The Formation Histories and Mass Profiles of Some Elliptical Galaxies Steve Zepf Michigan State University Overview – Three Questions 1. Metal-poor globular clusters and early structure formation (the biasing of the metal-poor GC population) 2. When were the major formation epoch(s) of elliptical galaxies? (ages of their metal-rich GCs) 3. What are the properties of the dark matter halos around elliptical galaxies? (radial velocities of the globular clusters) topics left out – GC formation, dynamical evolution, low-mass X-ray binary GC conncetion Why Globular Cluster Systems? • GCs have ~106 stars within a few pc - remarkable densities - readily observable • some GCs are very old – hope of constraining very early structure formation. • GCs are true simple stellar populations (to first order), so determining their age and metallicity is much simpler than for the composite populations in integrated light. • GCs are observed to form in all major star formation events in galaxies. • GCs are found out to very large radii around galaxies – useful dynamical probes of galaxy halos. one example – NGC 3256 Zepf et al 1999 Another example – the Antennae, Whitmore et al. Why Globular Cluster Systems? • GCs have ~106 stars within a few pc - remarkable densities - readily observable • some GCs are very old – hope of constraining very early structure formation. • GCs are true simple stellar populations (to first order), so determining their age and metallicity is much simpler than for the composite populations in integrated light. • GCs are observed to form in all major star formation events in galaxies. • GCs are found out to very large radii around galaxies – useful dynamical probes of galaxy halos. Bimodality in GC Systems of Elliptical Galaxies • Distribution because individual GC colors/metallicities can be determined. • Early results from Zepf & Ashman 93 and subsequent work suggest bimodality in optical colors of GC systems of elliptical galaxies. • Large samples establish bimodality as the norm (>60%) for ellipticals, S0s consistent but less certain. e.g. Kundu & Whitmore 01 ~30 Ellipticals and ~30 S0s Color and Metallicity Distributions • For mostly old systems, color traces metallicity, use Milky Way and models to do this (age-met degeneracy still present). • Can’t get the two peaks in the observed colors from a singlepeaked metallicity distribution. GCs come in two metallicity groups . • Metal-rich GCs - closely tied to bulk of host galaxy - trace galaxy formation/assembly. • Metal-poor GCs - Low metals, extended spatial distribution, and old ages in Milky Way halo constraints on early structure formation? Note this division in abundance also seen in Galaxy GC system Formation Sites of Metal-Poor GCs Metal-poor GCs include the oldest surviving fossil structures. When and where they formed constrains early star formation in the universe. Formation redshift can be addressed by determining the “biasing” in GC numbers vs galaxy mass (Rhode & Zepf 2004, Rhode, Zepf, & Santos 2005). Basic idea (Santos 2003)… • More massive galaxies have a greater fraction of their mass collapsed at a given redshift than lower mass galaxies on average. • If metal-poor globular cluster formation has a preferred early formation epoch, then more massive galaxies will have more metal-poor GCs per stellar mass. The earlier the formation epoch, the greater the “bias” in the metal-poor GC population today. • Need observational determination of total number of metal-poor GCs in galaxies with a range of masses Rhode & Zepf 01,03,04 Observational Requirements • Both wide-field and deep imaging (many extant data only cover ~10% of the GC system). • Need colors to separate blue and red GCs to isolate metalpoor GCs for very early formation. Katherine Rhode’s PhD thesis at Yale. • Utilize large-area CCDs, multiple colors, new analysis techniques to make both these advances for sample of 4 nearby ellipticals and 9 spirals (ellipticals in Rhode & Zepf 2004, 2001, spirals in Rhode & Zepf 2003 and in prep). NGC 4472 – 34’ x 34’ field. Ground-based Mosaic image and GC identification Rhode & Zepf 2001, 2004; WFPC2 images in blue, ACS in red Results: Early-type Sample (Rhode & Zepf 2004, AJ, 127, 302) NGC 4594: 1748 GC candidates selected in BVR NGC 4406: 1400 GC candidates NGC 3379: 321 GC candidates Mosaic image of NGC 4594 FOV = 38’ x 37’, FWHM ~ 1” Results: Early-type Sample (Rhode & Zepf 2004) Cluster Es Field E/S0s Results: Early-type Sample (Rhode & Zepf 2004) Color Distributions • Two (or more) Gaussians fit better than one at >99.9% confidence • NGC 4472, NGC 4406, and NGC 4594 have 60% blue, 40% red GCs • NGC 3379 has 70% blue, 30% red Disagrees with collapse+accretion and multi-phase SF models, which predict larger proportions of blue GCs in more luminous galaxies Observational Results T NGC/109MSun Determined observationally for metal poor GCs in galaxies with a range of galaxy masses. Some biasing observed Rhode, Zepf, & Santos 2005 ApJL Formation Redshift Hierarchical model with Greg Bryan Most systematics tend to flatten the trend in the data. most likely not extremely early formation zform < ~11 Ages of Formation Epoch(s) of Elliptical Galaxies – Why “Metal-Rich” GC Systems? • Metal-rich GC systems trace elliptical galaxy populations spatially and in color. • GC formation observed in all starbursts in local universe – trace major formation episodes. • Each individual GC is a simple stellar population, most basic determination of ages and abundances. GC systems reveal distribution in age and metallicity of major formation events of ellipticals. Ages of the Formation Epoch(s) How to get ages of GC populations? Optical colors good for finding metallicity peaks, but bad for ages because of age-metallicity degeneracy – older age and higher metal content have same effect on optical colors (colors can be red because of old age or high metal content). Ideas… • luminosity function (assumes mass function) • Balmer absorption lines • optical to near-infrared colors efficient, effective Breaking Age-Metallicity Degeneracy with Near-IR to Optical Photometry I-H color primarily sensitive to metallicity, g-I color more sensitive to age. True of any set of optical to near-infrared color vs optical color – consequence of basic stellar properties. Difference between 3 Gyr (int. age) and 13 Gyr (old age) is about 0.3 mag. NGC 3115, NGC 4365 in VIK First try with this technique – Puzia, Zepf, Kissler-Patig, Hilker, & Minniti 2002 NGC 3115 – primarily consistent with typically old GCs NGC 4365 – surprise! – many GCs with VIK colors only explained by intermediate ages. Important result indicating some elliptical galaxies had substantial formation activity at t <~ 5 Gyr. Implications Not without controversy • Most galaxy light studies suggest older ages for integrated stellar light of NGC 4365. • Spectroscopy of GCs in NGC 4365 confused – initial confirmation of int. age by Larsen et al. 2003, similar data from same instrument, different result, Brodie et al. 2005. • Only case of possible inconsistency between optical to near-ir colors and optical spectroscopy for GC systems (agree for NGC 1316, NGC 1399, NGC 3115, NGC 4472) Get better data! Three deep NIC3 pointings for NGC 4365, very high S/N New NICMOS data (Kundu, Zepf, Hempel et al 2005 ApJ, 634, L41 ) Independent, deep data – data deep enough to separate optical and near-ir colors. Same result as before – substantial population of NGC 4365 GCs with optical/near-ir colors only accounted for with int. age. Comparison with other archival NIC3 data for NGC 1399 GCs. Mostly old GCs in NGC 1399, with a few int. ages, same as seen in previous spectroscopy. Simple Fraction of Int. Age GCs Simple test, adopts two populations, one at 13 Gyr, gets best fitting age and fraction of 2nd population. NGC 1399 result of ~10% int. age population same as seen in small spectroscopy sample. NGC 4365 substantial int. age result confirmed Is it age? • Test colors by getting much better, independent data. Deep NICMOS H-band imaging of NGC 4365. Much reduced photometric uncertainties. Colors