A&A 407, 913-917 (2003)
DOI: 10.1051/0004-6361:20031019
E. I. Vorobyov
Institute of Physics, Stachki 194, Rostov-on-Don, Russia, and Isaac Newton Institute of Chile, Rostov-on-Don Branch, Russia
Received 26 March 2003 / Accepted 19 June 2003
Abstract
The gas hydrodynamics modeling of the Cartwheel ring galaxy is performed
with the purpose to reproduce the measured intensity and radial
distribution of
emission in the Cartwheel's disk.
The star formation efficiency of
is derived for
an assumed Schmidt law with power index N=1.5 and a full
thickness of the Cartwheel's gas disk of
z0=250-1000 pc, respectively.
Simulations with a Schmidt law of star formation yield a higher
surface brightness
(
)
in the Cartwheel's inner ring as compared to the observed values.
The Toomre criterion for star formation fails to account for the observed
sharp drop of massive star formation (MSF) in the inner ring.
Numerical simulations indicate that a large shear near/at the position of the inner ring
can raise the gas threshold for star formation and suppress MSF in
the Cartwheel's inner regions.
Hence, the Schmidt law has to be supplemented with a shear criterion for
star formation in order to reproduce the observed radial distribution of
in the Cartwheel's inner regions.
Key words: galaxies: individual: the Cartwheel - galaxies: starburst
Collisional ring galaxies provide the means of studying star formation
processes on a kiloparsec scale. The gravitational perturbation of
a companion galaxy drives a ring density wave through the disk of the target
galaxy. The expanding-ring density wave is expected to trigger high
rates of massive star formation (MSF) along its perimeter.
On the other hand, strong suppression of
star formation is expected behind the wave where the surface density of
the gas drops substantially below the initial unperturbed values.
imaging of a sample of ring galaxies by Marston & Appleton
(1995) and Higdon (1995) has demonstrated that in most ring galaxies
star formation is indeed localized in the rings.
Observations of both normal and starburst disk galaxies suggest that on
scales of a few kiloparsecs star formation may be well
represented by a Schmidt law with index
(Kennicutt 1998).
Can star formation in ring galaxies, particularly in the Cartwheel ring
galaxy, be described by a Schmidt law as well?
A general correlation between the azimuthally averaged
radial profiles of the H I surface density (hereafter,
)
and the
surface brightness (hereafter,
)
in the Cartwheel's disk seems to favor a global Schmidt law.
However, a pronounced anticorrelation
between the azimuthal distributions of
and
found by Higdon (1996) in the Cartwheel's outer ring
implies that the Schmidt law is not locally observed.
Further, emission-line maps of the Cartwheel
(Amram et al. 1998; Vorobyov & Bizyaev 2003, hereafter VB) suggest
that the inner ring is at best much weaker than the outer ring in terms of MSF.
This is unexpected, since numerical simulations of the Cartwheel predict
gas surface densities in the inner ring
exceeding those of the outer ring (Struck-Marcell & Higdon 1993;
VB). Low-level star formation activity in the regions of high gas surface density
implies that there may exist a mechanism or mechanisms suppressing the MSF in the Cartwheel's
inner ring.
Figure 1 shows the
image of the Cartwheel galaxy obtained
from the CFHT archive (VB). Since this image
was intended to show the relative input of different parts of the Cartwheel
to the total
emission, the flux calibration was not applied.
The contrast in
the
image is enhanced to show faint line-emitting regions in
the inner ring. The
emission is mainly restricted to the
outer ring. The inner ring shows much fainter
emission.
The
luminosity
of the Cartwheel was found to be
ergs s-1
(Higdon 1995).
At the distance of 140 Mpc (H=65 km s-1 Mpc-1) assumed in this paper,
it becomes
ergs s-1.
![]() |
Figure 1:
The Cartwheel galaxy in
![]() |
Open with DEXTER |
The
emission line provides a measure of the star formation rate (SFR)
over the past 6-7 Myr. Calibrations have been published by several
authors, including Gallagher et al. (1984), Kennicutt et al.
(1994), and Leitherer & Heckman (1995).
There is a 30% variation among published calibrations, with most of
the dispersion reflecting differences in stellar evolution and atmospheric
models. In this paper the calibration of Kennicutt et al. (1994)
derived for solar metallicity and the Salpeter IMF (
)
is used:
Actually, the largest source of uncertainty in Eq. (1) is the
amount of internal extinction at
,
(Cardelli et al. 1989).
