Active Galactic Nucleus Pairs from the Sloan Digital Sky Survey. III. Chandra X-ray Observations Unveil Obscured Double Nuclei

TL;DR

This study uses Chandra X-ray observations to confirm and analyze dual active galactic nuclei in merging galaxies, revealing obscured nuclei and differences in luminosity ratios compared to single AGNs, thus advancing understanding of galaxy mergers.

Contribution

First X-ray confirmation of obscured dual AGNs in a systematically selected sample from SDSS, providing insights into their properties and merger-induced gas inflows.

Findings

Confirmed two dual AGNs with X-ray observations.

Observed lower X-ray-to-[O III] luminosity ratios in dual AGNs.

Identified high nuclear gas columns likely from merger inflows.

Abstract

We present Chandra ACIS-S X-ray imaging spectroscopy for five dual active galactic nucleus (AGN) candidates. Our targets were drawn from a sample of 1286 [O III]-selected AGN pairs systematically selected from the Sloan Digital Sky Survey Seventh Data Release. Each of the targets contains two nuclei separated by $\sim$3-9 kpc in projection, both of which are optically classified as Type 2 (obscured) AGNs based on diagnostic ratios of the narrow emission lines. Combined with independent, empirical star formation rate estimates based on the host-galaxy stellar continua, the new Chandra X-ray observations allow us to evaluate the dual-AGN hypothesis for each merging system. We confirm two (SDSS J0907+5203 and SDSS J1544+0446) of the five targets as bona-fide dual AGNs. For the other three targets, the existing data are consistent with the dual-AGN scenario, but we cannot rule out the possibility of stellar/shock heating and/or one AGN ionizing both gaseous components in the merger. The average X-ray-to-[O III] luminosity ratio in our targets seems to be systematically smaller than that observed in single AGNs but is higher than that seen in dual AGNs selected from AGNs with double-peaked narrow emission lines. We suggest that the systematically smaller X-ray-to-[O III] luminosity ratio observed in dual AGNs than in single AGNs is due to a high nuclear gas column likely from strong merger-induced inflows. Unlike double-peaked-[O III]-selected dual AGNs, the new sample selected from resolved galaxy pairs are not subject to the orientation bias caused by the double-peak line-of-sight velocity splitting selection, which also contributes to lowering the X-ray-to-[O III] luminosity ratio.

Figures

Tables

Table 1. Table 1: [O  III ]-Selected Dual AGN Candidates with New Chandra ACIS-S X-ray Observations.

Full Name	Name	Redshift	$r_p$	Plate	Fiber ID	MJD	$m_r$	${\sigma }_{\ast }$	log$M_{\mathrm{BH}}$	log$M_{\ast }$
J090714.45+520343.4	J0907+5203a	0.0596	8.9	553	208	51999	16.27	177	7.9	10.6
J090714.61+520350.7	J0907+5203b	0.0602	8.9	552	603	51992	16.95	132	7.4	10.3
J080523.29+281815.8	J0805+2818a	0.1284	5.6	929	570	52581	15.99	316	9.0	11.2
J080523.40+281814.1	J0805+2818b	0.1286	5.6	930	285	52618	18.18	169	7.8	9.8
J133031.75$-$003611.9	J1330$-$0036a	0.0542	4.4	298	264	51955	18.36	72	6.2	8.8
J133032.00$-$003613.5	J1330$-$0036b	0.0542	4.4	297	61	51959	15.17	153	7.6	10.7
J105842.44+314457.6	J1058+3144a	0.0728	4.1	1990	342	53472	17.27	101	6.9	10.0
J105842.58+314459.8	J1058+3144b	0.0723	4.1	2026	17	53711	15.79	159	7.7	10.9
J154403.45+044607.5	J1544+0446a	0.0420	3.4	2950	278	54559	17.82	153	7.6	9.8
J154403.67+044610.1	J1544+0446b	0.0416	3.4	1836	586	54567	14.21	201	8.1	11.1

Table 2. Table 2: Optical Emission-line Properties of the Five [O  III ]-Selected Dual AGN Candidates.

Name	$H\beta$	$H\alpha$	$\left[O\mathrm{III}\right]$	$\left[N\mathrm{II}\right]$	$\left[S\mathrm{II}\right]$	$\left[OI\right]$	$\frac{\left[O\mathrm{III}\right]}{H\beta }$	$\frac{\left[N\mathrm{II}\right]}{H\alpha }$	$\frac{\left[S\mathrm{II}\right]}{H\alpha }$	$\frac{\left[OI\right]}{H\alpha }$
J0907+5203a	$98.8\pm 70.9$	$361.6\pm 81.5$	$632.4\pm 141.8$	$213.0\pm 81.5$	$172.8\pm 80.2$	$69.1\pm 67.4$	$6.4\pm 4.8$	$0.6\pm 0.3$	$0.5\pm 0.2$	$0.2\pm 0.2$
J0907+5203b	$130.9\pm 29.9$	$581.7\pm 40.6$	$492.3\pm 49.1$	$348.1\pm 36.3$	$296.2\pm 33.2$	$98.0\pm 25.6$	$3.8\pm 0.9$	$0.6\pm 0.1$	$0.5\pm 0.1$	$0.2\pm 0.0$
J0805+2818a	$390.0\pm 102.2$	$1578.6\pm 180.2$	$3808.5\pm 266.3$	$1023.7\pm 161.4$	$554.2\pm 135.7$	$249.4\pm 72.6$	$9.8\pm 2.6$	$0.6\pm 0.1$	$0.4\pm 0.1$	$0.2\pm 0.0$
J0805+2818b	$52.7\pm 43.0$	$150.4\pm 40.2$	$360.1\pm 103.3$	$201.1\pm 60.3$	$74.6\pm 70.1$	$29.3\pm 40.2$	$6.8\pm 5.9$	$1.3\pm 0.5$	$0.5\pm 0.5$	$0.2\pm 0.3$
J1330$-$0036a	$111.6\pm 8.2$	$398.8\pm 49.3$	$264.2\pm 12.2$	$182.0\pm 10.2$	$146.1\pm 56.9$	$28.5\pm 46.9$	$2.4\pm 0.2$	$0.5\pm 0.1$	$0.4\pm 0.1$	$0.1\pm 0.1$
J1330$-$0036b	$367.6\pm 144.1$	$1568.2\pm 227.7$	$1120.1\pm 251.0$	$1426.4\pm 218.5$	$539.2\pm 151.3$	$110.7\pm 106.9$	$3.0\pm 1.4$	$0.9\pm 0.2$	$0.3\pm 0.1$	$0.1\pm 0.1$
J1058+3144a	$129.7\pm 15.3$	$377.3\pm 50.1$	$228.2\pm 30.0$	$239.2\pm 20.8$	$132.9\pm 51.0$	$25.9\pm 38.8$	$1.8\pm 0.3$	$0.6\pm 0.1$	$0.4\pm 0.1$	$0.1\pm 0.1$
J1058+3144b	$238.7\pm 83.8$	$921.0\pm 127.2$	$759.7\pm 145.9$	$762.2\pm 117.9$	$316.4\pm 85.6$	$64.2\pm 62.1$	$3.2\pm 1.3$	$0.8\pm 0.2$	$0.3\pm 0.1$	$0.1\pm 0.1$
J1544+0446a	$46.1\pm 38.5$	$209.1\pm 40.6$	$153.7\pm 57.7$	$613.8\pm 68.4$	$238.2\pm 54.4$	$97.9\pm 44.9$	$3.3\pm 3.1$	$2.9\pm 0.7$	$1.1\pm 0.3$	$0.5\pm 0.2$
J1544+0446b	$165.8\pm 62.6$	$749.4\pm 122.0$	$285.9\pm 85.7$	$643.3\pm 105.5$	$364.2\pm 86.3$	$105.7\pm 65.9$	$1.7\pm 0.8$	$0.9\pm 0.2$	$0.5\pm 0.1$	$0.1\pm 0.1$

