3 Results

3.1 Optical images

The brightest central part of HCG 54 (Fig. 2) shows a dominant knot (HCG 54a) embedded in an elongated feature visible in the blue and red bands at a PA of 65$^{\circ}$. It is more extended to the SW, where a very compact bright knot is located (HCG 54b) which is even more prominent in H$\alpha$ (see Vílchez & Iglesias-Páramo 1998). The elongated area defined by HCG 54a and b is connected to the NE with the weaker but still prominent knots HCG 54c and d.

Large variations in the B-R color index are apparent throughout the whole system (from 0.15 in HCG 54d to 0.70 around HCG 54a, Fig. 3). The elongated structure centered in HCG 54a has a blue color (0.5), likely a signature of a recent star forming episode, and is embedded in a rounder region that shows the reddest colours in HCG 54 (B-R = 0.7).

The B-R color index for HCG 54b is surprisingly red for a strong star formation burst like the one hosted by this galaxy. However, we argue that this is due to the contamination by the H$\beta$, [O III] and H$\alpha$ emission lines. After computing the equivalent widths of these lines $(EW({\rm H}\beta) = 155$ Å, $EW({\rm [OIII]}4959) = 246$ Å, $EW({\rm [OIII]}5007) = 728$ Å, $EW({\rm H\alpha + [NII]}) = 798$ Å), we estimated that the corrected B-R color should be 0.46 mag bluer than the one measured directly from the color map. Thus, we would obtain B - R = 0.20, which is more consistent with a very young stellar population.

Three shell-like structures are found (noted as t1 to t3 in Fig. 2), centered roughly on HCG 54a and b. A fourth optical feature is located to the SW and marked as t4. We detect all of these features in a R-band surface brightness range of 24 to 27 mag/( $^{\prime \prime }$)². The outer shells (t2 and t3) although redder than the internal knots, have blue colours ($\sim$0.5) typical of irregular galaxies (see Fukugita et al. 1995). The inner shell (t1) is slightly redder (0.60), probably due to contamination from the central parts of HCG 54. Numerous unresolved knots are detected everywhere in the area, such as the one marked in Fig. 2 with a "k''. No stellar counterpart to the HI long tail is detected down to 27  mag/( $^{\prime \prime }$)² in R.

We found three faint galaxies in our large R-band frame (Fig. 1), that are named A1126+2051, A1127+2057 and A1127+2054. The first two are background galaxies (Sect. 3.2) while the last one is at the velocity of the system according to the HI data (Sect. 3.3). The optical image of this galaxy shown in Fig. 11 reveals that the inner and the outer isophotes are off-centered. When measured at an R isophote of 26 mag/( $^{\prime \prime }$)² the optical size is 2.6 kpc and the total magnitude in R is M_R = -14.2 mag. The surface brightness derived as a function of radius is exponential, consistent with that of a disk system (see Fig. 11b).

   
3.2 Description of the spectra

From the Grism#4 spectra sp1 (Fig. 4) and sp2 (Table 2) we have determined that A1126+2051 and A1127+2057 are not at the group redshift. A1126+2051 shows absorption lines of CaII, G band, Mg I and Na lines and has a redshift of z = 0.15 (see also Sect. 3.3). A1127+2057 also shows an absorption spectrum typical of early type galaxies and gas at a redshift of z = 0.052.

