COLLISIONAL DEBRIS AS LABORATORIES TO STUDY STAR FORMATION

The goal of this paper is not to present a complete picture of intergalactic SF. The systems we have studied were mainly chosen based on the presence of prominent star-forming regions in their collisional debris. As shown later, they are not representative of the Local universe. Nevertheless, they offer the best laboratories where SF can be studied in detail in this environment. All of them benefit from extensive multiwavelength observations. In particular, they all have been observed in the UV by GALEX and in the MIR by Spitzer. Besides the space-borne data, ground-based spectrophotometry is available. A high oxygen abundance, typical of the outer regions of spiral disks (about half solar), has been measured in the H ii regions throughout their collisional debris (Duc & Mirabel 1994, 1998; Duc et al. 2000, 2007). Despite common characteristics—active SF, high metallicity, environment—the six systems presented in this study turn out to present a variety of origins (early tidal interaction, ongoing and final merger, direct collision) and environments (field, groups and cluster of galaxies). For most of them, numerical simulations able to account for their perturbed morphology have already been performed. Details on individual systems are given below in their description. Additional details are summarized in Table 1. Optical images of all systems are shown in Figures 1–6.

Table 1. Position and Distance of the Selected Systems

System	R.A. (J2000)	Decl. (J2000)	Distance (Mpc)	Spatial Resolution for 1 (pc)
Arp 105	11h11m131	+28°41'57	115	557.5
Arp 245	09h45m453	−14°20'50	31	150.3
NGC 5291	13h47m245	−30°24'25	62	300.6
NGC 7252	22h20m448	−24°40'42	64	310.3
Stephan Quintet	22h35m575	+33°57'36	85	412.1
VCC 2062	12h48m000	+10°58'15	17	82.4

Note. The coordinates have been obtained from the NASA/IPAC Extragalactic Database (NED).

Download table as:  ASCIITypeset image

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

2.1.1. Arp 105

2.1.2. Arp 245

2.1.3. NGC 5291

2.1.4. NGC 7252

2.1.5. Stephan's Quintet

2.1.6. VCC 2062

The properties of the individual star-forming regions within collisional debris have been systematically compared to those located in the disks of four spirals. This broadband reference sample consists of the following galaxies. (1) M51, a well studied spiral galaxy located at 8.2 Mpc (see for instance Calzetti et al. 2005), (2) M81, the main galaxy of the M81 group, located at 3.63 Mpc; its star-forming regions have been studied by Pérez-González et al. (2006), (3) Arp 24, an interacting system at 27.6 Mpc which hosts massive star clusters studied by Cao & Wu (2007), and (4) Arp 82, an interacting system at 57 Mpc exhibiting a long tidal arm containing multiple star forming knots, as illustrated in Hancock et al. (2007).

While Arp 24 and Arp 82 are interacting systems, they are still in an early phase. Contrary to all objects in the primary list, SF still mainly occurs in the galaxy under the "classical" regime and not along a tidal tail. This explains why they are included in the reference sample.

Since the most luminous star-forming objects in collisional debris have the luminosities of dwarf galaxies, it is also relevant to make a comparison with the integrated, broadband properties of individual galaxies. Large reference samples are available. In this study we made use of the following ones: (1) the 75 spiral galaxies of the SINGS survey (Kennicutt et al. 2003; Dale et al. 2007), (2) the 19 dwarf galaxies with M_B > − 19 from the sample by Rosenberg et al. (2006), and (3) the 66 starburst galaxies from the sample by Engelbracht et al. (2008).

As we also study the MIR spectral properties of tidal debris we have gathered several samples having Spitzer Infrared Spectrograph (IRS) observations, namely (1) galaxies from the BCDG sample of Wu et al. (2006), (2) the starburst sample of Engelbracht et al. (2008), and (3) individual star-forming regions in M101 (Gordon et al. 2008).

Three star-forming tracers—the UV, Hα, and MIR emission—were used for the present study. Their observation is described here and summarized in Table 2.

