Properties of Elliptical Galaxies – I. Stellar Populations and the Interstellar Medium Brightest stars are red giants and AGB stars w/ spectral type K, M Very few stars produced in the last 1 – 2 Gyr Stars at the center have ~solar metallicity (differ from the Milky Way’s globular clusters but similar to the bulge star metallicities) Color of integrated galaxy light correlates with galaxy luminosity Fainter/smaller galaxies are bluer either less massive ellipticals are younger or metal poor

Some “bluer” E’s are called “E+A galaxies” Model of galaxy that makes all stars in 10 8 yr burst strong emission lines and He absorption (OB stars) seen after 10 Myr After 1 Gyr, deep H absorption lines of A stars are seen (E+A galaxy) After 10 Gyr, looks like typical E Galaxies with more recent SF (younger stars) are bluer Luminous galaxies contain more old stars Study of ~90,000 galaxies from SDSS based on strength of Halpha line and H&K break Largest, most luminous galaxies are forming stars at a much slower rate than necessary to build up their current stellar content. Faint galaxies are making stars faster…

Why might fainter, less massive galaxies be metal poor? Metallicity measured in Z  (solar metallicity units), (O/H) or [Fe/H] = log(N Fe /N H ) gal – log(N Fe /N H )  (see xy/Structure/metals.html) Luminous galaxies - richer in heavy elements - currently have below average star formation rates (SFR) Metal absorption lines are strongest in the biggest Ellipticals which are redder Color gradient in Ellipticals  redder toward the center, more metal rich (based on strength of Mg abs lines) Galaxies that are more metal poor are bluer  Unable to retain most of the metal-rich gas shed by aging stars

Ellipticals have lots of globular clusters (about twice that of disk galaxies) these fall into two groups based on color color determined by metallicity, with more metal-rich GCs (redder) possibly the result of galaxy mergers Ellipticals have much less cool, atomic gas than spiral galaxies < 10 8 – 10 9 M sun while an Sc galaxy contains ~10 10 M sun sometimes (though rare) seen as dust lanes orbiting E’s - from merger or capture? Ellipticals have hot, ionized gas stars lose mass through shedding of stellar envelopes as they age this gas is ~few x 10 7 K  radiates in X-rays luminous, most massive E’s have the hottest gas extends to R > 30 kpc with 10 9 – M sun (10 – 20% stellar mass) smaller E’s have less hot gas – weaker gravity cannot prevent escape gas not produced through SNe – low metallicity (gas from stars with large random motions  kinetic motion heats the gas)

X-ray “surface brightness” is proportional to optical SB in many Ellipticals j opt ~ j x-ray - surprising! Because... j opt ~ n * (stellar density) j x-ray ~ n e 2 (plasma/gas density) (depends on collisions between e - and ions) And this means that the gas density falls with radius more slowly than stellar density to make SBs so similar. This happens because plasma/gas T is higher than the stellar/kinetic T. If n * ~ r -α n e ~ r -αβ to fit with observations β = T * /T ~ 0.5 (which is about right) T gas = 7x10 6 K T stars (kinetic temp) = 3x10 6 K (Jeltema et al 2003) X-ray – crosses (ROSAT) optical – line (HST) normalized at R=10”

Properties of Elliptical Galaxies – II. Kinematics Disordered, random motions are important for Ellipticals v/σ = ordered motion/disordered motion v = systematic velocity change across region of galaxy V rot or V r = rotational velocity σ = velocity dispersion of stars at one position or along line-of-sight (los) σ = 1/2 Many Ellipticals have v/σ << 1 Spirals generally have v/σ ≥ 1 cD galaxy NGC 1399 notice (V r -V sys )/σ << 1

Measuring Stellar Velocities: The spectrum measured from some point in an external galaxy consists of the sum of spectra from all unresolved stars along that los. Each star has a different V los, so the composite spectrum will be broadened as well as shifted most likely. Define los velocity distribution (LOSVD) as the function F(V los ) such that the fraction of stars with velocities between V los and V los + dV los is given by F(V los )dV los mean velocity V los = ∫dV los V los F(V los ) dispersion σ 2 los = ∫dV los (V los -V los ) 2 F(V los ) F(V los ) can be approximated by a Gaussian distribution, but its true form is modeled by closely examining line profile shapes (e.g. Gauss- Hermite model – BM – more on this later…)