confirmed, intermediate ages show up in independent, deep data Is it age? • Test colors by getting much better, independent data. Deep NICMOS H-band imaging of NGC 4365. Much reduced photometric uncertainties. Colors confirmed, intermediate ages show up in independent, deep data • What about the stellar populations models? M87 – deep NIC3 GO and AR data Kundu, Zepf & Hempel 2006 New NIC3 data in center, archival NIC data in outer field with new ACS data. Models lines appear to be mostly correct. Dominant old age in agreement with optical color bimodality and spectroscopic results for M87 Is it age? • Test colors by getting much better, independent data. Deep NICMOS H-band imaging of NGC 4365. Much reduced photometric uncertainties. Colors confirmed, intermediate show up in independent, deep data • What about the stellar populations models? Other galaxies match old model prediction lines very well. Can’t shift model lines. More NICMOS Results (Kundu, Zepf, Hempel 2006 in prep) Comparison of same NIC data for NGC 4365 and M87 reveals clear difference – only known way to account for this is intermediate age for substantial fraction of NGC 4365 GCs. Is it age? • Test colors by getting much better, independent data. Deep NICMOS H-band imaging of NGC 4365. Much reduced photometric uncertainties. Colors confirmed, intermediate show up in independent, deep data • What about the stellar populations models? Other galaxies match old model prediction lines very well. Can’t shift model lines. only extant explanation that works for NGC 4365 optical to near-ir is intermediate age. • Optical to near-ir color technique is efficient and effective. What about a galaxy sample? Optical to near-ir to date (Hempel, Kundu, Puzia, K-P, Geisler, Zepf) Sample in words NGC 4472 – old NGC 3115, NGC 7192, M87 – old with possible “frosting” NGC 1399 – old with definite ~10% intermediate age NGC 5846 – probable substantial int. age component NGC 4365 – clear intermediate age component Data in hand – Cen A, NGC 3379, NGC 4594 Notes: old for colors and spectra just means t > ~8-10 Gyr, z >~1 Tests higher metallicity GCs, not lower metallicity ones. Expect higher fraction of younger ones at brighter limits. Stochastic effects from finite number of stars come into play. Hempel, Kundu, Puzia, Kissler-Patig, Geisler, Zepf, et al. Overview of Ages • Current data indicate a modest but significant fraction of Es have substantial intermediate-age GC population in metal-rich population (major formation even at z < 1). Majority have major formation events at z > 1. Of this majority many have small population of intermediate-age GCs (e.g. NGC 1399). • Overall seems consistent with hierarchical merging – some merging today, increasing significantly to the past. • Large optical, near-ir GCS survey enable much more detailed tests (NIC3, SOAR,...). Combine with targeted optical spectroscopy. Kinematics of Globular Cluster Systems Address two questions 1. Dark matter around elliptical galaxies – how much, spatial distribution? Globular clusters one of best probes way out into the halos. 2. Angular momentum in elliptical galaxies. Why don’t they have any (in their inner regions anyway). Search for angular momentum transported outwards predicted by models. The Utility of GCs for Halo Kinematics Globular Clusters provide observable individual objects at very large radii for which the sky dominates the integrated light. e.g. NGC4472 giant Virgo E Photometry KPNO Mosaic Camera (Rhode & Zepf 2001, 2004) Obviously useful as dynamical probes of halo mass distribution Do Lower Luminosity Ellipticals have Expected Dark Matter Halos? • A few recent results suggest maybe lower luminosity ellipticals are lacking dark matter halos or have halos with either low concentration or modest M/L (e.g. Romanowsky et al). • Test this with VLT multi-fiber observations of globular clusters over a very wide field around best case L* elliptical galaxy, NGC 3379. 1. Confirm or reject previous result. 