2.3.1. Ultraviolet Imaging

2.3.2. Hα Imaging

2.3.3. Mid-Infrared Imaging

2.3.4. Aperture Photometry

Unpublished archival Spitzer/IRS MIR spectroscopic observations are presented here of two TDG candidates in Arp 105 and Arp 245. These observations were carried out using the Short–Low (SL: 5.2–14.5 μm; R = 64 ∼ 128) and Short–High (SH: 9.9–19.6 μm; R = 600) modules of the Infrared Spectrograph (IRS; Houck et al. 2004). Arp 105S and Arp 245N were observed for three cycles of 60 s in the SL mode and three cycles of 120 s in the SH mode on 2005 December 12 and on 2005 May 30, respectively. The data were processed with the Spitzer Science Center (SSC) pipeline version S13.0.1. The reduction started from the intermediate pipeline products (droop files), which only lacked stray light and flat-field correction. Individual points to each nod position of the slit were coadded using median averaging. Sky background from the SL spectral data was taken out by differencing the images of the two orders in this module. Then the differenced images were extracted with the spectral modeling, analysis, and reduction tool (SMART; Higdon et al. 2004) using a variable width aperture, which scales the extraction aperture with wavelength to recover the same fraction of the diffraction-limited instrumental point-spread function. The SH data were extracted with the full slit extraction method in SMART. Finally, the spectra were flux-calibrated by multiplying by the relative spectral response function (RSRF), which was created from the IRS standard stars αLac for SL and ξDra for SH for which accurate templates are available (Cohen et al. 2003). The SL spectra are presented in Figure 7 for Arp 105S (corresponding to region 1 in Figure 1) and 8 for Arp 245N (corresponding to region 2 in Figure 2)

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Tables 12 and 13, we present the measurements of broad and narrow emission-line features from the IRS spectra of Arp 105S and Arp 245N. We have also measured the equivalent width (EW) of NGC 5291N and NGC 5291S independently of Higdon et al. (2006). Indeed, this allows an easier comparison with the data of Wu et al. (2006) whose EWs have been measured using the same method. The EW of the PAH features are measured from the SL spectra. The PAH EW were derived by integrating the flux of the feature in the mean spectra of both nod positions above an adopted continuum and then dividing by the average adopted continuum. The baseline was determined by fitting a spline function to selected points. The wavelength limits for the integration of the features were approximately 5.95–6.55 μm for the 6.2 μm PAH, 7.15–8.20 μm for the 7.7 μm PAH, 8.20–8.90 μm for the 8.6 μm PAH and 10.80–11.80 μm for the 11.2 μm PAH features. The integrated fluxes for the fine structure lines of [S iv] 10.51 μm, [Ne ii] 12.81 μm, [Ne iii] 15.55 μm, and [S iii] 18.71 μm were measured by fitting a Gaussian profile using the line-fitting function in SMART. Note that no sky background has been subtracted from the SH spectra. This has not affected our measurements of the spectral lines, however.

Table 13. IRS Fluxes of Mid-Infrared Ionic Lines

Region	$F_{\left[{\rm S \,\mathsc{iv}}\right]@10.51\;\mu {\rm m}}$ (10−18 W m−2)	$F_{\left[{{\rm Ne}\, {\mathsc{ii}}}\right]@12.81\;\mu {\rm m}}$ (10−18 W m−2)	$F_{\left[{{\rm Ne}\, {\mathsc{iii}}}\right]@15.55\;\mu {\rm m}}$ (10−18 W m−2)	$F_{\left[{{\rm S}\, {\mathsc{iii}}}\right]@18.71\;\mu {\rm m}}$ (10−18 W m−2)
Arp 105S	5.41 ± 1.16	4.71 ± 0.58	9.47 ± 0.44	9.49 ± 2.47
Arp 245N	⋅⋅⋅	4.19 ± 1.21	5.24 ± 0.48	4.99 ± 0.61

Download table as:  ASCIITypeset image

Mid-infrared PAH emission at 8.0 μm is observed within all the collisional debris of our sample; it is enhanced in or close to H ii regions, as traced by the Hα maps.

PAH bands in collisional debris have been directly detected for the first time in the NGC 5291 system (Higdon et al. 2006). The new MIR spectra of the TDG candidates in Arp 105S and Arp 245N presented in this paper also exhibit strong PAH emission (see Figures 7 and 8), a property which may result from their relatively high metallicity. It has been established (e.g., Engelbracht et al. 2005; Walter et al. 2007, references there in) that the PAH emission is correlated with metallicity. In Figure 9 (respectively 10), we plot the 6.2 μm (respectively 11.2 μm) band EW of the PAH features versus oxygen abundance. Galaxies from the sample of Wu et al. (2006) and individual star-forming regions in the spiral M101 (Gordon et al. 2008) are plotted.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

These plots show a clear correlation between the strength of the PAH emission and metallicity for the BCDGs and the M101 individual star-forming regions. The four intergalactic star-forming regions exhibit an aromatic emission at the expected level. Indeed, at a given metallicity, the EWs of the PAH bands emitted by the intergalactic star-forming regions are very similar to those of star-forming regions within galaxies.