For Elliptical galaxies, σ los is determined by the “width” of the absorption features. Absorption lines are compared with local G & K type stars (templates) to determine dispersion-related broadening. K0 Giant star NGC 2549 S0 galaxy note Mg b absorption feature at 518 nm G(λ) = F(V los ) convolved with S(λ) where F(V los ) is LOSVD

Correlations among global parameters: 1) σ o ∝ color ∝ line strength higher central velocity dispersion = redder E = stronger metal absorption lines Somewhat non-intuitive because optical color comes mainly from light at R e while Mg strength and σ o are based on light primarily from the galaxy center.  more massive galaxies are more metal rich (stronger Mg 2 )  deeper potentials hold ISM longer allowing metals to build up

2) Radius - Surface Brightness Relation RecallI(R) = I e exp{-7.67[(R/R e ) 1/4 -1]} ThenL tot = ∫ o ∞ I(R) 2πR dR = 7.22 π R e 2 I e Then, mean surface brightness within R e is: = (∫ o Re I(R) 2πR dR)/(∫ o Re 2πR dR) = (½ L tot )/π R e 2 = (½ (7.22) π R e 2 I e )/(π R e 2 ) = 3.61 I e = easier to measure than I e R e ∝ -0.83±0.08 Larger galaxies have fainter effective Surface Brightness (Kormendy 1977; Djorgovski & Davis 1987) Observations show…

Since L e (light interior to R e ) = π R e 2  ∝ L e -1.5 More luminous Ellipticals have larger R e and fainter mean SB  larger and more luminous galaxies are fluffier with lower densities One inference: low-luminosity ellipticals formed with more gaseous dissipation than giant ellipticals. Consistent with Kormendy 09 interpretation for core ellipticals. 3)σ o 4 ∝ L e Faber-Jackson relation (F&J 1976) More luminous Ellipticals have higher central velocity dispersions brightest E’s  σ o ~ 500 km/s faintest E’s  σ o ~ 50 km/s *Can be used as a distance indicator!!

So, we have correlations between R e and I e and between σ o and L e. Is there a sensible way to combine this information? Before describing the observations, let’s explore what we might expect from dynamics. Virial Theorem: -If Ellipticals are self-gravitating and in dynamical equilibrium, they should obey the Virial Theorem 2KE + PE = 0 Scalar VT VT relates average stellar speeds to the depth of the gravitational potential well in which they move.

KE = ½ M PE = -GM 2 /R g Mean velocity squared of all stars in system Gravitational radius – similar to “effective radius” or half-mass r and depends on detailed mass distribution In VT, we get M = GM 2 /R g  M = R g /G Expressing VT in terms of mass surface density M/A = η = R g /(AG) Assume 1) ∝ σ 2 2) R g ∝ R e 3) A ∝ R e 2 η ∝ σ 2 /R e

If an elliptical galaxy obeys VT, then these 3 parameters contain most of the information about its structure. All E’s should obey the same relation if E’s are a homologous family (i.e. E’s have the same SB shape and velocity dispersion distributions). Then E’s should populate a plane in η, σ, R space. Since η is not directly observable, η = M/A = (M/L) ∝ (M/L) I e (M/L) I e ∝ σ 2 R e -1 Mass-to-light ratio

What do observations show? Ellipticals do occupy a plane in 3D space of I e, σ, R e Observations show… R e ∝ σ 1.4 I e (Dressler et al. 87) I e ∝ σ 1.6 R e -1.2 The Fundamental Plane VT gives (M/L) I e ∝ σ 2 R e -1 Both are true if (M/L) ∝ σ 0.4 R e 0.2 From F-J σ 4 ∝ L e  σ 0.4 ∝ L e 0.1 From R-SB R e 0.8 ∝ L e  R e 0.2 ∝ L e 0.25 M/L ∝ L 0.35  M/L is higher in more luminous, bigger galaxies M/L ∝ L 0.27+/-0.08 (Pragniel and Simien 1996)

