A Triple AGN in a Mid-infrared Selected Late-stage Galaxy Merger

Observational campaigns and theoretical studies have shown both that supermassive black holes (SMBHs) reside at the centers of most galaxies and that galaxy interactions are ubiquitous in the universe. As a result, galaxies grow and evolve hierarchically through collisions (e.g., Toomre & Toomre 1972; Schweizer 1982, 1996; Barnes & Hernquist 1992; Hibbard & van Gorkom 1996; Rothberg & Joseph 2004), during which gravitational torques drive gas reservoirs toward the centers of each galaxy, potentially fueling the SMBHs at their centers (Barnes & Hernquist 1992; Mihos & Hernquist 1996; Hopkins et al. 2008; Blecha et al. 2018). Recent simulations predict that late-stage galaxy mergers—with nuclear pair separations ≲10 kpc—facilitate the most rapid black hole growth and represent the merger stage during which both SMBHs are expected to begin accreting as active galactic nuclei (AGNs; Hopkins et al. 2008; Blecha et al. 2018). Such dual AGNs are predicted to be highly obscured by gas and dust, consequently exhibiting red mid-infrared colors (Blecha et al. 2018). Indeed, observational studies have demonstrated that late-stage galaxy mergers (pair separations <10 kpc) and post-mergers host higher fractions of dust obscured AGNs than do isolated galaxies in rigorously matched control samples, and in fact mid-infrared selection identifies a larger quantity of obscured AGNs in mergers than traditional optical selection techniques (Satyapal et al. 2014; Weston et al. 2017; Ellison et al. 2019). Despite their important role in black hole and galaxy evolution, however, dual AGNs are rare systems, and less than 30 have been robustly verified in the literature (see Table 8 in Satyapal et al. 2017 and references therein). The rarity of dual AGNs emphasizes the need for efficient selection methods when conducting systematic searches for dual AGNs.

In our previous studies of dual AGNs, we preselected the late-stage galaxy mergers (pair separations < 10 kpc) based upon their mid-infrared Wide-field Infrared Survey Explorer (WISE) colors (Satyapal et al. 2017; Pfeifle et al. 2019), selecting systems that exhibit a WISE $W1[3.4\,\mu {\rm{m}}]-W2[4.6\,\mu {\rm{m}}]\gt 0.5$ color, a color cut that has been demonstrated in simulations to be the most effective at identifying AGNs in late-stage mergers (Blecha et al. 2018). We obtained follow-up high spatial resolution X-ray observations of these mergers with the Chandra X-ray Observatory (Chandra) and longslit near-infrared spectroscopic observations with the Large Binocular Telescope (LBT) to look for dual nuclear X-ray sources and high ionization coronal emission lines. Not only did a majority of the mergers host dual AGNs or dual AGN candidates (Ellison et al. 2017; Satyapal et al. 2017; Pfeifle et al. 2019), but one system exhibited three nuclear X-ray sources and—upon closer inspection—was realized as a clear case of a triple merger: SDSS J084905.51+111447.2 (henceforth SDSS J0849+1114).¹⁰ This study demonstrated the effectiveness of using mid-infrared colors as a preselection strategy for finding AGNs in advanced mergers.

Dual AGNs in advanced mergers represent the most observationally accessible progenitors of the SMBH binary phase, which likely produces the main source of gravitational waves (Wyithe & Loeb 2003; Sesana et al. 2004; Kelley et al. 2017; Mingarelli et al. 2017) detectable by Pulsar Timing Array campaigns (Verbiest et al. 2016) and future spaced-based observations from the Laser Interferometer Space Antenna (Amaro-Seoane et al. 2017). However, SMBH binaries in realistic astrophysical environments can stall on parsec-scale orbits, resulting in merger timescales that exceed the age of the universe. An intruding third SMBH has been shown to significantly shorten coalescence timescales (Ryu et al. 2018) and can also result in slingshot ejections of one of the black holes from the host, resulting in either ejected or wandering SMBHs (e.g., Hoffman & Loeb 2007; Bonetti et al. 2018, 2019). Triple mergers are therefore of particular interest, because recent cosmological simulations predict that 16% of binary SMBHs resulting from major mergers will undergo an interaction with a third SMBH prior to coalescing (Kelley et al. 2017). Triple AGN-hosting mergers are exceedingly rare, with only a handful of triple AGN candidates in mergers reported in the literature (Liu et al. 2011; Koss et al. 2012; Kalfountzou et al. 2017) so far.^{11 ,12 ,13}

