Elliptical galaxies

•The brightest galaxies in the Universe are ellipticals, but also some of
the faintest.
•Elliptical galaxies appear simple: roundish on
the sky, the light is smoothly distributed, and
they lack star formation patches or strong
internal obscuration by dust.

•But are they so simple? Detailed studies

reveal great complexity:

•shapes (from oblate to triaxial);

•large range of luminosity and light
concentration;
•fast and slowly rotating;
•cuspy and cored …
 Classification by luminosity
• There are three classes of ellipticals, according to their luminosity:

• Giant ellipticals have L > L*, where L* is the luminosity of a large galaxy,
L* = 2 × 1010 L¯ or MB = - 20 (the Milky Way is an L* galaxy).

• Midsized ellipticals are less luminous, with L > 3 × 109 L¯, or MB < -18

• Dwarf ellipticals have luminosities L < 3 × 109 L¯

•These luminosity classes also serve to describe other properties of E

galaxies (in contrast to disk galaxies, where each class be it Sa … Sd
contains a wide range of sizes and luminosities).

•Ellipticals come in one-size sequence: their internal properties are

correlated with their total luminosity (or mass)
 Galaxy photometry
The surface brightness of a galaxy I(x) is the amount of light on the sky at a particular
point x on the image.

Consider a small patch of side D in a galaxy located at a distance d. It will subtend an

angle α = D/d on the sky. If the combined luminosity of all the stars in this region is L,
its apparent brightness is F = L/(4πd2).
Therefore the surface brightness

I(x) = F/α2 = L/(4π d2) * (d/D)2 = L/(4π D2).

The surface brightness is independent of distance (as long as the objects are not at
cosmological distances, in which case the geometry of the Universe plays a role).

The appropriate units of I(x) are L¯/pc2. However, quite often the magnitude is quoted
instead of the flux at a given point on an image. In this case,
one also speaks of a surface brightness:
µλ(x) = -2.5 log10 Iλ(x) + cstλ

The units of µ are [mag/arcsec2.]

 Galaxy photometry (cont)
The constant is set to have a value of 26.4 mag/arcsec2 in the V-band. This magnitude
.
corresponds to 1 L¯V/pc2. Therefore, IV = 10 0 4 (26.4 - µV) L¯V/pc2

The total (apparent) magnitude of the galaxy is simply:

which for a circularly symmetric galaxy, reduces to

Given that galaxies have no sharp edges, it is customary to measure integrated

magnitudes up to the radius of a given magnitude, usually 26 µΒ.

Note that usually I(R) will be given in units of energy/sec/arcsec2, so that R is given in
arcsec. Only if the distance to the object were known, could we derive its absolute
magnitude.

Contours of constant surface brightness on a galaxy image are known as isophotes.

 Surface brightness profiles of elliptical galaxies
The quantity I(R) denotes the surface brightness profile on the major axis of an image.

The figure shows the surface brightness profile of the giant elliptical galaxy NGC 1700
as function of the projected distance to the center R (left) and as function of R1/4 (right).

An R1/4 profile is a straight line in this

diagram. Note that it fits the data very
well over 2 decades in radius.

The surface brightness falls 9 magnitudes

from the centre to the outskirts, implying
a 109 fall-off in projected luminosity!

The light in elliptical galaxies is quite

centrally concentrated.
 De Vaucouleurs profile
The light distribution of NGC 1700 can be fitted with a function which can be
conveniently written as:

with
•Re the effective radius
•the factor 3.33 is included so that half of the total light is emitted inside a radius Re
•the parameter Ie is the surface brightness at R = Re
•the central brightness of the galaxy is I0 ~ 2000 Ie.

It is remarkable that such a simple 2-parameter profile, can fit the profiles of ellipticals
so well.

This profile is a particularly good description of the surface brightness of giant and
midsized elliptical galaxies.
 Other common profiles
Sersic law:

where bn is chosen such that half the luminosity comes from R < Re.
This law becomes de Vaucouleurs for n=4, and exponential for n=1.
Dwarf ellipticals are better fit by exponential profiles.

Hubble-Oemler law:

with I0 the central surface brightness, and r0 the radius interior to which the
surface brightness profile is approx. constant.
For r0 < R < Rt the surface brightness changes as I ~ R–2.
For R > Rt the surface brightness profile decays very quickly and predicts a finite
total luminosity.

