A&A 422, 941-950 (2004)
DOI: 10.1051/0004-6361:20040283
V. Casasola1 - D. Bettoni 2 - G. Galletta1
1 - Dipartimento di Astronomia, Università di Padova, Vicolo
dell'Osservatorio 2, 35122 Padova, Italy
2 - Osservatorio Astronomico di Padova, Vicolo dell'Osservatorio 5,
35122 Padova, Italy
Received 17 February 2004 / Accepted 28 April 2004
Abstract
A study of the gas content in 1038 interacting galaxies, essentially selected from Arp,
Arp & Madore, Vorontsov-Velyaminov catalogues and some of the published literature, is presented here.
The data on the interstellar medium have been extracted from a number of sources in the literature
and compared with a sample of 1916 normal galaxies. The mean values for each of the different ISM tracers
(FIR, 21 cm, CO lines, X-ray) have been estimated by means of survival analysis techniques,
in order to take into account the presence of upper limits.
From the data it appears that interacting galaxies have a higher gas content than normal ones.
Galaxies classified as ellipticals have both a dust and gas content one order of magnitude higher
than normal. Spirals have in most part a normal dust and HI content but an higher molecular
gas mass. The X-ray luminosity also appears higher than that of normal galaxies of same morphological type,
both including or excluding AGNs.
We considered the alternative possibilities that the molecular gas excess may derive from
the existence of tidal torques which produce gas infall from the surrounding regions or from
a different metallicity which affects the X conversion factor between the observed CO line
luminosity and the H2 calculated mass. According to our tests, it appears that interacting
galaxies possess a higher molecular mass than normal galaxies but with a similar star formation
efficiency.
Key words: galaxies: catalogues - galaxies: ISM - galaxies: interacting
Inside galaxies, gas is expected to reflect the effects of the interaction more strongly than stars. A recent study on the interstellar medium (ISM) in 104 peculiar galaxies (Bettoni et al. 2001) shows that polar ring galaxies have a gas content one order of magnitude higher than normal ones. In many cases, the higher gas content is visible even if the interaction with the environment has ceased for long time.
We wonder if a peculiar gas content is detectable in those galaxies where interaction is currently ongoing, such as some systems contained in the Vorontsov-Velyaminov (1959), Arp (1966) and Arp & Madore (1987) catalogues. Here we present the results of a selection of 1038 objects extracted from these catalogues. For these galaxies data on dust, HI, molecular gas and X-ray luminosities are available in the literature. The properties of the different components of the interstellar medium are studied here and compared with those of a sample of 1916 normal galaxies (Bettoni et al. 2003).
To analyze the ISM of these sample galaxies, we searched in the literature for survey
papers (e.g., Casoli et al. 1998; Fabbiano et al. 1992; Beuing et al. 1999; Roberts et al. 1991; O'Sullivan et al. 2001; Burstein et al. 1997; Knapp et al. 1989)
and single studies in four different tracers: warm dust from 60 and 100 m
observations; atomic gas from 21 cm data; molecular gas from CO line observations
and X-ray luminosities from Einstein and Rosat satellite observations. Many data come
from LEDA
galaxy catalogue (Paturel et al. 1997) and from the
NASA/IPAC extragalactic database (NED
).
The list of sample galaxies has been compared with the Véron-Cetty & Véron (2003) catalogue, to
check the eventual presence of active galaxy nuclei (AGN). The number of galaxies
with data (detections and upper limits) and the number of detections only are listed
in Table 2, binned according to the morphological type.
To homogenize all these data, we reduced the published values to the distances extracted from a single source, the LEDA on-line archive. The LEDA distance moduli are mainly derived from redshifts, corrected for Virgocentric inflow and adopting H0=70 km s-1 Mpc-1. When redshift was not available, we used the photometric distance modulus, if present. For each galaxy we also collected the morphological classification, the diameter and the blue luminosity, corrected for cosmological reddening and galactic- and internal-dust absorptions. The blue corrected luminosity LB in solar units and the square of the face-on corrected diameter in kpc Log D2 were used as normalization factors for the gas masses and the X-ray luminosities. However, due to the similarity in the results obtained with one or other normalization factor, we present only the data normalized to LB.
The LEDA morphological type t codes derive from RC3 (de Vaucouleurs et al. 1991) with some changes, as visible in our Tables 2 and 3. Type 5 includes systems with uncertain classification (type S?). In our sample 25 galaxies are unclassified in LEDA. We attributed the morphological type code t=11 to these systems in order to plot them in our graphs as well. We remember that the morphological classification may be misleading for very perturbed objects, and when present in interacting galaxies reflects the present-day observed morphology and not the original one. The difference we found in the comparison with normal galaxies then will reflect the difference between the present, may be transient, morphology and that of a stable galaxy with similar appearance.
