Transcript Star Clusters in Mergers

Massive Stars and Star Clusters in the
Antennae Galaxies
Brad Whitmore
2006, May Workshop
Principle Collaborators:
- Rupali Chandar
- Francois Schweizer
- Mike Fall
- Qing Zhang
- Barry Rothberg
OUTLINE
• The Formation of Young Massive Clusters
• Three Questions about Massive Stars
• The Big Picture
The Formation of Young Massive
Clusters (i.e. Super Star Clusters)
• Most stars are formed in groups and clusters, (e.g., Lada &
Lada, 2003)
• Star formation is enhanced in merging galaxies, and most of
the star formation is in the form of young massive clusters.
• Hence, understanding what triggers the formation of star
clusters in mergers is important for understanding the formation
of stars in general.
NGC 4038/9 is the youngest and
nearest galaxy in the Toomre
sequence, hence perhaps our best
chance for understanding the
formation of massive compact
star clusters in mergers.
Toomre, “The Evolution of
Galaxies and Stellar
Populations”, 1977
Are They Really Young Globular Clusters ?
Some of the young clusters we see in the Antennae (and other merging
galaxies) have the:
• Colors
(-0.2 < V-I < 0.6)
• Luminosities
(-15 < Mv < ?, power law LF with index ~ -2)
• Sizes
(Reff ~ 4 pc)
• Distributions (similar to the field stars)
• Spectra
(~ 10 objects age dated at 3 - 20 Myr)
• Velocity Dispersions (10 - 15 km/s)
• Masses
(104 - 106)
to be globular clusters with ages in the range 1 to ~ 500 Myr.
Perhaps the most convincing evidence that some clusters will
survive to become regular globular clusters is the presence of these
~500 Myr clusters in the NE extension.
Three Questions about Massive Stars
• What triggers the formation of stars (clusters) ?
• What fraction of stars are in clusters, and is this consistent with
the idea that essentially all stars form in clusters ?
• What can we say about the stellar content of the super star
clusters ?
What triggers the formation of clusters ?
It is clear that shocks are
important (e.g., along
spirals arms).
In mergers, one popular
model (e.g., Kumai et al.
1993) is that cloud-cloud
collisions with velocities ~
50 –100 km/s are required.
We have used STIS longslit spectra in 3 positions
angle of the Antennae to
test this idea.
Whitmore et al. 2005
We find the the velocity fields are
remarkably quiescent.
RMS dispersions are ~10 km/sec,
essentially the same as in the disks
of normal spiral galaxies.
This does not favor high velocity
cloud-cloud collision models.
Instead, models where a high
pressure interstellar medium
implodes GMCs without greatly
altering their initial velocity
distribution are favored (e.g, Jog and
Solomon, Elmegreen, ….)
.
What Fraction of Massive Stars in the Antennae are in Clusters?
Studies of starburst and merging galaxies
find that 10 – 50 % of the UV light (i.e.,
young stars) are in clusters (Meuer 1995).
Whitmore, Chandar & Fall (2006)
model (details in big picture part of
talk)
The initial fraction of stars in clusters is
even higher since most clusters don’t
survive.
Our Antennae model predicts
~ 8 % of the UV luminosity is from
clusters, in agreement with observations
( ~10 %).
Hence, our data is consistent with the idea
that all stars are born in clusters.
See Fall, Chandar, Whitmore (2005) for a
related calculation using total H_alpha flux.
~ median age of observed
clusters
Even clusters that survive loose a
large fraction of their stars from their
outer halos.
Whitmore et al. 1999
~ 50 % of the light is beyond 50 pc in
Knot S, a typical tidal radius for a
globular cluster.
~ 10 Myr
Bastian & Goodwin (2005) find similar
profiles in M82, N1569, and N1705,
compatible with N-body simulations of
clusters with rapid removal of mass
due to gas expulsion.
Fall, Chandar & Whitmore (2005)
make a similar argument to explain the
disruption of ~90 % of clusters in the
first 10 Myr (infant mortality).
~ 500 Myr
Other observations that support this basic picture are:
Chandar et al. (2005) find that the integrated spectrum of the field
stars in several local starburst galaxies is consistent with formation
from clusters which have dissolved, with typical time scales of
7 – 10 Myr. (NOTE: Similar result by Tremonti et al. 2001).