In this work we present a suite of multiwavelength observations providing convincing evidence for a triple AGN in the merging system, SDSS J0849+1114. SDSS J0849+1114 is a late-stage merger at a distance of ∼350 Mpc (z = 0.077) comprising three interacting galaxies, with nuclear pair separations of 23 (3.4 kpc, Galaxy 1 − Galaxy 2), 36 (5.3 kpc, Galaxy 1 − Galaxy 3), and 50 (7.3 kpc, Galaxy 2 − Galaxy 3) based upon archival (PI: X. Liu) Hubble Space Telescope (HST) imaging data (Figure 1). Morphologically, the system exhibits strong tidal features indicative of an advanced merger. With an integrated 8–1000 μm infrared luminosity of log(L_IR/L_⊙) = 11.43 ± 0.03, this merger falls within the class of luminous infrared galaxies. Based on the all-sky WISE, it displays a strong mid-infrared continuum with colors ($W1[3.4\,\mu {\rm{m}}]\,-W2[4.6\,\mu {\rm{m}}]=1.69$) often associated with powerful AGN (Stern et al. 2012), and was therefore included in our previous dual AGN sample. As reported in Pfeifle et al. (2019), SDSS J0849+1114 exhibited three nuclear X-ray sources with luminosities in excess of that expected by star formation contributions, and a high ionization [Si vi] coronal emission line was identified in the near-infrared spectrum of Galaxy 1. Furthermore, the system exhibited signs of high absorption along the line of sight based on its X-ray spectral properties.

Zoom In Zoom Out Reset image size

Interestingly, this object was also selected independently in a catalog of optically selected AGN pairs in Liu et al. (2011), and was identified as a triple AGN candidate. We combine the archival Chandra data from this program with our own in this analysis.

In this manuscript, we present new NuSTAR hard X-ray observations and X-ray spectral fitting results, new LBT optical spectroscopic observations along with archival SDSS optical spectra, we refine the source statistics using the combined Chandra data sets, and we show for the first time the near-IR spectra for all three galaxy nuclei. The manuscript is laid out as follows: In Section 2 we provide details on the observations and data reduction steps. In Section 3 we discuss the results obtained from the X-ray, near-IR, and optical spectroscopic observations. In Section 4 we explore alternative scenarios for the observed data and discuss this target in the broader context of our mid-infrared selection technique. In Section 5 we summarize our conclusions. Throughout this paper we adopt H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_M = 0.3, and Ω_Λ = 0.7.

During our original dual AGN campaign, we obtained Chandra X-ray observations and near-infrared spectroscopy from the LBT of SDSS J0849+1114. As part of a follow-up effort we also obtained hard X-ray NuSTAR observations and longslit optical spectroscopic observations from the LBT.¹⁴ During this analysis, we also retrieved the archival HST and Chandra observations. The Chandra data were reduced using the same procedure outlined in-depth in Pfeifle et al. (2019). Here we describe all other observations.