We have studied the kinematics of the central parts of the group with the spectra obtained with Grism#7 and #8. The spectrum along the HCG 54a-b direction (sp5, Fig. 4) shows emission lines along the 249 spatial sections (47 $^{\prime \prime }$), and the obtained velocity curve is plotted in Fig. 5a, where the zero position corresponds to the center of HCG 54b. We mark in the figure the continuum extent of HCG 54a and mark with an arrow the position of the emission line maxima. The spatial sections closer to HCG 54a (5 $^{\prime \prime }$ diameter) traces a distorted rotation curve with an amplitude of 45 km s^-1. At both sides of the center of HCG 54a we find discontinuities in the velocity that correspond in the R band to distorted isophotes. The sections in the direction of HCG 54c (R1 in Fig. 5a) have three well differentiated knots with emission lines, whose left end is the external part of HCG 54c. The strongest knot presents a velocity gradient of 50 km s^-1 in 4 $^{\prime \prime }$ that corresponds to the contact area between HCG 54a and HCG 54c (R3). The velocities in the direction of HCG 54b show irregular changes of up to 70 km s^-1, as is the case of R2 (Fig. 5a) that appears to be an HII region detached from the general trend of the velocity curve. The area of HCG 54b has a peculiar kinematics (Fig. 5a) that could be consistent with infalling gas, in the case that the background emission is hidden by extinction so that we are only observing the motions of the foreground component. In Fig. 5b we show the velocities in the slit through HCG 54c-d (sp6, Fig. 4), and the regions corresponding to HCG 54c and d in the R image are also marked. These regions have a weak continuum, and none of them show a rotation curve. HCG 54d has irregular motions, while HCG 54c shows a nearly constant velocity following a U-shape with an amplitude 30 km s^-1. We do not detect continuum emission toward the third emitting region in the direction of HCG 54a (Fig. 5b), while the velocities increase continuously until it reaches the values characteristic of HCG 54a. Finally we find a region with a decoupled velocity (R4) that is located between HCG 54c and d

In the lower resolution spectrum crossing HCG 54d (sp1, Fig. 4) we detect a velocity gradient of only $\sim$70  km s^-1 within 7 $^{\prime \prime }$. In Fig. 5c we show the sections where we have detected emission, where the center corresponds to the continuum peak of HCG 54d along the slit. The presence of several knots in our R band image along the slit direction suggests that the continuity of the observed velocity gradient might be due to smearing of individual components.
We have extracted individual spectra for 9 zones from the data taken with Grism#4 (sp3 and sp4, Fig. 4). Six of them are along the slit position joining HCG 54a-b and three along the HCG 54c-d direction (Fig. 4). The spectrum of HCG 54a (sp3, Fig. 4) shows strong Balmer absorption lines as well as lines of CaII, Gband, and possibly MgI, while Mg$_{\rm 2}$ is barely detected. The spectrum shows a blue continuum, characteristic of a post-starburst population (Fig. 6). The spectrum corresponding to knot b is remarkable (Fig. 7), with a high excitation and the presence of WR features at $\sim$4650 Å over a flat continuum, indicative of a young and strong burst of star formation. Figure 7a shows the full spectrum of HCG 54b, while Fig. 7b shows a detail of the spectrum around the Wolf-Rayet (WR) feature. The temperature sensitive [OIII] line at $\lambda$4363 Å was measured in this spectrum, giving a t([OIII]) temperature of 10 485 K, implying an oxygen abundance of $\rm 12+LogO/H = 8.26$ (see Table 3). According to Schaerer & Vacca (1998), the measured equivalent width of the WR $\lambda$ 4650 Å feature in knot b, $EW({\rm WR}) = 9\pm0.9$ Å, implies an age of the burst between 3 and 4 Myr.

For the zones showing emission line spectra (sp3, sp4, Fig. 4) we present in Tables 3 and 4 their fluxes relative to H$\beta$ as well as the derived physical conditions of the ionized gas. Reddening corrected line fluxes relative to H$\beta$ are presented for the 9 individual spectra extracted along slit positions sp3 and sp4 (Table 2, Fig. 4). For each zone given in Tables 3 and 4 the following ionization structure parameters have been derived in order to perform the abundance analysis (as detailed below):  $R_{\rm 23}$, denotes the abundance parameter after Pagel et al. (1980) which is defined as [I(3727) + I(5007) + I(4959) ]/I(H$\beta$), P denotes the abundance parameter defined by Pilyugin (2000), as quoted in Tables 3 and 4, $P = [1.3 \times I(5007)/I({\rm H}\beta)]/R_{\rm 23}$. The excitation, defined as $1.3 \times I(5007)/I(3727)$, and the nitrogen to oxygen abundance indicator, $1.3 \times I(6584)/I(3727)$, are also quoted in Table 2. Note the small range of variation of  $R_{\rm 23}$ in contrast with the large variations shown by the excitation along the slit. The electron temperatures t[OIII] and t[OII], corresponding to the ionization zones of [OIII] and [OII] respectively, have been derived for region $\char93$1 and listed in Table 3. For this region the ionic and total abundances of oxygen, O²⁺/H⁺, O⁺/H⁺ and O/H respectively, as well as the abundance ratio of neon to oxygen, Ne²⁺/O²⁺, have been calculated and are presented in the table. For the rest of the studied regions, only the total abundance of oxygen has been estimated (see Tables 3 and 4).

