Nuclear activity versus star formation: emission-line diagnostics at ultraviolet and optical wavelengths

TL;DR

This study develops CLOUDY-based photoionization grids for AGN narrow-line regions and star-forming galaxies, validating optical diagnostics and introducing robust ultraviolet emission-line diagnostics to separate AGN activity from star formation in distant galaxies. By exploring line combinations with He II $1640$ and leveraging Ne- and O- based UV transitions, the work identifies high-signal discriminants that remain informative across a broad range of gas densities, metallicities, and ionization parameters. The results show that UV diagnostics, especially Ne-based diagrams and He II–normalized ratios, provide strong separation between AGN and SF, offering practical tools for interpreting JWST/ELT-era spectra and constraining interstellar gas properties and ionizing radiation fields. The framework aligns with existing UV observations and lays groundwork for automated classification and Bayesian inference of ionizing sources in the early Universe.

Abstract

In the context of observations of the rest-frame ultraviolet and optical emission from distant galaxies, we explore the emission-line properties of photoionization models of active and inactive galaxies. Our aim is to identify new line-ratio diagnostics to discriminate between gas photoionization by active galactic nuclei (AGN) and star formation. We use a standard photoionization code to compute the emission from AGN narrow-line regions and compare this with calculations of the nebular emission from star-forming galaxies achieved using the same code. We confirm the appropriateness of widely used optical spectral diagnostics of nuclear activity versus star formation and explore new diagnostics at ultraviolet wavelengths. We find that combinations of a collisionally excited metal line or line multiplet, such as CIV 1548,1551, OIII]1661,1666, NIII]1750, [SiIII]1883+[SiIII]1892 and [CIII]1907+CIII]1909, with the HeII 1640 recombination line are individually good discriminants of the nature of the ionizing source. Diagrams involving at least 3 of these lines allow an even more stringent distinction between active and inactive galaxies, as well as valuable constraints on interstellar gas parameters and the shape of the ionizing radiation. Several line ratios involving Ne-based emission lines, such as [NeIV]2424, [NeIII]3343 and [NeV]3426, are also good diagnostics of nuclear activity. Our results provide a comprehensive framework to identify the sources of photoionization and physical conditions of the ionized gas from the ultraviolet and optical nebular emission from galaxies. This will be particularly useful to interpret observations of high-redshift galaxies with future facilities, such as the James Webb Space Telescope and extremely large ground-based telescopes.

Figures (17)

• Figure 1: Examples of spectral energy distributions (in units of the luminosity per unit frequency at the Lyman limit) of the incident ionizing radiation in the AGN and SF models of Table \ref{['tab:parameters']}. The grey shaded area indicates the location of AGN ionizing spectra with power-law indices between $\alpha=-2.0$ (bottom edge) and $-1.2$ (top edge). The blue and orange lines show the ionizing spectra of stellar populations of metallicities $Z=0.001$ and 0.03, respectively. Vertical lines indicate the ionizing energies of ions of different species.
• Figure 2: Predictions of the AGN and SF models described in Section \ref{['sec:models']} in the standard [O iii]$\lambda5007$/H$\beta$ versus [N ii]$\lambda6584$/H$\alpha$ BPT diagnostic diagram, for different assumptions about the gas density ($n_{\rm H}=10^2$, $10^3$ and $10^4$ cm$^{-3}$ from left to right) and metallicity ($Z=0.0001$, 0.001, 0.01774, and 0.03 from top to bottom). Also indicated in parenthesis in each panel is the interstellar oxygen abundance, $12+\log{\rm (O/H)}$, which includes the components in the gas and dust phases. In each panel, we show AGN models corresponding to wide ranges in power-law index, $-2.0\leq\alpha\leq-1.2$, and ionization parameter, $-4.0\leq\log U_{\rm S}\leq-1.0$ (both increasing from bottom to top as indicated in the lower-right panel, and colour-coded as indicated in the top-right panel), and SF models (magenta stars) corresponding to the same range in $U_{\rm S}$. In each panel, the data are the SDSS observations described in Section \ref{['sec:SDSS']}, while the long-dashed and dot-dashed black lines indicate, respectively, the criteria of kewley01 and kauffmann03 to distinguish AGN from star-forming galaxies.
• Figure 3: Predictions of the AGN models described in Section \ref{['sec:models']} in the [O iii]$\lambda5007$/H$\beta$ versus [N ii]$\lambda6584$/H$\alpha$ BPT diagram, for a gas density $n_{\rm H}=10^3$ cm$^{-3}$, a metallicity $Z=0.03$ and dust-to-metal mass ratios $\xi_{\rm d}=0.1$, 0.3 and 0.5, from left to right. In each panel, we show AGN models corresponding to wide ranges in power-law index, $-2.0\leq\alpha\leq-1.2$, and ionization parameter, $-4.0\leq\log U_{\rm S}\leq-1.0$. The data and colour coding of the models are the same as in Fig. \ref{['fig:BPT1']}. The long-dashed and dot-dashed black lines indicate, respectively, the criteria of kewley01 and kauffmann03 to distinguish AGN from star-forming galaxies.
• Figure 4: Predictions of the AGN and SF models described in Section \ref{['sec:models']} in the [O iii]$\lambda5007$/H$\beta$ versus [S ii]$\lambda6724$/H$\alpha$ (top) and [O iii]$\lambda5007$/H$\beta$ versus [O i]$\lambda6300$/H$\alpha$ (bottom) diagrams of veilleux87, for two values of the metallicity, $Z=0.008$ (left) and 0.03 (right), and a dust-to-metal mass ratio $\xi_{\rm d}=0.3$. In each panel, we show AGN models corresponding to wide ranges in power-law index, $-2.0\leq\alpha\leq-1.2$, and ionization parameter, $-4.0\leq\log U_{\rm S}\leq-1.0$ (both increasing from bottom to top as indicated in the upper-left panel, and colour-coded as in Fig. \ref{['fig:BPT1']}), and SF models corresponding to the same range in $U_{\rm S}$. AGN and SF models have gas densities $n_{\rm H}=10^3$ and $10^2$ cm$^{-3}$, respectively. The long-dashed black line indicates the criterion of kewley01 to distinguish AGN from star-forming galaxies.
• Figure 5: Predictions of the AGN and SF models described in Section \ref{['sec:models']} in the C iv$\lambda1550$/C iii]$\lambda1908$ versus C iv$\lambda1550$/He ii$\lambda1640$ diagnostic diagram, for different assumptions about the gas density ($n_{\rm H}=10^2$, $10^3$ and $10^4$ cm$^{-3}$ from left to right) and dust-to-metal mass ratio ($\xi_{\rm d}=0.1$, 0.3 and 0.5 from top to bottom). In each panel, we show AGN models corresponding to wide ranges in power-law index, $-2.0\leq\alpha\leq-1.2$ (connected by solid lines), and ionization parameter, $-4.0\leq\log U_{\rm S}\leq-1.0$ (connected by dashed lines; $\alpha$ and $U_{\rm S}$ increasing as indicated in the upper-left panel), and SF models (stars connect by dotted lines) corresponding to the same range in $U_{\rm S}$, for different metallicities $Z$ (colour-coded as indicated in the top-right panel). Also shown in each panel are the observations of AGN (crosses with error bars) and star-forming galaxies (large diamonds with upper and lower limits) described in Section \ref{['sec:UVdata']}.