2.1. NuSTAR Imaging Observations

2.2. Chandra and NuSTAR Spectral Analysis

2.3. LBT Near-infrared Spectroscopic Observations

2.4. LBT Optical Spectroscopic Observations

2.5. LBT Optical Spectral Fitting to Determine Fluxes and Stellar Velocity Dispersion

In order to measure accurate line fluxes, it is necessary to account for stellar absorption, which primarily affects the Balmer lines. The Penalized Pixel-Fitting method (pPXF; Cappellari 2017) is an algorithm developed to extract stellar kinematics using a maximum penalized likelihood approach to match stellar spectral templates to absorption features in galaxy spectra. A large (>150) number of stellar templates of various spectral type were chosen from the Indo-US Library of Coudé Feed Stellar Spectra (Valdes et al. 2004) in order to decrease the possibility of template mismatch. To ensure that the Balmer absorption is adequately constrained, we model the stellar population using the full spectrum (4000–7500 Å) with pPXF. After subtracting the best-fitting stellar population model (red, see Figure 2)¹⁵ from the continuum, the residual emission lines were fit using Gaussian models. Broadened wing components, which can be indicative of high-velocity gas outflows, are present in narrow lines of all three objects, and thus an additional Gaussian component is added (see Figure 2 for examples of emission line decompositions). The presence of Fe emission can also affect measurements of the stellar population and [O iii]λλ4959,5007 doublet, and therefore we include the Fe ii template from Véron-Cetty et al. (2004) and Fe i lines from the NIST Atomic Spectra Database in our fits. Fitting was performed using a custom maximum-likelihood routine implemented in Python (Sexton et al. 2019), which uses the affine-invariant MCMC ensemble sampler emcee (Foreman-Mackey et al. 2013). Line fluxes are determined from the best-fit Gaussian models of the emission lines. Stellar-velocity dispersions were fit using a similar method described, except we fit the region from 4400 to 5800 Å, which includes the Mg ib region used to estimate stellar velocity dispersion. The [O iii] doublet is best fit by a combination of two Gaussian components. The line widths from the [O iii] doublet fit are then used to constrain the widths of the rest of the emission lines (Rodríguez Zaurín et al. 2013), so careful modeling of their profiles is necessary. The fitting routine can actually overestimate the flux and width of the broad component of the[O iii]λ5007 emission line due to contaminating flux from the Fe i λ4985, λ4986, λ5016 emission lines. To address this contamination, we model the Fe lines with independent amplitudes and with widths equal to that of the narrow component of [O iii] (Manzano-King et al. 2019).

Zoom In Zoom Out Reset image size

3.1. Chandra/ACIS-S Imaging Results

We combined our 2016 observation with an archival 2013 observation (PI: X. Liu), and assessed source significance using the binomial no-source probability statistic (P_B) (Lansbury et al. 2014), which is suitable for the low-count regime; we point the reader to Pfeifle et al. (2019) for further discussion. The Chandra source statistics are reported in Table 1. The Galaxy 1, 2, and 3 sources were detected with a statistical significance of 13.8σ [log(P_B) = −371.6], 3.5σ [log(P_B) = −17.3], and 2.9σ [log(P_B) = −12.0], and based upon the derived P_B values it is highly unlikely that these sources are the result of spurious background activity.¹⁶ We used the ciao modelflux package to uniformly estimate the hard 2–10 keV (rest frame) luminosities for the sources, assuming Γ = 1.9 and correcting for only Galactic ${N}_{{\rm{H}}}$ along the line of sight. Galaxies 1, 2, and 3 exhibit 2–10 keV luminosities of 2.37 ± 0.17 × 10⁴¹ erg s⁻¹, ${2.4}_{-0.7}^{+0.8}\,\times {10}^{40}$ erg s⁻¹, ${1.5}_{-0.5}^{+0.6}\times {10}^{40}$ erg s⁻¹ respectively.

Table 1.  Chandra X-Ray Source Statistics

Galaxy	${N}_{{\rm{H}}}$	αχ	δχ	Counts	Counts	Counts	HR	σ	log(PB)
	(1020 cm−2)			0.3–8 keV	0.3–2 keV	2–8 keV
1	3.80	8h49m05529	+11deg14'47876	196 ± 14	112 ± 11	84 ± 9	−0.14	13.8σ	−371.6
2	3.80	8h49m05381	+11deg14'45747	${15}_{-4}^{+5}$	${13}_{-2}^{+4}$	${2}_{-1}^{+3}$	−0.76	3.5σ	−17.3
3	3.80	8h49m05448	+11deg14'51646	${13}_{-4}^{+5}$	${12}_{-4}^{+5}$	${1}_{-1}^{+2}$	−0.84	2.9σ	−12.0