In the limit Rt → ∞ this one reduces to the Hubble law:

 Central regions: the effect of seeing
The observed profile for NGC 1700 and other giant E follow rather closely de
Vaucouleurs profiles, except in the very bright core, where atmospheric turbulence or
seeing, blurs the image.

At small radii, the light is suppressed

and redistributed at large radii.

Therefore seeing transforms a power-

law profile into a profile with a core
(a central region with nearly constant
surface brightness).

Due to seeing, stars are not observed as point sources but have finite extent. Their
light-profile can often be expressed as a Gaussian with a characteristic extent,
quantified by the dispersion σ, or half-peak intensity radius. This is known as the PSF:
Point Spread Function.
 Effect of Seeing - PSF
The effect of the seeing is to blur an otherwise sharp image. If in absence of seeing the
surface brightness of an object at a position R’ is It(R’), the measured brightness at a
location R will be:

where P(d) is the PSF. Note that in the absence of seeing, P would be the Dirac delta-
function δ( R – R’ ), and one recovers the original (true) profile.

In the simplest case, the PSF can be treated as a circularly symmetric Gaussian

P(d ) = 1 e − d 2 / 2σ 2

2πσ 2

It is then possible to show that, for a circularly symmetric surface brightness

distribution It(R'):

where I0 is the modified Bessel function of order 0.

 The outskirts of galaxies
•The surface brightness profile of giant ellipticals often shows an excess of light in the
outer parts compared to the de Vaucouleurs profile.

•Such galaxies are known as cD Galaxies. They are usually located at the center of
clusters of galaxies, or in areas with a dense population of galaxies.

•This excess emission indicates the presence of an extended halo. The cD halos could
belong to the cluster rather than to the galaxy.

M87, the central cD

galaxy in the clusters
of galaxies inVirgo.
Note the extended,
low-surface brightness
halo.
 Photometric properties of E galaxies
•The central surface brightness of an elliptical is tightly correlated with the total
luminosity.

•The plot shows the central brightness IV(0), the core radius rc (the radius at which the
surface brightness has dropped to half its central value) as functions of the total
luminosity or absolute magnitude

•Note that for the giant and

midsized E, the more luminous the
galaxy, the lower its central
brightness, and the larger its core.

•This shows, that just like with stars,

the properties of galaxies are not
random: there is a certain degree of
coherence.
 Cores and cusps
•Space observations, where seeing does not play any role, show that midsized galaxies
have central cusps, and not cores.

•The measured central surface brightness from the ground is therefore, only a lower
bound to the true value.

•The “core radius” does not always have the same physical meaning: it could be a point
where the brightness profile changes slope, rather than the outer limit of a region where
the stellar surface brightness is constant.

•Midsize ellipticals are usually cuspy,

while giant ellipticals tend to have cores
 Isophotes
•Isophotes are contours of constant surface brightness.

•The ratio of semi-major to semi-minor axis measures how far the isophote deviates
from a circle ε = 1 – b/a.

•This allows to classify E galaxies by type En, where n = 10 (1-b/a)

•Note that this classification depends strongly on viewing angle

 Deviations from ellipses
Isophotes are not perfect ellipses. There may be an excess of light on the major axis
(disky), or on the “corners” of the ellipse (boxy).

The diskiness/boxiness of an isophote is measured by the difference between the real

isophote and the best-fit ellipse:

δ(φ) = <δ > + Σ an cos nφ + Σ bn sin nφ

•If the isophotes have 4-fold symmetry, and the ellipse has been correctly fitted, then
terms with n < 4 and all bn should be small. The value of a4 tells us the shape:
•a4 > 0 is a disky E,
•a4 < 0 corresponds to a boxy E.
 Isophote twisting
In the case the true intrinsic shape of a galaxy is triaxial, the orientation in the sky of
the projected ellipses will not only depend upon the inclination of the body, but also
upon the body’s true axis ratio.

This is best seen in the projection of the following 2-D figure:

Since the ellipticity changes with radius, even if

the major axis of all the ellipses have the same
orientation, they appear as if they were rotated in
the projected image.

This is called isophote twisting.

Unfortunately it is impossible, from an observation

of a twisted set of isophotes, to conclude whether
there is a real twist, or whether the object is
triaxial.
 Here is an example of twisted isophotes in a satellite galaxy of Andromeda (M31).

The first, shallower, exposure shows the

brightest part of the galaxy. The second,
deeper, exposure shows the weaker more
extended emission. A twist between both
images of the same galaxy are apparent (the
orientation in the sky is the same in both
figures).