Table 3. Table 3: Chandra  Observations of the Five [O  III ]-Selected Dual AGN Targets.

Name	ObsID	Exp.	Counts	CR	CR1	CR2	HR	$\Gamma$	$N_H$
J0907+5203a	14965	14.5	33.9	$2.76\pm 0.44$	$0.33\pm 0.15$	$2.43\pm 0.42$	${0.76}_{-0.10}^{+0.12}$	-	-
J0907+5203b*	14965	14.5	108.9	$7.63\pm 0.73$	$1.57\pm 0.32$	$6.06\pm 0.66$	${0.56}_{-0.07}^{+0.08}$	${1.43}_{-0.75}^{+0.84}$	${2.6}_{-1.1}^{+1.5}$
J0805+2818a	14964	14.0	28.9	$2.04\pm 0.38$	$1.16\pm 0.28$	$0.89\pm 0.26$	$-{0.17}_{-0.19}^{+0.18}$	-	-
J0805+2818b	14964	14.0	12.4	0.87	0.88	0.45	-	-	-
J1330$-$0036a	14967	14.9	8.4	0.57	0.43	0.59	-	-	-
J1330$-$0036b	14967	14.9	15.9	$1.05\pm 0.27$	$0.57\pm 0.19$	$0.48\pm 0.18$	$-{0.13}_{-0.28}^{+0.27}$	-	-
J1058+3144a	14966	14.5	12.1	0.81	0.72	0.57	-	-	-
J1058+3144b*	14966	14.5	79.8	$5.99\pm 0.66$	$0.26\pm 0.13$	$5.73\pm 0.64$	${0.91}_{-0.04}^{+0.05}$	${1.13}_{-1.50}^{+1.64}$	${6.4}_{-4.5}^{+6.3}$
J1544+0446a	14968	14.9	5.8	$0.38\pm 0.16$	$0.32\pm 0.14$	0.58	$-{0.62}_{-0.38}^{+0.09}$	-	-
J1544+0446b*	14968	14.9	49.9	$3.93\pm 0.52$	$0.45\pm 0.17$	$3.48\pm 0.49$	${0.76}_{-0.08}^{+0.10}$	${1.66}_{-1.35}^{+1.47}$	${3.9}_{-2.7}^{+3.2}$

Table 4. Table 4: X-ray Luminosity of the Five [O  III ]-Selected Dual AGN Targets.

Name	log$L_{X,0.5-8}$	log$L_{X,0.5-2}$	log$L_{X,2-10}$	log$L_{X,2-10,\mathrm{obs}}$	log$L_{\left[O\mathrm{III}\right],\mathrm{obs}}$	log$L_{\left[O\mathrm{III}\right],\mathrm{cor}}$	SFR	log$L_{0.5-2}^{\mathrm{SF}}$	log$L_{2-10}^{\mathrm{SF}}$
J0907+5203a	${41.67}_{-0.06}^{+0.08}$	${40.68}_{-0.16}^{+0.26}$	${41.75}_{-0.07}^{+0.08}$	${41.71}_{-0.07}^{+0.08}$	${40.76}_{-0.11}^{+0.09}$	${40.94}_{-0.11}^{+0.09}$	${0.59}_{-0.52}^{+1.61}$	${39.42}_{-0.94}^{+0.57}$	${39.47}_{-0.94}^{+0.57}$
J0907+5203b*	${42.35}_{-0.04}^{+0.04}$	${41.89}_{-0.08}^{+0.10}$	${42.27}_{-0.04}^{+0.05}$	${42.20}_{-0.04}^{+0.05}$	${40.66}_{-0.05}^{+0.04}$	${40.98}_{-0.05}^{+0.04}$	${1.28}_{-0.94}^{+2.95}$	${39.76}_{-0.58}^{+0.52}$	${39.81}_{-0.58}^{+0.52}$
J0805+2818a	${42.24}_{-0.07}^{+0.09}$	${41.92}_{-0.09}^{+0.12}$	${42.01}_{-0.11}^{+0.15}$	${41.97}_{-0.11}^{+0.15}$	${42.27}_{-0.03}^{+0.03}$	${42.53}_{-0.03}^{+0.03}$	${11.68}_{-6.95}^{+15.90}$	${40.72}_{-0.39}^{+0.37}$	${40.77}_{-0.39}^{+0.37}$
J0805+2818b					${41.25}_{-0.15}^{+0.11}$	${41.25}_{-0.15}^{+0.11}$	${2.24}_{-1.38}^{+3.31}$	${40.00}_{-0.41}^{+0.39}$	${40.05}_{-0.41}^{+0.39}$
J1330$-$0036a					${40.29}_{-0.02}^{+0.02}$	${40.45}_{-0.02}^{+0.02}$	${0.35}_{-0.21}^{+0.49}$	${39.20}_{-0.39}^{+0.38}$	${39.24}_{-0.39}^{+0.38}$
J1330$-$0036b	${41.16}_{-0.10}^{+0.13}$	${40.84}_{-0.13}^{+0.18}$	${40.96}_{-0.14}^{+0.21}$	${40.92}_{-0.14}^{+0.21}$	${40.92}_{-0.11}^{+0.09}$	${41.21}_{-0.11}^{+0.09}$	${3.98}_{-2.57}^{+6.69}$	${40.25}_{-0.45}^{+0.43}$	${40.30}_{-0.45}^{+0.43}$
J1058+3144a					${40.50}_{-0.06}^{+0.05}$	${40.51}_{-0.06}^{+0.05}$	${1.95}_{-1.19}^{+2.73}$	${39.94}_{-0.41}^{+0.38}$	${39.99}_{-0.41}^{+0.38}$
J1058+3144b*	${42.61}_{-0.05}^{+0.05}$	${42.12}_{-0.18}^{+0.31}$	${42.61}_{-0.05}^{+0.05}$	${42.47}_{-0.05}^{+0.05}$	${41.02}_{-0.09}^{+0.08}$	${41.24}_{-0.09}^{+0.08}$	${2.44}_{-1.75}^{+5.33}$	${40.04}_{-0.55}^{+0.50}$	${40.09}_{-0.55}^{+0.50}$
J1544+0446a	${40.49}_{-0.15}^{+0.24}$	${40.35}_{-0.16}^{+0.26}$			${39.82}_{-0.20}^{+0.14}$	${40.16}_{-0.20}^{+0.14}$	${0.37}_{-0.34}^{+1.09}$	${39.22}_{-1.07}^{+0.59}$	${39.27}_{-1.07}^{+0.59}$
J1544+0446b*	${41.87}_{-0.05}^{+0.06}$	${41.43}_{-0.14}^{+0.21}$	${41.75}_{-0.06}^{+0.07}$	${41.63}_{-0.06}^{+0.07}$	${40.08}_{-0.15}^{+0.11}$	${40.42}_{-0.15}^{+0.11}$	${0.91}_{-0.72}^{+2.27}$	${39.61}_{-0.68}^{+0.54}$	${39.66}_{-0.68}^{+0.54}$

