3 Galaxy interaction

3.1 Classification of galaxy interactions

A number of CCD images of ULIRGs are now available, such as deep optical images (Leech et al. 1994), near-infrared images (Murphy et al. 1996) and HST images (Surace et al. 1998). Unfortunately, the overlap number of galaxies with CCD images with our sample is small. Our morphological classification and the study of environments of VLIRGs are based on the Digital Sky Survey (DSS) images (see Paper I). For the overlap galaxies with CCD images, our classification do not conflict with those based on the near-infrared or HST images. We use the classification scheme for VLIRGs of Lawrence et al. (1989), who define isolated galaxies as class 0, galaxies with far and near companions as class 1 to 4, interacting pairs as class 5 and mergers as class 6. Morphological classifications and environment parameters for our sample galaxies are listed in Table 3. In practice, we find it difficult to separate class 5 from class 6, because there is no exact dividing line between interacting pairs and mergers, it is therefore better to join classes 5 and 6 into a single class. Also one should be careful to define class 0. Our statistics show that this type has similar maximum infrared luminosity to that of class 5 and 6 and higher than that of class 1-4. Moreover, the percentage of AGN in class 0 is even higher than that in class 5 and 6 (see Sect. 4.3). Therefore, it is likely that the class 0 is in the advanced or post merging stages.

  

Figure 8: Distributions of the equivalent widths for $\rm H\alpha$of each spectral type

As described in Sect. 2, Tables 2a-c list the statistics of morphological types for the whole sample and two subsamples. It is clear from Table 2a that 56% (41/73) of VLIRGs are interacting or merging systems (class 5 or class 6). 18% (13/73) are isolated (class 0). The fraction of class 1-4 objects are about 26%.

  

Table 3: Environmental and morphological parameters of VLIRGs

 

Table 3: continued

 

Table 3: continued

For the subsample with $\rm Log({\it L}_{IR}/{\it L}_{\odot}) \geq 12.0$, nearly all the objects, namely 91% of ULIRGs, are strong interaction or merging systems (there are 10 ULIRGs belong to class 6 and the rest belong to class 0). These results are consistent with those of Sanders et al. (1988); Melnick & Mirabel (1990) and Clements et al. (1996a,b); the fractions are, however, higher than those found by Zou et al. (1991, 1993) and Leech et al. (1994). The different results for ULIRGs can be understood as follows: most of ULIRGs in Sanders', Melnick's and our sample are nearby galaxies with redshifts z < 0.13, for which the disturbed features could be detected more easily than the ULIRGs in Zou et al.'s and Leech et al.'s samples, which are most far-away. In order to avoid the large optical reddening, Murphy et al. (1996) obtained near-infrared images of 56 ULIRGs, 95% of these show evidence for current or past interaction. Recently, Surace et al. (1998) present HST images of 9 ULIRGs, in which 8 sources are advanced mergers. These near-infrared and high-resolution HST images show that there may exist very few truly isolated ULIRGs with no disturbed features. It should be very interesting to study the small class of "isolated'' galaxies with high resolution and/or infrared imagings. It is also obvious from Table 3 and DSS images that at least 14% (10/73) of VLIRGs are in groups of galaxies or in multiple-merging systems: 7 VLIRGs are located in groups of galaxies which include at least 3 galaxies with consistent redshifts, and 3 VLIRGs with multiple-nuclei. For example, IRAS 23532+2513 (Zou et al. 1995) is in a compact group with one Seyfert 1 galaxy and one starburst galaxy. Mrk 273 is a multiple-merging system with $\sim\! 10$ faint objects (within a projected distance of $\sim\! 100$ kpc), one of which is a companion dwarf galaxy showing an unusually high soft X-ray luminosity (Xia et al. 1998). ULIRGs and VLIRGs seem to be evolutionally connected with compact groups of galaxies; studying this class of galaxies may provide important clues for galaxy evolution.

3.2 Projected separation

The separation between an infrared galaxy and its nearest neighbor in double or multiple-nuclei is one important parameter that describes the intensity of galaxy interaction. It is also a good indicator of evolution in interacting galaxies. Since we can not obtain the physical separation between two objects we have to use projected separation instead. Bushouse et al. (1988); Telesco et al. (1988) and Wei (1990) have studied this parameter on optically selected interacting pairs. Wei (1994) also extended this to far-infrared galaxies. But since there was no redshift data for each source, these results are prone to errors induced by chance alignment. In contrast, the redshift information obtained from our observed spectra completely eliminate this error and allow us to study the relation between projected separations and activities of VLIRGs.

  

Figure 9: Distribution of projected separations between VLIRG and their nearest companions. The hollow histogram indicates the isolated VLIRGs, which could be advanced mergers

The angular separations measured from the DSS images are transformed to projected separations using the redshift of infrared galaxies. For groups or multiple-nuclei, the minimum separation is used. Figure 9 shows the distribution of projected separation. The hollow box includes the isolated galaxies, which are assumed as advanced merger with only one remaining nucleus. It can be seen that most VLIRGs have companion within 50 kpc. This is consistent with previous works by several authors that more active far-infrared galaxies have closer companions. If we include the isolated galaxies, we find that for large number of VLIRGs the project separations are smaller than 10 kpc.

  

Figure 10: Infrared luminosities as a function of projected separation. Dotted line is the upper envelope of data. 10 kpc is the character separation for an IRAS galaxy to be ultraluminous in the infrared

Figure 10 shows the relation between infrared luminosities and the projected separations of VLIRGs. Though there is no obvious correlation between them, the upper limiting boundary line (dotted line) decreases dramatically as projected separations increase. This trend can be easily understood as a result of projection effects. It is reasonable to assume that the fitted envelope line represents the relationship between infrared luminosity and true separation. We define a character separation as the separation when the infrared luminosity is about half of the peak value of very close sources (separation $\sim\! 0$). The separation is found to be about 10 kpc with a corresponding infrared luminosity is $\rm 2\;10^{12}\ {\it L}_{\odot}$.This result agrees with that of Melnick & Mirabel (1990) and Murphy et al. (1996).

  

Figure 11: $\rm H\alpha$ equivalent width vs. projected separation. The dotted line is the upper envelope of data

A similar behavior can be also seen in a plot of $\rm H\alpha$ equivalent width versus projected separation (Fig. 11). The envelope line goes up rapidly as the separation decreases. Considering that both infrared luminosity and $\rm H\alpha$ equivalent width are good indicators of star formation, Figs. 10 and  11 strongly suggest that interactions trigger starbursts.

3.3 Relative radial velocity

Besides the projected separation, the relative velocity is another important parameter that quantifies the interaction between galaxies. Since the true relative velocity can not be obtained, we use the relative radial velocity inferred from the redshift difference between an infrared source and its nearest companion. The uncertainty in the relative radial velocity is $\rm 30\,km\ s^{-1}$ in our data. Figure 12 shows the distribution of relative radial velocity. It is clear that the companions for most VLIRGs have small relative radial velocities. Figure 13 shows that the $\rm H\alpha$ equivalent width increases as the relative radial velocity decreases. This relation is very interesting because it imply that small relative radial velocity among interaction galaxies should be the favorate condition for triggering starburst.

  

Figure 12: The distribution of relative radial velocities of VLIRGs

  

Figure 13: $\rm H\alpha$ equivalent widths vs. relative radial velocities