In the present simulations, a possible
differential extinction in the Cartwheel's disk is taken into account by estimating
from the model's known gas surface density,
,
via the standard
gas-to-dust ratio (Bohlin et al. 1978) and
(see
Vorobyov & Bizyaev 2001 for details).
Observations of both normal and starburst disk galaxies
suggest that on the scales of a few kiloparsecs star formation may be
represented by a Schmidt law (Kennicutt 1998)
The numerical hydrodynamics model used to simulate the gas
dynamics in the Cartwheel is discussed in detail in VB.
Two modifications are made as compared
to the code described in VB. First,
a gravitational potential
of the pre-collision rigid stellar disk is added.
Second,
distribution is constructed using
Eq. (1). To compute the SFR,
stellar particles are
formed according to the Schmidt law (Eq. 3) each 0.5 Myr in each computational cell.
The average mass of a stellar particle is
,
with
the maximum mass amounting to
and the minimum mass as
small as the mass of a single massive star.
Further, stellar particles are assigned positions and velocities
drawn from the parent gas and are evolved as collisionless particles
in the combined gravitational potential of the halo, stellar
disk, and gas disk.
The local self-gravity among stellar particles is not computed. Since each
stellar particle carries information on the SFR at the time of its formation,
the SFR averaged over the past 7 Myr in each cell is easily constructed,
and then substituted in Eq. (1) to obtain
.
The conversion factor in Eq. (1) decreases linearly from
in the nucleus to
in the outer ring
to account for the radial metallicity gradient ranging from
in the inner regions
(R < 6 kpc) to
in the outer regions (Vorobyov & Bizyaev
2001; Fosbury & Hawarden 1977).
The gas consumption due to star formation is taken into account. Possible feedback effects of
star formation on the system are neglected in the present simulations.
The parameters of the numerical simulations are mostly identical to those in VB.
A few modifications are made, namely the halo mass is reduced to
because of the inclusion of the stellar disk of
,
so that the total halo+stellar mass is comparable to the
indicative mass of the Cartwheel,
(Higdon
1996).
The total gas mass is increased from
(VB)
to
to take account of
detected by
Horellou et al. (1998) in the Cartwheel's inner regions and
a possible
contribution of molecular hydrogen in the star-forming outer ring.
In order to make unambiguous conclusions about
the applicability of a Schmidt law
for describing star formation in ring galaxies, the gas distribution in
the Cartwheel's disk has to be modeled as accurately as possible.
Several test runs have shown that the Cartwheel's
H I kinematics and morphology are best reproduced for an off-center
collision with
kpc, in contrast to
kpc in
VB.
The companion's mass is chosen to be
in
the present simulations.
This mass is about 5 times larger than what is actually measured by Davies & Morton (1982) and Higdon (1996) for a companion G3, the most massive in the Cartwheel
group. Most studies indicate
that the well-defined rings are formed only for companions with masses no less than
20% of the primary galaxy's mass (see e.g. Hernquist & Wail 1993).
However, there is good reason to believe
that the "missing'' mass problem is a consequence of the ring galaxy formation
(see discussion in VB).
The present study addresses two main questions:
1) What is the star formation efficiency ()
in the Cartwheel?
2) Which modifications to the Schmidt law are required to
reproduce the observed azimuthally averaged
distribution
in the Cartwheel?
This approach is different from that of Mihos & Hernquist (1994),
who fixed
so that their model galaxy formed stars
at roughly a rate of
yr-1.
Firstly, an isothermal equation of state is considered with the gas temperature
104 K. The coefficient
in Eq. (3) is varied
to obtain the observed value for the integrated
luminosity, 4.0
1042 ergs s-1.
This is achieved for
5.7
10-3. Assuming that the full thickness
of the Cartwheel's gas disk, z0, is in the range of
250-1000 pc, the efficiency of
star formation becomes
,
respectively. Note that numerical
simulations (Struck 1997), as well as observations (VB), indicate
that the Cartwheel's disk is rather thick.
Either star formation in the Cartwheel is very efficient or the
luminosity reported by Higdon (1995) is overestimated.
In the following simulations, z0=500 pc or
is assumed.
The solid line in Fig. 2 shows the radial
profile obtained by azimuthally averaging the modeled
around
the dynamical center at r=0 kpc.
The dotted line gives the observed
radial profile of the Cartwheel derived from Fig. 1
(see VB for details).
Since the flux calibration was not originally applied to Fig. 1,
the Cartwheel's
radial profile in Fig. 2 was calibrated
to the integrated
luminosity of the Cartwheel,
ergs s-1.