Given that the temperature sensitive line [OIII]lambda 4363A was measured only for knot b, we have to rely on the empirical calibration in order to derive their abundances. We have used the calibration of the oxygen abundance versus  $R_{\rm 23}$ (cf. Pagel et al. 1980) as parameterized by McGaugh in Kobulnicky et al. (1999), and following the P-method (Pilyugin 2000), in order to provide an estimation of 12+Log O/H. The [NII]/[OII] line ratio, when observed, was used to discriminate between the lower and upper branch, though often the 6584 line was severely blended with H$\alpha$ and was not measured. For those zones with $R_{\rm 23}\ge 0.9$ we have assumed an average abundance of 12+logO/H = 8.2 as indicated by the calibration. All the zones selected are consistent with 12+Log O/H = 8.2, within the errors of the empirical calibration. This abundance is typical of galaxies like the Large Magellanic Cloud and the outer disks of late type spirals.

   
3.3 Neutral hydrogen

The integrated emission of neutral hydrogen (Fig. 8a) shows a NE-SW distribution with extensions to the SE and SW, and a long tail with a projected size of 20 kpc to the NE. The velocity field is quite perturbed, but still shows a velocity gradient similar to a rotating disk with a twisted major axis (Fig. 8b). This reflects itself also in the asymmetry of the HI line integrated profile (Fig. 10).

The situation is more complex when the channel maps (Fig. 9) are examined. The sizes of the optical knots are small compared with the VLA synthesized beam, and tracing the HI kinematics with respect to the optical features is difficult. Nevertheless these channel maps reveal the details that are lost in the integrated emission image shown in Fig. 8a. Except for the large HI tail to the northeast, most of the HI emission arises within the faint optical extent of HCG 54. Bright HI features directly associated with the bright optical ridge of emission are seen in the channel maps with velocity range between 1365-1490 km s^-1. The overall velocity field is along the length of the bright optical ridge in a manner consistent with that of rotation, but clear evidence for a velocity gradient in the perpendicular direction is also present, increasing in velocity from NW to SE. Since stars and gas inside the tidal radius are generally unaffected by a tidal interaction, the observed kinematic disturbance suggests an involvement of a deeply penetrating interaction or a merger.

Most of the high surface brightness HI features associated with the fainter outer optical envelope occur on the west side of the optical galaxy, closely associated with the optical tidal features t1, t2, and t4 (see Fig. 2). Both the Y-shaped HI morphology in Fig. 8a and the shifting of the focus of the iso-velocity contours to the west of the optical peaks in Fig. 8b are direct results of the large amount of HI associated with these tidal features. This one-sided appearance may indicate that only one of the progenitor systems was HI-rich if HCG 54 is mainly a product of a merger involving two late type galaxies - i.e. the progenitor responsible for the tidal feature t3 had relatively little HI associated with its stellar disk. The HI extension to the southwest, closely associated with the optical feature t4, occurs mainly in the velocity ranges of 1334 to 1428  km s^-1 and contains $1.4 \times 10^8~M_\odot$of HI. The long HI tail associated with either t1 and/or t2occurs in the velocity range of 1376 to 1459  km s^-1 and contains about $10^8~M_\odot$ of HI. Combined together, these two outer HI features account for more than 40% of the total HI detected in this system.

