A&A 383, 755-762 (2002)
DOI: 10.1051/0004-6361:20011785
Y. P. Wang
Purple Mountain Observatory, Academia Sinica, China; National Astronomical
Observatories of China
National Astronomical Observatory of Japan, Mitaka, Tokyo 181-8588, Japan
Received 20 July 2001 / Accepted 29 November 2001
Abstract
Recent far-infrared and submillimetre waveband observations
revealed a large number of Ultraluminous
Infrared Galaxies (ULIGs) with infrared luminosities >
.
These sources are
proposed to lie at redshifts above one, and in normally interacting systems with
very dusty environments. We discussed in a previous paper that a population with a
fast evolving infrared burst phase triggered by gas-rich mergers
at
predicted successfully the steep slope of faint IRAS
60
m source counts within the flux range of 100mJy
1Jy, still leaving
the infrared background level at this wavelength compatible with the upper limit from recent
high energy TeV
ray detection of Mrk 501. To extend the model to mid and far infrared wavelengths, we adopt a reasonable
template spectral energy distribution typical for nearby-infrared-bright starburst galaxies (
),
such as Arp 220. We construct the SED
for the dusty starburst mergers at
by a simple dust extinction law and a thermal continuum assumption for the far-infrared
emission. Since the radiation process at mid-infrared for these starburst merging systems is still uncertain, we assume
it is similar to the MIR continuum of Arp 220, but modify it by the observed flux correlation of ULIGs from IRAS and ISOCAM
deep surveys.
We show in this paper that the strong evolution of the European Large Area ISO Survey (ELAIS)
at 90
m, ISO 170
m and the Submillimeter deep survey at 850
m could be sufficiently accounted for by
such an evolutionary scenario, especially the hump of the ISOCAM 15
m source count around 0.4mJy.
From current best fit results, we find that the dust temperature of those extremely bright starburst merging system at
would be higher than that of Arp 220 for a
reconciliation of the multi-wavelength infrared deep surveys. We thus propose
that the infrared burst phase of dusty starburst galaxies or AGNs from gas-rich mergers
at
could contribute significantly to the strong evolution of the IRAS
60
m, the ISO 15
m, 90
m, 170
m, as well as the SCUBA 850
m number
counts, while being compatible with the current observational limits of the cosmic infrared background and the redshift distributions.
The major difference of our current model prediction is that we see a fast convergence of the differential number counts
at 60
m
below 50mJy, which is about a factor of two brighter than other model predictions. Future infrared satellites like Astro-F or
SIRTF would give strong constraints to the models.
Key words: galaxies: evolution - galaxies: interaction - galaxies: starburst - galaxies: Seyfert
On the other hand, the source counts from present infrared and
submilimetre surveys, such as IRAS, ISO and SCUBA all
significantly exceed the non-evolving predictions. The extremely
strong evolution is seen from the differential counts of the
ISOCAM at 15m, with the remarkable upturn at
mJy and a fast convergence when
mJy. This
striking feature is based on the data from several independent sky
surveys (Elbaz et al. 1999; Chary & Elbaz 2001; Mazzei et al.
2001; Serjeant et al. 2001). Although there are many other
possible evolutionary scenarios which could explain the present
observations, the reason that we are encouraged to explore here a
merger-driven galaxy evolution picture with binary aggregation
dynamics is simply because the IRAS database and recent ISO and
sub-mm deep surveys indicate that most of the luminous infrared
sources are actually interacting/mergering systems. Also, the
local IR luminosity function shows an excess over the
Press-Schechter formula (Press & Schechter 1974; Lonsdale 1995;
Pearson & Rowan-Robinson 1996; Guiderdoni et al. 1998;
Roche et al. 1998; Rowan-Robinson et al. 1998; Dey et al. 1999;
Sanders 1999; Dole et al. 2000; Efstathiou et al. 2000; Silk & Devriendt 2000; Serjeant et al. 2001; Takeuchi et al. 2001).
