NGC 4696

More Clues to Galaxy Formation: Massive Globular Clusters,

Stochastic Self-Enrichment, and Mass/Metallicity Correlations

NGC 4696

HST/ACS

Harris && 2006

Starburst dwarf NGC 5253 (ESO/HST)

Pregalactic dwarf

Proto-GCs

GMCGC MM 210

Young massive star clusters (YMCs)

forming at ~105 M0 in

starburst dwarfs today

Harris && 2008 Harris && 2006

Bimodal or not?

Harris && 2008 Harris && 2006

Bimodal or not?

Disagreements ahead --

Is this effect caused by ---

(1) A gradual shift of the blue sequence to redder color at higher luminosity? (Mass/Metallicity relation)

(2) The disappearance of bimodality altogether at the highest masses? (Threshold enrichment effect)

(3) An artifact of photometric measurement procedures? (i.e. not real)

Serious questions persist!

If it’s a true, physical MMR then Z ~ M1/2 at high mass, and it may smoothly connect upward to the UCD regime.

Does it continue to low mass? Why no red-sequence MMR?

Is it present in all galaxies? What is its astrophysical origin?

The systematic properties of globular clusters begin to change for M > 2 x 106 M0 …

- Appearance of the MMR

- Multiple populations within a single GC

Cen (Villanova && 2007)

- Different scaling of size vs. mass

Evstigneeva et al. 2008

Category 1: MMR is present and measurable

M87, NGC 1399, several other BCGs and gE’s

Category 2: MMR is not present

M49; any others?

Category 3: presence of MMR not decidable; GC sample too small or does not extend to high enough luminosity

Milky Way; M31; dwarf galaxies; most spirals; GC-poor E’s

The basic feature of bimodality is a first-order and (probably) universal effect. The MMR is a second-order effect and harder to trace. Though new, much confusion already exists:

Can be helped (partially) by constructing composite samples; e.g grouping Virgo Cluster Survey galaxies into 4 luminosity groups (Mieske && 2006) or combining several supergiants (Harris && 2006)

But if amplitude of MMR differs from one galaxy to another, net effect will be diluted in composite samples

1: strong MMR 2: no MMR 3: Not decidable

[Fe/H]

MV

Milky Way GCs

Most galaxies do not have clusters in the 106 – 107 M0 range

First, let’s get the measurements straightened out.

NGC 5128: d=3.8 Mpc

Globular clusters are easily resolved at <1’’ seeing

Photometry must account for individually different scale sizes

GC profile as seen on image =

PSF Intrinsic GC profile

rh ~ 1 – 5 parsecs; averages 3 pc 0.3” width

NGC 3311/3309 (A1060)

d = 50 Mpc

2 rh ~ 6 pc 0.025”

fwhm(PSF) = 0.5”

starlike! psf-fitting photometry is fine

Gemini-S + GMOS, Wehner & Harris

Several regimes determined by distance; no single photometric method is suitable for all regimes

4 distinguishable regimes: compare fwhm of stellar PSF

with intrinsic cluster size D (= 2 rh), half-light diameter

Well resolved: D >> fwhm(PSF)

Partially resolved: D ~ fwhm

Marginally resolved: D ~ 0.1 – 0.3 fwhm

Unresolved (starlike): D < 0.1 fwhm

Aperture photometry r(ap) adjusted for D

PSF-fitting photometryAll this is subject to S/N considerations …

HST/ACS imaging of GCs around 6 central supergiants in Abell-type clusters (Harris et al. 2006, 2008)

(B,I) bandpasses metallicity-sensitive

Thousands of GCs per galaxy, thus good statistical samples and big luminosity (mass) range

NGC 1407 Eridanus d=23 Mpc MV = -22.35

NGC 3258 Antlia 41 Mpc -21.87

NGC 3268 Antlia 41 Mpc -21.96

NGC 3348 CfA69 41 Mpc -22.13

NGC 4696 Centaurus 42 Mpc -23.31

NGC 7626 Pegasus I 49 Mpc -22.58

(M87 Virgo 16 Mpc -22.4)

