3 Results

Figure 1 shows a moderately deep optical image of the field around IRAS03158+4227. The data for the IRAS error ellipse were taken from the IRAS Point Source Catalogue, the coordinates for the objects on the optical image are from the astrometric solution for the field of the Perseus cluster of galaxies (Brunzendorf & Meusinger 1999). The IRAS error ellipse clearly overlaps with the optical image of a faint galaxy (G1) with z = 0.13(NASA Extragalactic Database, NED). At the distance of the galaxy, the size of the field shown in Fig. 1 corresponds to about 0.5 Mpc $\times$ 0.5 Mpc. There is no sign of a dense galaxy cluster or a rich group of galaxies around IRAS0315+4227. Several small and faint galaxy images are seen in the field, but a substantial fraction may be dwarf galaxies of the Perseus cluster in the foreground.

Murphy et al. (1996) presented a K-band image taken with the Palomar 200 inch telescope at a seeing of $0\hbox{$.\!\!^{\prime\prime}$ }8$that shows neither evidence of a neighbour galaxy outshined by the brighter foreground star S1 on a less resolved image, nor for a double nucleus of G1. As mentioned in Sect. 1, IRAS03158+4227 was assumed to be a late merger with nuclei separation of less than $0\hbox{$.\!\!^{\prime\prime}$ }8$, even though it was noted by Murphy et al. (1996) that there is a galaxy (G2) at a distance of 18''. The two galaxies G1 and G2 have approximately the same apparent magnitudes; after calibrating our B-band image using stars from the USNO-A2.0 Catalogue (Monet et al. 1998) we find B =18.6 and 18.4 for G1 and G2, respectively. This corresponds to $M_{B} \approx -21.4$ and -21.6(for A_B = 0.8 mag from the NED, and a k-correction k_B = 0.5 mag from Coleman et al. 1980) and to a very strong FIR excess of $\log L_{\rm FIR}/L_{B} = 2.3$ for G1 (see e.g., Soifer et al. 1987 for the definitions of $L_{\rm FIR}$ and L_B).

Figure 1: IRAS error ellipse of IRAS03158+4227. Coaddition of R- and I-band images taken with the Tautenburg Schmidt camera. The size of the image is $200''\times 200''$; N is up, E is left. S1 to S3 designate stars, G1 and G2 indicate galaxies.

The spectral energy distribution (SED) of IRAS 03158+4227 is shown in Fig. 2. The SED is typical for ULIRGs, with the bulk of the luminosity radiated at far-infrared wavelengths. In addition to the data points shown in Fig. 2, we note that IRAS 03158+4227 is identified with a radio continuum source in the NVSS (Condon et al. 1998). The flux density of 12.4 mJy at 1.4 GHz corresponds to a radio-to-infrared flux ratio (for definition see Helou et al. 1985) q=2.64. This is in agreement with the well-known correlation between the flux densities in the infrared and the radio continuum (see Sanders & Mirabel 1996, and references therein) where $\langle q \rangle \approx 2.35$ for most of the luminous infrared galaxies in the BGS, but is larger for the galaxies with highest infrared luminosities. (For instance, Helou et al. found q=2.60 for Arp220.)

The optical images (Fig. 3) clearly reveal that G2 has a faint, but very extended material arm on the side opposite to G1. This feature is seen in all optical bands, and the Lucy-Richardson-deconvolved image shows that it is quite narrow, as expected for tidal tails. In the optical bands, the arm is about 3 mag fainter than the main body of G2. The surface brightness of the brightest part of the tail is estimated to be $\mu_{B} \approx 25$ magarcsec^-2, the projected linear extent is about 70 kpc. Illustrative examples for similar tidal structures with one dominating arm can be found among the Arp (1966) and Arp & Madore (1987) galaxies (e.g., Arp107, 110, 129, 173, 252, 255, AM2350-302, AM0552-324, AM0427-475). For the host galaxy of the ULIRG itself, tidal signatures are much weaker, although there appears to be some fuzz around G1. Several brighter spots may represent tidal debris, perhaps including huge star-forming regions; but clearly, none of these spots is bright enough to be considered a second nucleus of G1. Unfortunately, it can not be excluded that faint morphological features of G1 are hidden behind the bright star S1.