Equations

$$\log\bigg{[}\frac{L_{\rm 2-10\,keV}}{{\rm erg}\,{\rm s}^{-1}}\bigg{]}=(1.22\pm 0.06)\log\bigg{[}\frac{L_{{\rm[O\,{\tiny III}]}}}{{\rm erg}\,{\rm s}^{-1}}\bigg{]}+(-7.34\pm 2.53),$$

$$\mathrm{HR}\equiv \frac{H-S}{H+S},$$

$$L_{0.5-2\mathrm{keV}}^{\mathrm{SF}}=4.5\times 10^{39}\frac{\mathrm{SFR}}{M_{\odot }{\mathrm{yr}}^{-1}}{\mathrm{erg}\mathrm{s}}^{-1},$$

$$L_{2-10\mathrm{keV}}^{\mathrm{SF}}=5.0\times 10^{39}\frac{\mathrm{SFR}}{M_{\odot }{\mathrm{yr}}^{-1}}{\mathrm{erg}\mathrm{s}}^{-1},$$

Full text

black hole physics – galaxies: active – galaxies: interactions – galaxies: nuclei – galaxies: Seyfert – X-rays: galaxies

1 Introduction

The observed growth of structures suggests that mergers of galaxies, and by extension, their central supermassive black holes (SMBHs; Kormendy & Richstone, 1995; Ferrarese & Ford, 2005), should be common throughout most of cosmic history. The final coalescence of merging SMBHs would produce low-frequency gravitational waves (GWs; e.g., Colpi & Dotti, 2011), providing a “standard siren” for cosmology (Hughes, 2009) and a direct testbed for strong-field General Relativity (Centrella et al., 2010). Unlike LIGO’s stellar mass black holes (Abbott et al., 2016) whose detection is still limited to the relatively nearby universe, merging SMBHs would be detectable close to the edge of the observable universe (Cornish & Robson, 2018). They are being hunted currently by pulsar timing arrays for the more massive, low-redshift population (e.g., Arzoumanian et al., 2018), and in the future by space-borne experiments for the less massive, high-redshift population (e.g., Amaro-Seoane et al., 2017).

While the GWs from merging SMBHs are yet to be detected, it is useful to study their progenitors – pairs of SMBHs selected by their electromagnetic signatures from accreting materials from the surroundings. These so-called “dual active galactic nuclei (AGNs)” are AGN pairs in merging galaxies with typical separations of a few kiloparsecs (Gerke et al., 2007; Comerford et al., 2009; Xu & Komossa, 2009). Dual AGNs and AGN pairs in general provide an exciting prospect for understanding massive black hole growth and their merger rates in galaxies in the era of multi-messenger astronomy (e.g., Bhowmick et al., 2019). The first concrete evidence for dual AGNs was a serendipitous discovery by Chandra in NGC 6240 (Komossa et al., 2003), a $z=0.02$ merging ultra-luminous infrared galaxy (ULIRG), which contains two X-ray nuclei separated by $\sim 1\arcsec$ ($\sim 0.7$ kpc) in projection. Until recently only a handful of other secure cases were known in the X-rays (e.g., 3C 75 (Hudson et al., 2006), Mrk 463 (Bianchi et al., 2008; Treister et al., 2018), Mrk 266 (Brassington et al., 2007; Mazzarella et al., 2012), and Mrk 739 (Koss et al., 2011); but see also Arp 299 (Ballo et al., 2004) for a candidate), all of which were confirmed by Chandra.

The past decade has seen significant progress in finding concrete evidence for dual AGNs at $z<0.5$ thanks to systematic searches using large surveys combined with dedicated follow ups, in particular in the X-rays, mid-IR, and radio (e.g., Green et al., 2010; Fabbiano et al., 2011; Liu et al., 2013; Comerford et al., 2015; Kocevski et al., 2015; Fu et al., 2015; Gatti et al., 2016; Koss et al., 2016; Satyapal et al., 2017; Ellison et al., 2017; Fu et al., 2018; De Rosa et al., 2018; Liu et al., 2018; Vignali et al., 2018; Pfeifle et al., 2019). Liu et al. (2011) identified a sample of 1286 [O III]-selected, spatially resolved AGN pairs from the Sloan Digital Sky Survey (SDSS) Seventh Data Release (DR7; Abazajian et al., 2009). Among the 1286 pairs, 94 have projected nuclear separation $r_{\rm p}<10$ kpc, making them the largest sample of [O III]-selected dual AGN candidates. Each of these galaxies contains two nuclei separated by a few arcsec, both of which have individual SDSS spectra. Due to the finite size of SDSS fibers (i.e., fiber collisions), two objects separated by $<55$” cannot both be spectroscopically observed unless being on overlapping plates. We have corrected for this spectroscopic incompleteness when calculating the frequency of dual AGNs (Liu et al., 2011). Based on the SDSS spectra, both nuclei in the candidates are optically classified as Type 2 (i.e., obscured) AGNs according to diagnostic ionization ratios of narrow emission lines (Baldwin et al., 1981; Veilleux & Osterbrock, 1987; Kewley et al., 2001). However, optical diagnostic line ratios only represent indirect evidence for dual AGNs; they cannot conclusively rule out alternative scenarios for the nature of the ionizing sources such as stellar/shock heating, and/or a pair of merging galaxy components ionized by a single active nucleus.

