The following text was automatically extracted from the image on this page using optical character recognition software:
SHEMMER & NETZER
TABLE 1
LINEAR REGRESSION PARAMETERS FOR N V/C IV VERSUS log UL,
Data Set Number of Pearson Spearman Slope Constant
Code" Objects (r) (r) (a) (b)
B 121 0.66 0.67 0.23 0.02 -3.22 0.29
B+N 130 0.51 0.54 0.16 0.02 -2.36 0.29
RQQ (B) 104 0.68 0.69 0.23 0.02 -3.17 0.30
RQQ (B+N) 113 0.52 0.56 0.16 0.02 -2.27 0.30
B+up.lim. 137 0.63 0.62 0.23 0.02 -3.28 0.30
B+N+up.lim. 146 0.49 0.50 0.16 0.02 -2.43 0.30
Data set codes are B for BLAGNs, N for NLS 1 s, RQQ for radio-quiet quasars,
and up.lim. for upper limits on the line ratio.
of objects in each bin. For each HF93 line ratio-PL, log-log
diagram, we performed a linear regression analysis and cal-
culated the Pearson and Spearman linear correlation coeffi-
cients. This analysis was carried out once for the BLAGN
population and a second time for the entire data set. The results
are presented in Tables 1 and 2; one can see that the exclusion
of radio-loud objects or the addition of flux ratio upper limits
(that were treated as real ratios) had little effect on the cor-
relations and slopes.
Inspection of Figure 2 and Tables 1 and 2 shows that the
N v/C iv ratio is strongly correlated with luminosity in the
BLAGN case. However, it is also apparent that this relation-
ship breaks down completely at low luminosities, when the
NLS1s are introduced into the sample. Several NLS1s that
belong in the lowest luminosity regime even have N v/C iv
ratios as high as those of the most luminous high-z quasars.
The extremely strong N v X1240 in some NLS1s has been
noted earlier by Wills et al. (1999), who did not investigate
the resulting N/C abundance ratio. The N v/He ii ratio behaves
somewhat differently when NLS1is are added (Fig. 2). We
find that the N v/He it ratio in BLAGNs is correlated with
luminosity, although not as strongly as in the N v/C iv case,
and that NL Sls only slightly deviate from the N v/He 11 Z-
L slope. We also discover that both line ratios are not cor-
related with luminosity for vL, at 1450 A < 1046 ergs s-',
neither for BLAGNs nor for the entire sample. We emphasize
once more that the N v/C iv ratio in NLS1s is very similar
to the one observed in high-L AGNs, and this result, which
is based on an easy-to-measure line ratio, is enough to com-
pletely change the original HF93 correlation, regardless of
the exact slope or the value of the correlation coefficient.
3.2. High N v /C iv at Low Luminosity
Our new results point at two distinct scenarios: either (1) the
HF93 line ratios overpredict N/C at least under some physical
conditions or (2) high metallicities at low luminosities are pos-
sible and are seen in NLS Is. We discuss briefly the implications
of these two scenarios and defer the more detailed analysis to a
later publication.
3.2.1. Is the N v /C iv Line Ratio a Reliable N/C Indicator?
Hamann et al. (2002, hereafter H02) have used state-of-
the-art photoionization calculations to investigate several
line ratios in order to select those that are robust abundance
indicators. The calculations span a vast range of densities
[7 log nH (cm-3) 14] that completely cover the typical
range attributed to the broad emission line gas in AGNs
TABLE 2
LINEAR REGRESSION PARAMETERS FOR N v/He II VERSUS log PL,
Data Set Number of Pearson Spearman Slope Constant
Code' Objects (r) (rs) (a) (b)
B 98 0.55 0.54 0.19 0.03 -1.85 0.37
B+N 107 0.48 0.48 0.15 0.02 -1.33 0.33
RQQ (B) 82 0.57 0.55 0.18 0.03 -1.67 0.35
RQQ (B+N) 91 0.51 0.50 0.14 0.03 -1.20 0.31
B+up.lim. 110 0.51 0.49 0.20 0.03 -1.91 0.39
B+N+up.lim. 119 0.43 0.43 0.15 0.03 -1.35 0.35
a Data set codes are identical to those in Table 1.
suspect that they lie outside the range covered by the H02 cal-
culations. According to the calculations, the N v/C iv line ratio
is a reliable N/C indicator over the range of interesting physical
conditions expected in the BLR. H02 have also considered sev-
eral of the weak intercombination lines, such as N iii] X1750/
O III] X1663 and N iv]/C iv, previously discussed by Shields
(1976), and they concluded that the first is more reliable than
the second, as it is less sensitive to the model assumptions.
Results for all relevant line ratios are shown in their Figure 4,
and a specific application to the locally optimally emitting clouds
(LOCs) model is shown in their Figure 5.
The only other line ratio available to us, except for the
N v/C iv ratio discussed above, is N iv]/C iv. We have therefore
investigated the N/C obtained from the two line pairs both
observationally and theoretically. Several BLAGNs of our sam-
ple have published flux values (or upper flux limits) of the
weak line N iv]. We added to this subsample measurements of
the N iv] line in eight of our nine extreme NLS1is. The N/C
abundance obtained from the N Iv]/C iv line ratio was then
calculated for the subsample, assuming that upper limits rep-
resent real ratios. We find that the two line ratios are well
correlated (r = 0.8 for 30 sources); i.e., if N v/C iv is a good
N/C indicator, so is N Iv]/C iv. However, the derived N/C,
assuming both ratios are reliable metallicity indicators, is very
different. The N v/C iv ratio gives systematically larger N/C,
sometimes by a factor as large as 3 or 4. We stress that the
results are still tentative because of the large number of upper
limits rather than real line ratios used in the analysis. Better
data are required to confirm this correlation.
Regarding the suitability of N iv]/C iv as an N/C indicator,
we note that the calculations presented in Figure 4 of H02 show
that the line ratio does not change by more than a factor of 2
over the range of conditions thought to be acceptable in AGN
BLRs (ionization parameter of 0.03-0.3 for the H02 continuum
and density below about 1012 cm-3). The H02 conclusion that
the line ratio is not a robust N/C indicator is based on regions
in parameter space that are different from the one specified
above. Moreover, the particular example shown in Figure 5 of
H02, applicable to the LOC model, clearly shows that under
such conditions (that produce well most of the observed line
ratios in AGNs; see Baldwin et al. 1995), the N iv]/C iv line
ratio is indeed a very good N/C indicator. A key issue is whether
or not the physical conditions in the BLR of NLSis are similar
to those in broader line AGNs. The idea that density and optical
depths may be different has been proposed in the past, and a
better assessment of the N iv]/C iv suitability in this case must
await a more detailed theoretical investigation of such sources.
At present we do not have a large enough sample and good
enough calculations to test the suggestion that the results shown
in Figure 2 are due to N v/C iv being an inadequate N/C
indicator.
[log nH (cm-3)~ 10]. Since there is no evidence for densities
as large as 1013-10"4 cm -3 in NLS 1s, we have no reason to
No. 1, 2002
L21