THE ORIGIN OF [O ii] EMISSION IN RECENTLY QUENCHED ACTIVE GALACTIC NUCLEUS HOSTS

There is growing evidence for a link between the evolution of supermassive black holes (SMBHs) and the galaxies that host them. This includes the observed correlation between SMBH mass and bulge velocity dispersion (Gebhardt et al. 2000; Tremaine et al. 2002). How such correlations are established remains to be determined, but they suggest that SMBHs and their host galaxies play a role in regulating one another's growth. Recent hydrodynamical simulations suggest that feedback from active galactic nuclei (AGNs) may provide a means for this regulation. In this scenario, a sufficiently energetic AGN can drive outflows that disrupt its host galaxy and effectively quench ongoing star formation by removing the galaxy's gas supply (Springel et al. 2005; Hopkins et al. 2005).

If AGNs are responsible for rapidly quenching star formation in their hosts, then signs of recent transformation should be present in these galaxies, including post-starburst spectral signatures. Post-starburst galaxies exhibit Balmer absorption features due to the presence of recently formed (<1 Gyr) A-type stars, but lack emission lines that are indicative of ongoing star formation, such as Hα at 6563 Å or the [O ii] doublet at 3727 Å. Similar, yet weaker, features are also found in galaxies that have experienced a rapid truncation of normal star formation (i.e., post-quenched galaxies; Poggianti 2004).

Several recent studies have found evidence that post-starburst features are detected in AGN hosts more often than in their non-active counterparts. For example, Goto (2006) finds that 4.2% of optically identified AGNs at low redshifts exhibit post-starburst spectral signatures, compared to only 0.2% of normal galaxies. At higher redshifts, Georgakakis et al. (2008) determined that 21% of X-ray-selected AGN hosts located on the red sequence at 0.68 ⩽ z ⩽ 0.88 exhibit post-starburst signatures. At slightly higher redshifts, Kocevski et al. (2009b, hereafter Koc09b) report that ∼44% of the X-ray-selected AGNs found in the Cl1604 supercluster at z ∼ 0.9 show strong Balmer absorption features.

One of the difficulties of extending such work to higher redshifts is that the Hα line moves out of the optical window, forcing surveys to rely on the [O ii] doublet as the primary indicator of star formation activity. However, a comprehensive study by Yan et al. (2006) suggests that [O ii] emission is a poor indicator of star formation activity in many galaxies. They show that in 91% of red, early-type galaxies that exhibit [O ii] emission, the line originates from AGN activity and not normal star formation processes. This poses significant problems for the correct identification of post-starburst/post-quenched galaxies at high redshift, which are often selected based on the absence of [O ii] emission (e.g., Dressler & Gunn 1983; Poggianti et al. 1999). If AGNs contribute significantly to their host galaxies' [O ii] line flux, then excluding all [O ii] emitters from such samples would severely underestimate the fraction of AGN hosts that are truly in a post-starburst or post-quenched phase.

In this study, we use emission-line diagnostics measured from optical and near-infrared spectra to investigate the origin of [O ii] emission in six X-ray-selected AGNs at z ∼ 0.9. These AGNs are a subset of active galaxies detected in the Cl1604 supercluster by Koc09b that were found to have both strong Balmer absorption lines indicative of post-starburst/post-quenched systems as well as moderate levels of [O ii] emission. If the observed [O ii] lines in these galaxies are of AGN origin, then the high incidence of Balmer features would indicate an increased post-starburst fraction among AGN hosts at high redshift. Using newly obtained Keck II Near-Infrared Echelle Spectrograph (NIRSPEC) spectroscopy, we examine the flux ratio of the [N ii] λ6584 to Hα lines to ascertain the source of the ionizing flux exciting the [O ii] emission found in these galaxies.

We present our analysis in the following manner. Section 2 describes the Cl1604 supercluster and our AGN sample in greater detail and Section 3 discusses the observation and reduction of the near-infrared spectroscopy. In Section 4, we describe our methods for measuring equivalent widths (EWs) and line flux ratios. In Section 5 we discuss our findings, while in Section 6 we summarize our conclusions.