Table 2:
Number of interacting galaxies observed (N) and detected ()
according to the different ISM
tracers. The column labeled "All'' represents the number of galaxies present in each morphological type
bin. The type 11 is assigned here to galaxies whose structure is not attributed to a specific morphological type in
LEDA.
We have estimated total gas masses and/or luminosities following standard procedures.
From the value of the IR fluxes at 60 and 100 m we estimated the temperature
(Kelvin) of the dust emitting at such a wavelengths:
![]() |
(1) |
![]() |
(2) |
Table 3: Mean Mass to light ratios of interacting galaxies for the different tracers, binned according the morphological types. For X-ray data, the corresponding luminosities are considered instead of the masses. All values are normalized to LB, the corrected blue luminosities and are in solar units. The values have been obtained with survival analysis technique.
HI gas masses were derived from 21 cm fluxes![]() |
(3) |
![]() |
(4) |
The formula used to derive molecular gas masses from CO(1-0) line fluxes (
in Jy km s-1) is:
![]() |
(5) |
X-ray luminosities may derive from diffuse gas and/or from discrete sources. Very recent observations with the Chandra satellite indicate there is the possibility for some galaxies that X-ray fluxes are "contaminated'' by the presence of X-ray binaries which must be taken into account in discussing the ISM physical properties (Kim & Fabbiano 2003). We shall discuss the data according to the different models.
When no observed fluxes but only mass values were available in the consulted references, these have been scaled to the distances assumed here. If data from different sources are available for a galaxy, mean values have been calculated. When both upper limits and detections are available, only detections have been used to compute mean values. Moreover, if several upper limits exist, only the lowest value has been adopted.
In order to define the mean content of each tracer according to the morphological
type, we made use of survival analysis methods (see Feigelson & Nelson 1985). This method
takes into account both detections and upper limits (UL) to derive representative
averages. Our survival analysis was performed by means of the statistical package
available in IRAF, using the Kaplan-Meier estimator. Mean values and errors listed
in Table 3 derive from survival analysis of our data. When all the galaxies
are detected, the survival analysis means coincide with arithmetic means. Conversely,
the method cannot be applied when both upper
and lower limits are present, for instance when considering the mass ratios (e.g.
/
). In such a case, we present ratios of mean values. No survival analyses
were made for unclassified galaxies (our type t=11).
![]() |
(6) |
![]() |
(7) |
Another parameter describing the stellar formation is the star formation efficiency, SFE,
indicating how much stars are produced by a unit of molecular mass. This value may be
estimated dividing the FIR luminosity by the total mass of
molecular hydrogen:
![]() |
(8) |
![]() |
Figure 1: Histograms for absolute blue magnitude ( left panel) and morphological type code ( right panel) of interacting galaxies (full lines) compared to that of normal galaxies (dotted lines). |
Open with DEXTER |
The mean values of gas content obtained with survival analysis for each tracer (dust, HI,
molecular gas, X-ray) and for each morphological type are presented in Table 3
and Fig. 2. We tested the statistical significance of the differences found
between interacting and normal galaxies samples applying a T-test to the mean values and
an F-test to the variances. The significance of the single values has also been checked by
a test, assuming the values of normal galaxies as expected values. The results of all
these tests are presented in Table 4.
Due to the different behaviour of galaxies with different t, we decided do not make the
comparison between the two samples mixing together galaxies of all morphological types, as
sometimes seen in the literature. For statistical tests only, we separated galaxies in two
large categories that present similar properties in all the tracers, as explained in the next
sections: early types (
)
and late types (
).
Table 4: Parameters of the statistical tests applied to our sample galaxies in comparison with normal ones (see text). Columns labeled P() represents the probability that the observed differences are statistically significants.
![]() |
Figure 2:
Comparison between mean values from Table 4 found for interacting galaxies and
corresponding means for normal galaxies (Bettoni et al. 2003). Bars represent 1 ![]() |
Open with DEXTER |
Looking at the Log
/LB values obtained with survival analysis for each morphological
type (Fig. 2a), one can see that the warm dust content is
quite similar to that of normal galaxies for types later than t=2 (Sab) but higher for early
type galaxies. Early type interacting galaxies have in mean a warm dust content one order
of magnitude higher than normal galaxies of the same type. According to statistical tests
(Table 4) the significance of this difference is high: the two samples of early type
galaxies have mean values that differ by more than 95% of significance. They show however a
similar dispersion of values (F-test). Late types show on the contrary a lower difference between
normal and interacting galaxies and the similarity in dust content is stronger if the comparison
is restricted to type from Sab to Sd, excluding Irregular galaxies (see the third line of
Table 4).