As shown for instance by Madden et al. (2006), the strength of the PAH features is also closely connected with the hardness of the radiation field. Gordon et al. (2008) recently claimed that this parameter, itself correlated with metallicity (see Figure 11), is the main parameter governing the evolution of the strength of the PAH features. Indeed, lower metallicity stars emit harder radiation due to the lack of metals in their atmosphere. As a result, the hard radiation field combined with a high luminosity density can destroy PAH carriers (Wu et al. 2006).

Zoom In Zoom Out Reset image size

Since we also have high-resolution IRS spectra for most regions in our sample we examine in Figures 12 and 13 the correlation between the neon line ratio and the EW of the 6.2 μm and 11.2 μm features. We find that the regions in the collisional debris fall broadly within the trend discussed by, e.g., Wu et al. (2006).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Unfortunately MIR spectra are available for only four TDGs. For the other collisional debris, we have used the emission in the IRAC 8.0 μm band as an indirect tracer. The validity of this approach has been discussed in more detail by several authors (see, e.g., Engelbracht et al. 2005, 2008) and is corroborated by our current results. In our IRS sample, PAH features dominate the emission in this broadband. Continuum emission from hot dust may, however, have a significant contribution which should be taken into account. To that purpose, we have adopted the method presented in Engelbracht et al. (2005, 2008): the 8.0 μm emission is normalized to the underlying hot dust continuum as follows: F_8.0/(F_4.5 − αF_3.6). This assumes that the 3.6 μm emission is purely stellar,¹⁵ and that the IRAC 4.5 μm band traces the hot dust, once the stellar continuum has been subtracted. The value of α—the conversion factor from the 3.6 μm to the 4.5 μm stellar flux—is given by a spectral synthesis code and depends on the metallicity. For those intergalactic H ii regions for which no oxygen abundance has yet been determined, we adopt α = 0.535, the mean of the coefficients for the 8.15 ⩽ Z < 8.50 and 8.50 ⩽ Z < 8.85 bins from Engelbracht et al. (2008).

In Figure 14, we plot the ratio of the MIR fluxes F_8.0/(F_4.5 − αF_3.6) as a function of oxygen abundance for intergalactic star-forming regions and the Rosenberg et al. (2006) dwarf galaxies, and the Engelbracht et al. (2008) starburst galaxies comparison samples. The aim is to see if intergalactic star-forming regions follow the same well-known relation between the strength of the aromatic emission and metallicity. This plot confirms that intergalactic star-forming regions lie along the correlation between the mid-IR emission, corrected for the hot dust and stellar contribution, and metallicity. The intergalactic star-forming regions seem to follow the same trend as spiral galaxies, although the dispersion might be slightly higher. This may also be seen in Figures 15 and 16 showing the F_8.0/(F_4.5 − αF_3.6) PAH indicator as a function of the 8.0 μm luminosity for the reference samples. The intergalactic star-forming regions describe approximately the upper envelope of the reference sample of integrated galaxies but the mean ratio is similar.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

From this study of the dust properties of collisional debris, we conclude that their MIR emission, dominated by PAH, is "normal" and comparable to that of the star-forming regions in regular galaxies that have the same metallicity. PAH are emitted in photodissociation regions associated with SF episodes and are thus considered as a reliable tracer of SF (except in a low-Z environment; Calzetti et al. 2007). Thus the calibrations between the IRAC 8.0 μm band and the SFR, as determined for regular spiral galaxies, should also be valid for collisional debris. This assessment is exploited in the rest of the paper.

We compare here the emission of the three SF tracers used in this study. (1) UV emission, due to the direct radiation of mostly massive stars, (2) Hα emission from regions ionized by young, very massive stars, and (3) MIR 8.0 μm emission due to the PAHs in photodissociation regions.

These tracers, although all somehow related to the SF process, have their own characteristics and biases, which are discussed in Section 4.2.

To put the UV, Hα, and MIR luminosities on a common scale, we have converted them into units corresponding to measures of the SFR, using the following formulas:

• 1.  
  SFR(UV) = 1.4 × 10⁻²¹L_ν [W Hz⁻¹].
• 2.  
  SFR(Hα) = 7.9 × 10⁻³⁵L(Hα) [W].
• 3.  
  SFR(8.0 μm) = 1.88 × 10⁻³⁶νL_ν [W Hz⁻¹].

