2024A&A...684A..53B

Context. Different studies have reported the so-called temperature problem of the narrow line region (NLR) of active galactic nuclei (AGNs). Its origin is still an open issue. To properly address its cause, a trustworthy temperature indicator is required.
Aims. To determine the temperature of an emission line plasma, the [O III] (λ4363Å/λ5007Å) line ratio is typically used. However, in the case of the NLR of AGNs, this ratio is not reliable when the electron density extends much above 10⁵ cm^–3 as collisional deexcitation strongly affects this ratio independently of the temperature. To verify the density regime, we need a density diagnostic that applies to high excitation plasma.
Methods. We propose that the weak [Ar IV] λλ4711,40Å doublet is the appropriate tool for evaluating the density of the high excitation plasma. We subsequently made use of the recent S7 survey sample to extract reliable measurements of the weak [Ar IV] doublet in 16 high excitation Seyfert 2s. As a result we could derive the plasma density of the NLR of our Seyfert 2 sample and compared the temperature inferred from the observed [O III] (λ4363Å/λ5007Å) ratios.
Results. It was found that 13 Seyfert 2s cluster near similar values as the [O III] (λ4363Å/λ5007Å) ratio, at a mean value of 0.0146 ± 0.0020. Three objects labeled outliers stand out at markedly higher [O III] values (> 0.03).
Conclusions. If for each object one assumes a single density, the values inferred from the [Ar IV] doublet for the 13 clustering objects all lie below 60 000 cm^–3, indicating that the [O III] (λ4363Å/λ5007Å) ratios in these objects is a valid tracer of plasma temperature. Even when assuming a continuous power-law distribution of the density, the inferred cut-off density required to reproduce the observed [Ar IV] doublet is in all cases < 10^5.1 cm^–3. The average NLR temperature inferred for the 13 Seyfert 2s is 13 000 ± 703 K, which photoionization models have difficulty reproducing. Subsequently we considered different mechanisms to account for the observed [O III] ratios. For the three outliers, a double-bump density distribution is likely required, with the densest component having a density > 10⁶ cm^–3.