Wit et al. (2005) use proper motions from Hipparcus to estimate that
only 4 +/- 2 % of the O and B stars in the Milky Way formed outside
of groups or clusters
•
Most of the O and B stars in the field are consistent with
being “runaway” stars from nearby groups and clusters.
What can we say about the Stellar
Component in the Antennae ?
In our 1999 paper, one of our primary difficulties was
differentiating stars from clusters.
This led us to conclude that the number of young star clusters in
the Antennae was between 800 and 8000, a pretty big range !
Our new ACS data provides a better opportunity for making this
determination, and for studying the stars in their own right.
NOTE: This is work in progress, and exploratory in nature.
Before starting on the Antennae …
Ubeda, Maiz-Appendiaz, and Mackenty (2006) recently completed an
impressive piece of work on NGC 4314, a nearby (3 Mpc) starburst dwarf
galaxy.
Using CHORIOS (publicly available) , a software package they wrote to
perform maximum likliehood fits to either clusters or stars, they have
taken the game to a new level of sophistication.
There main conclusions are:
•
Extinction is quite patchy, but relatively low around all but the
youngest clusters.
•
10 of the 12 clusters they study have ages < 10 Myr.
•
The blue-to-red supergiant ratios are consistent with theory.
•
The stellar IMF in the field is steeper than –2.8.
They isolate 12 apertures (cluster regions) and also individual field stars.
They use up to 6 filters (F170, F336W, F555W, F814W, J, H) and make fits
assuming
bothmost
stellar
and cluster
1. The
massive
youngSEDs.
clusters in merging galaxies, such as the
appear tocontour
be excellent
They alsoAntennae,
make confidence
plots. candidates for proto-globular
clusters.
Examples for region I-As
Below is an example of a cluster where 2 individual stars (1 RSG and 1
BSG) within the cluster appear to contribute a significant fraction of
the total luminosity.
Ubeda et al. (2006) use this example to demonstrate the stochastic nature
of low mass clusters, where a single star or two can greatly affect
the results.
What would 30 Dor (core Mv ~ -10, 10**5 Msolar) look like at the
distance of the Antennae?
30 Dor
Region S
• Is it possible to distinguish clusters from stars just based on a
concentration index (i.e., Vmag(3 pix) – Vmag(1 pix))?
Conclusions:
Past procedure of using Mv < -9
as criteria ( based on Humphreys
1983) to define clusters is ~ 90 %
accurate.
Using point-like vs. resolved
concentration indexes to define
stars is only partially successful.
(color-color is better).
clusters
clusters
> 50 % of the faint objects are
clusters (important !)
stars
stars
Chorizos Fits:
if cluster
if star
•
Help distinguish between stars and clusters
•
Provide estimates of Mbol and Log Te.
Radial Profiles
•
Help distinguish between stars and clusters
Some evidence for triggered
star cluster formation?
Red = greater than 10 Myr
Green = 3 – 10 Myr
Blue = less than 3 Myr
Attempt to make a Mbol vs. Log Te diagram for stars.
•
Selecting stars based on size or color alone does not work very
well.
•
Using both size and color appears to work pretty well.
How far can a runaway O star get?
250 pc (40 km/s for 5 Myr)
Maximum runaway O star
definition . Wit et al. (2005)
Tentative Summary for Exploratory Study of
Region S in the Antennae
1.
It is possible to study individual stars brighter than Mv = –6 mag in
the Antennae (using both “size” and color to differentiate stars from
clusters).
2.
The brightest (Mv ~ -9.1) and most massive (~100 Msolar) stars
around region S are typical of stars found in other environments
(e.g., MW, 30 Dor).
3.
There is tentative evidence for some triggering of star clusters
around Region S.
4.
> 50 % of the faint objects with Mv brighter than –9 are clusters.
Hence, the total number of clusters in the Antennae is closer to 8000
than 800. (NOTE: Need to sort out color selection effects).
Other Studies Comparing Stellar and Cluster
Components in External Galaxies.
Saviane, Hibbard, Rich (2004) studied the stellar component in the
dwarf at the end of the tidal tail in the Antennae. They conclude:
•
There is a young component of stars with ages 2 – 100 Myr.