It is generally thought that the X-ray band can provide more direct evidence for nuclear activity. In particular, hard X-rays (defined here as 2-10 keV) are transparent to column densities of $N_{\rm H}\lesssim 10^{24}\ {\rm cm^{-2}}$, while Compton-thick AGNs can be revealed at energies $\gtrsim$10 keV (LaMassa et al., 2011; Lansbury et al., 2015; Nardini, 2017).

We here present a new Chandra ACIS-S X-ray imaging of five dual AGN candidates drawn from the parent sample of 94 closely separated, dual AGN candidates (Figure 1). Figure 2 shows their optical and X-ray images. Chandra ’s high image quality (FWHM $\lesssim 1^{\prime\prime}$) and its capability of moderate spectral resolution imaging spectroscopy in the X-rays make it ideal for assessing the dual-AGN hypothesis. Unlike previous Chandra ’s searches for dual AGNs in galaxy pairs hosting single AGNs (e.g., Teng et al., 2012), we target merging galaxies where both nuclei are [O III]-selected AGNs.

The paper is organized as follows. Section 2 describes the target selection and their optical properties. Section 3 presents the new Chandra ACIS-S X-ray observations, data reduction, and data analysis. Section 4 shows the results on the X-ray luminosities and spectral properties, the X-ray contribution from star formation, and the nature of the nuclear ionizing sources. Finally, we discuss the implications of our results in Section 5 and summarize the main conclusions in Section 6. Throughout this paper, we assume a concordance cosmology with $\Omega_{m}=0.3$, $\Omega_{\Lambda}=0.7$, and $H_{0}=70$ km s*-1* Mpc*-1*, and use the AB magnitude system (Oke, 1974).

2 Target Selection and Optical Properties

Our targets are drawn from a parent sample of 1286 spectroscopic AGN pairs with $r_{\rm p}<100$ kpc and line-of-sight velocity offsets $<$ 600 km s*-1* selected from a heterogenous sample of 138,070 optical AGNs from the SDSS DR7. The optical AGN sample consists of 129,277 Type 2 (i.e., narrow-line) AGNs, 5,564 Type 1 (i.e., broad-line) AGNs, 3,117 Type 1 quasars, and 112 Type 2 quasars (Liu et al., 2011). The parent AGN pair sample is dominated by Type 2 AGNs whose optical narrow emission-line ratios are characteristic of Seyferts, LINERs, and/or AGN-H II composites.

From the parent AGN pair sample, we select kpc-scale dual AGN candidates by first requiring $r_{\rm p}<10$ kpc. We further select systems in which both nuclei have high enough [O III] fluxes to expect at least 50 hard (2-10 keV) X-ray counts in 15 ks for the weaker component (see details below). We additionally exclude systems with existing Chandra observations to avoid duplicate observations (e.g., Mrk 266). For systems with similar [O III] fluxes, targets which are classified as Seyferts based on the BPT diagram are prioritized over those that are classified as AGN-H II composites to maximize the probability of X-ray detections. Therefore, our final target sample consists of five dual AGN candidates. The distribution of our sample is shown in Figure 1. Tables 1 and 2 list their basic photometric and spectroscopic properties. The stellar velocity dispersion was measured by fitting the host-galaxy stellar continuum using the penalized Pixel-Fitting (pPXF) method (Cappellari & Emsellem, 2004) (see the Appendix for details). The total stellar mass was given by the MPA-JHU DR7 catalog from fitting the photometry (Kauffmann et al., 2003a; Salim et al., 2007). Figure 3 shows the narrow emission-line ratios measured from the SDSS spectra subtracted for stellar continua using the pPXF fits. It illustrates that the nuclei in our targets are optically classified as Type 2 Seyferts, LINERs, or AGN-H II composites.

3 Observations, Data Reduction, and Data Analysis

3.1 Chandra ACIS X-ray Imaging Spectroscopy

We observed the five dual-AGN candidates with the ACIS-S on board the Chandra X-ray Observatory between 2012 December and 2013 April (Program GO-14700264, PI: X. Liu). All targets were observed on-axis on the S3 chip, 5.2*′′* to 15.0*′′* away from the aimpoint. Each target was observed for 15 ks (Table 3). The exposure time was set to obtain $\sim$50 counts in the 2–10 keV range from the [O III] weaker nucleus in each merger, although the estimate was too optimistic. We estimated the X-ray counts from the [O III] luminosity for each nucleus (corrected for [O III] emission due to star formation in AGN-H II composites (Kauffmann & Heckman, 2009)), assuming an empirical correlation between the 2–10 keV (unabsorbed) and [O III] luminosities. Measurements of $L_{\rm 2-10\,keV}/L_{{\rm[O\,{\tiny III}]}}$ for optically selected Type 2 AGNs span a wide range (Mulchaey et al., 1994; Heckman et al., 2005; Panessa et al., 2006), with values from a few to a few hundred. For the baseline assumption, we adopted the mean calibration of Panessa et al. (2006) given by,

[TABLE]

where the [O III] luminosities have been corrected for the Galactic and intrinsic NLR extinction by using the Balmer decrement method via H$\alpha$/H$\beta$  ratio. We accounted for systematic uncertainties using the Heckman et al. (2005) relation based on optically selected (single) Type 2 AGNs. An X-ray power-law spectrum was assumed with an absorbing column density $N_{\rm H}=10^{22}\ {\rm cm^{-2}}$ (typical for our targets for which enough counts were detected for spectral analysis (see below) and for Type 2 Seyferts; Bassani et al., 1999) and a photon index $\Gamma=1.7$ (typical for unabsorbed Seyferts; Green et al., 2009).

We reprocessed the data using CIAO v4.8 and the corresponding calibration files following standard procedure111http://cxc.harvard.edu/ciao/. We examined the light curve of each observation and found no time interval of high background. We produced counts and exposure maps with the original pixel scale ($0\farcs 492{\rm~{}pixel^{-1}}$) in the 0.5–2 keV ($S$), 2–8 keV ($H$), and 0.5–8 keV ($F$) bands. The exposure maps were weighted by the above fiducial incident spectrum.

