The Origin of Multiple Populations within Stellar Clusters An Unsolved Problem Nate Bastian (Liverpool, LJMU) Ivan Cabrera-Ziri (LJMU), Katie Hollyhead (LJMU), Florian Niederhofer (LMU/ESO)
V (brightness) old, > 10 Gyr metal poor V-I (colour)
V (brightness) Some of the oldest luminous objects in the Universe Globular cluster formation intimately linked with galaxy formation V-I old, > 10 Gyr metal poor (colour)
Globular Clusters as Tools: GCs have been used to... Determine the structure of the Milky Way (Shapely 1918) Understand (and calibrate) stellar evolution and stellar populations (Eddington 1926) Constrain the formation and evolution of the Milky Way (Searle & Zinn 1978)...as well as other nearby galaxies (Brodie & Strader 2006) Find and study stellar exotica
Introduction to Stellar Clusters Pleiades Open Clusters few - 10 4 Msun few Myr - few Gyr ~solar metallicity disk of the Galaxy Globular Clusters 10 4-10 6 Msun 10-13 Gyr low metallicity bulge/halo of the Galaxy M80
Clusters historically viewed as simple stellar populations All stars have the same age (very small spread, < 1-2 Myr) All stars have the same abundances Range of stellar massses (and potentially rotation rates) Quintesential simple stellar populations GCs could only form in the special conditions of the early Universe
Surprise #1 V V-I
Surprise #1 V V-I
Surprise #1 I V multiple populations B-I V-I
Surprise #2 few million Msun ~12 Gyr Omega Cen - ESO
Surprise #2 few million Msun ~12 Gyr Omega Cen - ESO Antennae colliding galaxies - HST
Surprise #2 few million Msun ~12 Gyr few million Msun ~7 Myr Omega Cen - ESO Antennae colliding galaxies - HST
Surprise #2 few million Msun ~12 Gyr similar masses & sizes (densities) few million Msun ~7 Myr Omega Cen - ESO Antennae colliding galaxies - HST
Surprise #2 few million Msun ~12 Gyr globular clusters are similar masses & sizes still forming in the local (densities) universe few million Msun ~7 Myr Omega Cen - ESO Antennae colliding galaxies - HST
Young Massive Clusters (YMCs) NGC 34 Schweizer & Seitzer 2007 Cabrera-Ziri et al. 2014 Maraston et al. 2004 Bastian et al. 2006 NGC 7252
Young Massive Clusters (YMCs) NGC 34 Schweizer & Seitzer 2007 Cabrera-Ziri et al. 2014 ~400 Myr 10 8 Msun ~400 Myr 10 7 Msun ~100 Myr 10 7 Msun Maraston et al. 2004 Bastian et al. 2006 NGC 7252
Young Massive Clusters (YMCs) NGC 34 Schweizer & Seitzer 2007 Cabrera-Ziri et al. 2014 ~15 Myr 10 6 Msun ~400 Myr 10 8 Msun ~400 Myr 10 7 Msun ~100 Myr 10 7 Msun NGC 1705 Maraston et al. 2004 Bastian et al. 2006 NGC 7252
Globular clusters are not simple stellar populations Milone et al. 2013 Cordero et al. 2014 UV [Na/Fe] NGC 6752 47 Tuc UV - U Mulitple (spread) CMD features [O/Fe] Chemical Spreads All globulars show anomalies, but all differ in the details
NGC 2808 I ~60% of stars on nominal main sequence ( first generation ) ~30% of stars are He enriched ( second generation ) Piotto et et al. 2007 not due to age or metallicity differences, only He abundance can explain it B-I
D Antona 2012 (Vatican Observatory Lectures) enriched primordial Generally, elements affected by hot hydrogen burning show deviations. Not elements related to SNe.