2. Extend kinematical information to larger radii. Results for NGC 3379 Bergond, Zepf, Romanowsky, Rhode, & Sharples 2006, A&A, 448, 155 • Red points and error bars GC dispersions with radius. • Models: CDM-like halos with standard concentrations in blue hatched region. Blue dashed lines region covered by halo occupation models. Purple line best fit from outer HI ring and CDM halo with most likely concentration. Bottom dashed line is mass traces light. GC velocity dispersions are in agreement with CDMlike halos. Angular Momentum in Galaxies • General idea – protogalaxies torqued up by tidal interactions with other protogalaxies, have “spin parameter” 0.05 ( = J E^1/2 / G M^1/2), goes back to Hoyle, confirmed in modern cosmological simulations, mostly independent of mass and environment. • Halos to galaxies – Baryons that make up the parts of the galaxies we see cool, collapse, and spin up. • Similar densities of ellipticals and spirals suggest ellipticals and spirals collapsed by roughly the same amount. But ellipticals have an order of magnitude or more less angular momentum. What happened to ellipticals? • Need to transfer angular momentum. This has long been realized, as has the utility of hierarchical structure formation and merging to transfer ang momentum. • Can we find the angular momentum at large radii around ellipticals? Kinematics of GC Systems NGC 4472 • Our new VLT Flames data for NGC 4472, up to ~350 velocities out to ~ 75 kpc, builds on 144 GCs (Zepf et al 2000, Sharples et al 1998) to ~ 40 kpc, and another 100 over same area from Cote et al. (2003). Extant results (Z2000 et al) • not much evidence for rotation anywhere (to ~50 kpc). • Metal-poor GC population may have modest rotation. Higher than more spatially concentrated metal-rich GCs. • No evidence for rotation in metal-rich GC population about any axis - Upper limit on v/ < 0.34 (99%) Metal-rich GC pop not rotationally supported. Significant angular momentum transport required. Implicates mergers, But where did it go? • M87 consistent with 4472, evidence for ang mom in outer regions, M104 may have counter-rotating outer GCs. Kinematics of GC Systems NGC 4472 • Our new VLT Flames data for NGC 4472, up to ~350 velocities out to ~ 75 kpc, builds on 144 GCs (Zepf et al 2000, Sharples et al 1998) to ~ 40 kpc, and another 100 over same area from Cote et al. (2003). Extant results (Z2000 et al) • not much evidence for rotation anywhere (to ~50 kpc). • Metal-poor GC population may have modest rotation. Higher than more spatially concentrated metal-rich GCs. • No evidence for rotation in metal-rich GC population about any axis - Upper limit on v/ < 0.34 (99%) Metal-rich GC pop not rotationally supported. Significant angular momentum transport required. Implicates mergers, But where did it go? • M87 consistent with 4472, evidence for ang mom in outer regions, M104 may have counter-rotating outer GCs. SUMMARY • Globular Cluster Systems show the episodic formation history of their host elliptical galaxies. • Not all of these episodes are old (intermediate-age GC populations in normal ellipticals). Broad agreement with hierarchical merging models. • Optical to near-infrared colors of GC systems efficient, straightforward way to constrain ages of major formation episodes. • Metal-poor GCs are “biased” with galaxy mass, potential to constrain their formation sites/epochs. • Dark matter halos ubiquitous around elliptical galaxies. • Angular momentum transport needed for elliptical galaxy formation mergers. Out to large radii in at least one wellstudied case. NGC 4472 – new data Hempel, Zepf, Kundu, & Geisler 2006 Ongoing survey, NGC 4472 results show old ages in one of strongest bimodal cases, as expected. Insensitivity to Models Models on previous page Bruzual-Charlot, here Anders & Fritze v. Alvensleben and Maraston. Basically same answers. Optical to near-ir approach is astrophysically straightforward. NGC 4365 – photo and spect. Kundu, Zepf et al optical to near-ir colors of objects with spectroscopy. Note optical to near-ir data around re , B05 at large r . Mass Profile of NGC 4472 Derived from • Spherical Jeans eqn and isotropic orbits • smoothed distribution • Observed GCS profile Compared to • Mass profile from hot gas observed in X-rays (Irwin & Sarazin 1997) Hydrostatic equilibrium for gas and isotropic orbits for stars give the same answer within the errors. (M/L)B 12 at 34 kpc, increases roughly as sqrt (r) to larger radii. Potential of near-IR photometry V-I colors have more age sensitivity than V-K colors V-I, V-K can break age-metallicity degeneracy. Plot on the right shows the Milky Way and M31 GCs and various models for 12 Gyr and a range of [Fe/H] Younger ages will have blue V-I for a given VK. Note good agreement of models with CMD ages. Low-Mass X-ray Binary Globular Cluster Connection • A LMXB is a compact object accreting from a companion low-mass star. • LMXBs are X-ray bright – typical L ~ 1037 ergs/s Extragalactic obs are more typically 1038 • In Milky Way, only ~13 LMXBs in Galactic GCs. This is 10% of total LMXB population, even though less than 0.1% of the stars are in GCs. 1) Need more numbers for stats and 2) LMXBs are clearly over-represented in GCs – dynamical formation in dense environment. Great progress in extragalactic studies Much of the X-ray emission from early-type galaxies is from LMXBs Many (~50%) of the LMXBs are in globular clusters NGC 4472 as an example Chandra image in Grayscale HST-WFPC2 dark outlines Chandra sources circled, black circles are matches with GCs found with HST. Kundu, Maccarone, & Zepf 2002, ApJL What observables determine LMXB presence? Kundu, Maccarone, & Zepf 2002, Sarazin et al. 2003, MKZ03 KMZP03, Jordan et al. 2004, KMZ 2004… • Globular cluster luminosity (mass) – probability of a LMXB scales roughly linearly with GC mass • Globular cluster color (metallicity) – probability of a LMXB roughly 3x larger for red (metal-rich) GCs relative to blue (metal-poor) GCs Statistically marginal correlations with globular cluster size and distance from galaxy center. Metallicity effect unexpected, what’s the evidence? NGC 4472 – KMZ02 Small points globular clusters Dark Points – LMXB – GC matches Shows LMXBs preferentially in higher L,redder GCs. Metallicity effect at very high significance. KMZ 2002 Another look and another galaxy (KMZ02, KMZP03) More galaxies…. (KMZ04) And another one…. (KMZ04) What About Age? - No Strong Age Effect in NGC 4365 Key is intermediate age globular cluster population in NGC 4365 (discovered using optical to near-ir photometry, will go deeper with NIC3). Known broad age range allows test of effect of age. GCs open circles, red dots GCs with LMXbs Result is no significant age effect! Kundu, Maccarone, Zepf, & Puzia 2003, ApJL Possible Explanation – Irradiation Induced Winds (Maccarone, Kundu, & Zepf 2004) • X-rays heat mass donor star. Energy can be dissipated by line cooling or winds. • Line cooling dominates for metal-rich stars, winds for metal-poor stars (Iben, Tutukov, & Federova 1997). • Lifetime then varies with metallicity - more winds means less mass to accrete and shorter lifetime, so fewer X-ray sources in metal poor systems. • Also consistent with observed spectral differences with metallicity (e.g. Maccarone, Kundu, & Zepf 2003, Irwin & Bregman 1999). LMXB- Globular Cluster Review 1. Much of the X-ray emission in early-type galaxies comes from Low-Mass X-ray Binaries (LMXBs), and many (~50%) of the LMXBs are in globular clusters. 2. Metal-rich globular clusters are 3x as likely as otherwise similar metal-poor clusters to host LMXBs (Kundu, Maccarone, & Zepf 2002, MKZ2004). 3. Age is not a significant parameter determining whether or not a LMXB is found in a globular cluster (Kundu, Maccarone, Zepf, & Puzia 2003). 