Note. Column 1: X-ray source and galaxy nucleus designation. Column 2: Galactic hydrogen column density along the line of sight. Column 3–4: right ascension (R.A.) and declination (decl.) of the source. Column 5–7: X-ray counts detected in the full, soft, and hard X-ray bands for the combined 2013 and 2016 Chandra observations. Note that these are not background-subtracted counts, but formal statistical significances of the sources are derived from background-subtracted statistics. Values are rounded to closest integer. Column 8: hardness ratio, given by the formula (H − S)/(H + S), where H and S refer to the Chandra hard and soft energy bands. Column 9: formal source statistical significance, derived from background-subtracted source statistics. Column 10: logarithm of the binomial source probability, which is a metric for source statistical significance more appropriate for the low-count regime. Note: sources which meet a threshold of $\mathrm{log}({P}_{B})\lt -2.7$ are considered to be true sources, and it is highly unlikely that these are the result of spurious background activity (Lansbury et al. 2014); we used this statistical metric for our dual sample in both Satyapal et al. (2017) and Pfeifle et al. (2019).

Download table as:  ASCIITypeset image

3.2. Chandra/ACIS-S and NuSTAR Spectral Analysis Results

Simultaneously fitting the Chandra and NuSTAR data for Galaxy 1 (see Figure 3), we find an absorbing column of log(${N}_{{\rm{H}}}$/cm⁻²) $=\,{23.9}_{-0.2}^{+0.2}$, photon index ${\rm{\Gamma }}={1.4}_{\mathrm{NC}}^{+0.3}$, and a scattered fraction ${f}_{S}={11.7}_{-6.4}^{+22.1}$%.¹⁷ Correcting for the intrinsic absorption, we find an unabsorbed 2–10 keV luminosity of ${4.6}_{-0.6}^{+0.6}\,\times {10}^{42}$ erg s⁻¹. We also report the NuSTAR source statistics in Table 2. To measure the equivalent width of the Fe Kα line, we repeated the fitting process with the phenomenological model from Pfeifle et al. (2019), which includes a Gaussian emission line component to account for the iron line fluorescent emission. Due to the signal-to-noise ratio of our X-ray data, we conservatively quote here only an upper limit for the Fe Kα emission. We find an equivalent width upper limit of ∼0.6 keV for the Fe Kα line, which is consistent with the high absorbing column along the line of sight (Brightman & Nandra 2011).

Zoom In Zoom Out Reset image size

Table 2.  NuSTAR X-Ray Source Statistics

Camera	${N}_{{\rm{H}}}$	αχ	δχ	Counts	Counts	Counts
				3–78 keV	3–8 keV	8–24 keV
FPMA	3.80	8h49m05510	+11deg14'47260	141 ± 12	20 ± 4	63 ± 8
FPMB	3.80	8h49m05510	+11deg14'47260	149 ± 12	49 ± 7	59 ± 8

Note. Column 1: NuSTAR camera. Column 2: Galactic hydrogen column density along the line of sight. Column 3–4: R.A. and decl. coordinates of the source apertures. Column 5: X-ray counts in the 3–78 keV NuSTAR band. Column 6: X-ray counts in the 3–8 keV NuSTAR band. Column 7: X-ray counts in the 8–24 keV NuSTAR band.

Download table as:  ASCIITypeset image

While we do find different values for the spectral parameters here than those reported in Pfeifle et al. (2019), this is not entirely surprising; in this manuscript we are fitting both sets of Chandra and NuSTAR observations simultaneously, and it is well known that NuSTAR observations aid significantly in constraining the values for spectral parameters such as the photon index, ${N}_{{\rm{H}}}$, and the equivalent width of the Fe Kα line (Marchesi et al. 2018), in some cases leading to refined values that are significantly different from those found with data in the 2–10 keV energy range.

As we discussed in Section 2.2, it is not possible to constrain whether the two remaining AGNs are Compton-thick, and it is still possible they contribute appreciably to the observed hard X-ray flux because they would fall within the NuSTAR extraction aperture. In any case, this does not affect our X-ray spectral fitting results; the results are consistent with the picture of mergers hosting obscured AGNs, as found by previous studies (Kocevski et al. 2015; Ricci et al. 2017, 2016; Donley et al. 2018; Goulding et al. 2018; Koss et al. 2018; Lanzuisi et al. 2018).

