Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 351, 47-58 (1999)

Previous Section Next Section Title Page Table of Contents

5. Discussion

5.1. The spatial distribution of the emission lines and continuum

All the objects in the sample present extended continuum and line emission. These structures are aligned with the radio axis at least in MRC2025-218 and MRC2104-242 (McCarthy et al. 1990, Pentericci et al. 1999). L1 and L2 in SMM J02399-0136 define a line with position angle 88o while the radio axis position angle is 71o. Therefore, the optical and radio axis are closely aligned in this object as well. We located the slit in MRC1558-003 aligned with the radio axis and therefore, strong line and continuum emission is extended in this direction.

Our spectra reveal the presence of several spatial components in all objects. A clumpy morphology has been observed in most HzRG both in continuum and Ly[FORMULA] (eg. Pentericci et al. 1999). According to the properties of the individual clumps, Pentericci et al. have suggested that we are witnessing the merging of several sub-units to form the host galaxy of the radio source.

5.2. The emission line ratios

We have compared the line ratios of the objects in our sample with measurements for other HzRG (Röttgering et al. 1997) by plotting them in four diagnostic diagrams involving the main UV lines (Fig. 7). We plot also some models predicting the UV line ratios of HzRG when a) AGN is the dominant ionization mechanism b) shocks dominate the emission line processes (see Villar-Martín et al. 1997 for a detailed discussion of the models). The AGN photoionization models were built with the photoionization code MAPPINGS Ic (Luc Binette), which is described in Ferruit et al. (1997). The shock models are taken from Dopita & Sutherland (1996). Villar-Martín et al. (1997) showed that most HzRG define a sequence which can be explained in terms of (AGN) photoionization by a power law of index [FORMULA]=-1 (with [FORMULA]) of a low density gas ([FORMULA]100 cm-3) with solar abundances. The parameter sequence is the ionization parameter U, so that the difference from object to object is due to a difference in the ionization level of the gas, produced either by geometric dilution or by differences in the quasar luminosity.

[FIGURE] Fig. 7. Diagnostic diagrams involving the strongest UV lines. Our objects are represented as open circles. Stars are HzRG in Röttgering et al. (1997) sample. The solid triangle is the hyperluminous radio galaxy FSC10214+4724 (data obtained from Goodrich et al. 1996). The solid line is an AGN photoionization sequence (in U, the ionization parameter) with a power law of index [FORMULA]=-1.0, solar abundances and low density ([FORMULA]100 cm-3). Similar sequences with [FORMULA]105 and [FORMULA]106 cm-3 are shown. The solid circles in each of these three sequences show the models with U=0.01 and 0.1. Notice that MRC2025-218 and L1 are better explained by the higher density models, with U[FORMULA]0.01. The shock models (dotted lines) and shock+precursor models (dashed lines) from Dopita & Sutherland (1996) are presented as well. The weakness of Ly[FORMULA] in the first diagram is probably due to absorption by neutral H possibly combined with dust.

MRC2104-242 and MRC1558-003 lie in the general sequence defined by most HzRG. This is not the case for the integrated spectrum (L1+L2) of the system SMM J02399-0136, which presents weak Ly[FORMULA] and weak HeII relative to CIV and CIII] 1 This behaviour is more pronounced if we extract the spectrum of the L1 component (the active galaxy). The line ratios locate the object far beyond the general trend, due to the weakness of HeII and Ly[FORMULA] (see the spectrum in Fig. 2).

MRC2025-218 lies also far from the general trend. Since CIV/HeII is similar to the values observed in other HzRG and CIV/CIII] and CIII]/HeII are too low, the discrepancy is probably due to the unusual strength of CIII] in this source, rather than HeII being too weak, as it might seem from the diagnostic diagrams. 2

We have also studied the NV[FORMULA]1240 emission. This line was not often detected in earlier spectra of HzRG because of limited S/N (see Röttgering et al. 1997). In our small sample, only MRC2025-218 and SMM J02399-0136 have detectable NV emission. We present in Fig. 8 the diagram NV/HeII vs. NV/CIV. Quasars define a tight correlation in this diagram (Hamann & Ferland 1993) which is represented as a inclined line. Fosbury et al. (1998, 1999) showed that HzRG follow a parallel correlation to the one defined by quasars. This is also shown in the diagram. We have plotted the position of SMMJ02399-0136, L1 and MRC2025-218. Interestingly, the NV/CIV and NV/HeII line ratios measured in these sources are the largest observed for HzRG. MRC2025-218 lies at the top of the correlation defined by HzRG, while L1 lies on the quasars correlation and also occupies the position of the largest values for the NV line ratios. Both standard AGN photoionization models (with solar abundances and density [FORMULA]100 cm-3) and shock models are unable to reproduce the position of the objects in this diagram.