Therefore, results of parameter correlations in ellipticals can be understood if: 1)VT is only strong constraint 2)E’s are a nearly homologous family 3)M/L ∝ L 0.3 How do we interpret this M/L relation? -Differences in stellar populations (brighter galaxies are also redder) -Differences in M dark /M baryonic (though Romanowsky 03 show evidence for only small amounts of DM inside R e ) Trujillo et al. 04 argue FP tilt due to a systematic variation of the structural nonhomology of E’s as a function of L ¾ tilt due to structural nonhomology (more massive galaxies have steeper velocity dispersion profiles (σ 1.6 rather than σ 2) & SB with higher n) ¼ tilt due to stellar pops

Core Fundamental Plane (Faber et al. 97) R b, I b and σ obey FP relationship also (where “b” refers to the “break” radius) VT rules parameters at break radius Thicker plane  central M/L ratios may vary more D n – σ Correlation (Dressler et al. 87) D n = diameter within which mean SB is This essentially replaces I e and R e with one parameter D n /R e is larger for galaxies of high SB than low SB but with same L D n ~ σ 1.4 I e only weak dependence on I e Primarily used as distance indicator (better than L ~ σ because that relation did not also include SB parameter).

Large Scale Kinematics Kinematic parameters of E’s display overall symmetry about the center of the system (reason to assume equilibrium models explain current state - VT) V los changes sign across galaxy center Systems display net rotation about minor axis System flattened by rotation? though same Hubble types can have very different net rotation rates… Observed rotation of E’s is too low to explain observed flattening To quantify this, define V max as maximum V los. Where model values are based on a system whose ellipticity is determined only by rotation (which is ~independent of inclination angle (see BT 4.3)) Thus, (v/σ)* ~ 1 for system flattened by rotation

However, elliptical galaxies have ratios mostly less than unity Intermediate-to-low luminosity E’s lie closest to unity – flattened by rotation Brightest (and faintest) have lower values – not flattened by rotation alone BM Fig 11.7

Velocity distribution profile shapes give a clue… Recall broadening of E absorption lines is due to random velocities of stars along a given line of sight through the galaxy G(λ) = F(V los ) convolved with S(λ) where F(V los ) is LOSVD To first order, we assume the distribution is Gaussian, though deviations (kurtosis, skew) have been observed and show correlations with the amount of rotational velocity. Use truncated Gauss-Hermite function to model LOSVD where skew is 3 rd order term, kurtosis is 4 th order term

E’s with highest V rot have most skewed LOSVD indicating 2 components Non-rotating broad component Fast rotating narrow component

Embedded disks Recall “disky” isophotes have larger V rot than “boxy” isophotes (though photometric modeling is difficult for detecting disks in E’s) V los profiles provide the best evidence for embedded disk components (Rix & White 1990; van der Marel et al. 1994) Disky E’s have (v/σ)* ~ 1  they are rotationally flattened If E’s harbor disks, the relative size of this component dictates kinematics along major axis

Dynamical models help to explain non-rotationally flattened systems. Slow rotation in luminous E’s implies that such objects are flattened by velocity anisotropy rather than rotation. Schwarzchild (1982) showed that: This argues for triaxial objects – many giant E’s have not relaxed enough to develop an axis of symmetry equilibrium triaxial systems could exist with slow rotation about the minor axis Further indication of only partial mixing of stellar orbits – KDC’s

Kinematically Distinct Cores in E’s 2 kinematic components identified in 25 – 30% of E’s In most cases, core rotates differently than outer galaxy Rapid rotation in center, slower further out Rotation direction different Central component often rotates around minor axis (Bender et al 1988)

(Efstathiou, Ellis & Carter 1992) Formation may be due to merger (e.g. NGC 5813) Core (r < 7”) had distinct SB, V rot, σ Core is remnant of dense, low-L E which may have been captured by a larger E (recall low-L E’s have brighter I e, higher V rot, lower σ) Others may be explained by accretion of gas-rich galaxy which settled to center and formed a disk (many KDCs appear to have disk-like profiles. Torques on gas disk by main galaxy cause gas to spin either parallel or anti-parallel. Star formation eventually turns gas into stars.

Observations support 2 categories of Ellipticals Boxy and Disky galaxies have the same : color-magnitude relation Mg 2 vs relation Fundamental Plane relation Modified Hubble diagram ? Disky Ellipticals form an extension of the S0s Boxy Ellipticals lie at the extreme left end