3.3. LBT Spectroscopic Results

To determine the nature of the photoionizing sources observed in the LBT optical spectroscopy, we employ "Baldwin, Phillips, and Terlevich" (BPT) optical spectroscopic diagnostic diagrams, shown in Figure 4. The optical emission in each nucleus is consistent with AGN photoionization, complementing the X-ray and near-IR observations, and is consistent with the presence of a triple AGN in this merger. We list the logarithmic emission line ratios in Table 3.

Zoom In Zoom Out Reset image size

Following the procedure outlined in Section 2.5, we derived stellar and gas velocity dispersions for each nucleus, which we report in Table 3. Gas velocity dispersions were measured from the narrow component of the [O iii]λ5007 emission line while the stellar velocity dispersions were measured by fitting the Mg ib region in the spectral range 4400–5800 Å. We find very similar values for the stellar and gas velocity dispersions within each respective nucleus (see Columns 6–7 in Table 3), but these values differ between the three nuclei; these similarities further suggest that each of the three regions are more-or-less dynamically independent—that is, the gas and stellar components are well-coupled—especially for Galaxies 1 and 3.¹⁸

Table 3.  LBT and SDSS Spectroscopic Results

Galaxy	log([O iii]λ5006/Hβ)	log([N ii]/Hα)	log([S ii]λλ6717,6733/Hα)	log([O i]λ6302/Hα)	σ*	σ[O iii]
					(km s−1)	(km s−1)
1	1.029 ± 0.009	−0.209 ± 0.003	−0.39 ± 0.01	−0.733 ± 0.006	180 ± 9	191 ± 1
2	0.55 ± 0.09	−0.232 ± 0.03	−0.31 ± 0.07	−0.9 ± 0.2	149 ± 11	101 ± 4
3	0.4 ± 0.3	−0.23 ± 0.06	−0.5 ± 0.5	−0.96 ± 0.06	91 ± 26	91 ± 26

Note. Column 1: galaxy nucleus designation. Columns 2–5: logarithmic optical emission line ratios for each nucleus, calculated using the derived optical emission line fluxes. See Section 2.5 for more details on how these line fluxes were derived from the optical spectra. Column 6: Stellar velocity dispersion. Column 7: gas velocity dispersion. The large uncertainties in gas and velocity dispersions for Galaxy 3 are due to the lower signal-to-noise of SDSS spectra compared to LBT-obtained spectra.

Download table as:  ASCIITypeset image

We observe strong blue-wing components [O iii]λ5007 emission lines in Galaxy 1 and 2 with FWHMs on the order of ∼1300–1600 km s⁻¹, which is suggestive of strong galactic-scale outflows (Heckman et al. 1981; Nelson & Whittle 1996; Woo et al. 2016). Galaxy 3 shows a slightly redshifted wing in the [O iii]λ5007 emission, with an FWHM on the order of ∼700 km s⁻¹. Many of the lower ionization emission lines show broad-wing components as well.

The LBT near-IR spectra revealed high excitation [Si vi] (Pfeifle et al. 2019) coronal line emission (${f}_{\lambda }\,=1.63\pm 0.07\,\times {10}^{-15}\ \mathrm{erg}\ {\mathrm{cm}}^{-2}\ {{\rm{s}}}^{-1}$) in the nucleus of Galaxy 1, providing robust confirmation for the AGN in that nucleus (see Figure 5). Moreover, the near-infrared spectra of Galaxy 1 and Galaxy 3 reveal broad components (FWHM ≳2400 km s⁻¹) in the Paα emission (see Figure 6), significantly in excess of the blueshifted components observed in the [O iii]λ5007 emission. These detections offer additional evidence for the existence of black hole accretion in these galaxies. We point out that this accretion activity is likely missed in the optical due to large obscuring columns along the line of sight. An exhaustive analysis of the possible outflows observed in the optical spectra is, however, beyond the scope of this paper, but will be featured in a future publication focusing on the environmental impacts of AGNs in mergers.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

While our multiwavelength observations point to each nucleus hosting an AGN, we considered alternative scenarios that might explain the observed characteristics of this system.

