Robustness of AGN selection with optical emission lines and the physical drivers of their emission line diversity
Reliably separating galaxies which do or do not host active nuclei is necessary for investigations into the formation and evolution of galaxies as the contamination from such objects may render measurements of gas-phase metallicity or star formation rate inaccurate. With studies that use optical emission lines, the [NII]-based BPT diagnostic diagram is the most widely used method for discriminating between galaxies which are purely star-forming and those which appear to have an extra source of ionization, perhaps from an active galactic nucleus (AGN), shocks, or hot evolved stars. We collectively refer to these objects as Seyferts/LINERs (S/Ls). Using the BPT diagnostic diagram, S/Ls and non-S/Ls can be studied statistically with large samples from SDSS, but the completeness of their selection and the extent to which the non-S/L selection may be contaminated are not well-parameterized. We place secure AGNs selected by excess X-ray emissions on the BPT diagram and determine the X-ray AGN fraction as a function of BPT position. We explore the possible reasons suggested in the literature for the misclassification of some X-ray AGNs as star-forming and find that neither hidden broad lines nor dilution by star formation can provide a satisfactory explanation. Motivated by this finding, we then explore what physical factors drive the observed emission line diversity of SDSS S/Ls. It has been suggested that the S/L branch of the BPT diagram is a mixing sequence between star formation and S/L emissions, and we test the merits of this scenario by matching S/Ls to non-S/L doppelgangers and empirically remove the host contributions to the S/L emission lines. We find the extent of the AGN branch remains after the removal of star formation contributions, suggesting that mixing cannot explain its wide variety. Among the "pure" S/Ls, we find multimodality in the emission line ratios, and, based on a k-means clustering analysis of seven emission lines, the S/L branch splits into three groups which we refer to as Seyfert 2s, Soft LINERs and Hard LINERs. This multimodality suggests that the three groups may have ionizing sources that are fundamentally different from each other. Among each group, we find that variations in the ionization parameter, the ionizing source's hardness, and the metallicity are sufficient to explain their considerable diversity.
Hosted by Prof. Mathews