This excess of mass content seen in early type galaxies is also evident for atomic gas
(log
/LB, see Fig. 2b) but does not seem to be present as far as the
molecular gas content is considered (log
/LB, see Fig. 2c). The statistical
tests show a significant difference for HI and a strong similarity for molecular gas. Ellipticals
and S0 galaxies (t<-1) have low values of Log
/LB with respect to the sequence
of spiral galaxies, but are based in total on 10 detections only, compared to 71 detections for
early type systems in the sample of normal galaxies.
The mean Log
/LB for late type interacting galaxies appears similar to that of normal
ones to an higher statistical significance. For these types, the differences between
interacting and normal galaxies appear when log
/LB is considered. The two samples
show the same dispersion of values but mean molecular gas contents significantly
different, being interacting galaxies richer than normal ones.
In the literature, Horellou & Booth (1997) find that the HI surface density in interacting
galaxies, mediated over all the morphological types, is equal to
Kpc-2. This value is lower of that found by Haynes & Giovanelli (1984) for "isolated'' galaxies,
corresponding to
.
In this latter sample most part of detected galaxies are
of morphological type from Sa to later. On the contrary, our values of Log
/D225
show that early type interacting galaxies have an higher HI content per Kpc2 than normal,
as seen for Log
/LB curves plotted in Fig. 2b, while the values are similar
for late-type. If we mediate our values from survival analysis over the morphological types
we obtain a mean value of Log
/D225 =
,
based on 518 detections and 38 upper
limits. This value increases to 6.9
0.2 if types from Sa to Irr only are considered. On the
contrary, mediating all the Log
/D225 values published by Bettoni et al. (2003) and based on 891
detections and 220 upper limits, we find a mean Log
/D225 value of
that
becomes
if late type galaxies only are considered. Both values of normal
galaxies are consistent with that found by Haynes & Giovanelli (1984) in their sample of 288 detected galaxies.
![]() |
Figure 3: The ratio between molecular and atomic gas according to the morphological type t. Mean ratios for interacting galaxies are compared with that of normal galaxies and with the "mixed'' sample of Casoli et al. (1998). |
Open with DEXTER |
The change of the molecular/HI gas ratio with morphological type has been studied
calculating (log
/
)
from the mean values of log
/LB and
log
/LB of Table 3. The resulting mol/HI ratio appears very low for
early-type systems, if compared with those of normal galaxies (Bettoni et al. 2003) and similar
to them for types later than -2. Similar low values for early type galaxies were found by
Casoli et al. (1998). Looking at the papers cited as sources of their data (see in Casoli et al. 1998),
we found that in their sample 126 galaxies out of 456 (28%) are peculiar, having disrupted
structures (e.g. NGC 520) or polar rings (e.g. NGC 2685) or counter-rotations, all phenomena
typical of interaction with the environment. This means that the Casoli et al. (1998) values
are contaminated by the presence of these peculiar galaxies and that early-type interacting
systems really have lower mol/HI values with respect to non-interacting, normal galaxies.
Resuming the above considerations, we find a difference in the colder component of those interacting galaxies classified early-type (t<0) with respect to those classified as late-type (t>0): the first ones have dust and HI overabundance, even with respect to their molecular gas, while the second are richer in molecular gas and are quite normal, on average, for dust and atomic gas content. This behavior may be due to the characteristics of the interaction and the subsequent evolution of the gas, as discussed later.
The fact that interacting galaxies have a luminosity function brighter than normal ones, as visisble in Fig. 1, strengthen our conclusion on gas overabundances. In fact, the adopted normalisation factor LB should reduce the log M/LB for interacting galaxies, contrary to what observed.
![]() |
Figure 4: The X-ray luminosity (W m-2) versus the blue luminosity (solar units) for interacting galaxies. Active galaxies are represented as open circles. A bar under the circle indicates an upper limit. The lines represent the values expected from hot discrete sources (dotted line) or hot diffuse gas (dashed-dotted line). The full line is the limit expected from steady-state cooling flows (see text for details). Most of detected sources lie near the line of hot diffuse gas. |
Open with DEXTER |
![]() |
Figure 5: The absolute far infrared luminosities vs. morphological type code of the single interacting galaxies (full circles) compared to the single normal galaxies included in the sample of Bettoni et al. (2003). |
Open with DEXTER |
![]() |
Figure 6:
The star formation efficiency per unit of molecular gas mass, indicated by the ratio
![]() |
Open with DEXTER |
To understand the origin of the emission, we plotted the log
of the single galaxies
versus log LB and we compared the observed values with theoretical expectations. In the plot
the representative points of galaxies hosting nuclear activity are distinguished by open symbols.