Simulations with a Schmidt law of star formation predict
a higher or comparable
in the inner ring (r<6 kpc) as compared to that of
the outer ring, while the Cartwheel's inner ring is an order of magnitude weaker
in terms of
.
This discrepancy is not due to gas-density-dependent
extinction in the Cartwheel's disk, this effect is accurately
taken into account as described in Sect. 2.
There should exist a mechanism or mechanisms that suppress
star formation in the Cartwheel's inner regions.
![]() |
Figure 2:
The azimuthally averaged
![]() ![]() |
Open with DEXTER |
Higdon (1996)
has suggested that a gas threshold effect described by the Toomre criterion
is responsible for the lack of
MSF interior to the Cartwheel's outer ring. As found by Kennicutt (1989)
and confirmed by Martin & Kennicutt (2001),
MSF occurs only if
,
where
is the speed of sound and k is the epicyclic frequency.
The simulations are repeated with a Toomre-modified Schmidt law, i.e.
star formation takes place only if the local
,
with
km s-1 and k computed from the
model's known velocity field. The results are shown in Fig. 2
by the dashed line.
The Toomre criterion affects
only in the inter-ring
region (6<r<20 kpc) where
drops below critical values and has almost
no influence on
in the inner and outer rings.
Another mechanism that might suppress star formation
is a large shear.
The effects of shear
can be parameterized in terms of the free-fall time defined as
,
and the local shear rate
defined as
.
Star formation is supposed to
occur only if the following condition is met
.
This inequality simply states that a gas cloud is stable against the shear
if the time it takes for the shear to tear apart a cloud
is bigger than the free-fall time. Solving this inequality for
and
noticing that the expression
is in
fact the Oort constant A results in the following shear criterion for star formation
The shear criterion (4) is taken into account when deriving
the azimuthally averaged
shown in Fig. 2 by the dotted-dashed line.
The Oort constant A is computed from the model's known velocity field.
A Schmidt law supplemented with the shear criterion (4) yields an azimuthally averaged
that agrees better with
the observed profile of the inner ring, which proves that the shear could
be effective in suppressing MSF in the Cartwheel's inner region.
On the other hand, the shear appears to have little effect on star formation in
the Cartwheel's outer regions.
Substituting
in Eq. (4)
results in another definition of the critical density
,
where
is a one-dimensional velocity dispersion and
.
This definition of
is
outlined in Martin & Kennicutt (2001) who refer to Elmegreen
(1993) and Hunter et al. (1998), the only difference
is in the value of
.
As discussed in Hunter et al. (1998), the coefficient
is derived from the assumption that the amplitude of shearing perturbations
must grow by a sufficiently large factor,
100, during the characteristic
time of 2/A for the gravitational instability to be significant.
If the gas disk of the Cartwheel was marginally stable before the collision
(the high star formation efficiency favors this assumption),
then a smaller increase
in
,
say
10, might be enough for
the instability to grow, which would result in
.
Simulations with different values of
indicate that
reproduces best
the observed azimuthally averaged
profile of the Cartwheel's inner regions.
Recently, Martin & Kennicutt (2001) have demonstrated that
the shear may play an important role in determining a gas
threshold for star formation in the inner parts of NGC 2403 and
M 33, whose gas surface densities are well below
,
yet they have numerous H II regions.
Indeed, if the local shear rate, rather than the Coriolis force (epicyclic
frequency k), best describes the destruction rate of
gas clouds, then the inner parts of galaxies with slowly rising rotation
curves should have a lower gas threshold than the outer
parts with flat rotation curves.
It appears that the shear in the strongly perturbed inner parts
of ring galaxies has the opposite effect on a gas threshold
for star formation. Simulations show that the velocity field at/near the
inner ring is strongly perturbed as a result of radial cross-motions
of gas, which may effectively raise the gas threshold
and suppress star formation there.
Simulations of Sect. 4.1 were repeated for an adiabatic equation
of state including the effect of cooling and heating. The gas disk was initially
set at T=104 K.
The equilibrium cooling curves of Wada & Norman (2001)
(their Fig. 1) were adopted. The cooling function
is parameterized in terms of metallicity, i.e. it linearly decreases
from solar metallicity in the nucleus
to one-fifth of solar in the outer ring.
An empirical heating function that initially balances the cooling is applied.
This approach may be thought of as a crude model for stellar energy input
and can imitate a constant heating of 10-24 ergs s-1 by the
background UV radiation field.