Following the source detection procedure detailed in Wang (2004) and Hou et al. (2017), we detect X-ray sources in the $S$, $H$ and $F$ bands in each image. With a local false detection probability $P\leq{10^{-6}}$ (empirically yielding $\sim$0.1 false detection per field), we detected a total of 124 sources in the field-of-view covered by the S3 and S2 CCDs. For each detected source, we derived background-subtracted and exposure map-corrected count rates in each individual band from within the 90% enclosed-energy radius (EER), taking into account the position-dependent point-spread function and the local background.

Figure 2 shows the ACIS images of the five targets in the $F$, $S$ and $H$ bands. Table 3 summarizes the X-ray measurements. Given the low count levels, we do not apply any smoothing to avoid artifacts. As we will show in Section 4.1, our targets are significantly weaker hard X-ray emitters than those predicted from both the Panessa et al. (2006) and Heckman et al. (2005) relations based on single optically selected AGNs, resulting in far fewer counts than expected. Seven of the ten nuclei in our targets were detected in the $F$-band. Six nuclei were detected in both $S$ and $H$ bands, whereas one nucleus (J1544+0446a) was detected in the $S$ band only. Three nuclei (J0805+2818b, J1330$-$0036a and J1058+3144a) were undetected in the X-rays.

3.2 Spectral Analysis and Hardness Ratio

Three of the seven X-ray detected nuclei have sufficient net counts (${\geq}50$) for spectral analysis. For these nuclei, we extracted spectra from within the 90% EER of each source of interest using the CIAO tool specextract and built the Response Matrix Files (RMFs) and Ancillary Response Files (ARFs); the corresponding background spectrum was extracted typically between two to five times the 90% EER. For nucleus J1544+0446b, with the other X-ray detected nucleus located in the background extraction region given the small projected nuclear separation (4.1″), we further removed the 90% EER of this other nucleus from the background to eliminate its contamination. We fit each spectrum with an absorbed power-law model. Due to the low net counts, we adopted the C-statistic (a Poisson log-likelihood function, taking into account the known Poisson background, pstat in XSpec) for the spectral fitting (Cash, 1979). Figure 4 shows the fitted spectra. Table 3 lists the spectral fitting results. The best-fit power-law spectral index and the absorption column density are consistent with Type-2 AGN for all three nuclei.

For the other four X-ray detected AGNs with $<50$ net counts, we estimate their spectral properties using hardness ratios (HR). The hardness ratio is defined as

[TABLE]

where $H$ and $S$ are the number of counts in the hard and soft X-ray bands, respectively. We adopt the Bayesian Estimation of Hardness Ratios (BEHR; Park et al., 2006) to measure the HRs and their uncertainties, which is appropriate for the low count regime. Table 3 lists the HR results.

For the three sources without X-ray detection and one nucleus (J1544+0446a) without X-ray detection in H band, we estimate the X-ray net counts 3$\sigma$ upper limit using the CIAO tool aprate.

4 Results

4.1 X-ray Luminosities

X-ray emission provides the most direct evidence for nuclear activity. The 2–10 keV hard X-ray band is transparent to column densities of $N_{{\rm H}}\lesssim 10^{24}$ cm*-2*. To infer X-ray luminosity, we assume a simple absorbed power-law model. For the three nuclei with enough counts for X-ray spectral analysis, we adopt the best-fit column density and photon index given in Table 3. For the other seven nuclei without enough counts or without X-ray detection, we assume a photon index of 1.7 and an absorption column density $N_{{\rm H}}=10^{22}$ cm*-2*. In Table 4 we list the unabsorbed/intrinsic X-ray luminosity or upper limit of each nucleus in the total, soft, and hard bands, respectively.

The adopted single absorbed power-law model is most likely too simple for the X-ray spectra of obscured AGNs, in which thermal emission from starburst components and/or scattered nuclear emission are often present (e.g., Turner et al., 1997a, b). However, the low counts of our detections do not allow us to test more realistic models. In addition, our estimates of the intrinsic absorbing column may not necessarily reflect the true values in cases of patchy obscuration and/or significant scattering off of an ionized medium in Compton-thick (i.e., $N_{{\rm H}}\sim 10^{24}$ cm*-2* or larger) AGNs [which represent about half of the local Type 2 Seyfert population (Risaliti et al., 1999)], as observed in NGC 6240 (e.g., Vignati et al., 1999; Ptak et al., 2003) and in NGC 1068 (e.g., Matt et al., 1997; Guainazzi et al., 1999), although, again, the quality of our data do not allow us to robustly test these possibilities.

The seven X-ray detected nuclei have estimated unabsorbed 0.5–8 keV luminosities ranging from $3.1\times 10^{40}{\rm~{}erg~{}s^{-1}}$ to $4.1\times 10^{42}{\rm~{}erg~{}s^{-1}}$. One nucleus J1544+0446a is not detected in the hard band, and the remaining six hard X-ray detected nuclei have estimated unabsorbed 2–10 keV luminosities ranging from $9.0\times 10^{40}{\rm~{}erg~{}s^{-1}}$ to $4.0\times 10^{42}{\rm~{}erg~{}s^{-1}}$. Figure 5 shows the comparison of [O III] luminosities (both observed and extinction-corrected) and 2–10 keV X-ray luminosities (both observed and unabsorbed). The estimated upper limits for the four hard X-ray undetected nuclei range from $6.5\times 10^{40}{\rm~{}erg~{}s^{-1}}$ to $5.1\times 10^{41}{\rm~{}erg~{}s^{-1}}$ in 2–10 keV. These luminosity estimates are comparable to or smaller than those of the previously known X-ray confirmed dual AGNs (e.g., Komossa et al., 2003; Ballo et al., 2004; Hudson et al., 2006; Bianchi et al., 2008; Mazzarella et al., 2012; Koss et al., 2011).

4.2 X-ray Contribution from Nuclear Starburst

The estimated intrinsic hard X-ray luminosities of our targets are close to or below $\sim 10^{42}$ ergs s*-1* – the characteristic upper limit for the most luminous star-forming galaxies (e.g., Zezas et al., 2001). Hence it is possible that much or all of the luminosity is due to star formation. X-ray spectral shape offers another diagnostic to discriminate between AGN and starburst scenarios. However, the uncertainties of our spectral estimates are too large to draw firm conclusions for the majority of the nuclei.

We use independent star formation rate (SFR) estimates to test the AGN scenario for each nucleus. To estimate the expected X-ray emission due to star formation within similar apertures used to perform our X-ray extraction (typically a radius of 2″), we use the SDSS fiber SFR given by the MPA-JHU DR7 catalog (Salim et al., 2007). For galaxies classified as AGNs or composites according to the optical BTP diagram, which are the case for our targets, the SFRs are estimated by constructing the likelihood distribution of the specific SFR as a function of the 4000 Å break Dn(4000) based on the star-forming sample (Brinchmann et al., 2004) multiplied by the stellar mass.

