Astron. Astrophys. 345, 73-80 (1999)

4. Description of the associated system

4.1. Overview

The absorption profiles produced by the associated absorbers are shown for the most important transitions on a velocity scale in Fig. 1. Complementary identifications in the G230L spectrum of lower quality because of the presence of the LLS break at 2700 Å are given in Fig. 2. The two spectra obtained in 1997 (end of october) and 1998 (beginning of october) are superimposed on Fig. 2 to look for variability. It can be seen that the modest signal-to-noise ratio prevents any firm conclusion about the variability of the Ne VIII absorption lines. There are absorption features near the expected position of Mg X 609,624 but, due to poor spectral resolution, this cannot be ascertained. On the contrary, O V and probably N III are present. The maximum column density of He I found in the models discussed below is 10¹¹ cm^-2. We thus do not believe that the possible line at 1875 Å seen in only one of the spectra can be He I 584.

Fig. 1. Absorption profiles of different transitions in the associated systems observed along the line of sight to J 2233-606. The zero velocity is taken at z = 2.20. The vertical dashed lines mark redshifts 2.1982, 2.2052, 2.2075 and 2.2215 from the left to the right. Note the component at +950 km s-1 ( = 2.21) with O VI and Ne VIII but no detectable H I absorptions.

Fig. 2. Possible identifications of lines from the associated system in the G230L spectrum. The vertical dashed lines mark the redshift range 2.198-2.210. The two spectra obtained in 1997 (end of october) and 1998 (beginning of october) are superimposed to look for variability. It can be seen that the modest S/N ratio prevents any firm conclusion about the variability of the Ne VIII absorption lines.

The emission redshift of QSO J 2233-606, derived from the high-ionization emission lines C IV , C III ] + Al III , z_em = 2.237, is smaller than the redshift derived from Mg II , z_em = 2.252, by about 1390 km s^-1 (Sealey et al. 1998). Considering the Mg II redshift as more representative of the intrinsic redshift (Carswell et al. 1991), the associated absorptions, seen over the redshift range 2.198-2.2215, have outflow velocities relative to the quasar of 2800-5000 km s^-1 which is modest compared to usual associated or BAL outflows.

4.2. z_abs = 2.2215

There is a Ly line at this redshift with flat bottom, consitent with line saturation, but with non-zero residual intensity. From the latter, it can be seen on Fig. 3 that the minimum covering factor for this line is   0.7. However before drawing any conclusions it is important to show that the feature is not due to blending of a few weaker lines. In Fig. 3 we plot the line profiles of the other detected Lyman series lines from this system. It can be seen that the Ly line is very strong. Moreover the residual intensity in the Ly line is smaller than the residual intensity in the Ly line, consistent with saturation of the Ly line.

Fig. 3. Analysis of partial coverage in the = 2.2215 system. The observed Ly profile is plotted in the bottom panel. The middle panel shows the observed Ly profile (solid) together with the predicted profiles computed from Ly assuming covering factors  = 1, 0.8 and 0.70 for the dotted, short-dashed and long-dashed lines respectively. The top panel shows the observed Ly profile with the predicted profiles computed from Ly assuming covering factors  = 1 and 0.9 for the dotted and dashed lines respectively.

We first assume that the covering factor is the same for Ly and Ly. In the middle panel we plot the observed Ly profile together with the predicted Ly profiles computed from the Ly profile for three values of the covering factor: dotted, short-dashed and long-dashed lines are for covering factors 1, 0.8, 0.70 respectively.

The predicted Ly profiles are inconsistent with the observed Ly profile and the latter seems too strong even for the minimum covering factor acceptable for the Ly line (  0.7). The numerous saturated lines present in this part of the spectrum assures that the error in the zero level determination cannot explain the discrepancy. Although we cannot reject the presence of weak Ly absorption lines superimposed with the Ly absorption, especially in the blue-wing, the good wavelength coincidence between Ly and Ly seems to indicate that the contamination cannot be large. One way to explain the apparent strength of the Ly line is to assume that the covering factor for Ly is larger than for Ly. Note that this is consitent with the QSO Ly emission line to be stronger than the Ly emission line. If true, then we can expect that the covering factor of the Ly line be even larger.