D Antona 2012 (Vatican Observatory Lectures)
Relation to the field Martell et al. 2011 Stars in GCs know about where they form [Na/Fe] 47 Tuc [O/Fe]
Relation to the field Martell et al. 2011 Stars in GCs know about where they form [Na/Fe]??? ~3% of halo stars ~50% of cluster stars ~97% of halo stars ~50% of cluster stars 47 Tuc [O/Fe]
Observables Na-O anti-correlation Large Al spread, small (or no) Mg spread. Some spread in other light elements (e.g C, N). Little/no spread in Fe. Discrete/spread main sequences/turn-offs, sub-giant branches, presumably due to discrete He abundances and/or CNO abudances Found in red (metal rich) and blue (metal poor) clusters Sources of the enriched material Only certain stars produce the right abudances. SNe can t do it. AGB stars (3-8 Msun - although wrong Na-O correlation), Rapidly rotating high mass stars, interacting massive binaries, extremely massive stars
Models of multiple populations in GCs Multiple epochs of star formation The ejecta of 1st generation stars, mixes with primordial material and forms a 2nd generation Only certain stars produce the correct abundance ratios - AGBs, massive stars Extremely ad-hoc, many adjustable parameters, mix of theory and fixes to fit observations Generally assume that a cluster can hold on to gas expelled from stars (and accrete new gas at just the right time) for 10s to 100s of Myr (Bekki & Norris 2006, Decressin et al. 2007, D Ercole et al. 2008, Vesperini et al. 2009, Conroy & Spergel 2011, Krause et al. 2013)
AGB scenario The 1st generation forms T > 30 Myr, AGBs begin shedding material which collects in the centre of the cluster The cluster accretes (a lot) of pristine gas from the surroundings. The 2nd generation forms in the cluster center Most of the 1st generation (~95%) is lost
Predictions of the AGB scenario The 2nd generation is more centrally concentrated in order to have enough material to form the 2nd generation, GCs must have been 10-100 times more massive that presently seen (mass budget problem) clusters can retain ejecta and accreted gas for long periods Massive clusters should show an age spread (or multiple bursts) - i.e. we should be able to find young massive clusters (>10 Myr) with ongoing star formation
Predictions of the AGB scenario The 2nd generation is more centrally concentrated -no, quite complicated profiles (Larsen et al. 2015, Bastian et al. in prep) in order to have enough material to form the 2nd generation, GCs must have been 10-100 times more massive that presently seen (mass budget problem) clusters can retain ejecta and accreted gas for long periods Massive clusters should show an age spread (or multiple bursts) - i.e. we should be able to find young massive clusters (>10 Myr) with ongoing star formation
Predictions of the AGB scenario The 2nd generation is more centrally concentrated -no, quite complicated profiles (Larsen et al. 2015, Bastian et al. in prep) in order to have enough material to form the 2nd generation, GCs must have been 10-100 times more massive that presently seen (mass budget problem) -directly contradicted by observations (Larsen et al. 2012, 2014) clusters can retain ejecta and accreted gas for long periods Massive clusters should show an age spread (or multiple bursts) - i.e. we should be able to find young massive clusters (>10 Myr) with ongoing star formation
Predictions of the AGB scenario The 2nd generation is more centrally concentrated -no, quite complicated profiles (Larsen et al. 2015, Bastian et al. in prep) in order to have enough material to form the 2nd generation, GCs must have been 10-100 times more massive that presently seen (mass budget problem) -directly contradicted by observations (Larsen et al. 2012, 2014) clusters can retain ejecta and accreted gas for long periods -not observed (Bastian & Strader 2014, Cabrera-Ziri et al. 2015) Massive clusters should show an age spread (or multiple bursts) - i.e. we should be able to find young massive clusters (>10 Myr) with ongoing star formation
Predictions of the AGB scenario The 2nd generation is more centrally concentrated -no, quite complicated profiles (Larsen et al. 2015, Bastian et al. in prep) in order to have enough material to form the 2nd generation, GCs must have been 10-100 times more massive that presently seen (mass budget problem) -directly contradicted by observations (Larsen et al. 2012, 2014) clusters can retain ejecta and accreted gas for long periods -not observed (Bastian & Strader 2014, Cabrera-Ziri et al. 2015) Massive clusters should show an age spread (or multiple bursts) - i.e. we should be able to find young massive clusters (>10 Myr) with ongoing star formation let s see...
Are Young Massive Clusters the Same as Globular Clusters? While Globular Cluster formation (at high-z) may have been fundamentally different from massive clusters forming today, all main theories for the origin of multiple populations predict that it should be happening in young clusters today. i.e. current theories do not invoke any special conditions/physics for GC formation.
Evidence for extended star formation histories in other clusters? 140 clusters with ages between 10-1000 Myr and masses between 10 4-10 8 Msun, from the literature, with integrated optical spectroscopy or resolved stellar photometry clusters in spirals, dwarfs, starbursts/mergers Look for emission associated with the clusters (Hβ, O[III]) or O-stars in the CMD Peacock et al. 2013, Bastian et al. 2013a
Bastian et al. 2009
Bastian et al. 2009
Bastian et al. 2009
Range where the AGB scenario expects the 2nd generation to be forming (30-200 Myr) Conroy & Spergel (2011) No ongoing starformation detected in any cluster Bastian et al. 2013a Given the relative sketchiness of the scenarios put forward, there is not an agreed upon age (or even continuous vs. discreet bursts) where the 2nd generation should form. The obs here rule out continuous (>7 Myr duration) and disfavour discreet bursts.