Considering mergers as a possible formation mechanism of
Ultraluminous Infrared Galaxies both at high and low redshift, as well as their
significant infrared emissions, Wang (1999) and Wang &
Biermann (2000) discussed the effects of galaxy mergers on the
strong evolution of the IRAS 60m deep survey within a
binary aggregation galactic evolutionary scheme. In this model, the bright tail of the infrared luminosity function is
simulated in a consistent way for both the density and luminosity
evolution due to the decrease of the merger fration with cosmic time and a merger-triggered infrared burcst phase.
They found that a
luminosity-dependent infrared burst phase is crucial for the
interpretation of the steep slope within a flux range of
10mJy
Jy by the IRAS 60
m deep survey. This means
dusty starburst galaxies or AGNs from gas-rich mergers at high
redshift may experience an infrared burst phase around a
transition redshift
,
and fade quickly within the merger
time scale of that epoch. The more massive merger systems could
have such infrared emission enhanced to a higher level and
decrease even faster. This kind of speculation is based on the
observation that ULIGs are usually more than a
factor of 20 brighter than normal starburst galaxies. Although the
detailed mechanism for such enormous infrared emission is still
unclear, it is believed to be related to a special stage of the
merger process when the dust mass and temperature are both
dramatically increased (Kleinmann & Keel 1987; Taniguchi &
Ohyama 1998). Recent numerical simulations of the evolution of
dusty starburst galaxies by Bekki & Shioya (2001) shows that
there is a very strong photometric evolution during the merger
process of two gas-rich disks, and a dramatic change of the
spectral energy distribution (SED) over a cosmic time scale
Gyr, when the two disks of the merger become very
close and suffer from violent relaxation and the star formation
becomes maximal (
378
yr-1). The infrared flux
in this case could increase by one magnitude, especially for the
far infrared wavelength range (
m
m) in the
emitting frame.
The redshift distribution of the contributing sources for the
steep slope at faint IRAS m counts in the model of Wang
& Biermann (2000) shows that the infrared burst phase around
could have comparable significance to the local IR
sources. The question is then whether such an
infrared burst phase, or such a population of ULIGs, could also
sufficiently account for the strong evolution seen in other infrared
wavelengths, especially at the ISOCAM
m, ISOPHOT
m,
m and
SCUBA
m. We thus try to make a reconciliatory evolution model which could fit at least the present statistics of the
multi-wavelength deep surveys.
In this paper, we will first review the binary aggregation galaxy
evolution model by Wang (1999) and by Wang & Biermann (2000) in Sect. 2, where starburst/AGN activities may be triggered during the
merger process, as well as an infrared burst phase from gas-rich
mergers around a redshift of one. We will discuss the SED template we adopt in our calculation for
the nearby starburst galaxies and a possible strong evolution of the spectral energy distribution
of the dusty starburst merging systems
at .
We thus could further investigate whether the infrared
burst phase from gas-rich mergers around redshift
is
sufficient to account for the strong evolution also detected by
ISO and submillimetre deep surveys. One set of cosmological parameters, namely
kms-1/Mpc,
and
is adopted in the
calculation.
Considering different evolutionary characteristics of different
morphologies, we adopt in our study a multi-component model that
contains starburst galaxies, dust shrouded AGNs and spiral
galaxies as three major classes of infrared emitting sources. The
local luminosity functions of the spiral and starburst galaxies at
m from Saunders (1990), and that of Seyferts from Ruch et al. (1993)
are used to normalize the Monte-Carlo simulation. We adopt the
mass-light relation of blue starburst galaxies, which is given by
Cavaliere & Menci (1997) in a study of the excess of faint blue
galaxies in optical surveys. The abundances of dust-shrouded AGNs
are set to be
and
at local and high redshift based
on the statistics from Hubble Space Telescope imaging survey of
nearby AGNs and the cosmic X-ray background (Malkan et al. 1998;
Gilli et al. 1999).
The modelling of a luminosity-dependent ultraluminous infrared
burst phase from gas-rich mergers is described in detail by Wang
& Biermann (2000). Here we give a brief review of the basic
dynamics and the template SED we adopt in our model for the infrared luminous sources.