[FIGURE] Fig. 8. NV/CIV vs. NV/HeII. The same models described in Fig. 7 are presented. The solid circle in each AGN sequence is the model with U=0.1. The U=0.01 model lies outside the diagram. Therefore the models able to explain the positions of the objects in the diagrams in Fig. 7 are unable to predict line ratios involving the NV line; NV is too weak. The inclined dot-dashed lines represent the sequence defined by high redshift quasars (Hamann & Ferland 1993) and HzRG (Fosbury et al. 1998, 1999). The numbers on the quasar line indicate the metallicity (in solar units) calculated by the models of Hamann and Ferland. FSC10214+4724 is indicated with the solid triangle.

The emission line spectrum of L1 in SMM J02399-0136

The nuclear spectrum of high z quasars presents weak or absent HeII and strong NV. (NV/HeII[FORMULA]5 and NV/CIV presents a large diversity with values sometimes [FORMULA]5). HeII is generally narrower (when detected) than other lines like CIV and CIII] (Foltz et al. 1988, Heckman et al. 1991). This has been interpreted as the origin of an important fraction of the HeII emission ([FORMULA]50%) in a lower velocity extranuclear region, possibly the ISM of the host galaxy of the quasar. This gas will have an important contribution to the Ly[FORMULA] emission, but lines like NV, CIII] and CIV will be dominated by the broad line region (BLR). A view of the BLR in L1 would explain the weakness of HeII and the strength of NV.

However, L1 is not a quasar since the lines are too narrow (FWHM[FORMULA]2500 km s-1 in quasars). CIII] is broad (FWHM[FORMULA]6100 km s-1), but this could be due to the contamination by the SiIII]1895 doublet. It is possible that the lines could appear narrower and asymmetric as the result of absorption by gas and/or dust, which could be very efficient at quenching the resonant lines (CIV, NV, Ly[FORMULA]). However, H[FORMULA]+[NII] presents also a narrow profile ([FORMULA]1060 km s-1) (IV98). Therefore, the emission lines are much narrower than in quasars and L1 is not a normal quasar.

The spectral properties of L1 rather suggest that it is a narrow line active galaxy (a Seyfert 2 or narrow line quasar). Why does it lie on the sequence defined by quasars in the NV/HeII vs. NV/CIV diagram? Why does it show like quasars weak HeII and strong NV relative to the C lines? Since the BLR is not visible the spectrum must be dominated by the intermediate density narrow line region ([FORMULA]) and/or the low density narrow line region ([FORMULA]100, usually named the extended emission line region, EELR). We showed above that the standard EELR models fail to reproduce the observed line ratios. We present in Fig. 7 two sequences (in U) of models similar to the AGN sequence described above, but for densities [FORMULA]105 and [FORMULA]106. The [FORMULA]106 sequence solves the discrepancies between the models and the observed line ratios. Therefore, the weakness of HeII can be explained if the intermediate density narrow line region dominates the line emission. However, these models still fail to reproduce the line ratios involving NV. Fig. 8 shows that the [FORMULA]106 models (only the higher ionization model is present) predict the NV line too weak. On the other hand, our models show that such high densities would produce [FORMULA]4, while the ratio measured by IV98 is [FORMULA]1.

One possibility is that the magnification due to the gravitational lens is not the same for all emission lines. The magnification depends strongly on the source size (Trentham 1995) being higher for a smaller size of the source. Those lines whose emission is dominant in a more nucleated region will be more magnified than those lines preferentially emitted in a more extended region. Since the high density models predict stronger metal lines relative to HeII compared to the EELR models (density [FORMULA]100 cm-3) we expect CIV, CIII] and NV to be more nucleated and therefore more amplified. In this scenario, the region emitting NV should at the same time be more nucleated than the regions emitting the C lines to explain the large N/C ratios. Differential magnification has been suggested by Lacy et al. (1998) to explain the spectroscopic properties of FSC10214+4724.

An alternative explanation is that N is overabundant. NV/HeII and NV/CIV have been used both in quasars (Hamann & Ferland 1993) and HzRG (Fosbury et al. 1998, 1999) as abundance indicators. By studying the NV/HeII vs. NV/CIV diagram and the tight correlation defined by high redshift quasars, the authors conclude that the BLR of quasars at high redshift ([FORMULA]2) present N overabundance and a large range in metallicities typically [FORMULA]1 to [FORMULA]10 times the solar values. N/C is enhanced compared to the solar values due to secondary production (N behaves in a different way than C and O. [FORMULA], where Z is the metallicity). Fosbury et al. proposed a similar explanation (but referred to the ionized gas outside the BLR) for the correlation defined by HzRG in this diagram. The failure of the solar abundance (both low and high density) models to explain the strength of NV (and maybe also the weakness of HeII) suggests metal enrichment also in the ionized gas of L1. We have calculated upper limits to the flux of the NIII][FORMULA]1750 and NIV][FORMULA]1486 lines assuming FWHM=1600 km s-1 (larger values are possible. In this case we will obtain higher upper limits and the conclusions will not vary). We obtain [FORMULA]0.7 and [FORMULA]0.8. These values are well above the photoionization model predictions ([FORMULA]0.1 and [FORMULA]0.25). If the large values are confirmed, the NIII] and NIV] lines will support the interpretation of N overabundance.