4. An irradiation induced wind model can account for the strong metallicity dependence and lack of an age dependence (Maccarone, Kundu, & Zepf 2004). An extragalactic globular cluster Spatial Distribution of GCs by Color Because of dissipation and enrichment in progenitor disks and during the merger itself, the metal-rich red GCs should be more concentrated towards the center of the galaxy than the metal-poor blue GCs. Confirmed by Geisler et al in 96, many other cases now. red blue M87, Kundu & Zepf 2004 Dynamical Information Velocities for 144 GCs around NGC 4472 (Zepf et al. 2000, Beasley et al. 2000, Sharples et al 1998) • From CFHT and WHT – note wide-field is valuable, GCs available to very large radii • These data extend to ~ 8´, factor of 4 larger than previous velocity data. (half-light radius of galaxy light is 1’ ~4 kpc, so 8’ ~32 kpc) How do Globular Clusters Form? First: PRESSURE – High pressure in the interstellar medium of starbursts can account for the formation of dense star clusters (Elmegreen & Efremov, Ashman & Zepf 2001). r M1/2 P-1/4 Molecular clouds that form in the high pressure environment of starbursts have to be very dense in order to exist. Second: SIZES – Zepf et al. 99 found little or no correlation between size (radius) and mass (r M0.1±0.1) for young clusters in NGC 3256 . But virialized clouds have r M1/2 P-1/4 This is observed for molecular clouds in the galaxy, and hard to avoid for virialized clouds. So how do you erase the r M1/2 ? Not easy! Best hope seems to be a star formation efficiency that is dependent on the Binding Energy of the parent cloud. In this case, low mass clouds are inefficient, lose mass, and expand. Fine line to keep numbers of low mass GCs around. See AZ01 for details! Angular Momentum in Galaxies • Where is the angular momentum in elliptical galaxies? • General idea – protogalaxies torqued up by tidal interactions with other protogalaxies, have a dimensionless spin parameter 0.05 . Goes back to Hoyle, confirmed in modern cosmological simulations, mostly independent of mass and environment. • Halos to galaxies – Baryons that make up the parts of the galaxies we see cool, collapse, and spin up. Spin-up works to zeroth order for spirals. What about Balmer Lines? Natural first step - carried out for - M87 GCs by Cohen et al - NGC 1399 GCs Kissler-Patig et al, Forbes et al. - NGC 4472 GCs by Beasley et al BUT Challenges with calibration for Galactic GCs - when the same techniques used for extragalactic GCs are applied to metal-rich Galactic GCs, get ages ~25% older than CMD ages (e.g. 47 Tuc, Gibson et al, Vazdekis). Also requires a long time on a 8/10-m telescope to get reasonable signal, so even when systematics are sorted out it is a painful way to go. Dynamical Information Velocities for 144 GCs around NGC 4472 (Zepf et al. 2000, also Beasley et al. 2000, Sharples et al 1998, Cote et al. 2003) • These data extend to ~ 8´, factor of 4 larger than previous velocity data. (half-light radius of galaxy light is 1’ ~4 kpc, so 8’ ~32 kpc) • New FLAMES data, another 100 GCs to ~ 15’ Which is the bimodal Milky Way? • One plot is the color distribution of the Galaxy GCS, the other is NGC 4406, a modestly bimodal elliptical galaxy GCS – which one is which? • Milky Way GCS known to be bimodal directly in metallicity. • Given bimodal Milky Way metallicity distribution, can NGC 4406 and most other ellipticals be so different? Plot from Katherine Rhode, data Rhode & Zepf 2004 Which is the bimodal Milky Way? • One plot is the color distribution of the Galaxy GCS, the other is NGC 4406, a modestly bimodal elliptical galaxy GCS – which one is which? • Milky Way GCS known to be bimodal directly in metallicity. • Given bimodal Milky Way metallicity distribution, can NGC 4406 and most other ellipticals not be similar? Plot from Katherine Rhode, data Rhode & Zepf 2004 Color and Metallicity Distributions • For mostly old systems, color traces metallicity, use Milky Way and models to do this (age-met degeneracy still present). • Can’t get the two peaks in the observed colors from a singlepeaked metallicity distribution. GCs come in two metallicity groups . • Metal-rich GCs - closely tied to bulk of host galaxy - trace galaxy formation/assembly. • Metal-poor GCs - Low metals, extended spatial distribution, and old ages in Milky Way halo constraints on early structure formation? Note this division in abundance also seen in Galaxy GC system