Star formation history of clusters NGC 34 Cluster 1 Schweitzer & Seitzer 2007
Star formation history of clusters NGC 34 Cluster 1 Schweitzer & Seitzer 2007
Star formation history of clusters NGC 34 Cluster 1 Single population - 100 Myr 2 x 10 7 Msun Cabrera-Ziri, NB et al. 2014 Schweitzer & Seitzer 2007
Star formation history of clusters NGC 7252: W3 570 Myr, ~10 8 Msun no star-formation for the past ~400 Myr Cabrera-Ziri et al. 2015b, in prep. 5 more clusters under study to sample the full 30-200 Myr range
ALMA observations of the Antennae Cabrera-Ziri, NB, et al. 2015 Whitmore et al. 2014 50-200 Myr 1-3 * 10 6 Msun No gas detected (<1-10% cluster mass)
Summary of young clusters Appear to be gas free at young ages (<2 Myr) -a problem for the FRMS scenario - Hollyhead et al. 2015 No evidence for ongoing star formation in *any* young massive cluster studied to date (older than 10 Myr) Integrated spectroscopy of YMCs (>10 7 Msun) shows no evidence for multiple bursts or extended SFHs (Cabrera-Ziri et al. 2014, 2015b) No gas/dust found in YMCs, which would be required to form further generations of stars Bastian & Strader 2014; Cabrera-Ziri et al. 2015a Models with multiple star formation events are disfavoured by observations of YMCs Niederhofer et al. 2015 Previous (popular) models do not agree with observations of YMCs
New Model: The Early Disc Accretion Scenario (Bastian et al. 2013) Single star-formation event Uses the nuclear burning products of high mass stars and accretes them onto some low-mass stars within the clusters Matches observations of YMCs
New Model: The Early Disc Accretion Scenario (Bastian et al. 2013) Single star-formation event Uses the nuclear burning products of high mass stars and accretes them onto some low-mass stars within the clusters Matches observations of YMCs
Can Self-Enrichment Scenarios Work? Basic Abundance Predictions Observations now routinely measure Na, O and He (Y) spreads in GCs. Can self-erichment scenarios match the observations?
Can Self-Enrichment Scenarios Work? Basic Abundance Predictions For a given position in Na-O space, dilution models give a direct prediction of He (Y) spreads Bastian, Cabrera-Ziri, Salaris 2015
Can Self-Enrichment Scenarios Work? Basic Abundance Predictions For a given position in Na-O space, dilution models give a direct prediction of He (Y) spreads AGB ejecta Y=0.38 Pristine material Y=0.25 Bastian, Cabrera-Ziri, Salaris 2015
Can Self-Enrichment Scenarios Work? Basic Abundance Predictions Constant He value For a given position in Na-O space, dilution models give a direct prediction of He (Y) spreads AGB ejecta Y=0.38 Pristine material Y=0.25 Bastian, Cabrera-Ziri, Salaris 2015
Can Self-Enrichment Scenarios Work? Basic Abundance Predictions Allowed spread based on He AGB stars overproduce He Expected He spread of 0.10 based on Na/O Observations Bastian, Cabrera-Ziri, Salaris 2015
Disagreement with observations of GCs
AGBs AGBs FRMS Binaries Bastian, Cabrera-Ziri, Salaris 2015
AGBs AGBs Allowed range FRMS Binaries Bastian, Cabrera-Ziri, Salaris 2015
AGBs AGBs Allowed range FRMS Binaries Bastian, Cabrera-Ziri, Salaris 2015
AGBs AGBs All suffer Allowed from the range same basic problem, over-production of He. FRMS Binaries Bastian, Cabrera-Ziri, Salaris 2015
Can Self-Enrichment Scenarios Work? Doomed to Fail. [O/Fe] [O/Fe] [O/Fe] Similar spread in Na-O, huge differences in their He abundance spreads. Bastian, Cabrera-Ziri, Salaris 2015
Can Self-Enrichment Scenarios Work? Doomed to Fail. [O/Fe] [O/Fe] [O/Fe] Similar spread in Na-O, huge differences in their He abundance spreads. Can never be accounted for in current self-enrichment scenarios. Bastian, Cabrera-Ziri, Salaris 2015
Take away messages Globular clusters host multiple populations within them, seen in chemistry and in their strange CMDs GCs still forming today (YMCs) No evidence for age spreads or multiple bursts in young clusters, either through resolved photometry or integrated spectroscopy Obs of YMCs appear to rule out the popular scenarios for the multiple populations in GCs. Obs of abundance trends in GCs appear rule out all selfenrichment scenarios. New ideas needed! (ISM physics? Crazy stellar evolution?...) Back to square one.