In the top panel we plot the observed wavelength range of the Ly line and the predicted profile using the Ly profile for covering factors 1 (dotted line) and 0.9 (dashed lines). The best match is obtained for complete coverage though this does not reproduce the Ly very well. Smaller values of the covering factor predict too strong a Ly line. We conclude that, in order to understand the residual intensities in the different Lyman series absorption lines, it must be assumed that the covering factor increases from Ly to Ly. It is thus likely that the absorbing cloud at z_abs = 2.2215 completely covers the continuum source and partially covers the BLR.

This system does not show absorption due to any detectable heavy element transitions either in the optical data (Outram et al. 1998; Savaglio 1998) or in the HST data. If partial coverage is a signature of physical association between the absorbing gas and the AGN, then the lack of metal lines in this system could suggest that there are large inhomogeneities in the chemical enrichment of the gas physically associated with central engines of QSOs. However, on the contrary, if we presume that the gas associated with the central regions of the quasar is more or less uniformly enriched, then this system could correspond to very highly ionized gas. The ionization state should be such that all observable metal transitions are weak and undetectable. This condition demands log H I /H to be smaller than -8 (e.g. Hamann 1997) and log N(H)22. Note that such a cloud could be related to the warm absorbers.

Another possibility is that the gas is extremely metal-poor and is produced by an intervening cloud with sizes less than the BLR (i.e. few pc). Not only this is much smaller than the dimensions derived for intervening Ly clouds using adjacent lines of sight (e.g. Petitjean et al. 1998) but also these clouds would have been detected by previous surveys far away from the QSO.

4.3. z_abs = 2.207

The C IV , N V and Ly absorption lines produced by this system are shallow and broad (500 km s^-1) like a miniaturised Broad Absorption Line system (BALs). Outram et al. (1998) discuss the N V and Ly absorptions from this system. They could not fit the N V doublet when assuming 100% coverage. However, they managed to obtain a consistent fit after correcting the continuum by subtracting a Gaussian centered at 3938 Å with FWHM = 670 km s^-1 and maximum depth 19 percent of the original continuum level. They used a three-component model with large velocity dispersions. Savaglio (1998) observed the C IV absorption doublet from this system. She could fit the doublet with six components. The O VI and Ne VIII doublets are detected in the G430M and E230M spectra respectively. Note that the O VI doublet is observed at slightly lower resolution than N V and C IV and the Ne VIII doublet is found in a low S/N region blueward the Lyman limit of the moderately thick system at z_abs = 1.87. Finally, strong O V 629 is seen at 2020 Å as expected and possibly Mg X 624 at 2000 Å.

The stronger line of the C IV , N V , O VI and Ne VIII doublets, together with the Ly line, are plotted on Fig. 4 on a velocity scale. The dotted and dashed lines are the predicted velocity profiles computed from the profile of the second transition of the doublets for different values of the covering factor. Here again we assume identical covering factors for both transitions. It can be seen that in order to reproduce the residual intensities of N V we need a covering factor of the order of 0.35 at v  +150 km s^-1 and 0.50 at v  -100 km s^-1. The difference between the two models with  = 0.35 and 0.50, is, at these places, of the order or larger than 0.1 (10% of the normalized continuum; see Fig. 4) when the rms deviation in the spectrum is 0.04.

Fig. 4. Analysis of partial coverage in the = 2.207 system. The observed Ly profile is plotted in the bottom panel. The solid curves in the other panels are the observed profiles of the stronger line of the doublets. The dotted and short-dashed lines are the predicted velocity profiles computed from the profile of the second transition of the doublet and assuming covering factors  = 0.35 (dotted line) and 0.50 (dashed line).