In Fig. 4 the lines corresponding to the X-ray emission foreseen for diffuse gas
(labeled
)
and from discrete sources inside the galaxy (
)
are plotted.
In the first case, the X-ray luminosity would be proportional to the square of blue luminosity
LB while in the second one it would be directly proportional to LB (Ciotti et al. 1991).
Most of our non-AGN galaxies are clustered around the line expected from emission of diffuse
gas. Their luminosity is lower than the values expected for cooling flows (line labeled
)
assuming a SN rate of 0.18 (Cappellaro et al. 1999). Both these aspects are consistent with
the hypothesis that this X-ray flux derives from gas stripped from the galaxies by ram pressure
during their motion or, alternatively, from gas emitted by SN winds in the intergalactic space.
The origin of this gas has been studied in detail in the literature for some of our systems: Davis et al. (1996) found a diffuse X-ray emitting gas confined near the center of the NGC 2300 group (Arp 114) in addition to point sources localized on the single galaxies. The presence of diffuse gas in a common potential well is a feature visible in other galaxy pairs (Henriksen & Cousineau 1999) but some cases (e.g. Arp 116 = NGC 4647/49) have X-ray emission coming from a single member of the pair only. Given the various X-ray morphologies, a common model in the origin of the X-ray emitting gas is difficult to apply to all our interacting galaxies.
In the first one, interacting galaxies often appear more luminous in infrared than normal ones, but with a large spread of values. High FIR luminosity is evident in those systems with no morphological classification (right edge of the figure). These latter are often systems with disrupted morphology, probably in a phase of strong merging, that appear to also be actively star-forming. We must note, however, that also normal galaxies (i.e. non-perturbed, with no polar ring or known kinematical decoupling between gas and stars) may exhibit high values of star formation rate, even those classified as ellipticals.
The star formation efficiency (Fig. 6) appears with a similar dispersion of values.
Despite a higher FIR luminosity and star formation rate, the interacting galaxies however
do not seem to be more efficient in forming stars. The two distributions intersect together but
with a clear deviation of interacting systems toward higher levels of
.
This effect is
different from that found by Solomon & Sage (1988); Horellou et al. (1999) and by Combes et al. (1994) which suggest, for
interacting pairs, an higher efficiency in forming stars than pairs with larger separation.
However, Combes et al. (1994) find that in their sample of paired galaxies log
/M(H2)
does not change in function of the separation between components.
The range of SFE values of our galaxies is larger than that of the previous authors
and covers the differences found by them. This may explain our lack of differences between our
samples of interacting and normal galaxies. In our data, even binning the interacting galaxies
in groups of different separation (i.e. merging,
semidetached components, separated components with tails or bridges, single galaxies), the plot of
Fig. 6 does not change and the groups have representative points mixed together. At the
same time, there are no apparent dependences from the morphological type t, including
the unclassified objects (t=11).
We find that the morphological type also has a role in determining the properties of these stellar systems and that may be not correct to mix together interacting galaxies of all morphological types.
In addition, the comparison with the literature is complicated by the fact that very often authors use as "normal'' galaxies a list of objects that contains known cases of peculiar galaxies. For instance, in Table 3 of Devereux & Hameed (1997) is visible NGC 660, a disrupted polar ring galaxy. Haynes & Giovanelli (1984) include several Arp and VV galaxies while Casoli et al. (1998) present a sample containing many dynamically peculiar galaxies. The data of these papers have been used by other authors to deduce properties of peculiar galaxies in comparison with "normal'' ones. The sample of normal galaxies by Bettoni et al. (2003) used here may also contains galaxies that in the future may be discovered as "peculiar'', but have been already cleaned by all the known cases of interacting and dynamically peculiar galaxies known.
From our data, we see that interacting or disturbed galaxies have always more gas than normal ones, and that this gas is mainly in the form of atomic hydrogen in early type systems, while it is in molecular form for galaxies of latest types. All have a higher X-ray luminosity. Because of these differences, we shall discuss the early type systems separately from the spiral ones, comparing our finding with that from literature.
![]() |
Figure 7: Distribution of the representative points for 179 galaxies in pairs or triplets present in our sample for which detections or upper limits are available. The higher mass member (primary) is shown with a full diamond while the lower mass ones (secondary) are indicated with a cross. |
Open with DEXTER |
A more complicated scenario appears from the data concerning galaxies classified as spirals. In these systems the mean HI and dust content seems to be similar to that of an unperturbed, normal galaxy of the same morphological type. In addition to the excess in X-ray luminosity, these galaxies seem to have a molecular gas content always one order of magnitude higher that that of normal galaxies. There are two possible explanation for this effect.