The emission line spectrum of MRC2025-218

Fig. 7 shows that models with density 105 cm-3 reproduce the MRC2025-218 line ratios involved in these diagrams. This suggests that also in this object the line emission is not dominated by the EELR, but by the intermediate density narrow line region. As before, these models predict too weak NV (see Fig. 8).

NV is rather broad (FWHM[FORMULA]2800 km s-1) compared to the other emission lines in this object. Broad NV (FWHM[FORMULA]3000 km s-1) was also reported by McCarthy et al. (1990). This suggests that the BLR emission could contaminate the line. The spectroscopic properties of this objects show that it is not a quasar (or BLRG). All the lines, including CIII] not susceptible to absorption, are narrow and therefore, the emission is not dominated by the BLR. The line ratios are also inconsistent with quasar values (CIV/HeII[FORMULA]7, CIII]/HeII[FORMULA]3). The high level of polarization of the continuum and the fact that the line emission is not dominated by an unresolved component (see the plateau presented by the Ly[FORMULA] in Fig. 3) suggest also that it is not a quasar.

However, a close look to the spectrum (see Fig. 1 top left panel and Fig. 5 bottom panel) shows the presence of a very broad underlying component to the CIV line (maybe also to HeII and CIII]) that suggests some contribution of the BLR (scattered or direct). The broad NV suggests that at least this line is contaminated by the BLR emission. While the fit to the CIV line to measure the FWHM and flux of the line neglects the broad wings seen in the spectrum, the contribution to NV (and CIII]) could be strong enough to broaden the line profile and enhance the NV emission relative to the other lines. This would also explain the anomalous strong CIII] emission.

An alternative possibility is that MRC2025-218 is richer in metals than other HzRG. MRC2025-218 lies at the top of the HzRG correlation. Unless we have a view on the BLR, the NV diagram suggests that MRC2025-218 is the most enriched HzRG observed. We have calculated upper limits to the flux of the NIII] and NIV] lines. After taking into account the possible FWHM values suggested by the other emission lines we obtain [FORMULA]0.6 and [FORMULA]0.8 which are again well above the model predictions.

In summary, the strength of NV is inconsistent in L1 and MRC2025-218 with the standard model (low density and solar abundances) predictions. In both objects the spectrum seems to be dominated by the intermediate density narrow line region. This is supported by the compact appearance of the dominant line emitting region. This could be the case of other HzRG. The models suggest that N is overabundant in the ionized gas.

5.3. The absorption lines

In Sect. 3.3 we reported the detection of several absorption features, and the presence of PCygni profiles for CIV and SiIV lines. We discuss here the nature of the absorption: is it stellar or interstellar? The detection of stellar features would be very important; the most convincing evidence for stars in a HzRG has been found in 4C41.17 (Dey et al. 1997). On the other hand, associated ([FORMULA]) narrow absorption line systems have been found in the spectrum of several radio loud quasars (e.g. Anderson et al. 1987) and HzRG. Röttgering et al. (1997) reported the existence of deep troughs in the Ly[FORMULA] velocity profile of many HzRG. In most cases the Ly[FORMULA] emission is absorbed over the entire spatial extent (up to 50 kpc). The authors interpret these results as absorption by HI, physically associated with the radio galaxy, and having a covering factor near unity. Narrow absorption troughs have been also found in the spectral profile of the CIV line in the radio galaxy 0949-242 (Röttgering & Miley 1997), which is likely to be due to associated absorption systems as well (Binette et al. 1999).

The safest way to confirm the presence of stars is the detection of purely stellar photospheric features, but we do not detect them in MRC2025-218. The identified features may have a dominant contribution from interstellar absorption.

We have compared the absorption line spectrum of MRC2025-218 with that of:

  1. 4C41.17

  2. 3C191 ([FORMULA]1.953), the first QSO found to have a rich absorption line spectrum ([FORMULA]1.947) (Burbidge et al. 1966). The absorption is produced by associated absorption systems, interpreted as the consequence of material flowing out of the nucleus of the QSO.

  3. the star forming knot B1 in the nearby starburst galaxy NGC1741 (and other nearby starburst galaxies).

  4. star forming galaxies at [FORMULA]3-4

The EWs of the absorption features are larger than the values measured in 4C41.17 and NGC1741 and other starburst galaxies (EW[FORMULA]2 Å, York et al. 1990). The values are consistent (except for CIV) with the EWs in star forming galaxies at [FORMULA]3 (EW[FORMULA]2-3.5 Å) (Steidel et al. 1996, Yee et al. 1996). The agreement is best with the absorption lines in 3C191. We cannot give a definitive answer on the nature of the absorption features in MRC2025-218. P Cygni profiles are characteristic of Wolf-Rayet and O star winds and have been observed in star forming galaxies at redshift [FORMULA]3 (Steidel et al. 1996). However, the features we detect are highly contaminated by interstellar absorption in normal starburst galaxies. P Cygni profiles have been observed both in high and low ionization lines in some HzRG (see Fig. 5 in Dey 1998). The author suggests that the absorption is produced by fast outflowing material moving at high velocity relative to the galaxy (this could not explain redshifted absorption features observed in some HzRG, though [Röttgering et al. 1997]).

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: November 2, 1999