To derive X-ray luminosites from the fiber SFRs, we adopt the empirical calibration of Ranalli et al. (2003, see also ) based on 23 nearby star-forming galaxies, which is given by

[TABLE]

[TABLE]

with an rms scatter of 0.27 dex and 0.29 dex. In Table 4 we list the fiber SFR estimates and the derived $L^{{\rm SF}}_{0.5-2\,{\rm keV}}$ and $L^{{\rm SF}}_{2-10\,{\rm keV}}$ estimates for each nucleus. Figure 6 compares the expected X-ray luminosities due to star formation against the observed X-ray luminosities in the soft and hard bands. In both bands, the predicted X-ray contribution from star formation for the majority of our targets’ nuclei is below the observed X-ray luminosity, suggesting an additional excitation source from the AGN. We caution, however, that there are significant systematic uncertainties of our estimates of the expected X-ray luminosities due to star-formation-related processes (e.g., uncertainties in the IMF, extinction correction; Liu et al., 2013).

We also test the contribution to hard X-ray luminosity from low-mass X-ray binaries, which is proportional to stellar mass ($M_{*}$). Based on the $L_{\rm X}-M_{\ast}$ relations from Gilfanov (2004) or Lehmer et al. (2010), the estimated contribution from stellar mass enclosed in the SDSS fiber is negligible compared to the contribution from star forming activity (typically $<$ 10% and never exceeding 30%).

4.3 Results on Individual Targets

4.3.1 SDSS J0907+5203

Both galaxies in the merger are optically classified as Type 2 Seyferts (Figure 3). Both nuclei were detected in both soft and hard X-ray bands. The northern galaxy (J0907+5203b) has enough counts for X-ray spectral analysis, which suggests moderate nuclear obscuration, with an estimated column density of $N_{{\rm H}}{\approx}2.6^{+1.5}_{-1.1}\times 10^{22}$ cm*-2*. The spatial profiles of the two nuclear X-ray sources are consistent with the AGN scenarios for both galaxies. For both galaxies, the expected star-formation-induced X-ray luminosities are too low to explain the observed values in both soft and hard X-ray bands (Figure 6), consistent with the dual AGN scenario.

4.3.2 SDSS J0805+2818

The NW nucleus (J0805+2818a) in the merger is optically classified as a Type 2 Seyfert whereas the SE nucleus (J0805+2818b) a Type 2 Seyfert or a LINER (Figure 3). Only the NW nucleus was detected in both soft and hard X-ray bands; the SE nucleus was detected in neither. The host galaxy of the SE nucleus shows strong Balmer absorption features in the SDSS spectrum characteristic of post-starburst galaxies (Figure 7). For the NW nucleus, both the spatial profile of the X-ray source and the comparison between the observed to the expected star-formation-induced X-ray luminosities (Figure 6) are consistent with the AGN scenario. For the SE nucleus, on the other hand, the upper limits of the X-ray luminosities (Figure 6) are still consistent with the presence of an additional AGN, although the possibility that one AGN in the NW nucleus ionizes gas in both galaxies cannot be ruled out.

4.3.3 SDSS J1330$-$0036

The NW nucleus (J1330$-$0036a) in the merger is optically classified as an H II/AGN composite whereas the SE nucleus (J1330$-$0036b) a Type 2 Seyfert (Figure 3). Only the SE nucleus was detected in both soft and hard X-ray bands; the NW nucleus was not detected in either. The host galaxy of the NW nucleus shows strong Balmer absorption features in the SDSS spectrum characteristic of post-starburst galaxies (Figure 7). For the SE nucleus, the observed X-ray luminosities are similar to those expected from star-formation-induced X-ray luminosities (Figure 6), although an additional AGN component cannot be ruled out given significant uncertainties in the estimates. For the NW nucleus, on the other hand, the upper limits of the X-ray luminosities (Figure 6) are still consistent with the presence of an additional AGN, although the possibility that one AGN in the SE nucleus ionizes gas in both galaxies cannot be ruled out.

4.3.4 SDSS J1058+3144

The SW nucleus (J1058+3144a) in the merger is optically classified as an H II/AGN composite whereas the NE nucleus (J1058+3144b) a Type 2 Seyfert (Figure 3). Only the NE nucleus was detected in both soft and hard X-ray bands; the SW nucleus was not detected in either. The NE nucleus has enough counts for X-ray spectral analysis, which suggests moderate nuclear obscuration, with an estimated column density of $N_{{\rm H}}{\approx}6.4^{+6.3}_{-4.5}\times 10^{22}$ cm*-2*. For the NE nucleus, both the spatial profile of the X-ray source and the comparison between the observed to the expected star-formation-induced X-ray luminosities (Figure 6) are consistent with the AGN scenario. For the SW nucleus, on the other hand, the upper limits of the X-ray luminosities (Figure 6) are still consistent with the presence of an additional AGN, although the possibility that one AGN in the NE nucleus ionizes gas in both galaxies cannot be ruled out.

4.3.5 SDSS J1544+0446

The SW nucleus (J1544+0446a) in the merger is optically classified as a LINER whereas the NE nucleus (J1544+0446b) a LINER, Type 2 Seyfert or composite (Figure 3). The NE nucleus was detected in both soft and hard X-ray bands; the SW nucleus was detected only in the soft X-ray band. The NE nucleus has enough counts for X-ray spectral analysis, which suggests moderate nuclear obscuration, with an estimated column density of $N_{{\rm H}}{\approx}3.9^{+3.2}_{-2.7}\times 10^{22}$ cm*-2*. For the NE nucleus, both the spatial profile of the X-ray source and the comparison between the observed to the expected star-formation-induced X-ray luminosities (Figure 6) are consistent with the AGN scenario. An additional AGN component is likely also for the SW nucleus because the observed soft X-ray luminosity is significantly larger than that expected from star-formation-related activity, despite significant uncertainties.

It is noteworthy that those target galaxies without X-ray detection tend to have a blue color in the composite SDSS image (Figure 2). These seem to be relatively small galaxies with young stellar populations. One possibility is that the X-ray emission from these galaxies is dominated by star formation and they are more likely to host a dwarf SMBH which is harder to detect. Whether this is a definitive trend can only be answered with a sizable sample of close pairs.