D = 6 pc at d ~ 40 Mpc half-light profile width ~ 0.03”

compare PSF fwhm = 0.1” marginally resolved

HST/ACS Imaging program for BCGs

(Partial list – biggest GCSs out of 12 studied)

Photometric technique:

- Uniform catalog of detected objects with DAOPHOT

- Construct PSF from average of many bright starlike objects

- For each individual source, convolve PSF with “King30” model GC profile and vary D(model) to obtain best match (ISHAPE; Larsen 1999)

- finally, use fixed-aperture photometry corrected for profile width to obtain final magnitude in each band

S/N=441

fwhm a=1.3 px

b/a = 0.91

S/N=24

fwhm a=0.82 px

b/a = 0.50

S/N=108

starlike

ISHAPE sample fits

1 px = 0.05”

HST/ACS

Growth curves for simulated GC profiles convolved with PSF

ISHAPE solve for best-fit D

Measure magnitude through 2.5-px aperture, corrected back to the growth curve for a starlike profile

Simulations show that the systematically correct intrinsic D (FWHM of GC profile) is returned for D > 0.1 (PSF) (transition boundary from unresolved to marginally resolved)

Our regime

More tests …

Measured size a not affected by modestly elliptical shape

b/a, returned correctly for a > 0.1 PSF

S/N > 50 !!

Full, profile-corrected aperture photometry for 6 supergiant ellipticals Previous PSF-fitting data

(Harris && 2006)

Trend lines:

- blue and red?

- linear slope? or top end only?

- how steep?

N=12000 brighter

than MI = -8.

Largest sample in existence!

RMIX fits of bimodal gaussians within selected magnitude intervals: forces two modes into the solution, but (a) less affected by field contamination, (b) avoids the strong assumption imposed by a ‘linear fit’

Red sequence: no trend

Blue sequence: gradual changeover to MMR toward higher mass

Z ~ M0.3+-0.1

The top end: uni- or bi-modal?

A detour: the measured cluster sizes

Trends (?) versus galactocentric distance and metallicity: projection effects, or intrinsic?

Low-metallicity GCs average larger size at any galactocentric zone

The MMR is not due to an unaccounted-for size-mass relation.

What is responsible for the metallicity distribution function (MDF)?

Is a proto-GC

- PRE-enriched from the surrounding GMC gas?

- internally SELF-enriched by its own SNe within the first few Myr?

- stochastic? (can self-enrichment be responsible for the internal dispersion of the MDF?)

Input assumptions to self-enrichment model:

SNe from >8 M0 stars enrich lower-mass stars while still in formation

Salpeter IMF 0.3 100 M0 and SF efficiency f* ~ 0.3

Woosley/Weaver SN yields, and fraction fZ of heavy elements retained in GMC

c

ZZc M

MfZ )(log38.0]/[ ZffHm and thus

Bailin & Harris 2008

NSN ~ 1 per 100 M0Pre-enrichment

level for fZ = 0.08

2/1

510059.0

GC

c

Z M

Z

Internal dispersion of MDF due

to statistical variation in NSN

Stochastic self-enrichment fails to explain the MDF

dispersion at any cluster mass higher than 104 M0

Two additional, major factors to add:

- reff ~ M1/2 at high mass

- fZ is a strong function of M(init) and thus reff as well

Proto-GC = truncated isothermal sphere logarithmic potential (R). All SNe go off while PGC is still highly gaseous; all ejected energy absorbed and thermalized.

Gas will leave if outside an “escape radius” defined by total energy > potential energy at edge of cloud.

Ejecta become efficiently retained at a characteristic mass (after star formation)

MGM

rfEretainM effSN

GC7

2* 10

100)(

Combined effects of pre-enrichment, self-enrichment, and mass/radius relation

Match to BCG data for 6 galaxies

-pre-enrichment of each “mode” (blue, red) tuned to match mean color

-self-enrichment drives shape of mean MDF at high mass

Basic features of the model:

- No MMR for cluster masses < ~106 M0 (i.e., sequences vertical)

- Very metal-poor, very massive GCs should be rare (anywhere)

- blue and red sequence converge at high-mass end

- Similar red-sequence MMR should exist at top end, but smaller amplitude

- Internal dispersion and mean metallicity of each mode driven by pre-enrichment

Wehner && 2008