Received 07th April 1998
"The Magellanic Clouds and Other Dwarf Galaxies
of the Bonn/Bochum-Graduiertenkolleg
WFPC2 Observations of Leo A: A Young Galaxy?
Tolstoy1, J.S. Gallagher2, A.A. Cole2,
A. Saha3, R.
Dohm-Palmer4, E. Skillman4, and
, Garching bei
2Department of Astronomy
University of Wisconsin, Madison, Wisconsin
NOAO, Tucson, Arizona
University of Minnesota, Minneapolis, Minnesota
, University of Michigan, Ann Arbour, Michigan
The unprecedented detail of the HST
of Leo A leads to a new distance determination which comes from the
position of the Red Clump, the Blue Loops and the tip of the Red Giant Branch.
We obtain a distance modulus, m-M
= 24.2±0.2, or
We suggest that a burst of star formation at some point in the past has led
to an anomalously high number of W Virginis or Anomalous Cepheid variable
stars in this galaxy, which were earlier misidentified as classical
δ-Cepheid variable stars.
From our interpretation of the features present in the WFPC2
CMDs at this
new distance we show a good match with extremely low metallicity young models
= 0.0004), and the strong possibility that we are looking at
a predominantly young galaxy, which is of order ≤2 Gyr old,
although we cannot rule out the possibility of an older, underlying
globular cluster age population.
Small galaxies are common and apparently structurally simple.
They may provide an important perspective on how luminous structures have
evolved in the Universe.
Despite a variety of theoretical models (e.g. Dekel & Silk 1986; Hensler
& Burkert 1990), we cannot predict how even the simplest galaxies
have changed over time.
In particular the internal clocks which set the time interval for major star
formation is seen to be highly variable in the Galactic retinue of dSph,
ranging from ancient systems where star formation was complete 10 Gyr ago
to galaxies that are mainly intermediate age (e.g. van den Bergh 1994).
The presence of numerous small, actively star forming field galaxies at
moderate redshifts of 0.3 < z < 1 suggests that such asynchronous
behavior may be the rule rather than the exception amongst smaller galaxies.
An HST program was initiated using
4 orbits per galaxy, in three filters (effectively B, V and
I), to study a sample of nearby dwarf irregular galaxies
(Skillman et al. 1998).
The sample consists of Sextans A, Pegasus, Leo A & GR 8.
The results have been dramatic and illustrate the tremendous advances possible
even with short exposures when crowding has been virtually eliminated
(Dohm-Palmer et al. 1997a,b; Tolstoy et al. 1998; Gallagher et al. 1998).
Here we present the star formation history (SFH) of the last few Gyr
for Leo A (≡ DDO 69), a Magellanic dwarf irregular galaxy,
and show how this helps us understand the properties of this galaxy.
There are several faint H II regions distributed along
the ridge of highest column density H I (Tolstoy 1996;
Hunter et al. 1993), which provides a very low limit to the current star
formation rate, SFR, (over the last 10 Myr) of
<10-4 Msun yr-1.
The brightest Halpha emission in the galaxy comes from
a planetary nebula, which yields an extremely low oxygen abundance of
∼2.4% solar (Skillman et al. 1989).
Deep ground based images also reveal the lack of an extended red star
(i.e. old) stellar halo (Tolstoy et al. 1998).
Recent H I observations (Young & Lo 1996), show that
the optical galaxy is surrounded by a large H I halo,
extending out about 3 times the optical diameter at a column density of
The detected H I flux corresponds to an
H I mass of
(8.1±1.5)·107 Msun, of which
∼30% is in the halo at column densities below
The observed velocity gradient across Leo A in H I
is small, and the changes in the velocity dispersion probably reflect
the conditions in the ISM rather than rotation.
2. Modelling the Colour-Magnitude Diagram
To properly model a Colour-Magnitude Diagram (CMD) it is important to first
decide upon the range of values that are acceptable for the numerous basic
parameters that affect the CMD models.
The most significant are the distance, the extinction and the metallicity of
a galaxy (Tolstoy 1996).
These properties can be determined independently from the CMD, and these
independent measures must be consistent with the findings in a CMD.
These three basic parameters make the most significant impact on the properties
of the CMD and hence the final SFH model.
The distance to Leo A was thought to be well determined from the detection
of 5 δ-Cepheid variable stars (Hoessel et al. 1994).