We introduce also in this section the construction of a SED for the dusty starburst mergers at ,
normally with the luminosity
,
in order to further
investigate such an evolutionary scenario at mid and far-infrared
wavelengths from the ISO deep survey. Considering that star formation is
triggered by mergers and proportional to
(
is the dynamical interaction time scale), Cavaliere & Menci
derived a mass-light ratio for dwarf galaxies,
(where
,
if the cross section is
purely geometrical).
could be used to
describe a redshift dimming, or a luminosity evolution. A power law prescription of
is adopted in the model. Simplifying the color and
K-corrections, they roughly get
.
We assume in
our model a luminosity ratio
,
which is consistent with current understanding
of the nature of ULIGs, where people normally believe that the
extremely infrared bright phase is due to the starburst merger
events with the far-infrared luminosity
enhanced both by the
accumulation of the dust mass and the increase of the dust
temperature. This burst phase could enhance the infrared
luminosity by a factor of about 20 over that of normal starburst
galaxies (Kleinmann & Keel 1987; Taniguchi & Ohyama 1998; Bekki
& Shioya 2001). We give
.
is adopted in the calculation, which not
only reasonably represents an infrared enhancement of about a
factor of 20 for a typical ULIG with a mass of
(the mass increases about one magnitude over that of normal starburst
galaxies), but successfully interprets the steep slope of IRAS
number counts. The scaling factor of the mass-light ratio here is
normalized by the local luminosity function of the IRAS deep survey. We found from our best fit results that a population of
infrared starburst sources, especially with spheroidal morphology, would have experienced very strong evolution
in the past. A rate of
in the luminosity evolution
since a transition redshift
indicates a very strong evolution for such a population of starburst
galaxies. This is at least comparable to, if not stronger than, QSOs (Roche et al. 1998; Lonsdale 1995; Pearson &
Rowan-Robinson 1996; Rowan-Robinson et al. 1998;
Sanders 1999; Franceschini et al. 1988, 2001;
Dole et al. 2000). A differential dimming is simulated by
below a
transition redshift
,
in order to match the observed
local luminosity function by the IRAS deep survey. The simulation
gives a best value
.
This power law
suppression also includes another physical reality, that the infrared
luminous galaxies at the bright tail of the luminosity function
become gas poor faster than the less luminous ones. Besides the merger rate decrease with cosmic time,
this physical effect is very important for a good fit of the steep
slope at IRAS
m number counts within a flux range of
100mJy
Jy.
We reviewed above the dynamics and some important physical
parameters in the current study, which are the same as those used
in the previous paper of fitting IRAS m number counts
(Wang 1999; Wang & Biermann 2000). In the following, we will
start to construct the spectral energy distribution for dusty
starburst mergers around redshift
,
in order to
extrapolate the calculation to mid- and far-infrared wavelengths.
Because of the unclear nature of the Ultraluminous Infrared Galaxies
at high redshift, we do not have a good understanding of the dust
environment and properties of these sources. An optically thin,
single-temperature dust model is adopted as a first order
approximation for a modified blackbody continuum of temperature
T at far-infrared wavelength in this calculation. The formula is
simply given by
.
is the dust opacity, and
the dust absorption
coefficient (
of
). In this case, the flux received at wavelength
is
,
where
is the scaling factor
for a conservation of the dust absorbed energy and re-emitting
energy,
is the wavelength in the
emitting frame. A flux ratio in the observer's frame could be
derived by
,
where h is Plank's constant, k is Boltzmann's constant and Cis the speed of light. With a reasonable assumption for the dust
emissivity power
and the dust temperature T, we can
easily extend our calculation to far-infrared wavelengths. The
mid-infrared emission is more complicated than that of the
far-infrared which could be well described by a single temperature
blackbody spectrum by cold, large grain dust. The MIR emission
properties are usually dominated by the radiation field of heated
small grains and PAHs. These dust grains are normally heated
stochastically, and are not in thermal equilibrium with the
ambient radiation field. Thus the MIR continuum is most like a
power law spectrum. In this calculation, we will not provide a
detailed modelling of the MIR emission feature.
