Astron. Astrophys. 354, 17-27 (2000)

4. Discussion

The obtained emission-line ratios of NV1240/CIV1549 and NV1240/HeII1640 were used to estimate the metallicity of the line emitting gas. The conversion of observed emission line ratios to relative abundances is affected with several severe uncertainties. One has to garantee for example that the emission lines which are studied originate in the same region of the BLR under comparable conditions of the gas. Even the input parameters of standard BLR photoionization models are uncertain within a factor of 2 - 3. A detailed discussion of the current limitations of the method can be found in Baldwin et al. (1996), Ferland et al. (1996), or Hamann & Ferland (1999). In spite of these uncertainties the emission line ratios of dedicated lines can be used to estimate relative abundance ratios. These results can be regarded as lower limits of the metallicity of the gas (e.g. Ferland et al. 1996).

Our estimate is based on the model calculations Hamann & Ferland (1992, 1993) presented. They computed for a large range of metallicities several evolutionary scenarios. They varied the slope of the IMF, the evolutionary time scale for the star formation, as well as the low mass cutoff of the IMF. They concluded that the high metallicities observed in high redshift quasars can be achieved only in models with rapid star formation (RSF) comparable to models of giant elliptical galaxies and a more shallow IMF than the IMF provided by studies of the solar neighborhood, i.e. more massive stars are present in RSF models.

To derive an estimate of the metallicity we compared our results with their models for two enrichment scenarios (Hamann & Ferland 1999) with a fixed density of the line emitting gas. The first scenario is characterized by scaled solar metallicities, i.e. with relative solar abundance ratios. The second model assumes abundance ratios which are given by the rapid star formation model. In contrast to the scaled solar metallicity model in the RSF model the abundance of nitrogen as a secondary element is built up quadratically with metallicity while primary elements like carbon or oxygen are built up only linearly.

Within the framework of the scaled solar abundance model the observed NV1240/HeII1640 ratio indicates supersolar metallicities. The derived abundance of the line emitting gas ranges from 7 (Q0044-273) up to 43 (HS1700+64). Computing the average of the metallicities indicated by NV/HeII yields Z 20 . In contrast to this result the observed NV1240/CIV1549 line ratios yield even higher abundances which are of the order of Z 100 . For several quasars we could calculate the OVI1034/HeII1640 line ratio. The corresponding abundances are in good agreement with the values obtained from NV1240/HeII1240. But the metallicity estimate based on the NIII]1750/OIII]1663 line ratio is consistent with metallicities of the order of 100 . The line ratios of NV/HeII and NV/CIV yield significantly different ranges for the metallicity of the line emitting gas assuming the scaled solar abundances model (Hamann & Ferland 1999). The line ratios combining nitrogen and carbon or oxygen indicate abundances of the order of 100 . This might be due to the strength of the nitrogen emission lines as a result of a stronger enhancement of N in comparison to C and O.

The rapid star formation (RSF) model yields higher than solar abundances for the broad emission-line gas, too. However, the range of the supersolar abundances is consistent based on the NV1240/CIV1549 and NV1240/HeII1640 line ratios. A supersolar metallicity ensues from the NV1240/ CIV1549 line ratio which ranges from 7 up to 17 with an average of Z 11 . The NV1240/ HeII1640 line ratio yields an average overabundance of the studied emission line gas of Z 5 ranging from 2 up to 10 for the individual quasars (Table 3). Both line ratios indicate an average enhancement of the metallicity of the line emitting gas for the studied quasars (2.4z3.8) of the order of Z 8 . This result is supported by the line ratios of OVI1034/HeII1640 and NIII] 1750/OIII]1663 which could be measured for a few quasars we observed. The OVI1034/HeII1640 line ratio also indicates a higher than solar abundances. However, the scatter of the derived metallicities is large due to the obvious difficulty in measuring the line flux of OVI1034 and HeII1640. The obtained estimate of the metallicity based on the OVI/HeII line ratio covers a range of 4 - 25 . This result is in good agreement with the high abundances provided by the NV/CIV and NV/HeII line ratios.

Table 3. The derived relative abundance of the line emitting gas given in units of solar metallicity . The estimates based on the calculations made with the rapid star formation scenario (Hamann & Ferland 1999). The metallicity is given in units of solar metallicities .

In four of the observed quasars, UM196, Q1548+0917, HS1700+64, and Q2231-0015, we could also detect the important intercombination line NIII]1750. This line was used together with OIII]1663 and CIII]1909. The line ratios NIII]/OIII] and NIII]/CIII] indicate only moderate oversolar abundances but again the metallicity of the gas is in the range of Z 3 .

In contrast to the scaled solar abundance model the RSF scenario provide consistent ranges for the metallicity of the line emitting gas based on different line ratios. Especially the line ratios of nitrogen as a secondary element with carbon or oxygen as primary elements provide evidence for an enrichment of nitrogen Z² while the portion of carbon and oxygen is enhanced Z as predicted by the RSF model (cf. Ferland & Hamann 1992, 1993).

The derived supersolar abundances of the line emitting broad-line region gas can be used to date the age of the first star formation epoch. We estimated the redshift z_f for several settings of , , and H_o. The evolutionary time scale was set to 0.5, 1.0, and 2.0 Gyr. With the diagrams given in Carroll et al. (1992), Perlmutter et al. (1999), Yoshii et al. (1998), and Hamann & Ferland (1999) we derived z_f for our quasar sample. For a universe with =1 (Einstein, de Sitter) the evolutionary time scale has to be of the order of less than or approximately 0.5 Gyr. For longer time scales the redshift of the first star forming epoch is larger than z10. Furthermore, for z4 the age of the universe starts to become less than the evolutionary time scale. Generally, the Einstein, de Sitter case requires z10 and km s^-1 Mpc^-1. However, current measurements of H_o and are in serious conflict with this.

More reasonable estimates are provided by models characterized by =0.3 (open universe), =0.3, =0.7 (flat, dominated universe, cf. Carlberg et al. 1999; Perlmutter et al. 1999), and H_o=65 km s^{- 1} Mpc^-1. Assuming an evolutionary time scale of = 0.5 Gyr the epoch of the first star formation can be dated to 4.5 for quasars with 3z4 and supersolar abundances. For longer evolutionary time scales of = 1 Gyr the first violent star formation should took place at 6 to 10 and at least 10 for = 2 Gyr. For redshifts of z4 or even larger already an evolutionary time scale of 1 Gyr indicates z_f 10. If 2 Gyr the detection of supersolar metallicity for quasars with z4 to 5 requires 0.3. Otherwise the universe would not be old enough for the assumed time scale to produce the observed metal enrichment of the gas. This might be taken as some evidence for short evolutionary time scales for the first violent star formation epoch. The quasar with the currently highest known redshift of z = 5.0 (SDSSp J033829.31+002156.3, Fan et al. 1999) provides some hint for a metal abundances comparable to the quasars observed at redshift z4 (Dietrich et al. 1999). Hence, the evolutionary time scale for this object should be less than 1 Gyr and the studied quasars favor models with 0.3, 0.7, and H_o=65 km s^{- 1} Mpc^-1.

© European Southern Observatory (ESO) 2000

Online publication: January 31, 2000
helpdesk.link@springer.de