•
No “super star clusters” are present. They suggest that the
environment is not conducive to their formation.
•
They use the tip of the red giant technique to estimate a distance
which is ~ 30 % closer than our previous estimate (we believe
unlikely since peculiar velocity would have to be ~ 500 km/s).
•
While there are no Super Star Clusters (i.e., with Mv < -9), many of the
point-like objects are clearly resolved clusters.
•
With this low a rate of star formation we wouldn’t expect a cluster brighter
than Mv = -9. The brightest cluster on the PC is Mv = –8.4 (-7.7 if we use
their distance), hence only slightly fainter. Statistics not different physics.
The Big Picture
Roughly 40 gas-rich mergers have now been observed in
detail by HST. All show young star clusters.
In addition, we find young, massive, compact clusters in:
starburst dwarf galaxies (e.g., Meurer et al., 1995),
barred galaxies (Barth et al., 1995),
spiral galaxies (Larsen & Richtler, 1999)
Milky Way and LMC (e.g., Walborn 2000)
These clusters have properties similar to those seen in the mergers,
but always fewer in number, and generally fainter in luminosity.
It appears that wherever you have regions of star formation, you
have young massive clusters being formed, not just in mergers !
Essentially all the cluster
luminosity functions in
merging and starbursting
galaxies are power laws
with index ~ -2.
Whitmore, 2000
Power law luminosity functions with index
-2 appear to be the norm, in spiral galaxies
as well as merging and starburst galaxies.
Larsen (2002)
Mergers and starburst
galaxies may have the
brightest clusters only
because they have the
most clusters (i.e., there
may be a “universal”
luminosity function with
the correlation simply
being due to statistics, not
special physics).
Whitmore, 2000
mergers
Best fit
(i.e., a size-of-sample
effect)
spirals
Larsen (2002) has shown
that a similar relation
holds versus the total star
formation in a galaxy.
Predicted if universal power-law,
index = -2
Fall, Chandar & Whitmore (2005) show that 90 % of the young clusters in the
Antennae are disrupted each decade of time (i.e, infant mortality).
The disruption rate appears to be independent of mass to first order.
Fall, Chandar & Whitmore 2005
The disruption rate
appears to be a power
law with index –1 for the
Antennae, the SMC
(Rafelski & Zaritsky
2005), and the MW
(Lada & Lada 2003).
Universal destruction
rate ?
The Big Picture - A General Framework for
Understanding the Demographics of Star Clusters
Ingredients (assume all stars form in clusters) :
1. A universal initial mass function (power
law, index -2)
2. Various star(cluster) formation histories
3. Various cluster disruption mechanisms
(e.g., T**-1 < 100 Myr, 2-body relaxation > 100 Myr)
4. Convolution with observational artifacts and
selection effects
Observations (luminosity and age distributions,
color-color diagrams, total luminosity of a
galaxy, fraction of field stars, …)
Here is an example of two
of our models based on
observations of the
Antennae.
Whitmore, Chandar, Fall
(2006)
Whitmore, Chandar, & Fall
Probably the most relevant
output from the model for
the present discussion is the
prediction of the fraction of
stars in the field, as discussed
earlier.
(2006) model
Our Antennae model predicts
~ 8 % of the UV luminosity is from
clusters, in agreement with
observations ( ~10 %).
Hence, our data is consistent with
the idea that all stars are born in
clusters.
~ median age of observed
clusters
Summary
1.
The small-scale velocity dispersion between clusters is remarkably
quiescent (e.g, ~ 10 km/sec), indicating that high-velocity cloud-cloud
collisions are not the triggering mechanism for cluster formation.
2.
Observations in both the Milky Way and external galaxies support the
idea that essentially all stars are formed in groups and clusters.
3.
The most luminous (massive) stars in external galaxies as far away as
the Antennae (~ 20 Mpc) can be differentiated from the clusters.
2. A general framework has been developed to explain the demographics
of stars and star clusters in galaxies. Important aspects are;
•
Assumption that all stars are born in groups and clusters.
•
An initial mass distribution which is a power law with index = -2
•
Infant mortality of ~90 % of the clusters each decade of time.