However, from the WFPC2 CMDs (see
Fig. 1), it is clear that
the Leo A Cepheid distance cannot be correct.
There is no way to reconcile either the He burning Blue Loops (BLs) or
the presence of a Red Clump (RC) with this large distance.
Thus, based on this new HST CMD,
we assume the new distance to Leo A to be m-M = 24.2±0.2,
with an error resulting from the range that allows a reasonable fit to
the CMD models.
This distance is consistent with the largest number of stars in the CMD.
It agrees with the tip of the Red Giant Branch (RGB), the luminosity
(and shape) of the BLs, a young RC and with the (low end) of the reanalysis
of the variable star distance (Tolstoy et al. 1998).
Burstein & Heiles (1984) give a low extinction of
E(B-V) = 0.02 towards Leo A, and there is nothing in
the IRAS 100 µm maps, the H II region or
H I analysis or CMDs to suggest this should be any
different (unlike Pegasus, Gallagher et al. 1998).
The young population, specifically the BLs, are the most sensitive global
feature in a CMD to population metallicity.
They do not require an assumed age as the RGB does.
The BLs in our CMD are consistent with Padua (Fagotto et al. 1994)
Z = 0.0004 (2% solar), and not with Geneva (Schaller et al. 1992)
Z = 0.001 (5% solar) for any reasonable distance.
This low value is also consistent with the Skillman, Kennicutt & Hodge
(1989) PN metallicity, but this galaxy is desperately in need of more detailed
To fit metallicity evolution into a CMD model is very uncertain, and it has
potentially a large impact upon the final result.
Without metallicity evolution a galaxy model will be younger than if
metallicity evolution is included.
This is because the age-metallicity degeneracy works in such a way that
an old metal poor star can lie in the identical position in a CMD to a young
more metal rich star.
At the low metallicity of Leo A the age-metallicity degeneracy is
particularly acute (Da Costa 1997), and a more straight forward practical
problem with determining a metallicity evolution model is that for the young
population of Leo A we are already using the lowest metallicity set of
consistent model stellar evolution tracks available and so there is no
parameter space left to model more metal poor older stars.
There is even some argument about what is the lowest metallicity that
an object can have and with Leo A we are pretty close to the perceived
limits for galaxies (Kunth & Sargent 1986).
Thus, even though it would be desirable to include metallicity evolution it is
not clear how to do this.
Having sorted out the basic parameters suitable for Leo A we can begin the
It is important to understand the theoretical models being used and their
Many different aspects of this are discussed in detail in Tolstoy (1996),
Tolstoy & Saha (1996), Tolstoy et al. (1998) and Gallagher et
Errors in the models have very serious implications for Leo A because of
the insensitivity of low metallicity models to age.
The isochrones crowd together in a very limited parameter space, and so small
changes in stellar physics can change a best model from a young 2 Gyr old
population to an old 10 Gyr population.
Another problem discussed in Tolstoy et al. 1998 is the uncertainty
both in the reliability of the RGB models and the
Both of these can push the RGB blueward and thus make it more consistent with
a uniformly young galaxy.
[Click here to see Fig. 1 - 3!]
3. The Star Formation History of Leo A
The modelling results in the combined final best model shown in Fig. 2,
from which we see that:
- We clearly fit the MS effectively.
The small scatter to the red side is presumed to be due to non-uniform
reddening intrinsic to Leo A, and perhaps binaries.
- The RC is modeled convincingly.
There are data not matched by the model which could not be helped by
any extra BLs or RC model (or mixture thereof).
This is probably due to underestimates of the errors or incompleteness
at fainter red magnitudes, and possibily also due to reddening and/or
- There is no counter part (in the
HST data) for
the younger RGB component.
We assume that this is either due to problems in the RGB models,
meaning that the galaxy could be modeled as entirely young, or
to problems with RC models in which case the galaxy could be
predominantly much older.
- An old population cannot be excluded from these data using the
available stellar evolution models.
In fact the RGB is clearly well fit by a very old population
Thus, we have managed to match the major features in the CMD with this model.
It is by no means a unique model and the mismatch of the observed and model
RGB is a cause of serious problems in finding the true best model using the
Tolstoy & Saha (1996) methods.
The MS and BL make it obvious that there has been fairly active SFR over the
last 500 Myr, and the RC makes it hard to avoid a fairly strong peak in
the SFR 0.9-1.5 Gyr ago.