In summary, the new Chandra X-ray observations support the dual-AGN scenario for two of our five [O III]-selected targets (SDSS J0907+5203 and SDSS J1544+0446). For the other three targets (SDSS J0805+2818, SDSS J1330$-$0036, and SDSS J1058+3144), the existing data are still consistent with the dual-AGN scenario, although the possibility of only one AGN ionizing both components in the mergers cannot be ruled out.

5 Discussion

5.1 Systematically Smaller X-ray-to-[O III]-luminosity Ratio in Dual AGNs than in Single AGNs

Figure 5 shows the relation between the hard X-ray luminosity and the [O III] luminosity for each nucleus in our targets. We compare the X-ray-to-[O III]-luminosity ratio of our new targets studied in this work to those observed in both single AGNs and dual AGNs systematically selected from double-peaked [O III] emission lines Liu et al. (2010a, see also ). We examine both the relation between the observed hard X-ray luminosity $L_{X,2-10\,{\rm keV,observed}}$ and the observed [O III] luminosity $L_{{\rm[O\,III],observed}}$, and that between the unabsorbed hard X-ray luminosity $L_{X,2-10\,{\rm keV,unabsorbed}}$ and the extinction-corrected [O III] luminosity $L_{{\rm[O\,III],extinction-corrected}}$. We use the appropriate comparison samples for the two cases separately, because usually either the observed or the corrected luminosity is available in any given literature sample.

For the $L_{X,2-10\,{\rm keV,observed}}$-$L_{{\rm[O\,III],observed}}$ relation (left panel of Figure 5), the comparison samples include 47 hard X-ray (3–20 keV) selected local AGNs and 55 optically selected local [O III]-bright AGNs (Xu et al., 1999; Whittle, 1992) studied by Heckman et al. (2005), and 8 optically selected Type 2 quasars from Ptak et al. (2006) at redshifts $z{\sim}$0.3–0.8. Heckman et al. (2005) showed that single, optically selected Type 2 AGNs (the grey dashed line) have systematically lower $L_{X,2-10\,{\rm keV,observed}}$ (by an average of 1.0 dex) at a given $L_{{\rm[O\,III],observed}}$ than hard X-ray selected AGNs (both Type 1 and Type 2) and optically selected Type 1 AGNs (the blue dashed-dotted-dotted line), as expected for heavily absorbed AGNs (see also Mulchaey et al., 1994; Panessa et al., 2006). Liu et al. (2013) has shown that the four dual AGNs (individually as the yellow standing glasses and collectively as the yellow dotted line) selected from the parent sample of Type 2 AGNs with double-peaked [O III] emission lines (Liu et al., 2010b) have systematically smaller $L_{X}/L_{{\rm[O\,III]}}$ (observed) ratios (by $\sim 0.8\pm 0.2$ dex on average) than even optically selected single Type 2 AGNs. Our new targets seem to have X-ray-to-[O III]-luminosity ratios that are on average in between that observed in single, optically selected Type 2 AGNs and that observed in the dual AGNs selected from double-peaked [O III] emission lines (with some ambiguities and uncertainties due to the upper limits of several measurements).

A similar trend is also seen in the $L_{X,2-10\,{\rm keV,unabsorbed}}$-$L_{{\rm[O\,III],extinction-corrected}}$ relation as shown in the right panel of Figure 5). Again, our new targets seem to have X-ray-to-[O III]-luminosity ratios that are on average in between that observed in single AGNs (the grey dashed-dotted line) and that observed in dual AGNs selected from double-peaked [O III] emission lines (the yellow dotted line). The comparison sample of single AGNs includes 47 Palomar Seyfert galaxies (optically selected Type 1 and Type 2 Seyferts drawn from the Palomar survey of nearby galaxies by Ho et al. 1995) from Panessa et al. (2006). Panessa et al. (2006) has demonstrated that after properly accounting for absorption correction (including for Compton-thick sources), optically selected Type 1 and Type 2 Seyferts follow the same $L_{X,2-10\,{\rm keV,unabsorbed}}$-$L_{{\rm[O\,III],extinction-corrected}}$ relation. In particular, optically selected Type 2 Seyferts, which were significantly X-ray weaker than Type 1 Seyferts, also obey the same relation, after the “Compton thick” luminosity correction. Liu et al. (2013) has shown that after correction for gas absorption and dust extinction, the unabsorbed hard X-ray luminosities of double-peaked-[O III]-selected dual AGNs appear to be $\sim 2.4\pm 0.3$ dex smaller (at log$L_{{\rm[O\,III]}}$ of 42.0) than those expected from the Panessa et al. (2006) relation, log$L_{X}$ = $1.22$log$L_{{\rm[O\,III]}}-7.34$, although the absorption correction of dual AGNs may have been significantly underestimated.

5.2 Interpretation: Enhanced Nuclear Absorption from Merger-induced Gas Inflows

Liu et al. (2013) has suggested that the observed X-ray weak tendency in dual AGNs selected in Type 2 AGNs with double-peaked narrow [O III] lines is caused by a combination of a higher nuclear gas column, which may be induced by merger events, and an orientation bias related to the double-peak narrow emission-line selection. In contrast to the Liu et al. (2013) sample, our targets are not subject to the orientation bias due to the line-of-sight velocity splitting requirement caused by the double-peaked [O III] selection. On the other hand, our sample is likely to have a higher nuclear absorption from merger-induced gas inflows than that in single local AGNs, similar to the case of double-peaked-[O III]-selected dual AGNs. Arising from the narrow-line regions that are much further out, the [O III] emission is less subject to nuclear gas absorption and dust obscuration than the hard X-ray emission from the black hole accretion disk corona, which would explain the systematically smaller hard-X-ray-to-[O III]-luminosity ratios observed in dual than in single AGNs. The fact that our targets seem to have hard-X-ray-to-[O III]-luminosity ratios that are smaller than that seen in single AGNs but larger than that observed in double-peaked-[O III]-selected dual AGNs is consistent with the conclusion of Liu et al. (2013) that a combination of two effects (i.e., both merger-enhanced absorption and obscuration and an orientation selection bias) are at work for double-peaked-[O III]-selected dual AGNs.