The radial intensity profiles of G1 and G2 in the R-band are shown in Fig. 4. G1 may be approximated by two components with different scalelengths. Alternatively, it can be classified as an "E-like'' ULIRG: the radial surface density profile is reasonably fit by a deVaucouleurs r^1/4-law over the range $R \approx 4$ ...30 kpc. (The innermost 1 $.\!\!^{\prime\prime}$5 were excluded from the analysis.) According to Sanders et al. (2000), about one third of the ULIRGs from the 1 Jy sample are classified as "E-like''. Contrary to G1, the profile of G2 is better approximated by an exponential law which suggests a "disk-like'' structure.

The images in the NIR-bands are shown in Fig. 5. These high-resolution images were taken to search for point sources indicating either a close double (or multiple) nucleus with small separations or a nucleus outshined by the bright star S1 in the optical image, rather than to evaluate extended structures of low-surface brightness. For such an aim, observations at longer wavelengths are needed, since the morphology of the central parts of ULIRGs is strongly affected by dust obscuration. As expected (cf. Sanders et al. 2000), the light distribution in K' is very compact for the galaxy G1 which appears as a point source. Figure 5 does not reveal any signs of a double nucleus in G1 or for a nucleus close to S1 down to the resolution (beam-deconvolved size) of 0 $.\!\!^{\prime\prime}$40. Of course, there remains the possibility of an incidental superposition of the image of S1 and a further nucleus, but the probability for such a configuration is very low.

The optical spectra of G1 and G2 (Fig. 6) show strong emission lines; the equivalent widths (EWs) are listed in Table 2.

Figure 2: Spectral energy distribution (SED) of IRAS 03158+4227. Asterisks denote our measurements while diamonds mark the IRAS fluxes. (The data point at 12 $\mu$m represents an upper limit.)

 

 

Table 2: Equivalent widths EW (in units of Å) of the emission lines measured for the galaxies G1 and G2.

	G1	G2
EW(H$\beta$)	$3.5\pm0.4$	$2\pm2$
EW([O III] 5007)	$8.0\pm0.4$	$59\pm1$
EW([O I] 6300)	$8.0\pm0.4$	$0.1\pm0.2$
EW(H$\alpha +[$N II] 6584)	$118.7\pm0.2$	$124.0\pm1.0$
EW([S II] 6717, 6731)	$3.7\pm0.3$	$1\pm1$

A radial velocity difference between G1 and G2 of 240 kms^-1 is estimated, but the uncertainty due to the low spectral resolution is quite large (about $\pm200$ kms^-1).

IRAS03158+4227 has sometimes been classified as a Seyfert 2, whereby strong absorption was invoked as the reason for the absence of indications of nuclear activity in the hard X-rays (Risaliti et al. 2000). In the spectrum from Fig. 6, we do not see clear-cut evidence for an AGN in IRAS03158+4227; in particular the conventional diagnostic EW ratio [O III] 5007/H$\beta$is too low for a Seyfert nucleus. However, the signal-to-noise ratio is low for these lines, and, owing to the low dispersion, we are not able to resolve H$\alpha +[$N II] 6584. The low intensity of the emission lines in the blue part of the spectrum is probably best explained as due to strong internal dust absorption and resultant reddening, though the effect of the underlying stellar absorption is difficult to estimate. If H$\beta$ is significantly affected by stellar absorption of an older, A-type burst population, the intrinsic [O III] 5007/H$\beta$ratio would be even less compatible with a Seyfert spectrum. On the other hand, the diagnostic line ratios of G2 are more consistent with an AGN. It should be noted that the K'-band image of G1 is more concetrated than that of G2 (Fig. 5), as seems typical for ULIRGs compared to AGN hosts (Sanders et al. 2000).

Figure 3: Optical images of IRAS03158+4227: a) B-band; b) R-band; c) I-band; d) co-addition of all optical images; e) the same image as d), but after Lucy-Richardson deconvolution with 10 iteration steps; f) the contour plot of image e) where successive contour lines correspond to a factor 3 in intensity. (The unfiltered image shows the same structures as the images a) to c) and is not displayed here.) The size of each image is $30''\times 42''$, N is up, E is left.