•Boxy galaxies are

•more likely to show isophote twists,
•more luminous in general,
•probably triaxial.

•Disky E are
•midsize,
•more often oblate,
•and faster rotators.
Some people have suggested disky E can be considered an intermediate class between
the big boxy ones and the S0s.
 Fine structure
About 10 to 20% of the elliptical galaxies seem to contain sharp steps in their
luminosity profiles. An example is the elliptical NGC 3923:

These features are called ripples

and shells.

Ripples and shells have also been detected

in S0 and Sa galaxies.

Since in early-type galaxies they are

detected because of the smooth profile of
the underlying galaxy, it is not clear
whether this is an universal phenomenon
also present in later-type galaxies but
difficult to detect there.
They are probably the result of the accretion/merger of a small galaxy on a very
elongated (radial) orbit (Quinn 1984).
 De-projection of galaxy profiles
So far we have discussed observed surface brightness profiles I(R), that is 3-
dimensional distribution of light projected onto the plane of the sky.

The question is whether we can, from this measured quantity, infer the true 3−D
distribution of light, j(r) of the galaxy. If I(R) is circularly symmetric, it is possible that
j(r) is spherically symmetric, and from the following figure it is apparent that:
 De-projection: Abel integral
This is an Abel integral equation for j as a function of I, and its solution is:
∞
1 dI dR
j (r ) = − ∫
π r dR R2 − r 2

An example of a simple pair (connected via the Abel integral), that approximately
represents the profiles of some elliptical galaxies, is:

This surface brightness profile is known as the modified Hubble law. Notice that for
R >> r0: I ~ R−2, and j ~ r−3.
 De-projection: Non-spherical case
•If the isophotes are not circularly symmetric, then the galaxy cannot be spherically
symmetric, but it can still be axisymmetric.

•In that case, if the observer looks along the equatorial plane of such object, it can be
shown that is still possible to de-project the surface density profile and obtain the spatial
profile.

•However, in general the line of sight will be inclined at an angle with respect to the
equatorial plane of an axisymmetric galaxy.

•In that case it can be shown that there are infinite de-projected profiles that match an
observation.

•It is easy to see that when observed from the pole, both a spherical galaxy as well as
any oblate or prolate ellipsoid will produce the same projected distribution (as long as
the 3-D radial profile is properly constructed).
 Shapes of elliptical galaxies
What can we learn from the distribution of observed apparent
ellipticities about the true (intrinsic) distribution of axis ratios?

In the most general case, the (luminosity) density ρ(x) can be expressed as ρ(m2),
where: 2 2 2
x y z
m2 = 2 + 2 + 2
α γ β

The contours of constant density are

ellipsoids of m2 = constant.

There are three cases:

α ≠ γ ≠ β: triaxial
α = γ < β : prolate (cigar-shaped)
α = γ > β : oblate (rugby-ball)
 Shapes of elliptical galaxies
Let’s assume that an elliptical galaxy is an oblate spheroid.

An observer looking down the z-axis will see an E0 galaxy, while

when viewed at an angle, the system will look elliptical, with axis ratio q0 =b/a.

How is q0 related to α and β ?

 Apparent axis ratios
The observed axis ratio is q0 = A/(m α).

The line of sight intersects the ellipsoid at (x0, z0)

at the constant-density surface m.

The segment A = C sin θ, while C = z0 + (- x0)/ tan θ; and tan θ = dx/dz.

Differentiating d(m2 = x2/α2 + z2/β2 ) = 0, we find tan θ = -z0/x0 α2/β2 .

Replacing, C = m2 β2 / z0. Finally q0 = m β2 / (α z0) sin θ, or

q02 = cos2θ + (β/α) 2 sin2θ

This implies that the apparent axis ratio is always larger than the true axis ratio, a
galaxy never appears more flattened than it actually is.
 Expectations
We can use the previous relation to find the distribution of apparent ellipticities q0
produced by a random distribution of oblate/prolate ellipsoids with axis ratios q = β/α.

If the ellipsoids are randomly oriented wrt line-of-sight with angle θ, then of the
N(q) dq galaxies with true axis ratio in the interval (q,q + dq), a fraction sinθ dθ will be
inclined with their axis in (θ , θ + dθ).

