Astron. Astrophys. 330, 19-24 (1998)

3. Results

3.1. Absorption lines in the spectrum of Q0151+048A

The line search algorithm described in Moller et al. (1994) was used to find, measure, and identify absorption lines. In this case the continuum against which absorption lines were measured included the absorption of a damped Voigt profile near , with parameters as described below. The resulting line list, excluding the DLA line, and complete to 4 , is provided in Table 2. Listed there are the observed vacuum wavelength, the observed equivalent width , and the error on the equivalent width , for each line. The error on the wavelengths of the line centroids is dominated by the uncertainty associated with centring the quasar in the slit, and is estimated to be .

Table 2. Absorption lines in Q0151+048A

Sargent et al. (1988) have identified lines 14 and 15 as the CIV(1548, 1550) doublet at . With our identification the wavelength difference between the OI(1302) and SiII(1304) lines, at , is , whereas with their identification the wavelength difference is only less. With these data it is not possible to distinguish between these two possibilities, but given the strength of the other SiII lines at we find the most natural identification is that given in Table 2.

The DLA line near absorbs part of the quasar Ly and NV emission lines, so to determine the best-fit parameters for the damped line it was necessary to allow for the fact that the unabsorbed spectrum (the "continuum") is not at all flat in the region of the DLA line. By dividing the spectrum by the Voigt profile for trial values of redshift and column density, a first solution for these parameters was found, which produced a realistic unabsorbed quasar emission line profile - disregarding the wavelength region over which the absorption line is saturated. A smooth continuum, including Ly and NV emission lines, was fitted to this corrected spectrum, interpolating across the saturated region, . For this continuum we now determined the Voigt profile which best fit the observed spectrum. Dividing again by the model, the procedure was iterated to a solution. The results of this process are illustrated in Fig. 2, which shows an expanded plot of the spectrum in the region of the damped line, together with the best fit to the absorption line.

Fig. 2. Observed spectrum of Q0151+048A. The smooth line shows the fit to the damped absorption line, for values of , . Note the residual in the blue side of the saturated part of the DLA line.

While there is a certain amount of arbitrariness in the details of the final model for the quasar emission lines, the same is not true for the DLA absorption line. The parameters of this line are strongly constrained by the saturated central part and by the steep sides. The best fit was obtained with , , but acceptable fits could be obtained for column densities in the range .

3.2. Emission from the DLA absorber

Inspection of Fig. 2 shows that the model provides a close match to the profile of the absorption trough, except in the bottom of the DLA line where there appears to be a weak narrow emission line at the blue edge of the saturated part of the profile. On the assumption that the model absorption profile is correct, as evinced by the excellent fit at all other wavelengths, this emission feature is significant at the level. Note that while it is possible to obtain other acceptable fits to the absorption trough by adjusting slightly the modelled Ly and NV quasar emission lines, and making corresponding changes to the DLA parameters, this can never significantly change the saturated part of the absorption line profile, so the flux in this emission feature is quite insensitive to the details of the fitting. To illustrate the emission feature more clearly we have subtracted the model absorption profile from the data, and divided the difference by the 1 error spectrum. The resulting residuals (smoothed for display purposes) are shown in Fig. 3.

Fig. 3. Illustration of the emission feature at 3563.4 . After subtraction of the model absorption profile, the residuals of the quasar spectrum were divided by the 1 error spectrum. Plotted is the resultant signal-to-noise spectrum, smoothed with a 5-pixel boxcar filter. The emission feature, if Ly , is blueshifted relative to the DLA line by .