These conversion factors are based on the values determined by Kennicutt (1998) for the UV and Hα luminosities, and Wu et al. (2005) for the 8.0 μm emission. One should note that they are only valid for specific SF histories, in particular a SFR which is assumed to be constant for 100 Myr and with a Salpeter initial mass function (IMF). As discussed in Section 4.2.3, these conditions may not apply to collisional debris, and thus the derived SFR may be off by some unknown factor, especially in the youngest regions. This scaling is done here mainly to facilitate comparison between the three tracers (i.e., their fluxes) and to compare the SFR in debris with that of other star-forming regions.

In Figure 17 (respectively 18) we compare the SFR (uncorrected for extinction) estimated from the FUV band with that estimated from the 8.0 μm (respectively Hα). Boquien et al. (2007) have shown that the intergalactic SF regions around NGC 5291 exhibit a strong, systematic, UV excess that can be clearly seen in Figure 18 where Hα appears depressed compared to the FUV emission. On average, this is not the case in the other interacting galaxies in our sample (see Tables 14 and 15). This UV excess is found only locally, in specific star-forming regions, which mostly are hosted by gas-rich collisional debris. The dispersion among different regions is important, but not dramatically larger than that measured in star-forming regions belonging to a single galaxy, like M51. The 8.0 μm emission is itself the strongest for regions which contain an old stellar population. This is probably due to the fact that older stars can also heat the dust and therefore create diffuse emission. This kind of emission is particularly visible in spiral galaxies and can account for a significant fraction of the 8.0 μm flux as suggested by Calzetti et al. (2007).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The objects in our primary list were selected for having prominent star-forming regions in their collisional debris. This may be quantified using the UV maps, a reliable tracer of recent SF episodes. Table 16 lists the UV fluxes, converted to M_☉ yr⁻¹ (using the relation given in Section 3.2), from regions located within the optical radius of the parent galaxies and outside it, within the collisional debris. The instantaneous fraction of stars born in the "intergalactic" medium is as high as 80% with a median value of nearly 21%. The SFR in this environment may exceed 1 M_☉ yr⁻¹, a value which corresponds to the measure of the SFR integrated over the entire disk of normal spiral galaxies! This SF is rate is a lower bound as it is not corrected for internal extinction.¹⁶ Even if the extinction may be higher in spiral galaxies than in intergalactic star-forming regions, Boquien et al. (2007) showed that using the 8.0 μm indicator, the intergalactic SFR still accounted for over 50% of the total SFR in NGC 5291.

For comparison, in less active (and more typical) nearby interacting systems, the tidal structures account for 10% at most of the total SFR (Struck 2007; Smith et al. 2007; Hancock et al. 2007). Their result, uncorrected for extinction, holds whether the systems are observed in infrared, Hα, or UV.

One of the main characteristics of the collisional debris studied here is their high gas content, reflecting the fact that their parent galaxies were gas rich. This situation is rather rare in the local universe but was much more common at high redshift when galaxies had higher gas fractions.

This paper makes use of three SFR diagnostics that are sensitive to a number of factors.

• 1.  
  Timescale: Hα is sensitive to the most recent SF (nearly instantaneous compared to UV), giving an estimate of the SFR averaged over a few million years. On the other hand UV emission gives an estimate of the SFR averaged over about 100 Myr,
• 2.  
  Metallicity: the relation between PAH luminosity and SFR is not linear and depends on the metallicity. The Hα luminosity is also metal-dependent, as shown by Bicker & Fritze-v. Alvensleben (2005),
• 3.  
  Extinction: UV radiation suffers the most from extinction (A_FUV/A_Hα ∼ 3.1),
• 4.  
  Geometric factors: Hα is sensitive to the effects of geometry. Young stars are enshrouded in dust clouds and their ionizing radiation is efficiently absorbed by dust, and
• 5.  
  Variations of the initial mass function: depending on the distribution of the mass of stars at birth, the amount of energetic radiation emitted per unit mass changes. Since UV emission is not only due to the most massive stars, changes in the shape of the IMF will affect differently the UV and Hα luminosities.

In principle, ages, extinction, or even the IMF of a region of SF may be constrained by measuring flux ratios of UV/[8.0] or UV/Hα, when coupled with evolutionary synthesis models. In practice, strong degeneracies make this exercise particularly delicate.

The aim here is to find which factors can explain the apparent diversity of intergalactic star-forming regions. This is particularly important to better understand the SF process in tidal debris and find what parameter is the driving factor. In this section, we verify the influence of the extinction, metallicity, or stellar populations. We have made here such a comparison between the various SF indicators for about 60 individual star-forming regions in collisional debris. We discuss below the effect of each parameter on the measure of the SFR, or more precisely on the scatter of the various estimates from one region to the other (see Section 3.2); we will try to determine which parameter best manages to reduce it.