The O VI profiles are consistent with a covering factor that cannot be reconciled with the values found for N V . Note that the O VI profiles could be affected by the relatively poor spectral resolution and the derived covering factors should be considered as lower limits. The maximum error in the covering factor for the O VI lines is about 0.1, computed from the error in the residual intensity. The S/N ratio over the C IV doublet is not good enough and consistent residual intensities are obtained for a wide range of covering factors. The covering factor required for the Ne VIII lines is in the range 0.8-1.0 with an error per pixel of 0.15. The covering factor derived from the Ne VIII lines is therefore significantly larger than the one derived from the N V doublet. There seems to be an anti-correlation between the covering factor and the ionization state or the wavelength. This again is consistent with clouds partially covering the BLR while covering most of the continuum emission region.

Another interesting observation concerns the doubly-ionized species. It can be seen on Fig. 1 that O III 832 is certainly blended with other lines; that there may be a shallow absorption at the expected position of C III 977 although most of it should be Ly; and that there is a strong absorption at the expected position of N III 989. This line cannot be Ly at z_abs = 2.093 as there is no corresponding Ly line. We cannot reject the hypothesis that this is an intervening Ly line as the Ly range is of very poor S/N ratio. Interestingly enough, there is an absorption feature at the expected position of N III 684,685 (see Fig. 2). Although such a strong N III absorption would be very surprizing (see next section and Fig. 6), the presence of doubly-ionized species cannot be ruled out. The corresponding absorptions from triply ionized species, although weak and noisy, could be present (see in particular O IV 787 and N IV 765). If true, this would imply that the medium has two phases of low and high-ionization (see next section).

Since the possible strong N III 989 line goes to zero, suggesting complete coverage; we note that there is a tendency for the absorption lines redshifted on top of the emission lines (here C IV and N V ) to have covering factors smaller than lines redshifted in parts of the spectrum free from emission-lines (Ne VIII and N III ). This further supports the conclusion that the gas covers the continuum emitting region but only part of the BLR.

4.4. z_abs = 2.198

Ly and N V absorptions produced by this system are detected by Outram et al. (1998) who already noted the incomplete coverage of the background source. They conjectured that the absorbing cloud is larger than the continuum emitting region and smaller than the BLR. Savaglio (1998) has noted that single as well as two component fits to the C IV doublet result in a very poor fit, again suggesting partial coverage. O VI and Ne VIII doublets are detected in the HST spectra. The stronger lines of the doublets together with Ly are plotted on Fig. 5 on a velocity scale. The dotted and dashed lines are the predicted velocity profiles computed from the second transition of doublets assuming different values of the covering factor (dotted, short-dashed and long-dashed lines are for = 0.7, 0.8 and 0.9 respectively). Here again we assume identical covering factors for both transitions. The N V 1242 line of this system is blended with N V 1238 of the system at 2.208. We have subtracted the contribution due to the N V 1238 line before doing the analysis. The covering factor required to fit the N V doublet is . However the O VI profiles require values larger than 0.85. It is quite likely that the Ne VIII 780 is blended with some other line in the blue wing. Also the Ne VIII 770 profile is very noisy and is consistent with a wide range of covering factors. It is interesting to note that inspite of the poor signal to noise ratio, the C IV profiles clearly suggest that the covering factor is less than 0.7 and we have obtained a consistent fit for = 0.5. Thus like the z_abs = 2.207 system, this system also shows different covering factor for different transitions; the covering factor beeing lower for C IV than for N V .

Fig. 5. Analysis of partial coverage in the = 2.198 system. The observed Ly profile is plotted in the lowest panel. The solid curves in other panels are the observed profiles of the strongest transition of the doublet. The dotted, short-dashed and long-dashed lines are the predicted velocity profiles computed using the profile of the weakest member of the doublets and assuming covering factor of = 0.7, 0.8 and 0.9 respectively.

© European Southern Observatory (ESO) 1999

Online publication: April 12, 1999
helpdesk.link@springer.de