The six AGNs targeted for NIRSPEC observations are spectroscopically confirmed members of the Cl1604 supercluster, a large-scale structure which consists of eight galaxy clusters and groups at a median redshift of z ∼ 0.9 (Gal & Lubin 2004). The supercluster has been extensively studied at a variety of wavelengths (Gal et al. 2008; Kocevski et al. 2009a, 2011; L. M. Lubin et al. 2011, in preparation) and in Koc09b we identified nine X-ray-selected AGNs within the Cl1604 system. These AGNs have moderate X-ray luminosities (L_X ∼ 10⁴³ erg s⁻¹) and their optical spectra are consistent with type 2 Seyferts. The average spectral properties of the Cl1604 AGNs targeted with NIRSPEC can be seen in Figure 1, where a high signal-to-noise ratio (S/N) composite spectrum created by co-adding the individual AGN spectra is shown. All of the AGNs exhibit moderate to strong [O ii] emission lines and a majority show strong Balmer absorption features, which are clearly visible in the stacked spectra.

Zoom In Zoom Out Reset image size

Our available observing time allowed for only two-thirds of the full Cl1604 X-ray AGN sample to be targeted with NIRSPEC. Preference was given to those galaxies whose Hα and [N ii] features would be detected blueward of the strong OH airglow lines at λ ≈ 1.27 μm (z ⩽ 0.93). The details of the NIRSPEC observations are given in Section 3.2. The coordinates, redshifts, X-ray luminosities, and spectral properties of the six Cl1604 AGNs observed with NIRSPEC are listed in Table 1.

3.1. Optical Spectroscopy

3.2. Near-infrared Spectroscopy

4.1. Spectral Line Measurements

4.2. Line Ratio Diagnostics

Several sources of photoionizing radiation can give rise to narrow emission lines in galaxies, including young stellar populations and AGN-related continuum emission. Fortunately, the different line species these two mechanisms excite provide a means to distinguish which is the dominant source of ionizing flux. Therefore, it has become common to use the relative strength of lines with high and low ionization potentials to differentiate these two excitation mechanisms.

Introduced by Baldwin et al. (1981, hereafter BPT), such line ratio diagnostics typically utilize the strength of the [O iii] feature relative to Hβ and the strength of a forbidden line (typically 6300 Å [O i], [N ii], or 6716 Å+6731 Å [S ii]) relative to Hα. An example of a BPT diagram using the [N ii]/Hα and [O iii]/Hβ line flux ratios as measured in ∼55,000 Sloan Digital Sky Survey (SDSS) galaxies is shown in Figure 2. In this parameter space, Seyferts and low-ionization nuclear emission-line regions (LINERs) dominated sources (hereafter LINER/Seyfert) have higher values of [N ii]/Hα. In a study of nearly 100,000 SDSS galaxies, Kewley et al. (2006) found that the logarithm of the F([N ii])/F(Hα) ratio (hereafter $F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }$) for star-forming galaxies varied from −1.5 in extremely metal-poor systems to −0.3 in those with super-solar abundances. This upper bound is similar to the maximal boundary of $\log (F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }) = -0.22$ found for star-forming galaxies in a sample of Kauffmann et al. (2003); this limit is denoted by the vertical line in Figure 2.

Zoom In Zoom Out Reset image size

For this study, we lack observations of the Hβ and [O iii] emission features due to the redshift of our target galaxies. Instead, we rely on a pseudo-BPT diagram that utilizes the $F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }$ ratio as the primary discriminator between star-forming galaxies and galaxies dominated by LINER/Seyfert emission. In addition, we use the EW of the [O ii] and Hα emission features as a replacement for the traditional ordinate of the BPT diagram (i.e., F([O iii])/F(Hβ)). Yan et al. (2006) previously demonstrated the effectiveness of this type of pseudo-BPT diagram in separating star-forming and AGN-like systems (they utilized the [N ii]/Hα and [O ii]/Hβ ratios; see their Figure 11).

