Deriving Star Formation Histories of Local Group Galaxies: The Magellanic Clouds

Jon Holtzman, New Mexico State University

• Collaborators:
  • J. Gallagher (U Wisc)
  • J.R. Mould (MSSSO)
  • M. Geha (UCSC)
  • A. Cole (U. Wisc.)

• What comprises a star formation history? Time history of:
  • Star formation rate
  • Metallicity distribution
  • IMF
  • Dynamical mixing and merger history

• Motivation:
  • Comparison with high redshift observations, e.g. SF history of the Universe
  • Understand star/galaxy formation by looking at similarities/differences between SF history in different locations within a galaxy, proximity to other galaxies, etc.
  • Constrain numerical models of SF, as modellers begin to be able to create ``mock'' Local Groups

• Specific questions:
  • Is the IMF universal?
  • Is star formation bursty? What might trigger this?
  • Is there more than one mode of star formation? IMF, cluster/field, etc.
  • How does SF history vary with location within galaxies, and why?
  • How important is gas inflow/outflow? Is it a function of galaxy type/ environment?

• Magellanic Clouds
  

  • Unique opportunities of Magellanic Clouds:
    • Nearby: get SF history both from brighter evolved stars and from less evolved stars; test results from brighter stars alone
    • Can reach unevolved main sequence stars, leading to initial mass function more directly
    • Contain populations of star clusters to probe whether cluster formation traces field star formation
    • Test SF history in environment which interactions are as strong as they are likely to get
    • Relatively rich, giving better number statistics (although not necessarily for fixed field size)

  • Previous results on SF history of the Clouds
    • Cluster age distribution. LMC star cluster age distribution suggest strongly episodic star formation. SMC clusters more continuously distributed in age.
      

    • In LMC, relative numbers of upper ms stars, giants suggest predominantly young (< 4 Gyr) field population (Bertelli et al. 1992). SMC field star studies quite sparse: ground-based data in outer fields suggest significant component of older population (Gardiner & Hatzidimitriou), leading to general impression that SMC stars are older than LMC stars. HST results presented below challenge these ideas.

    • SF histories constrained to some extent by recent models of chemical evolution (Pagel and Tautvaisiene 1998)
      

    • Location dependence of SF history:
      • Ground-based results (Vallenari et al.) suggest variation across LMC, perhaps systematically.

      • However, from HST, three ``outer'' fields have comparable CMDs (Geha et al. 1998)

      • Some degree of mixing for given galactocentric radius seems likely based on kinematics

      • Elson et al. (1997) suggest bar formed more recently than outer disk, $\sim$ 1 Gyr ago, based on CMD morphology (but listen further!)

      • More remote field appears older (Walker et al.)

      • SMC data suggest gradient in stellar population, with younger component more concentrated towards bar

• New approach: use deep HST color-magnitude diagrams to probe SF history using main sequence stars
  

• Determining star formation histories

  • Coarse/qualitative SF history can be determined from presence/absence of key types of stars: horizontal branch stars, RR Lyraes, red clump stars, AGB stars, carbon stars, upper main sequence stars, supergiants, distinct main sequence turnoffs, etc.

  • Quantitative analysis can be important. Structure in a Hess diagram does not necessarily imply bursts! (Continuous SF CMD)

  • In principle, can derive full SF histories based on locations of stars in CMDs. Basic principle: what combination of simple stellar pops provide best match to observed data? Need to define best match. Also important to consider whether best match is in fact a good match.

  • Deriving SF history complicated by possible uncertainties in stellar models, photometric calibration, reddening, distance, binarity. LF more robust, but non-unique.

  • Sensitivity to age decreases as age increases.
    

    Age-metallicity relation makes these problems more severe for older ages.

    

  • Various groups have been working on these techniques (Tolstoy et al.; Gallart, Aparicio et al.; Dolphin; Ng; Hurley-Keller, Mateo et al.; Han; Hernandez, Valls-Gabaud & Gilmore; Holtzman et al.). Detailed approaches differ (results too??)