6 Conclusions

Dual AGNs are crucial to our understanding of the accretion and dynamical evolution of SMBHs in mergers, the effects of merger-induced activity on galaxy evolution, and the initial conditions of close binary SMBHs. Building on Chandra ’s previous success on its unique power in resolving dual AGNs, here we have studied the X-ray properties of a sample of five optically selected dual AGN candidates. Our targets were drawn from a sample of 1286 [O III]-selected AGN pairs (both Type 1 and Type 2 sources) systematically selected from the SDSS DR7. Each of the targets contains two nuclei separated by 3–9 kpc in projection, both of which are optically classified as Type 2 (obscured) AGNs based on diagnostic ratios of the optical narrow emission lines. While being systematically selected from the largest sample of dual AGN candidates, the optical classification was inconclusive. Furthermore, because the double nuclei are close (with physical projected separations of a few kpc), there may be only one AGN ionizing both galaxies, producing two optical emission-line nuclei. Arguments based on the spatial distribution of ionization parameters estimated from optical emission lines cannot conclusively discriminate between the single- and dual-AGN scenarios (e.g., Liu et al., 2010b). The new Chandra ACIS-S X-ray imaging presented here helps solve the problem by resolving and localizing the ionizing sources directly in the X-rays. The X-ray confirmation of a systematically selected sample also helps place the optically inferred dual-AGN frequency on a firmer ground (Liu et al., 2011). Our main findings are summarized as the following:

• •

Chandra ’s superb spatial resolution and sensitivity in the X-rays allowed us to localize the ionizing sources and determine their X-ray properties. Seven of the ten nuclei were detected in the full 0.5–8 keV band. Six were detected in both soft (0.5–2 keV) and hard (2–8 keV) bands, whereas one nucleus was detected in the soft band only. Three nuclei were undetected in the X-rays (Table 3).

• •

The hard X-rays directly probe the accretion disk corona of the accreting SMBHs, providing a more robust estimate of the intrinsic AGN luminosity than using [O III] $\lambda$5007 luminosity as a surrogate. In three of the ten nuclei we observed enough counts to perform spectral fittings to constrain the X-ray spectral properties and absorption column densities. We fit each X-ray spectrum with an absorbed power-law model. The best-fit power-law spectral indices and the absorption column densities are consistent with them being Type 2 AGNs for all three nuclei. For the other four X-ray detected nuclei, we have estimated their spectral properties and luminosities using hardness ratios. For the three sources without X-ray detection, we estimate the X-ray net counts 3$\sigma$ upper limit using the CIAO tool aprate (Section 3.2).

• •

Combined with independent star formation rate estimates empirically calibrated based on the host-galaxy stellar continua, the new Chandra X-ray observations allowed us to evaluate the dual-AGN hypothesis for each target. We have confirmed two (SDSS J0907+5203 and SDSS J1544+0446) of the five targets as bona-fide dual AGNs. For the other three targets, the existing data are consistent with the dual-AGN scenario, but we cannot conclusively rule out the possibility of stellar and/or shock heating and/or one AGN ionizing both gaseous components in a merger (Section 4.3).

• •

The average X-ray-to-[O III] luminosity ratio in our targets seems to be systematically smaller than that observed in single AGNs but is higher than that seen in dual AGNs selected from AGNs with double-peaked narrow emission lines. We suggest that the systematically smaller X-ray-to[O III] luminosity ratio observed in dual AGNs than in single AGNs is due to a high nuclear gas column likely from strong merger-induced inflows. Unlike double-peaked-[O III]-selected dual AGNs, the new sample selected from resolved galaxy pairs are not subject to the orientation bias caused by the double-peak line-of-sight velocity splitting selection, which also contributes to lowering the X-ray-to-[O III] luminosity ratio (Figure 5).

Our sample size is still too small for a statistical analysis to compare with theoretical predictions from simulations in a meaningful way (e.g., Capelo et al., 2017; Blecha et al., 2018; Rosas-Guevara et al., 2019; Solanes et al., 2019). To put the conclusions on a firm statistical ground, future wide-field, high-resolution, and high-sensitivity X-ray telescopes (such as Lynx X-ray Surveyor; The Lynx Team, 2018) may detect thousands of dual AGNs, which will be needed to fully understand black hole growth in mergers and dual AGNs (e.g., Burke-Spolaor et al., 2018; Koss et al., 2019).

Acknowledgements

We thank M. Strauss for his generous support and guidance on the project. M.H. and Z.L. acknowledge support by the National Key Research and Development Program of China (2017YFA0402703). X.L. and H.G. acknowledge support by NASA through Chandra Award Number GO3-14103X issued by the Chandra X-ray Observatory Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of NASA under contract NAS 8-03060. Y.S. acknowledges support from the Alfred P. Sloan Foundation and NSF grant 1715579.

This research has made use of software provided by the Chandra X-ray Center in the application packages CIAO, ChIPS, and Sherpa.

Funding for the Sloan Digital Sky Survey IV has been provided by the Alfred P. Sloan Foundation, the U.S. Department of Energy Office of Science, and the Participating Institutions. SDSS-IV acknowledges support and resources from the Center for High-Performance Computing at the University of Utah. The SDSS website is www.sdss.org.

SDSS-IV is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS Collaboration including the Brazilian Participation Group, the Carnegie Institution for Science, Carnegie Mellon University, the Chilean Participation Group, the French Participation Group, Harvard-Smithsonian Center for Astrophysics, Instituto de Astrofísica de Canarias, The Johns Hopkins University, Kavli Institute for the Physics and Mathematics of the Universe (IPMU) / University of Tokyo, Lawrence Berkeley National Laboratory, Leibniz Institut für Astrophysik Potsdam (AIP), Max-Planck-Institut für Astronomie (MPIA Heidelberg), Max-Planck-Institut für Astrophysik (MPA Garching), Max-Planck-Institut für Extraterrestrische Physik (MPE), National Astronomical Observatories of China, New Mexico State University, New York University, University of Notre Dame, Observatário Nacional / MCTI, The Ohio State University, Pennsylvania State University, Shanghai Astronomical Observatory, United Kingdom Participation Group,Universidad Nacional Autónoma de México, University of Arizona, University of Colorado Boulder, University of Oxford, University of Portsmouth, University of Utah, University of Virginia, University of Washington, University of Wisconsin, Vanderbilt University, and Yale University.

Facilities: Chandra X-ray Observatory (ACIS), Sloan

Appendix

In this appendix, we present details of our spectral fitting analysis to carefully measure the host-galaxy stellar continuum and to model the emission-line fluxes over the host-subtracted spectrum. For the host galaxy spectral fitting, we adopt the penalized Pixel-Fitting (pPXF) method222https://pypi.org/project/ppxf/ (Cappellari & Emsellem, 2004). The method works directly in the pixel space and uses the maximum penalized likelihood formalism to extract as much information as possible from the spectra while suppressing the noise in the solution. After subtracting the host-galaxy continuum using the pPXF best-fit solution, we then model the emission-line-only spectrum using the spectral fitting code qsofit333https://github.com/legolason/PyQSOFit (Shen et al., 2018). Figures 7 and 8 show the fitting results for all the 10 nuclei in our targets.