Instead, we only
modify the template SED of the starburst galaxy Arp 220 by Silva et al. (1998), with the observational correlations of
S15/S60by IRAS and ISOCAM deep surveys for the ultraluminous case
(
)
to represent the dusty starburst
merging system around redshift one. The color ratio of
S15/S60 for the ULIGs is a factor of about 5 lower than
the mean value of the whole sample, which may imply a very
complicated process to heat small grains during the merger process
(Aussel et al. 2000; Dunne et al. 2000; Saunders et al. 2000;
Chary & Elbaz 2001). Although there are many indications that the
MIR continuum is correlated with the temperature of large grain
dust (Dale et al. 2001), there is still no exact modelling for
such a process. Our goal in this paper is to construct a simple
SED for those luminous starburst mergers based on various
observational correlations, which not only represents the observed
trend for individual samples of a certain luminosity bin
(
), but matches the statistical results
from the multiwavelength deep surveys. The number of sources
in a comoving volume
within the flux range
to
,
measured at wavelength
,
is
defined by:
,
,
where
is the K-correction, and
is the luminosity distance in a
dominated universe.
The infrared background in this calculation gives 2.4
at 15
m,
1.9
at 60
m, 3.8
at 90
m, 10.6
at 170
m, which are all consistent
with current upper limits from TeV detections, COBE results and the resolved fraction of the CIRB by the deep ISO
surveys (Funk et al. 1998; Guy et al. 2000; Hauser & Dwek 2001).
The redshift distribution of the ISOCAM 15 m contributing
sources within the detected flux range (
)
from our model calculation is shown in Fig. 6. It gives a
rough statistical finding that these luminous infrared sources
cover a wide redshift range of
,
peaking at
.
Comparing our model prediction and the redshift distribution
of 15
m sources with
in the HDF North and
the z-distribution of sources in the CFRS field (Flores et al.
1999; Cohen et al. 2000; Aussel et al. 2001), we found that the
starburst mergers at
in our model are good candidates
for a strongly evolving population that results in the strong
evolution in mid- and far-infrared deep surveys. Recent redshift
estimation from sub-mm follow-up of 10 known FIRBACK 170
m
ISO sources by Scott et al. (2000) suggests that they are in a
redshift range of
,
still consistent with our current
model predictions. However, these redshift determinations strongly
depend on the assumption of the dust properties. We need further
accurate measurements for the robust constraints of the models. We
discussed in a previous paper that shifting the peak redshift of
these ULIGs by a factor of 2 could affect the source count fitting
of the IRAS 60
m deep survey, especially for a low redshift
peak (z<0.5). Strong evolution of the ULIGs to
may be
the most reasonable case for the existing model constraints from
both the infrared deep surveys and the cosmic infrared background
upper limits from high energy TeV detections, as well as the
indicated star formation history by UV/optical deep surveys (Lilly
et al. 1996; Connolly et al. 1997; Madau et al. 1998).
We plot the redshift distribution of ULIGs (
)
to understand the evolutionary
properties of the ULIGs from mergers in our model. A rapid
increase in the number density of ULIGs up to
is seen in
Fig. 7, which is actually consistent with a scenario
where galaxy merger rates increase dramatically during that epoch
as seen from various observations and theoretical considerations
(Zepf & Koo 1989; Carlberg 1992; Burkey et al. 1994; Carlberg et al. 1994). However, the number density of ULIGs decreases beyond
2-3, which may reflect a stage when merger pairs are
mostly dwarfs and the infrared emissions are less than
even with intensive starburst activities
triggered by mergers. In this scenario, an infrared luminous tail
of the luminosity function may form at
,
with enormous
infrared emission enhancement.
There is still no firm statistical basis for the classification of starbursts and AGNs from current spectroscopies. We thus adopt the observed AGN local luminosity function of Rush et al. (1993) as a model constraint, and assume in our calculations that the observed starburst galaxies and Seyferts follow the same evolutionary track, based on naive thinking that starburst/AGN may both be triggered by galaxy interactions. We know that the subtle differences in the dust emission properties could result in a different fraction of their contribution. This is still far too uncertain to discuss this here. We thus adopt only the SED of the Cloverleaf quasar which represents a phase poor in cold gas, as well as the dust enshrouded phase of F10214+4724 as two typical AGN templates in our calculation. We know from the result that the AGN contribution is only a small fraction of the whole and our current model prediction is within the present understanding of this issue, i.e. the starburst powered ULIGs are dominating over the AGN powered ones (Fig. 7) and may take over at higher redshifts and in the higher luminosity case (Lutz et al. 1998; Tran et al. 2001).