  • Philosophy: fit for everything vs external constraints.

  • Healthy skepticism required. Comparison of results from different techniques limited so far (Carina)

  • Differences in SF history between different locations are easier.

  • Broad-brush SF histories vs. detailed. What do you want to know?

• LMC Results

  • HST data obtained in F555W/F814W ($\sim$ 4000s per filter) in three fields: two outer, one bar. Other data also available (Elson et al, Olsen, Walker, Smecker-Hane, others...)

  • HST allows observations of stars to M $\sim$ 0.7M_sun

  • lower main sequence (HST) constrains IMF without assumptions about SF history. Unfortunately, older data not quite deep enough, IMF slope constrained within -3 < $\alpha$ < - 2.1 (Holtzman et al 1997), but new data soon available.
    • Note: IMF has now been measured in systems covering 2 dex in metallicity with no obvious variations:
      • Draco (Grillmair et al.)
      • Ursa Minor (Wyse et al.)
      • SMC (Holtzman et al.)
      • LMC (Holtzman et al.)
      • Galactic Bulge (Holtzman et al.)

  • History of SF rate : HST color magnitude diagrams from outer fields suggest significant component of ``older'' stars, i.e. $\sim$ 50% of stars older than 4 Gyr, based on luminosity function (Holtzman et al. 1997). However, LF solution not unique; steeper IMF with younger population also allowed (plus probably other SF histories!)
    

  • Easier to reject proposed SF histories than to derive a correct one!
    

  • Derived SF histories from full CMD, assuming observed age-metallicity relation and $\alpha$ = - 2.35, qualitatively comparable to those derived from LF:
    • LMC outer fields
      

    • LMC bar
      

    • ``older'' component of stars still needed.

    • Metallicity distributions are not totally implausible (although only for low reddening if stellar models and calibration are correct).

    • Best fits obtained for m-M=18.5 and E(B-V) = 0.07 (bar) and E(B-V) = 0.04 (outer)

  • Fits with single-valued age-metallicity relation, however, are poor in absolute sense. Discrepancies with observations plausibly explained by missing metallicity dispersion in models
    
    • Remarkably, age-metallicity relation is recovered without any assumptions!
      

    • Note that these solution still assume discrete metallicities and thus are still not completely realistic

  • Full fits to CMD, as well as CMD differences between field and bar suggest bar is, on average, older

• SMC results
  • HST data obtained in F555W/F814W ($\sim$ 4000s per filter) in two fields

  • As in LMC, metallicity spread at a given age must be included to fit CMDs. Again, CMD analysis recovers something like the age- metallicity relation of the star clusters.
    

  • To first order, SF history in SMC inner regions comparable to that of the LMC
    

  • Best fits obtained for m-M=18.8 and E(B-V) = 0.06

• Implications
  • Metallicity dispersion must be included in derivation of star formation history, also in chemical evolution models

  • Realistic errors on SF rate relatively large if systematic errors are included, and correlated
    

  • Current results make crude predictions for metallicity distributions. Note difference between overall metallicity distribution and metallicity distribution of giants. Independent metallicity determinations will significantly help to confirm validity of derived SF histories and constrain them

  • Does field SF trace cluster SF? PROBABLY NOT
    
    Are there different modes of SF? MAYBE

  • To what extent is SF bursty? MAYBE NOT SO MUCH

  • Do the SF histories of the two clouds differ significantly? MAYBE NOT
    
    Is star formation in the Clouds triggered by mutual interactions or interaction with the Milky Way?

  • Is the IMF variable? MAYBE NOT!

• Directions
  • Uniform compilation/analysis of data sets: to what extent is SF bursty?
    
  • Independent metallicity determinations
  • Couple chemical evolution models with CMD information
  • To study Local Group, need to get past dwarf galaxies: Milky Way, M31, M32
  • More distant objects: gEs
  • SF histories in Local Group likely provides best place to tune star formation parameterizations in hydrodynamical cosmological simulations.