Radial profile of the multiple populations in M15 Larsen et al. 2015 Primoridal population more centrally concentrated
log (delta(y)) Galactic GCs Other problems for selfenrichment scenarios Milone 2014 MV M31 GCs He spread and [N/Fe] are functions of cluster mass Schiavon et al. 2013
NGC 1806 2*10 5 Msun 1.5 Gyr LMC cluster No abundance spreads! Mucciarelli et al. 2014
Testing the Fast Rotating Massive Star Scenario First generation forms at t=0, cluster remains embedded for 20-30 Myr 2nd generation stars form in the decretion discs around high mass stars, also fed from the primordial material that is left in the embedded cluster (5-10 Myr) Prediction: YMCs with ages < 20 Myr should still be embedded Krause et al. 2012, 2013
ESO 338-IG04 - Cluster 23 t = 6 +4-2 Myr Av = 0 M~ 1x10 7 Msun Rbubble ~ 120-200pc Z = 0.2 Zsun 100pc 200pc Östlin et al. 2007 Bastian, Hollyhead, & Cabrera-Ziri 2014
ESO 338-IG04 - Cluster 23 t = 6 +4-2 Myr Av = 0 M~ 1x10 7 Msun Rbubble ~ 120-200pc Z = 0.2 Zsun Bubble began expanding 1-3 Myr after formation Efficiently removed any pristine material out to hundreds of parsecs (still expanding at 40 km/s) Metallicity below that of Galactic globular clusters that show anomalies 100pc 200pc Östlin et al. 2007 Bastian, Hollyhead, & Cabrera-Ziri 2014
No clusters, older than 10 Myr, were found with signs of ongoing star formation integrated spectroscopy resolved photometry new integrated spectra Bastian et al. 2013a
New model One burst of star formation (i.e. an SSP) - as observed in young clusters High mass stars (binaries) mass segregated interacting binaries and spin-stars eject (low velocity) material into the cluster - this material has been processed by the high mass stars (70% of high mass stars are in binaries that will interact) Sana et al. 2012 low mass stars keep their discs for 5-10 Myr, which can entrain material as they move in the cluster Throop & Bally 2008 the material eventually accretes onto the young star (<< stellar mass) Bastian et al. 2013b
cluster interacting high-mass binaries core low mass PMS star + disc
1) Source of the enriched material Interacting high mass binaries The ejected envelope is very He rich de Mink et al. 2009 Many/most abundances reproduced (trends and quantitatively) (Cassisi & Salaris 2014) - may have a lithium problm (Salaris & Cassisi 2014) Ejected at low velocities (<20-30 km/s)
Summary of the model The entire volume core is swept out every ~2 Myr Without invoking multiple episodes of SF, the model 1) accounts for the enrichment patterns observed, 2) Potentially quantised abundances (and fractions), 3) doesn t have a mass budget problem (other models miss the mass budget by 10-100, even in the best cases) Explains why young massive clusters are observed to be gas free from young ages (~1-2 Myr) and don t show evidence for extended SFHs Same effects should be visible in young massive clusters (haven t been tested yet)
Caveats and Predictions The model requires a low coupling between SNe and dense cool matter in the discs of young stars and in the ejecta material from binaries. Supported by radiation-hydrodynamical models - e.g., Rogers & Pittard (2013) Circumstellar discs in massive clusters need to survive for 5-10 Myr. Leads to different kinematic predictions (in progress) Different radial profiles, with the degree of concentration a direction function of the amount of enrichment.
fraction of current mass that has been processed by AGB stars prediction Conroy 2012
ALMA observations of the Antennae Cabrera-Ziri, NB, et al. submitted Whitmore et al. 2014 50-200 Myr 1-3 * 10 6 Msun No gas detected (<10% cluster mass)
2) Mass budget problem? M dn/dlogm 0.6 0.5 0.4 0.3 0.2 0.1 0.0 stars that are still on the main sequence today stars that accreted processed material 0.8 M 2.0 M enriched gas accreted by pre main sequence stars (~10%) stars that have evolved off the main sequence AGB stars 19% 8% 13% enriched gas potentially returned to the inter stellar medium (13%) binaries, dynamical stripping spin stars 0.1 1.0 10 100 M (M ) But since a star is already in place to accrete the material, we can make the approximation that a star accretes <50%. In that case, <10% of the initial cluster mass needs to be accreted
3) Distribution of the stars in the cluster
Problems with the spin-star scenario Need primordial and processed gas to stay within the cluster for 5-10 Myr Should not find any massive young clusters (<5 Myr) that have cleared their gas Hollyhead et al. in prep. 10438 10597 11164
2) Quantisation of He and other elements? only 45% of stars ever enter the core
Caveats and Predictions The model requires a low coupling between SNe and dense cool matter in the discs of young stars and in the ejecta material from binaries. Supported by radiation-hydrodynamical models - e.g., Rogers & Pittard (2013) Circumstellar discs in massive clusters need to survive for 5-10 Myr. Need to invoke stellar mergers to explain Mg depletion in ~15% of clusters (although all models have some difficulties with the chemistry) Leads to different kinematic predictions (in progress) Different radial profiles, with the degree of concentration a direction function of the amount of enrichment.
