Export citation and abstract BibTeX RIS
IOP Publishing regrets that the published article was released before all revisions were included. The text and figures of a few sections should be updated as follows:
In Section 1.
Numerous attempts have been made to analyze the narrowline spectrum of AGNs in the context of photoionization models (e.g., Ferland & Osterbrock 1986; Ho et al. 1993; Groves et al. 2004, 2006; Stern et al. 2014). This paper focuses on a limited goal: using realistic AGN photoionization models to predict the maximum range of the strengths of [Ne ii] 12.81 μm and [Ne iii] 15.56 μm relative to [Ne v] 14.32 μm. We show that in the radiation pressure-dominated regime, AGNs produce a relatively stable, restricted range of mid-IR neon emission-line spectra, paving the way for an updated calibration of [Ne ii] and [Ne iii] as a new SFR indicator for active galaxies. Our neon-based SFRs show good consistency with independent SFRs based on SED fitting.
In Section 2.2.
As we are interested in powerful AGNs with detected [Ne v] emission, sources with a relatively high ionization parameter, we restrict our investigation to log U ≈ −2.5 to −0.1, in steps of 0.3 dex.
In Section 3.1.
Figure 2 shows the influence of the ionization parameter on the strengths of the low-ionization lines [Ne ii] and [Ne iii] relative to the high-ionization line [Ne v]. Remarkably, all the line ratios remain stable when log U ≳ −1.6, with a scatter of ≲0.16, within the range of hydrogen densities and ionizing spectra considered in our model grid. This behavior is consistent with that of the dusty, radiation pressure-dominated isobaric photoionization models of Dopita et al. (2002), in which the ionization parameter-sensitive line ratio [O iii] λ5007/Hβ shows only mild variation with U in the regime of high U. When U ≳ 0.03 (Stern et al. 2014), dust becomes the dominant source of opacity and effectively couples radiation pressure to gas pressure. Dust influences the temperature structure of the NLR via photoelectric heating and excitation of the high-ionization gas. Radiation becomes the dominant source of pressure contributing to the static gas pressure, and the gas density scales linearly with ionizing flux, making the local ionization parameter independent of the initial ionization parameter (Dopita et al. 2002). In any event, the line ratios depicted in Figure 2 are affected only mildly by the range of densities explored in our calculations. As shown in Dopita et al. (2002) and Stern et al. (2014), the small dispersion of derived U for Seyferts suggests that the NLRs of most AGNs are radiation pressure-dominated. In the radiation pressure-dominated regime (−1.6 ≤ log U ≤ −0.1) and over the range of parameters investigated here, we find
This indicates that AGNs that span typical NLR conditions (in terms of ionizing SED, ionization parameter, and density) emit a relatively restricted range of [Ne ii] and [Ne iii] for a given level of [Ne v]. Similar trends are found by Groves et al. (2006) and Stern et al. (2014). The absolute value of their line ratios differs somewhat from ours mainly because of the choice of different input ionizing spectra. As described in Section 2.1, our adopted incident radiation field is based on observed intrinsic SEDs of unobscured AGNs that span a wide range in bolometric luminosity.
Figure 2. Dependence of (top) [Ne ii]/[Ne v], (middle) [Ne iii]/[Ne v], and (bottom) ([Ne ii]+[Ne iii])/[Ne v] on ionization parameter U, for hydrogen particle densities nH = 102 cm−3 (solid lines) and nH = 103 cm−3 (dashed lines), and the three input SEDs shown in Figure 1: log (Lbol/erg s−1) (red) > 46.3, (green) 45.8–46.3, and (blue) <45.8.
Download figure:
Standard image High-resolution image
Figure 3. Distribution of [Ne iii]/[Ne ii] ratios for star-forming galaxies (orange), AGN host galaxies (blue), and SINGS emission-line regions with projected extraction area <0.3 kpc2 (SINGS small; red) and ≥0.3 kpc2 (SINGS large; green), normalized by the total number of objects in each subsample. The measurements for the AGN hosts have been corrected for AGN contamination.
Download figure:
Standard image High-resolution image
Figure 5. Comparison between SFRs derived from SED fitting from GSWLC-2 (Salim et al. 2018) with those from our neon method assuming (a) solar metallicity and (b) individual metallicities. Circles and squares represent star-forming galaxies and AGNs, respectively. Error bars indicate lower and upper 1σ uncertainties. In panel (b), the SFRs from the neon method are computed assuming individual metallicities derived from the mass–metallicity relation. We adopt the mean of two metallicity estimates (from the N2O2 and N2 methods), and the vertical error bars show the difference between the two metallicity calibrations. The data points are color-coded with M* from GSWLC-2, according to the color bar on the right. The median and standard deviation of the difference in SFRs (logSFRNe−logSFRGSWLC−2) are shown in the bottom right corner of each panel for star-forming galaxies (black) and AGNs (red).
Download figure:
Standard image High-resolution image
In Section 3.2.
Figure 3 shows the distribution of [Ne iii]/[Ne ii] for star-forming galaxies, AGN hosts, and emission-line regions extracted from SINGS. A Kolmogorov–Smirnov test of the distributions of [Ne iii]/[Ne ii] for AGN hosts and the "SINGS small" sample indicates that the null hypothesis that the two distributions are drawn from the same parent population cannot be rejected with a probability of 71.9%.
In Section 4.2.
Equations (7) and (8) should be updated as follows:
In Section 4.3.
We compare the SFRs derived using our neon method with those from SED fitting. With the metallicity fixed to solar (Figure 5(a)), the neon-based SFRs are slightly but systematically higher than those from GSWLC-2, for both star-forming and AGN host galaxies, with a clear tendency for the magnitude of the excess to correlate with the stellar mass of the system (see the color bar). The offset is marginally higher for AGNs (0.24 ± 0.22 dex) than for star-forming galaxies (0.16 ± 0.20 dex), because AGN hosts tend to be more massive (Ho et al. 2003; Kauffmann et al. 2003), and hence more metal-rich. After adopting individual metallicities (Figure 5(b)), the neon-based SFRs are shifted downward and come into much better agreement with the SED-based SFRs. The systematic offset between the two methods essentially disappears, and the scatter also is reduced somewhat (star-forming galaxies: 0.02 ± 0.19 dex; AGNs: 0.05 ± 0.18 dex). We also run models with twice solar metallicity. Although slight differences exist for [Ne ii]/[Ne v] and [Ne iii]/[Ne v] ratios, the resulting SFRs for AGNs are only affected by 0.04 dex (0.09 dex), much smaller than the scatter.