In order to rule out star formation as an alternative explanation for the observed X-ray emission, we calculated star formation rates (SFRs) and possible X-ray contributions from a population of X-ray binaries using the Paα emission observed in the LBT near-IR spectra. We refer the reader to Satyapal et al. (2017) and Pfeifle et al. (2019) for a discussion of the SFR and XRB contribution calculations. With this calculation, we estimate an upper limit for the star formation activity in the merger nuclei; in reality, because the nuclei all exhibit BPT AGN-like optical spectroscopic line ratios, some fraction of the Paα emission is in fact attributable to the AGNs and therefore the true SFR, and hence stellar contribution to the X-rays, is much lower.

We obtained SFRs of 13.16 M_⊙ yr⁻¹, 0.48 M_⊙ yr⁻¹, and 1.79 M_⊙ yr⁻¹ and predicted hard 2–10 keV X-ray luminosities of 2.27 ± 0.4 × 10⁴⁰ erg s⁻¹, 0.12 ± 0.11 × 10⁴⁰ erg s⁻¹, and 0.43 ± 0.20 × 10⁴⁰ erg s⁻¹ for Galaxy 1, 2, and 3, respectively. Contrasting these values with the observed X-ray emission (uncorrected for intrinsic absorption) in Galaxies 1, 2, and 3 (see Section 3.1), there is a clear order of magnitude disparity between the observed and predicted values, strongly suggestive of three AGNs. Once again we note that in reality—due to the expected contributions from the AGNs to the Paα flux—we are likely overestimating the X-ray contributions from SF. Moreover, we are underestimating the X-ray fluxes of the three sources as the quoted luminosities are not corrected for absorption; as indicated by our joint NuSTAR–Chandra spectral fitting results, any absorption corrections made to the observed X-ray luminosities are expected to be large.

While we have confirmed the AGN in Galaxy 1 via the detection of a near-IR coronal line, Galaxies 2 and 3 require closer examination. If the SFR within Galaxies 2 and 3 were indeed ∼30 M_⊙ yr⁻¹ and ∼9 M_⊙ yr⁻¹, respectively—a requirement to explain the observed X-ray emission—the optical spectra should be dominated by a very young (<10 Myr) and blue stellar population, when the population of high mass X-ray binaries (HMXBs) is expected to be high (von der Linden et al. 2010). The most luminous source of X-ray emission in galaxies arises from HMXBs.¹⁹ Our near-infrared spectra allow us to also test this hypothesis by constraining the ages of the nuclear stellar populations. Specifically, the equivalent width of hydrogen recombination lines show a steep decline as the most massive stars evolve off the main sequence, causing a simultaneous decrease in the ionizing photon flux and an increase in the K-band continuum flux (Leitherer et al. 1999). However, the equivalent width of the Br–γ line (or upper limit in the case of Galaxy 2 and 3), a strong indicator of the starburst age (Leitherer et al. 1999, 2014), suggests that the stellar population in all three galaxies is greater than ∼6 million years (Figure 7), well after the time when the population of HMXBs is expected to drop dramatically (von der Linden et al. 2010).

Zoom In Zoom Out Reset image size

We also considered the possibility that the optical emission lines are produced by ionizing shocks powered by starburst driven winds or tidal flows (Veilleux et al. 2002; Colina et al. 2005). Shocks are known to arise in mergers and become an increasingly important component of the optical emission lines as a merger progresses (Rich et al. 2011), generating [O iii]/Hβ and [N ii]/Hα ratios consistent with AGN when in fact only star formation is present (Allen et al. 2008). However, shock models have shown that when there is a significant contribution from shocks to the ionization, it often results in enhanced [O i]/Hα and [S ii]/Hα emission line ratios compared to AGN photoionization models (Dopita & Sutherland 1995; Allen et al. 2008), which is not seen in any of the three nuclei. Furthermore, detection of three X-ray point sources coincident with the nuclei supports an AGN origin rather than the result of shock-heated gas.