![]() |
Figure 1:
The model prediction of the differential number counts of ISOCAM ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
The fitting of the ELAIS
differential source count at ![]() |
Open with DEXTER |
![]() |
Figure 3:
The result of FIRBACK ![]() |
Open with DEXTER |
![]() |
Figure 4:
The fitting of the IRAS ![]() |
Open with DEXTER |
![]() |
Figure 5:
Integral number counts at ![]() |
Open with DEXTER |
![]() |
Figure 6:
The redshift distribution of three infrared contributors (starburst galaxies, spiral galaxies and
AGNs) at a flux range of
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 7:
The redshift distribution of ultraluminous infrared sources (starbursts or AGNs)
with
![]() ![]() |
Open with DEXTER |
Lacking a comprehensive understanding of the gas and dust
environment of faint ISO sources, especially the starburst merging
system at ,
we adopt the template spectral energy
distribution of Arp 220 by Silva et al. (1998) as
typical for nearby starburst galaxies. Meanwhile, we construct a
simple SED for the starburst mergers at
.
The
far-infrared emission in such a system is modelled by a single
temperature, optically thin dust law with modified black body
emission. The MIR emission feature is assumed to be similar to Arp 220, but modified by the flux correlation from IRAS and ISOCAM
observations. In this case, we can further investigate such a
merger-driven galaxy evolutionary scenario at other infrared and
submillimeter wavelengths with ISO and SCUBA deep surveys. Our
calculations shows that the current results of multi-wavelength
deep surveys at ISOCAM 15
,
ELAIS 90
,
FIRBACK 170
,
IRAS 60
and SCUBA 850
number counts could be
sufficiently accounted for by the merger-triggered infrared
enhancement at
from our model with the dust temperature
(
,
and
), slightly higher than the
local starburst galaxy Arp 220. Future accurate redshift
measurements and multiband photometries would provide a robust
model check.
The background levels at these wavelengths are estimated from our
model calculation, which gives 2.4
at
15
,
3.8
at
90
,
10.6
at 170
,
still compatible with the cosmic infrared
background level both from the upper limit of high energy TeV
ray detection of nearby Blazars
and from COBE and ISO results.
The redshift distribution of the luminous infrared sources within the
ISOCAM
detection flux range (
)
from our calculation is plotted in Fig. 6. The redshift
distribution of these sources cover a wide redshift range from
and
peak around a mean redshift of
.
We also plot the redshift distribution of ULIGs (
)
in Fig. 7. It shows a strong
increase in the ultraluminous infrared population until a mean
redshift
,
and decreases by a factor of about 2 by
2-3. This is probably the major difference between our current
calculation and other models, and indicates that the infrared
luminous tail may be produced at the cosmic epoch of
,
when the merger rate and the size of parent galaxies are suitable
for such an infrared emission enhancement.
A brief discussion about the fraction of contribution from AGNs
and starbursts from our calculation is given in Sect. 3. We assumed in the model a similar evolutionary
track for the starburst galaxies and AGNs based on the idea that
both AGNs and starbursts may be triggered by galaxy interactions,
where the AGN population is constrained by the observed Local
Luminosity Function of Seyferts from Rush et al. (1993). Given the
uncertainty of the dust properties of ULIGs, especially for those
harboring an AGN in the center, we adopt here only two typical SED
templates of the Cloverleaf QSO and F10214+4724. In this case, we give
a rough estimation of the relative abundance of AGN and
starburst-powered ULIGs (
)
of
1/5, which seems to be close to the recent submillimeter
observations of Chandra X-ray sources (Almaini et al. 1999; Barger
et al. 2001; Gunn & Shanks 2001).
Acknowledgements
We acknowlege helpful discussions with Profs. P. L. Biermann, M. Harwit, N. Arimoto, and Drs. K. Kawara, Y. Taniguchi, T. Yamada. YPW would like to thank the anonymous referee for the kind suggestions and comments, which helped to improve the paper significantly. YPW is supported by NSFC 10173025 and Chinese post-doctoral science foundation. YPW is also very grateful for the hospitality of NAOJ staff and COE fellowship of Japan, where the final stage of this work was finished.