Cooling and heating were treated numerically using Newton-Raphson iterations.
The star formation efficiency
in the non-isothermal gas depends on the gas temperature.
The relevant physics is poorly understood and different modifications
based on empirical relations have been proposed to account
for the gas temperature dependence of a Schmidt law (e.g. Samland et al.
1997).
To be consistent with the isothermal simulations in Sect. 4.1,
the Schmidt law is modified so that the SFR
is equal to that for the isothermal gas at T<104 K and rapidly declines
at T>104 K. The modified Schmidt law can now be written as
,
where
The model
distribution is found to be
similar to that of
the isothermal case if an unmodified Schmidt law is employed,
the azimuthally averaged
of the inner regions
is higher or comparable
to that of the outer ring. However, the measured azimuthally averaged
profile
of the Cartwheel's inter-ring region (10 <r<20 kpc) is better reproduced
in non-isothermal simulations, indicating that the temperature dependence
of a Schmidt law rather than the Toomre criterion is responsible for the
observed low-level star formation activity in the inter-ring region.
Thus, simulations with the effects of cooling and heating show that
shocks and compressional heating are not effective in preventing MSF in the inner ring.
The characteristic cooling times in the inner ring are still shorter than
the gas dynamical times.
A better correspondence between the modeled and observed azimuthally averaged
in the inner regions (
kpc) can only be achieved
for a Schmidt law supplemented by the shear criterion for star formation
defined in Eq. (4).
Hence, both isothermal and non-isothermal simulations indicate
that the large shear may prevent MSF in the Cartwheel's
inner ring.
Gas hydrodynamics modeling and a Schmidt law of star formation are used
to simulate the total
luminosity from the Cartwheel ring
galaxy,
1042 ergs s-1.
The efficiency of star formation
is derived for the
assumed full thickness of the Cartwheel's gas disk of
z0=250-1000 pc,
respectively.
The corresponding efficiency in Kennicutt's sample of isolated and
interacting galaxies (Kennicutt
1998) is
,
which implies that star formation in the Cartwheel is very
efficient.
Indeed, according to Marston & Appleton (1995) the Cartwheel galaxy has an
exceptional
,
at least 6 times higher than
of the other
11 ring galaxies in their sample. The reasons for such a star
formation activity are uncertain.
A more accurate determination of the Cartwheel's total
luminosity is highly desired to prove the exceptional status of this
galaxy.
Simulations with a Schmidt law of star formation predict
higher or comparable azimuthally averaged
in the inner ring than in the outer ring.
Observations indicate the opposite,
the Cartwheel's inner ring is much weaker in terms of
.
Simulations with the effects of cooling and heating
show that the shocks and compressional heating are not effective
in preventing MSF in the inner ring. Instead,
high-density
and low-temperature
150<T<400 K clumps are formed in the inner ring, which
may favor H I
transitions and account for the large
amount of molecular hydrogen detected by Horellou et al. (1998)
in the Cartwheel's inner regions.
A Schmidt law supplemented with the Toomre criterion for star formation
cannot explain a sharp drop of star formation in the inner ring.
Failure of the Toomre criterion is also reported by Martin & Kennicutt
(2001) for NGC 2403 and M 33 where there is evidence for strong
radial flows of gas. With radial gas velocities between 60 km s-1(Higdon 1996) and 100 km s-1 (VB), the Cartwheel appears
a classic case of failure of the Toomre criterion.
Numerical simulations indicate
that the large shear near/at the position of the inner
ring may raise the gas threshold for star formation
and effectively suppress MSF in the inner ring.
Thus, the Schmidt law has to be supplemented with the shear criterion for star formation
in order to reproduce the measured azimuthally averaged
profile in the Cartwheel's inner regions (r < 6 kpc).
The existence of large shear in the Cartwheel's inner regions is observationally supported by the
detection of strong perturbations in the
rotation curve near/at the position
of the inner ring (Amram et al.
1998).
The measured azimuthally averaged
profile of the Cartwheel's inner
regions is best reproduced for a 2.5 times lower
than that outlined in
Martin & Kennicutt (2001) (refering to Elmegreen
1993 and Hunter et al. 1998). The difference implies
that the gas disk of the Cartwheel was marginally stable before the collision.
Acknowledgements
The author is thankful to the anonymous referee for suggestions that helped to improve the paper. The author is grateful to D. Bizyaev for providing theimage of the Cartwheel galaxy.