Finally, we explored the possibility that there are fewer than three distinct sources of ionization (one located at each galactic nucleus). In particular, given the [O iii] luminosities of each nucleus, the typical narrow line region size is expected to be comparable to the nuclear pair separations between Galaxy 1 and Galaxy 2 (Bennert et al. 2006; Hainline et al. 2013, 2014), raising the possibility that only one AGN is ionizing the gas at both locations. However, the detection of three separate luminous X-ray sources at the location of each nucleus makes this scenario unlikely. Furthermore, the velocity offset of the [O iii] lines between the two nuclei is consistent with the velocity offset measured for the Hβ lines, suggesting a lack of ionization stratification (Zamanov et al. 2002) expected for a single AGN, and therefore also favoring a scenario in which there are two independent sources of ionization. Moreover, each galaxy has unique, well-coupled gas and stellar velocity dispersions, suggesting three kinematically distinct sources consistent with galaxy nuclei.

We note that SDSS J0849+1114 was targeted for multiwavelength follow-up as a result of mid-infrared color preselection (Satyapal et al. 2014, 2017; Pfeifle et al. 2019). While optical preselection strategies (Liu et al. 2011; Comerford et al. 2015) are extensively used in campaigns to identify dual AGNs, these strategies have yielded very few confirmed cases. In fact, very few late-stage galaxy mergers are known to host dual AGNs (see Table 8 from Satyapal et al. 2017), most of which have been discovered serendipitously or through hard X-ray surveys (Koss et al. 2012). While there have been a couple of triplet quasars reported in the literature (Djorgovski et al. 2007; Farina et al. 2013), they are at much larger separations and it is unclear if they are hosted in mergers. A few candidate triple AGNs in mergers have been reported (Liu et al. 2011; Koss et al. 2012; Kalfountzou et al. 2017) but thus far there have been no cases exhibiting triple X-ray point sources and optical narrow line signatures consistent with AGNs. Emission from the nuclei of advanced mergers at these pair separations are expected to be highly obscured (e.g., Kocevski et al. 2015; Ricci et al. 2017; Blecha et al. 2018), making it difficult to identify dual AGNs using traditional techniques. While it is true that SDSS J0849+1114 exhibits optical signatures of AGNs in each nucleus, it is unlikely if all triple AGNs will exhibit such optical spectroscopic signatures. In contrast, mid-infrared preselection has proven an effective strategy in finding dual AGNs and now a triple AGN candidate in follow-up multiwavelength studies (Ellison et al. 2017; Satyapal et al. 2017; Pfeifle et al. 2019), suggesting that similar cases to SDSS J0849+1114 can be found in future studies.

In this investigation, we employed a suite of multiwavelength diagnostics which yield a compelling case for the existence of an AGN triplet in the triple galaxy merger SDSS J0849+1114. This merger, preselected in our dual AGN campaign for follow-up observations based on its WISE mid-infrared colors, was only later realized as a triple AGN candidate (Pfeifle et al. 2019). The nature of SDSS J0849+1114 can be summarized as follows:

• 1.  
  Chandra revealed three nuclear X-ray point sources which exhibit luminosities in excess of any expected contributions from stellar activity.
• 2.  
  High excitation [Si vi] coronal line emission was detected in the nucleus of Galaxy 1 via the near-infrared longslit spectra obtained with the LBT. We also find evidence for broad components in the near-IR Paα emission of Galaxy 1 and 3, with significantly greater widths than those observed in the broad components of the optical [O iii]λ5007 emission.
• 3.  
  New longslit optical spectroscopic measurements shown for the first time here for Galaxy 1 and Galaxy 2, along with the archival SDSS spectrum for Galaxy 3, reveal optical spectroscopic emission line ratios consistent with AGN photoionization in all three nuclei.
• 4.  
  Simultaneous spectral fitting of the Chandra and new NuSTAR observations confirms at least one of the nuclei (Galaxy 1) hosts a heavily absorbed AGN and we further refine our previously reported spectral parameters: log(${N}_{{\rm{H}}}$/cm⁻²) $=\,{23.9}_{-0.2}^{+0.2}$, photon index ${\rm{\Gamma }}={1.4}_{\mathrm{NC}}^{+0.3}$, scattered fraction ${f}_{S}={11.7}_{-6.4}^{+22.1} \%$, and Fe Kα equivalent width upper limit of ∼0.6 keV. We find an unabsorbed 2–10 keV luminosity of ${4.6}_{-0.6}^{+0.6}\times {10}^{42}$ erg s⁻¹.
• 5.  
  Based upon our multiwavelength diagnostics, alternative scenarios—such as star formation activity, shock-driven emission, or photoionization by fewer than three AGNs—cannot effectively explain our observations.