Without the additional information that $F_{[{\rm O\,\mathsc{iii}}]/{\rm H}\beta }$ provides, we have chosen to divide star-forming galaxies from those dominated by a LINER/Seyfert component at the Kauffmann et al. (2003) limit of $\log (F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }) = -0.22$. This value is similar to cuts used in other studies when Hβ and [O iii] are weak or unobservable (e.g., Miller et al. 2003; Stasińska et al. 2006; Lem10). Furthermore, we believe this value to be a conservative boundary between star-forming galaxies and LINER/Seyferts as previous studies have found evidence for substantial AGN contamination out to as far as $\log (F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha })= -0.35$ (Stasińska et al. 2006). In the following section we discuss where the six Cl1604 X-ray AGNs targeted with NIRSPEC fall on our pseudo-BPT diagram.

In Figure 3, we plot the ratio of [O ii] and Hα rest-frame EWs as a function of $F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }$ for the six Cl1604 AGN hosts observed with NIRSPEC. In four of these six galaxies we detect strong (>3σ) [O ii], [N ii], and Hα emission lines, while in two galaxies (X1 and X5) [N ii] was detected in our NIRSPEC spectroscopy, but Hα was not. For these two galaxies we make use of 3σ upper limits on the flux and luminosity of the Hα line. We find that five of the six Cl1604 AGN hosts have [N ii]/Hα flux ratios greater than $\log (F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }) = -0.22$, the maximum value found for star-forming galaxies by Kauffmann et al. (2003). This suggests that the dominant source of ionizing flux in these galaxies is AGN-related emission and not radiation from young stellar populations.

Zoom In Zoom Out Reset image size

The single Cl1604 AGN found to lie in the star-forming region of our pseudo-BPT diagram is hosted by a spiral galaxy with clear signs of ongoing star formation within its disk. This galaxy has the latest morphological type among the Cl1604 AGN hosts. An ACS-F814W image of the galaxy can be seen in Figure 4 of Koc09b (see source 5). The galaxy also appears to be in the early stages of an ongoing merger, as it has a nearby companion that is spectroscopically confirmed to be at the same redshift (Koc09b). Given these properties, it is not surprising that our emission-line diagnostic indicates the presence of ongoing star formation in the system.

To further investigate whether the [O ii] emission lines observed in these galaxies could originate from star formation processes, we compared the extinction-corrected [O ii] and Hα line luminosities measured in each galaxy. The L([O ii])/L(Hα) (hereafter $L_{[{\rm O\,\mathsc{ii}}]/{\rm H}\alpha }$) ratio, plotted against $F_{[{\rm N\,\mathsc{ii}}]/{\rm H}\alpha }$, is shown in Figure 4. Kewley et al. (2004) previously demonstrated that the maximum $L_{[{\rm O\,\mathsc{ii}}]/{\rm H}\alpha }$ ratio that can be achieved by star-forming regions is roughly 2.1. We find that four of the six Cl1604 AGN hosts have $L_{[{\rm O\,\mathsc{ii}}]/{\rm H}\alpha }$ ratios that exceed this limit. On average, these galaxies have over twice the [O ii] line luminosity that could be generated by star formation processes alone given their observed Hα line luminosity. The two galaxies that fall below the Kewley et al. maximum are sources X0, the star-forming spiral host previously discussed, and source X3. This latter source is the only Cl1604 AGN in which the highly ionized [Ne v] emission line is visible. The strength of the [O ii] line is predicted and observed to decrease in high-ionization AGNs (Ho 2005), which likely explains the lack of [O ii] contamination in this galaxy.