The probability, P(q0|q) dq0, to observe a galaxy with true axis ratio q to have apparent
axis ratio between q0 and q0 + dq0 is:
P(q0|q) dq0 = sinθ dθ thus P(q0|q) = sinθ / | dq0 /dθ |.
(where q0=q0 (θ) is known)

If there are f(q0)dq0 galaxies with (observed) axis ratios in (q0,q0+ dq0 ), then
f(q0)dq0 = N(q) dq P(q0|q) dq0 .
 Distribution of ellipticities
We have then that N(q) dq
f(q0 ) = ∫ N(q) dq P( q0 | q) = q0 ∫
2
1 - q 2 q0 − q 2

This is an integral equation for N(q), which can in principle be solved

from the observed distribution of ellipticities.

This is the observed distribution of apparent

ellipticities (ε = 1 – q0), for 2135 E galaxies

Conclusion: It is not possible to reproduce the

observed distribution if all galaxies are either
prolate or oblate axisymmetrical ellipsoids.
 More on shapes
•The apparent shapes of small E are more elongated than for large E

•On average, midsized ellipticals (M > -20), have q0 ~ 0.75. If they are oblate, this
would correspond to 0.55 < q < 0.7

•Very luminous E, with M < -20,

have on average q0 ~ 0.85. But since
there are so few that are spherical on
the sky, it is very likely that most of
these are actually triaxial.
 The kinematics of stars in E galaxies
Stars in E galaxies do not follow ordered motions, but most of their kinetic energy is in
the form of random motions. Moreover, the more luminous an E, the larger the velocity
dispersion (which, as we have seen can be used to derive distances to E).

To measure the orbital velocities

of stars within galaxies is a
difficult task.

Absorption lines are used, which

are the result of the light of all
stars. Each star emits a spectrum
which is Doppler shifted in
wavelength according to its
motion. This orbital motion
makes the resulting spectral lines
be wider than those of an
individual star.
 Spectra and motions of stars in E
Let the energy received from a typical star, at rest with respect to the observer, be
F(λ) ∆λ. Ιf the star moves away from us with a velocity vz<< c, the light we receive at
wavelength λ, was emitted at λ (1 - vz/c).

To find the spectrum on the galaxy image, one needs to integrate over all stars along
the line of sight (z-direction). If the number density of stars at position r with velocities
in (vz, vz + dvz) is f(r, vz)dvz, the observed spectrum is

∞ ∞
Fgal ( x, y, λ ) = ∫ dv z F(λ[1 − v z / c]) ∫ dz f(r, v z )
−∞ −∞

If the distribution function f (r, v) was known, and the spectra were the same for all
stars, we could derive the spectrum of the galaxy. In practice, one makes a guess for the
f(r, vz), which depends on a few parameters, and fixes those in order to reproduce the
observed spectrum.
 Spectra and motions of stars in E
∞

∫ z ∝ − − σ
2 2
A common choice is the Gaussian: f(r, v ) dz exp[ (v z Vr ) / 2 ]
−∞

where σ(x,y) is the velocity dispersion of stars, while Vr(x,y) is the mean radial
velocity at that position.
2D-maps of the kinematics
It has recently become possible to take
multiple spectra over the image of a galaxy.
This is known as integral field spectroscopy.
Each pixel in the image has a corresponding
spectrum.

This is an example from the SAURON team

for 4 elliptical galaxies.

It is now possible to see how the velocity

field and its higher moments vary across the
galaxy, and to find departures from axial
symmetry.

This study has shown that E galaxies are far

less regular than originally believed. One
finds counter-rotating cores, misalignments,
minor axis rotation…
 Do Elliptical galaxies rotate?
It was initially thought that the flattened shape of some E galaxies was due to them
being fast rotators. However, observations have shown that

•Bright ellipticals (boxy) rotate very slowly, particularly in comparison

to their random motions. Their velocity ellipsoids must be anisotropic
to support their shapes.

•Midsized ellipticals (disky) seem to be fast rotators

boxy disky
 Link between shape and internal kinematics
In the case of a star like the Sun, the frequent random encounters between gas particles
produces a round smooth sphere. When this sphere is given some angular momentum, it
will change its shape and become more flattened near the poles.

In the case of E galaxies, we know that collisions play no role (timescales are much
larger than a Hubble time). So the argument that their flattened shape of some E
galaxies would be due to rotation is flawed.

However, one still expects to find a correlation between the shape of a galaxy and the
degree of anisotropy (rotation + random motions) of the velocities of the stars.