Follow-up radio observations of SDSS J0849+1114 obtained with Very Long Baseline Interferometry (VLBI) are presented in Gabányi et al. (2019).

The presence of triple AGNs in hierarchical galaxy mergers is a natural consequence of ΛCDM predictions. Since late-stage mergers are predicted to be highly obscured (Blecha et al. 2018; Hopkins et al. 2008), the apparent dearth of dual and triple AGNs may in fact be due to the preselection strategy adopted to find them. With the serendipitous identification of one compelling case for a triple AGN originally selected using our WISE preselection strategy, the natural next step is to extend our study to identify further cases of triple AGNs using mid-infrared color selection. In a forthcoming investigation, we hope to examine a sample of WISE selected ($W1[3.4\,\mu {\rm{m}}]-W2[4.6\,\mu {\rm{m}}]\gt 0.5$) triple mergers in a systematic search for triple AGNs using an arsenal of optical, near-infrared, and X-ray diagnostics.

The authors thank the anonymous referee for a thoughtful review and constructive criticism of the manuscript. R.W.P. and S.S. gratefully acknowledge support from the Chandra Guest Investigator Program under NASA Grants GO6-17096X and GO7-18099X. R.W.P thanks the Chandra Help Desk for support during this project. R.W.P. thanks P. Boorman for the helpful discussions and assistance in troubleshooting the NuSTAR pipeline while R.W.P. became acquainted with the software. C.M.-K., R.O.S., and G.C. acknowledge support from the National Science Foundation, under grant number AST 1817233. C.R. acknowledges support from the CONICYT+PAI Convocatoria Nacional subvencion a instalacion en la academia convocatoria año 2017 PAI77170080. L.B. acknowledges support from the National Science Foundation under grant number AST-1715413.

This research made use of APLpy, an open-source plotting package for Python (Robitaille & Bressert 2012), as well as the Ned Wright Cosmology Calculator (Wright 2006). We also gratefully acknowledge the use of the software TOPCAT (Taylor 2005) and Astropy (The Astropy Collaboration et al. 2018).

The LBT is an international collaboration among institutions in the United States, Italy, and Germany. LBT Corporation partners are The University of Arizona on behalf of the Arizona Board of Regents; Istituto Nazionale di Astrofisica, Italy; LBT Beteiligungsgesellschaft, Germany, representing the Max-Planck Society, The Leibniz Institute for Astrophysics Potsdam, and Heidelberg University; The Ohio State University, and The Research Corporation, on behalf of The University of Notre Dame, University of Minnesota and University of Virginia.

This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration. Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Office of Science. The SDSS-III website is http://www.sdss3.org/.

SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration including the University of Arizona, the Brazilian Participation Group, Brookhaven National Laboratory, Carnegie Mellon University, University of Florida, the French Participation Group, the German Participation Group, Harvard University, the Instituto de Astrofisica de Canarias, the Michigan State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley National Laboratory, Max Planck Institute for Astrophysics, Max Planck Institute for Extraterrestrial Physics, New Mexico State University, New York University, Ohio State University, Pennsylvania State University, University of Portsmouth, Princeton University, the Spanish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

In this section we show the full dynamic range of the optical spectra for the three galaxy nuclei within SDSS J0849+1114.

Zoom In Zoom Out Reset image size