Zoom In Zoom Out Reset image size

To determine the degree to which current star formation activity would be overestimated in the galaxies showing excess [O ii] emission, we calculated the star formation rate (SFR) of each host using the relations of Kennicutt (1998) and the observed [O ii] and Hα line luminosities. On the right axis of Figure 4, we compare the SFR derived from the extinction-corrected Hα line to that of the extinction-corrected [O ii] line using a constant E(B − V) = 0.3 (adopted from Lemaux et al. 2010). In the four AGN hosts where [O ii] is elevated relative to Hα, the SFR would, on average, be overestimated by a factor of five. We find that varying the assumed extinction over a reasonable range of E(B − V) values (i.e., 0.1–0.55) does not significantly change this result. Even if we make no correction for extinction, which attenuates [O ii] more than Hα, the [O ii]-derived SFRs in these galaxies would be over twice the rate derived from Hα.

Our finding that the [O ii] line can be significantly contaminated bolster the results of Koc09b, who reported that a large fraction of AGNs in the Cl1604 supercluster appear to be associated with recently quenched hosts. A major caveat in that study was the assumption that the observed [O ii] emission in the spectra of these galaxies was due to AGN activity and not ongoing star formation. We have now demonstrated that this assumption is supported by our new observations. This finding provides significant insight into the nature of these systems. In the absence of substantial star formation, the properties of these galaxies generally agree with AGN feedback scenarios in which AGN-driven outflows rapidly suppress the star formation activity of their hosts. For example, we find both a high incidence of Balmer lines and a large A/K ratio⁶ in the composite spectra of the Cl1604 hosts, which suggests that these systems are either in a post-starburst phase or have recently had their star formation rapidly truncated. Furthermore, these galaxies have optical colors that place them between the red sequence of early-type galaxies and the blue cloud of star-forming objects, while their morphologies show tell-tale signs of interactions in the recent past, including fading tidal features and multiple nuclei. In most regards, these galaxies resemble a transition population within the Cl1604 structure that has been caught in the process of transforming from actively star-forming galaxies to passively evolving systems. The fact that a large fraction of AGNs within the Cl1604 supercluster show such signatures, coupled with recent findings that high-redshift AGNs are more common in denser environments than they are in the field (Gilli et al. 2003; Cappelluti et al. 2005; Eastman et al. 2007; Kocevski et al. 2009a), suggests that AGN-related feedback may play an important role in accelerating galaxy evolution in large-scale structures at high redshift.

We have used emission-line ratios measured from DEIMOS and NIRSPEC spectroscopy to determine the origin of the [O ii] emission line observed in six AGN hosts at z ∼ 0.9. These six galaxies are a subsample of nine AGNs detected in the Cl1604 supercluster that exhibit strong Balmer absorption lines in their spectra. If not for the presence of the [O ii] line, many of these galaxies would be classified as post-starburst or post-quenched systems.

Examining the flux ratio of the [N ii] to Hα lines, we find that in five of the six AGN hosts the dominant source of ionizing flux is AGN continuum emission and not radiation from young stellar populations. Furthermore, we find that four of the six galaxies have, on average, over twice the [O ii] line luminosity that could be generated by star formation processes alone given their Hα line luminosities. This strongly suggests that AGN-excited narrow-line emission is contaminating the [O ii] line flux. A comparison of SFRs calculated from the [O ii] and Hα line luminosities indicates that the former yields a five-fold overestimate of the star formation activity of these galaxies.

Our findings reveal the [O ii] line to be a poor indicator of current star formation activity in a majority of these moderate-luminosity Seyferts; a result that agrees with the conclusions of Yan et al. (2006) and Lem10. This has important ramifications on the study of post-starburst and recently quenched AGN hosts at high redshifts. Since post-starburst systems are selected based on the absence of current star formation, any studies that rely solely on the [O ii] line as a measure of current activity in AGN hosts will severely underestimate the fraction of AGNs found in post-starburst galaxies. Furthermore, our results confirm the increased connection between AGN activity and post-starburst/post-quenched systems at high redshifts reported by Koc09b. Although we do not observe direct evidence for AGN-driven quenching, the relatively high fraction of AGN hosts in the Cl1604 supercluster that shows signs of recently truncated star formation activity may suggest that AGN feedback plays an increasingly important role in suppressing ongoing activity and accelerating galaxy evolution in overdense environments at high redshift.