For example, if a system is very flattened in one direction, this means that the velocities
of the stars along that direction (or axis) must be smaller than those along a
perpendicular direction (otherwise they would reach larger distances and lead to a
rounder shape).

This is why the internal kinematics of the stars (i.e. velocity anisotropy) will be related
to the intrinsic shape of a Galaxy.
 Scaling relations for E galaxies
We have seen before that E galaxies follow the Faber-Jackson relation

σ
LV ≈ 2 × 10 Lsun (
10
)4
200 km/s

which can be used to derive distance to a galaxy. Note that this relation implies that
more luminous galaxies have larger velocity dispersions (the typical range is from 50
km/s for dE up to 500 km/s for the brightest Es).

Another scaling relation E galaxies follow is the fundamental plane. Elliptical galaxies
all lie close to a plane in the 3D space defined by (velocity dispersion σ, effective radius
Re, surface brightness Ie). Approximately

−0.82
Re ∝ σ 1.24
Ie
 The fundamental plane
These plots show some
of the scaling relations
that E galaxies satisfy.

They are all projections

of the fundamental
plane relation, that is
shown in the last panel.

They express some

fundamental property
of E galaxies.

We will show in an exercise that the

such correlations are not surprising if
Kormendy & Djorgovski 1989 one uses the virial theorem
 Stellar populations in E galaxies
It is not possible to see individual stars in galaxies beyond 10-20 Mpc.

Hence the stellar populations characteristics of E galaxies must be derived from the
integrated colours and spectra.

As we saw before, the spectrum of an E galaxy resembles that of a K giant star.

E galaxies appear generally red:

•very few stars made in the last 1-2 Gyr (recall that after 1 Gyr, only stars with
masses < 2 M¯ are still on the main sequence)
•Most of the light is emitted while stars are on the giant branch
•The stars in the centres of E are very metal-rich, similar to the Sun

This would seem to suggest that E galaxies are very old

 CMDs for E galaxies
There is a relation between the colour, and total luminosity for E galaxies.

Sandage 1972

These plots show, for galaxies in the Coma and Virgo clusters, that

•Brighter galaxies are redder

•Fainter systems are bluer

This could be explained if small E galaxies were younger or more metal-poor than the
large bright ones. The fact that metallicity can mimic age effects, prevents one from
concluding which of the two cases (or both) is closer to the truth (without resolved
deep colour-magnitude diagrams it is not possible to solve this problem)
•We understand why stars occupy certain regions of the color-magnitude diagram: the
luminosity and temperature are controlled by the stars mass, and the nuclear processes
occurring inside the stars.

•Explaining the correlations in the properties of E galaxies is harder, because they

almost certainly reflect the conditions under which the galaxies formed, rather than their
internal workings at the present day. The well-defined patterns tell us that galaxy
formation also had some regularity, and that the process must somehow be related to its
mass.
 Gas in E galaxies
The lack of young stars in E galaxies is not surprising given that most contain very
little cool gas.

Roughly 5 - 10 % of normal ellipticals contain HI or molecular gas to be detectable (for

the big elliptical galaxies this is less than 108 - 109 M¯. This should be compared to a
large Sc which has 1010 M¯).

There are some exceptions, but are usually peculiar (i.e. dust lanes, recent merger, etc).

However, the average elliptical contains very large amounts of hot ionized gas.
This gas
•emits in X-rays (T ~ 107 K)
•located in a halo of ~ 30 kpc radius
•roughly 10 - 20% of the mass in stars is in this component

Boxy ellipticals tend to have more luminous X-ray halos and are also radio-loud (i.e.
they emit in the radio wavebands).
 Black holes in the centres of E
Some E galaxies have rising velocity dispersions as one moves closer to the centres. To
keep fast moving stars within the centres, requires large masses concentrated in small
regions of space, beyond what can be accounted for by the stars themselves.

For example, for M32, a dwarf E satellite

of M31, 2 million solar masses needed
within the central 1 parsec!!

The current preferred explanation is that

these large masses are actually due to
super-massive black holes. Some believe
that every galaxy harbours a SMBH.
 Black holes in the centres of E
There is some evidence that the mass of the SMBH correlates with the total luminosity
or velocity dispersion of the E galaxy.

Mbh ∝ σα

α ∼ 4.8

Magorrian et al 1998
Ferrarese & Merritt 2000
Gebhardt et al 2000