe] tells us star formation timescale
Thomas et al. (2002a,b) find that the biggest ellipticals are the most -enhanced
 they formed their stars quickly, before Sne Ia polluted the ISM with Fe.
After that, they formed few stars, so L-weighted [/Fe] did not get depleted.