The exact duration and intensity of these star formation periods are dependent
on small number statistics.
If there was significant metallicity evolution the conclusions based on the RC
If the RC were more metal poor than the MS (e.g. Z = 0.0001) then it's
maximum age would be older than that based on Z = 0.0004 models, but
only by a few hundred million years (Castellani & Degl'Innocenti 1995).
In our final best model we also include an old population.
This population appears to fit the RGB stellar distribution very well, but
it is strongly reliant on the stellar evolution models, which are not well
tested in this regime. There is also uncertainty in the reliability of
the HST calibration.
In Fig. 3 we show the SFH
for Leo A which created the CMD model in
We use pseudo-SFR because we have not calibrated these values onto an absolute
The SFH is split into 2 parts with differing time resolution.
The error bars give a rough estimate of the reliability of a given SFR.
During quiescent periods they give an estimate of the intensity of SFR
that would be difficult to hide.
This is a difficult error to make absolute.
We can hide a <.0005 pseudo SFR units star burst of short duration
(<108 yr) between 2 and 10 Gyr quite easily.
But if there are too many of them, or one lasts too long then there will
be problems for the model (the RGB will become over populated and perhaps
The error bars make no allowance for errors in distance, reddening, metallicity
or model inaccuracies.
These are the major error sources of uncertainty in creating this SFH.
If one of them is wrong the whole scenario presented here could change
Thus Leo A remains an excellent candidate for a young galaxy.
There are uncertanities in the modelling process presented here meaning that
we cannot unambiguously confirm the presence of any stars older than 2 Gyr.
We can say that if such stars are present they represent a small fraction of
the star formation in Leo A.
The extremely low metallicity and the lack of an extended red halo
in this galaxy are also suggestive of a predominantly young galaxy.
- Burstein D., Heiles C., 1984, ApJS 54, 33
- Castellani V., Degl'Innocenti S., 1995, A&A 298, 827
- Da Costa G.S., 1997, 'Stellar Astrophysics for the Local Group:
A First Step to the Universe', Canary Islands winter school,
Aparicio A., Herrero A. (eds.), in press
- Dekel A., Silk J., 1996, ApJ 303, 39
- Dohm-Palmer R.C., Skillman E.D., Saha A., Tolstoy E., Gallagher J.S.,
Hoessel J.G., Mateo M., Chiosi C., 1997a, AJ 114, 2527
- Dohm-Palmer R.C., Skillman E.D., Saha A., Tolstoy E., Gallagher J.S.,
Hoessel J.G., Mateo M., Chiosi C., 1997b, AJ 114, 2514
- Fagotto F., Bressan A., Bertelli G., Chiosi C., 1994, A&AS 104, 365
- Gallagher J.S., Tolstoy E., Dohm-Palmer R.C., Skillman E.D., Cole A.A.,
Hoessel J.G., Saha A., Mateo M., 1998, AJ 115, 1869
- Hensler G., Burkert A., 1990, Ap&SS 170, 231
- Hoessel J.G., Saha A., Krist J., Danielson G.E., 1994, AJ 108, 645
- Hunter D.A., Hawley W.N., Gallagher J.S., 1993, AJ 106, 1797
- Kunth D., Sargent W.L.W., 1986, ApJ 300, 496
- Schaller G., Schaerer D., Meynet G., Maeder A., 1992, A&AS 96, 269
- Skillman E.D., Kennicutt R.C., Hodge P., 1989, ApJ 347, 875
- Skillman E.D., Dohm-Palmer R.C.,
Kobulnicky H.A., 1998, this volume
- Tolstoy E., 1996, ApJ 462, 684
- Tolstoy E., Saha A., 1996, ApJ 462, 672
- Tolstoy E., Gallagher J.S., Cole A.A., Hoessel J.G., Saha A.,
Dohm-Palmer R., Skillman E., Mateo M., Hurley-Keller D., 1998,
AJ, in press (September issue) [astro-ph/9805268]
- van den Bergh S., 1994, in 'The Local Group: Comparative and Global
Properties', CTIO/ESO Workshop, Layden et al. (eds.), p. 3
- Young L.M., Lo K.Y., 1996, ApJ 462, 203
Links (back/forward) to:
|First version: ||16th||August,||1998
|Last update: ||08th||October,||1998
Jochen M. Braun &