The wavelength centroid of the emission line is 3563.4 , and the measured line flux is . However the spectrum of Fig. 1 has not been corrected for relative slit losses between the quasar and the spectrophotometric standard. These were substantial because the quasar was observed at large zenith distance. Therefore we calibrated our spectrum by firstly scaling to the spectrum of Osmer, Porter, and Green (1994), and then scaling their spectrum to the literature UBV magnitudes. This brings the line flux to , and indicates that we only captured about 35% of the flux. The new value for the line flux may still be an underestimate of the total line flux, dependent on the angular size of the emission-line region relative to the slit width. If the line is interpreted as Ly emission the redshift of the object is , which is to the blue of the absorber. The line luminosity would be several times that for the DLA absorber towards PKS0528-250 (object S1,
Paper I).

3.3. The emission redshifts of Q0151+048A, B

In considering the nature of the DLA absorber, and the object responsible for the emission line, it is important to measure accurately the redshift of the quasar. The Balmer lines or narrow forbidden lines can be used to measure the systemic redshift, but have yet to be observed for this object. Instead we must use lines in the restframe UV. However it is well documented that the high ionization lines (e.g. CIV here) are blueshifted relative to the quasar systemic redshift by typically several hundred (e.g. Espey 1989, Tytler & Fan 1992, Espey 1997). Any blueshift for the low ionisation lines (e.g. SiII, OI, MgII here) fortunately is small.

In Table 3 we collect values of the redshift measured for a number of lines in the spectrum of the quasar Q0151+048A, as well as the MgII line in the spectrum of the quasar Q0151+048B. All estimates are from fits to the line peak. The values for the weak SiII and OI lines are our own measurements from the spectrum of Fig. 2. For both of these lines we used rest wavelengths of both multiplets under the assumption that the lines are optically thick. The values for the Ly and NV lines are also our own measurements from the spectrum of Fig. 2, but here after division by the model DLA line. This correction does not strongly add to the uncertainty of the NV emission redshift, but the error on the Ly emission redshift is dominated by the uncertainty due to the absorption correction. The data for the SiIV and CIV lines are taken from Sargent et al (1988). The SiIV redshift is not very useful however, as the line is chopped up by strong absorption lines, and it is only included in the table for the sake of completeness. The values for the MgII lines for the two quasars were measured by us from the plots of the spectra provided by Meylan et al (1990).

Table 3. Emission lines and derived emission-line redshifts

In a large study Tytler & Fan (1992) found that, after accounting for measurement errors, the intrinsic scatter of the blueshift of any particular line relative to the systemic redshift is small, no more than , and they tabulated mean values of the blueshift for several lines. The corrections are smallest for the low ionisation lines. For example for OI and MgII they found mean values of 50 and 100 respectively. In the last column of Table 3 we list the redshifts (TF) after applying the corrections suggested by Tytler & Fan.

As discussed below, the corrections for the high-ionisation lines may not be suitable for bright quasars. Therefore to estimate the quasar systemic redshift we have formed a weighted mean of the redshifts for the three lines SiII, OI, MgII, and added 100 to the final error as an estimate of the systematic uncertainty. Our best estimate of the systemic redshift for Q0151+048A is then . For Q0151+048B our best estimate for the redshift is based on the MgII line only, and is .

From Table 3 it can be seen that for the two high ionization lines, CIV and NV, the corrections suggested by Tytler & Fan are too small. The CIV line in Q0151+048A is blueshifted by relative to the systemic redshift, and this is much larger than the mean blueshift for this line of quoted by Tytler & Fan. Espey (1997), and Moller (1997) have found several other cases of quasars where the CIV line is blueshifted by a similar, or larger, amount. Both Corbin (1990) and Espey (1997) find a correlation between blueshift of the CIV line and quasar brightness, and this is likely to be the explanation for the discrepancy, since the sample of Tytler & Fan contains very few optically bright quasars. In fact the correlation is visible in their Fig. 27. The correlation found by Espey would suggest a correction of for the CIV line for Q0151+048A, bringing it in line with the low ionisation lines.

To summarise, our best estimate for the systemic redshift of Q0151+048A is , which is lower than the redshift of the DLA absorber.

© European Southern Observatory (ESO) 1998

Online publication: January 8, 1998
helpdesk.link@springer.de