Do clusters have extended star formation histories? Young massive (10 4-10 5 Msun) clusters (Westerlund 1, NGC 3603, R136) do not, < 1-2 Myr (Kudryavtseva et al. 2012) Resolved clusters in the LMC (200-300 Myr) and 10 5 Msun do not, < 30 Myr (Bastian & Silva-Villa 2013) MV V-I Bastian & Silva-Villa 2013
A Critical Look at Globular Cluster Formation Theories: Constraints from Young Massive Clusters July 14th - 18th 2014
[Na/Fe] Cordero et al. 2014 [O/Fe] r/rh
Metallicity distribution of stars in Fornax Larsen, Strader, Brodie 2012
~15 Myr 10 6 Msun NGC 1705
Do clusters have extended star formation histories? Young massive (10 4-10 5 Msun) clusters (Westerlund 1, NGC 3603, R136) do not, < 1-2 Myr (Kudryavtseva et al. 2012) There have been suggested of extended SFHs (200-500 Myr) in intermediate age (1-2 Gyr) clusters in the LMC/SMC based on extended main sequence turn-offs Mackey & Brobie Neilson (2007) Mackey et al. (2008) - 3 clusters Milone et al. (2008) - 15 clusters Glatt et al. (2008) - 1 cluster in SMC Goudfrooij et al. (2009, 2011a,b)
mv Intermediate age clusters in the LMC ~1.5 Gyr V-I Mackey & Brobie Neilson 2007
mv Intermediate age clusters in the LMC ~1.5 Gyr V-I Mackey & Brobie Neilson 2007
Intermediate age clusters in the LMC ~1.5 Gyr mv 300 Myr age difference V-I Mackey & Brobie Neilson 2007
observed modelled Goudfrooij et al. 2011a
Goudfrooij et al. 2011b
Goudfrooij et al. 2011b
If an age spread, and a common feature of clusters, this should be seen in younger clusters with similar properties < 500 Myr 1-3 Gyr clusters with claimed age spreads Bastian & Silva-Villa 2013
Cabrera-Ziri et al. in prep VLT/X-Shooter mini-survey of 6 YMCs (10 6-10 8 Msun) with ages between 15-500 Myr to search for age spreads
NGC 1856 NGC 1866 Age ~ 280 Myr mass ~ 10 5 msun Av ~ 0.8 mag Age ~ 180 Myr mass ~ 10 5 msun Av ~ 0.15 mag published HST photometry in Brocatto et al. 2001 and 2003
MV V-I To avoid background contaminations, only inner 4.8pc were used in both clusters
MV NGC 1856 B-V Quantitative results: use the FITSFH (Silva-Villa & Larsen 2010) code to derive the SFH of the clusters -Takes into account photometric errors -Adopts Salpeter IMF (above 1 Msun) -Uses Padova isochrones (Bressan et al. 2012) at z=0.008
upper limits on the width Intermediate age clusters shifted to 200 Myr Bastian & Silva-Villa 2013
upper limits on the width Intermediate age clusters shifted to 200 Myr No evidence for extended star formation histories Bastian & Silva-Villa 2013
Reported Age Spreads in Clusters Young clusters (<10 Myr): Δ(age) = few Myr High mass stars - little or no age spread (Kudryavtseva et al. 2012) Spread in PMS luminousities - although may not need an age spread Intermediate age clusters (1-3 Gyr): Δ(age) = 100-500 Myr
91 Konstantopoulos et al. 2009 Westmoquette et al. 2009 clusters with ages from 1-300 Myr and masses from 10 3 to 10 6 Msun also found in spirals, dwarfs, irregulars Mayya et al. 2008
Problems with the spin-star scenario Need primordial and processed gas to stay within the cluster for 5-10 Myr Should not find any massive young clusters (<5 Myr) that have cleared their gas Hollyhead et al. in prep. 10438 10597 11164