4.2.1. Extinction

To study the influence of extinction, we corrected the SFRs using the Calzetti et al. (2000) extinction curve (A_FUV/A_V = 2.54 and A_Hα/A_V = 0.82). The optical extinction was derived from the Balmer decrement.¹⁷, as measured from spectra of the H ii regions. No correction was applied in the MIR regime. The results are presented in Table 17.

Table 17. Ratio of the SFRs $\Delta \left({\rm{SFR}}\left(8.0\right)/{\rm{SFR}}\left({\rm{FUV}}\right)\right)/\langle {\rm{SFR}}\left(8.0\right)/{\rm{SFR}}\left({\rm{FUV}}\right)\rangle$ and $\Delta \left({\rm{SFR}}\left({\rm H}\alpha \right)/{\rm{SFR}}\left({\rm{FUV}}\right)\right)/\langle {\rm{SFR}}\left({\rm H}\alpha \right)/{\rm{SFR}}\left({\rm{FUV}}\right)\rangle$ Raw and Corrected for the Extinction

System	$\langle \mathit {A_V}\rangle$	$\frac{\Delta \mathit {A_V}}{\langle \mathit {A_V}\rangle }$	$\frac{\Delta \left({\rm{SFR}}\left(8.0\right)/{\rm{SFR}}\left({\rm{FUV}}\right)\right)}{\left<{\rm{SFR}}\left(8.0\right)/{\rm{SFR}}\left({\rm{FUV}}\right)\right>}$	$\left(\frac{\Delta \left({\rm SFR }\left(8.0\right)/{\rm SFR }\left({\rm FUV }\right)\right)}{\left<{\rm SFR }\left(8.0\right)/{\rm SFR }\left({\rm{FUV}}\right)\right>}\right)_{\rm corr. }$	$\frac{\Delta \left({\rm SFR}\left({\rm H}\alpha \right)/{\rm SFR}\left({\rm FUV}\right)\right)}{\left<{\rm SFR}\left({\rm H}\alpha \right)/{\rm SFR}\left({\rm FUV}\right)\right>}$	$\left(\frac{\Delta \left({\rm SFR}\left({\rm H}\alpha \right)/{\rm SFR}\left({\rm FUV}\right)\right)}{\left<{\rm SFR}\left({\rm H}\alpha \right)/{\rm SFR}\left({\rm FUV}\right)\right>}\right)_{\rm corr. }$
Stephan's Quintet	1.2	0.7	1.3	1.3	0.8	0.8
NGC 5291	0.2	0.8	0.7	1.0	0.3	0.4
NGC 7252	1.6	0.4	0.7	1.3	1.2	1.4
Debris with older stars	1.4	0.4	0.5	0.7	0.8	1.0
Debris without older stars	0.4	1.2	1.5	1.0	2.0	0.6

Note. Only regions having a SFR of at least 0.02 M_☉ yr⁻¹ have been selected.

Download table as:  ASCIITypeset image

We see that the scatter in the extinction correction is stable among the intergalactic star-forming regions in Stephan's Quintet, but increases in NGC 5291, as already mentioned by Boquien et al. (2007), and in NGC 7252. On average, however, when a significant reduction of the relative scatter is observed (which is the case only for debris without older stars), it is comprised between ∼30% and ∼70%. Given these results, we cannot claim that the extinction is the main parameter governing the discrepancies between the measures of SFR(Hα), SFR(NUV), and SFR(8.0).

Figure 19 plots the ratio of the SFR(8.0) over SFR(FUV) as a function of extinction. We are only including regions with equivalent SFR exceeding 0.02 M_☉ yr⁻¹ in order to ensure that the measure of the Balmer decrement, and thus of the extinction, is reliable. The correlation between SFR(8.0)/SFR(FUV) and A_V is weak. The reduction of the scatter mentioned above is largely due to the regions most affected by extinction, such as SQ-B in Stephan's Quintet, where A_V∼3.0 magnitudes.

Zoom In Zoom Out Reset image size

4.2.2. Metallicity

4.2.3. Stellar Populations

As noted before, the discrepancies between the different SFR indicators indicate variations of the local properties of star-forming regions. We measure some scatter from one intergalactic region to another. This is, however, similar to what is observed in individual star-forming regions of a single galactic disk. Besides, the MIR to UV ratio is on average similar in the collisional debris and in spiral disks of the same metallicity. More generally, despite the differences in the large-scale environments, it is remarkable that intergalactic and galactic regions look so similar, at least when studied via the three main SF diagnostics. The presence of strong PAH features in the collisional debris perfectly illustrates this.