Several HI clumps that are detached from the main body of HI are also seen at $3-5~\sigma$ levels in these channel maps, and they indicate a more extensive debris field associated with this system. A velocity coherent string of HI clumps forming a nearly complete loop or a ring is seen in the channel maps within the velocity range of 1438 to 1469  km s^-1, largely to the northeast. Some of these features make up part of the 20 kpc long HI tail to the north, but their appearance in the channel maps, particularly at 1438  km s^-1 and 1448  km s^-1, suggest a more ring-like morphology, similar to those seen in collisional ring galaxies (e.g. Higdon 1996). The stream of HI clumps extending to the west of the main body seen in Fig. 8a form a second distinct, velocity coherent structure appearing at velocities of 1386-1407  km s^-1, and these channel maps suggest its origin being more to the south of the main body rather than a linear east-west structure. The total HI mass associated with this feature is about $3 \times 10^7~M_\odot$.

The overall velocity field of the bulk of HI delineates an elongated structure with an axis ratio that would correspond, if intrinsically circular, with an approximate inclination of 50$^{\circ}$ and a position angle of 70$^{\circ}$, similar to the one traced by the direction of HCG 54a-b ( $\rm PA=65^{\circ}$). The amplitude derived from the velocity field (140 km s^-1 ) and deprojected according to this inclination gives an overall velocity gradient of $\sim$183 km s^-1. If this is interpreted as a Keplerian rotation, we estimate a dynamical mass for the system of $M_{\rm tot} \sim 10^{10}~M_{\odot}$.

A dwarf companion galaxy A1127+2054 is found at a projected distance of 27 kpc northeast of HCG 54a (see Sect. 3.1, Figs. 1 and 11). Its location also corresponds to a distance of 8 kpc from the tip of the 20 kpc long HI tail, in the same projected direction. The HI channel maps show associated emission in 3 channels covering velocities of 1386-1407  km s^-1. A close examination suggests a central depression in the integrated emission (see Fig. 11c). The velocity field is consistent with a slowly rotating disk with a velocity amplitude of 30  km s^-1 (Fig. 11d). The HI extent at 3$\times$ 10¹⁹ at cm^-2 (3$\sigma$ level) is 45 $^{\prime \prime }$ $\times$ 25 $^{\prime \prime }$ (4.3 kpc $\times$ 2.4 kpc) and contains $1.9 \times 10^7~M_{\odot}$ of HI. The atomic component seems to be perturbed since it is not well centered on the optical component.

In the superposition of the integrated HI emission over the deep optical image shown in Fig. 8a, the HI isolated peak located about 2$^\prime$ east of HCG 54a appears to have an optical counterpart in a bright, compact galaxy. From the optical spectrum we obtain, however, that this is a chance superposition as this galaxy A1126+2051 is found to be a background object with z=0.15 (see Sect. 3.2). At the position of this galaxy we find HI emission in two channel maps, with no signature of systematic motions. Since the gas is located close to the center of the group, it is possibly part of the tidal debris. This is supported when a sharpening of the optical image of A1126+2051 is performed, suggesting the existence of two objects, a symmetric disk structure whose peak is located in the slit, and then corresponds probably to the background galaxy, plus a blob that might be a stellar counterpart of the HI emission.

3.4 Radio continuum emission

Using the data from the line-free channels, a 1.4 GHz continuum image was constructed. No significant continuum emission associated with the optical galaxy is detected at the $3\sigma$limit of about 0.5 mJy beam^-1. The New VLA Sky Survey image also gives a comparable upper limit of about 1.5 mJy for about a 3 times larger beam. HCG 54 was detected by IRAS in the 12 $\mu$m, 60 $\mu$m, and 100 $\mu$m bands with flux densities of 0.24, 0.50, and 0.84 Jy, respectively, with the derived FIR luminosity of $L_{\rm FIR}=3\times 10^8~L_\odot$ (Verdes-Montenegro et al. 1998). The upper limit of 1.5 mJy for 1.4 GHz radio continuum makes this object slightly underluminous in the radio with a lower limit on the q-value of about 2.7 (see Yun et al. 2001). The enhanced IR luminosity resulting from the relatively high dust temperature, inferred from the FIR ( $S_{\rm 60}/S_{\rm 100}=0.60$) and mid-IR ( $S_{\rm 25}/S_{\rm 60}=0.48$) color, offers a natural explanation for the relatively weak radio continuum. These warm infrared colors are also consistent with the evidence for significant WR activity detected in the optical spectra (see Sect. 2.2).