The first possibility is that during the collision the galaxies stimulate a gas inflow
by means of gravitational torques, that enhance the CO luminosity because of a massive
accumulation of molecular gas. This hypothesis has been presented by Combes et al. (1994)
to explain the observed characteristics of a sample of 51 paired galaxies, whose interacting
subsample appears more luminous in CO line than the remaining binaries with higher separation.
We tested the idea that the gas is simply exchanged by the two components of the same
pair, enriching one of them and depauperating the other. We call these galaxies primary and
secondary respectively. In our sample there are 179 galaxies in interacting pairs or triplets
where the single galaxies are separated and have literature data at 21 cm or CO lines. Plotting
in Fig. 7 (upper panel) the values of log
/LB of these galaxies only and
identifying with different symbols the primary and the secondary member, one can see that spirals
which are primary members have - on mean - higher values than normal galaxies of same morphological
type. On the contrary, the secondary members, of whatever morphological type, have values generally
lower, below that of normal galaxies. This is in agreement with the idea that part of atomic gas
has been transferred between them. However, this is not true when log
/LB values are
considered (Fig. 7, lower panel): in this case, the primary members of the pair show
an excess of molecular mass and the secondary members have in general values near or over the mean
curve of normal galaxies. We may suppose that less dense, atomic gas, may be exchanged during an
encounter more easily than the molecular gas, condensed in clouds. The existence of a molecular
mass excess in these paired galaxies is then difficult to explain supposing a simple transfer
during the close encounter.
The second, completely different possibility is that the molecular gas excess is not real but due
to different physical conditions inside the galaxy. In the totality of cases presented here
the
is determined from CO line luminosities, transformed in solar masses adopting a X conversion factor, as explained in Sect. 2.1. This value depends on the metallicities
or on the physical conditions of the molecular gas and in different kind of galaxies may assume a wide
range of values (see the discussion in Sect. 2.2). An overestimate of the molecular mass
may derive by the assumption of a X factor lower than real. Among the various possibility, the
presence of gas of high temperature may overestimate the H2 mass. However, in some well known
cases present in our sample, such as the already cited "Antennae'' or "Atoms for Peace'' (PGC 68612),
believed to be the result of a two spirals merger, the H2 mass excess is accompanied by an HI
mass excess. Similarly the "Whirlpool galaxy'' M 51 and its satellite (PGC 47404 and PGC 47413), show
an excess of gas mass. In cases like these, we may think that the gas excess is real for the
molecular component like it is for the atomic one.
Another suggestion on the reality of the higher molecular masses comes from the analysis of the star formation rates and efficiencies: interacting galaxies appear more luminous in the infrared (Fig. 5) because of the higher number of newly formed stars heating the dust (Thronson & Telesco 1986). This higher star formation rate does not correspond to a different efficiency in star formation per unit of calculated H2 mass. This may be explained if the higher star formation is simply due to a higher molecular gas quantity. Higher H2 mass and concentration in perturbed galaxies was already suggested by Braine & Combes (1993). In our data, the large spread of data in Fig. 6 may indicate that the X conversion factor is different between galaxies but even in this hypothesis, the range of variation inside interacting galaxies does not appear different from that of normal, non interacting stellar systems. Our conclusion are different with the higher SFE found in interacting galaxies by previous authors (Solomon & Sage 1988; Horellou et al. 1999; Combes et al. 1994). As discussed before, their comparison sample contains peculiar galaxies and the large spread of values found in our sample cover the differences found by them. We must conclude that, with the present data, our sample of interacting galaxies appears richer in molecular gas and in some cases also in atomic gas.
It is not clear from where this mass excess originates. An excess of gas, both atomic and molecular, of one order of magnitude is known in the literature for another class of peculiar objects: the polar ring galaxies (Bettoni et al. 2001). In that case, the matter forming the ring is accreted from outside. May this also be the case of interacting galaxies?
Acknowledgements
The authors thanks Dr. C. Horellou for her useful suggestions during the revision of this paper. This research made use of LEDA database, the Vizier service (Ochsenbein, Bauer, & Marcout 2000), the SIMBAD database (operated at CDS, Strasbourg, France), the NASA/IPAC Extragalactic Database (NED) (which is operated by JPL, California Institute of Technology, under contract to NASA) and NASA's Astrophysics Data System Abstract Service (mirrored in CDS of Strasbourg). G.G. made use of funds from the University of Padova (Fondi 60%-2003).