6. The influence of the physical properties of the gas
6.1. The metallicity
So far, we have considered that differences in the ionizing mechanism determine those in the emission line spectra. The physical conditions of the gas can also influence strongly the emission line processes. An interesting and reasonable possibility is that the metallicity of the gas at high redshifts is lower than solar. We have calculated models in which the abundances of the heavy elements are 0.4 times the solar values. The ionizing continuum is the traditional -1.5 PL.
The agreement with the data is excellent as the diagrams in Fig. 4 show. The models are represented by the solid line connected with solid circles. Again, these models can also explain the data points in the CIV/Ly vs. CIV/CIII] diagnostic diagram.
Since the CIV/CIII] line ratio is strongly dependent on U (and not so much on metallicity in 0.4 solar to solar range of values considered) it is interesting to compare the U values required to reproduce the CIV/CIII] line ratio with the values derived for low redshift EELR from the optical line ratios. Most high redshift objects lie in the logU range [-2,-0.5]. The ionization level of the gas is similar to that observed in highly ionized EELR at low redshift (Robinson et al. 1987). However, this is not the trend observed in low redshift jet-cloud interaction objects, where shocks are important (Clark 1996, Clark et al. 1996). In these cases the average level of ionization of the gas is lower than that measured for the EELR of radio galaxies where AGN photoionization dominates. If a -1.5 PL is responsible for the ionization of the gas, these results suggest that we are biased towards very powerful objects in which the gas is highly ionized.
If lower metallicities are confirmed at high redshift a harder AGN ionizing continuum is no longer necessary and the classical -1.5 PL is still valid. An interesting consequence is that this would be the first clear evidence for chemical evolution with redshift in the host galaxies of powerful radio galaxies.
There is other evidence for chemical evolution with redshift, with decreasing metallicities for younger objects: damped Ly systems (DLA) at large redshifts ( 2-3), which are thought to be produced by protogalactic disks, are measured to have a metallicity 1/10 (Pettini et al. 1994). Although the metallicity evolution revealed by the DLA might not be valid for all kind of objects, it is suggestive and might also happen in radio galaxies.
6.2. Matter bounded clouds
In spite of the good general agreement between -1.5 PL models and the observed optical line ratios of low redshift radio galaxies, there are three problems which the classical photoionization models cannot explain: a) too weak high ionization lines; b) too low electronic temperatures; c) too small range in the ratio HeII 4686/H . Several authors have proposed in the past that a contribution of matter bounded clouds could help to solve some of the discrepancies (e.g. Viegas & Gruenwald 1988). In a recent paper Binette et al. (1996) develop a detailed study around this possibility. They consider two distinct populations of line emitting clouds: a matter bounded component and an ionization bounded component, with the ionization bounded clouds illuminated by the ionizing spectrum escaping from the matter bounded component. A variation in the ratio of the two components can explain the sequences in the optical line ratios on the diagnostic diagrams, and they can also solve the three problems mentioned above.
We have considered these models as a possible solution to the discrepancies between the -1.5 PL models and the data. The authors present the predictions for the CIV/CIII], CIV/H and HeII 4686/H ratios. Asumming as before, HeII 1640/HeII 4686 7.7 we can compute the UV line ratios of interest to us.
The models are represented in Fig.4 (short thick solid line). The authors adopt as continuum energy distribution a power law of index -1.3 and select a constant ionizing parameter 0.04. The variable parameter is , which represents the collecting area ratio of the ionization bounded clouds to radiation bounded clouds. The values vary from 0.06 to 4.
These models are unable to explain the wide range in UV line ratios observed for the high redshift sample. The reason is that the UV line ratios are not so sensitive to as the optical line ratios, which involve lines produced in partially ionized regions inside the clouds. The UV lines studied here are produced mainly near the illuminated face of the emitting clouds. Unlike at low z, where a simple variation of can explain the observed trend in the optical line ratios, the UV line ratios of the high redshift sample suggest that changes in U play a more important role.
The harder continuum might be the reason why the models lie closer to the data than the -1.5 PL models (the -1.3 lies between the -1.0 and -1.5 sequences). Adding a matter bounded component to the previous -1.5 PL models will not help to fit the data: CIV/HeII will remain nearly constant and CIII] /HeII will decrease slightly (although overlapping in part with the HeII region, CIII] is formed deeper inside the clouds).
Therefore, a mixture of radiation and matter bounded clouds, cannot explain the discrepancies between the -1.5 PL models and the observed data.
6.3. A diagnostic in the optical
Among the possibilities studied here to explain the UV data, low metallicities and a hard power law ionizing continuum produce a good fit to the data. As the models overlap in the UV, a way to discriminate between them is to compare the predicted optical (rest frame) line ratios with the observations. One of the few examples of objects with infrared (optical rest-frame) spectroscopy is the radio galaxy 4C40.36 (Iwamuro et al. 1996). We have compared the optical line ratios with the values derived from a) the -1.0 power law models and b) the -1.5 power law and 0.4 solar metallicity values, both of which explain the UV line ratios (Table 2). This object is also included in our sample. The UV line ratios (CIV/CIII] =1.05, CIII]/HeII=1.05 and CIV/HeII=1.11) indicate that, for a hard PL of index =-1, the ionizing parameter U 0.008 and for the -1.5 PL and low metallicity models, U 0.02. The ratios predicted in the optical are given in Table 2:
Table 2. Predicted and observed optical line ratios for the radio galaxy 4C40.36 (the models presented are the ones which reproduce the UV line ratios).
The flux of some of the lines is poorly determined and the errors in the line ratios are large. The most accurate ratio is OIII 5007/OII 3727 whose value is in very good agreement with the -1.0 power law model while the low metallicity model can not account for it. The high values of the ratios OII/H and OIII/H , if confirmed, are also in better agreement with the hard continuum model.
It is clear that optical line ratios are able to discriminate between these two possibilities. The acquisition of good quality optical (rest-frame) spectra of the HZRG will be crucial to test if the models valid for the UV lines can also fit the optical line ratios.
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998