CHANDRA DISCOVERY OF A BINARY ACTIVE GALACTIC NUCLEUS IN Mrk 739

Mrk 739 was imaged by the Sloan Digital Sky Survey (SDSS) on 2005 March 10. Using a Sérsic profile with a fixed bulge (n = 4), we fit the optical nuclei using two-dimensional surface brightness fitting (GALFIT; Peng et al. 2002). In Mrk 739E, a point source component was used to measure the AGN light since it has a broad-line region (BLR).

We observed Mrk 739 with Gemini on 2011 February 7. Both nuclei were observed simultaneously in the B600-G5307 grating with a 1'' slit in the 4300–7300 Å wavelength range. The exposure totaled 37 minutes. We follow Winter et al. (2010) for correcting Milky Way reddening, starlight continuum subtraction, and fitting AGN diagnostic lines. To correct our line ratios for extinction, we use the narrow Balmer line ratio (Hα/Hβ) assuming an intrinsic ratio of 3.1 and the Cardelli et al. (1989) reddening curve.

Mrk 739 was observed in the UV with XMM-Newton in 2009 June (PI: Brandt; R. V. Vasudevan 2011, in preparation). We follow the XMM ABC guide for photometry. We also analyzed archival Very Large Array (VLA) observations with times of 33 minutes at 1.49 GHz and 38 minutes at 4.86 GHz.

Chandra observed Mrk 739 on 2011 April 22 with an exposure time totaling 13 ks. Subpixel event repositioning was applied to improve the resolution of the image. Two extraction regions of 15 radius were used for spectral fitting and timing analysis with CIAO version 4.3. To fit the X-ray spectra, we used a fixed Galactic photoelectric absorption (Kalberla et al. 2005), a floating photoelectric absorption component at z = 0.0297, and a power law. For the eastern source (Mrk 739E), we also include a pileup model because mild pileup (10%–20%) is expected based on pixel count rates.

The ¹²CO (2–1) and (3–2) molecular lines of Mrk 739 were observed with the James Clerk Maxwell Telescope (JCMT) on 2011 March 12–13. The A3 (211–279 GHz) and HARP (325–375 GHz) receivers were used. The spectra were co-added, binned, and fitted with linear baselines. To calculate velocity-integrated line flux densities, we assumed an aperture efficiency of 0.60 and 0.53 for the A3 and HARP receiver and followed Greve et al. (2009).

Two hard X-ray point sources coincide with the best-fit model of the optical light from the bulge components (Figure 1). The bulge magnitudes are m_r = 14.03 ± 0.15 for Mrk 739E and m_r = 13.75 ± 0.15 for Mrk 739W. The small difference in apparent magnitudes suggests a major merger between the two galaxies.

Zoom In Zoom Out Reset image size

The optical spectra for both sources are shown in Figure 2. Mrk 739E shows broad lines (FWHM Hβ = 2960 km s⁻¹ and Hα = 2120 km s⁻¹) consistent with a Seyfert 1. Mrk 739E also has strong [Fe vii] 5721 and 6087 Å emission, a feature of some Seyfert 1 galaxies indicative of highly ionized material near the central AGN (e.g., Veilleux 1988). In Mrk 739W, there are narrow lines at the spectral resolution of instrument (FWHM = 280 km s⁻¹). Based on the Balmer decrement, E(B − V) = 0.26 for Mrk 739E and E(B − V) = 0.43 for Mrk 739W. Mrk 739W is classified as a starburst using the [O i]/Hα and [S ii]/Hα line diagnostics and a composite galaxy using the [N ii]/Hα diagnostic (Kewley et al. 2006). Assuming that all of the Hα luminosity is from star formation and using Kennicutt (1998), the estimated star formation rate (SFR) = 0.3 M_☉ yr⁻¹.

Zoom In Zoom Out Reset image size

The UV image from XMM-Newton Optical Monitor shows sources coincident with the hard X-ray sources (Figure 3). In the UV, m_UVW1 = 15.4 ± 0.1 and m_UVM2 = 16.6 ± 0.1 for Mrk 739E and m_UVW1 = 16.6 ± 0.1 and m_UVM2 = 16.4 ± 0.1 for Mrk 739W. The spectral index connecting 2500 Å and 2 keV, α_OX, is −1.10 ± 0.04 for Mrk 739E, typical of an AGN (−1.15 ± 0.24; Steffen et al. 2006). In Mrk 739W, α_OX = −1.53 ± 0.1, suggesting the UV is dominated by star formation. Assuming all the UV emission in Mrk 739W is from star formation and following Kennicutt (1998), we estimate SFR = 0.6 M_☉ yr⁻¹.

Zoom In Zoom Out Reset image size

Mrk 739 was detected in all bands by IRAS. The measured luminosity is log L_{☉, IR} = 10.9 and log L_{☉, FIR} = 10.6. Following the Kennicutt (1998) relationship between far-infrared (FIR) luminosity and SFR, we estimate SFR = 6.9 M_☉ yr⁻¹.

Mrk 739E was detected at both 1.49 and 4.86 GHz (Figure 3), with a flux density of 2.6 ± 0.2 mJy at 1.49 GHz and 0.5 ± 0.2 at 4.86 GHz (after convolving to match the 1.49 GHz resolution). This is a spectral index (S_ν = K × ν^−α) of α = 1.2 ± 0.5.

The VLA data for Mrk 739W are consistent with resolved star formation and show no signs of an AGN. The emission at 1.49 GHz has an integrated flux density of 2.6 ± 0.2 mJy. The 4.86 GHz convolved data also show resolved emission with an integrated flux density of 1.3 ± 0.3 mJy. The spectral index for Mrk 739W is α = 0.8 ± 0.3, consistent with optically thin synchrotron emission from supernovae (SNe) found in star-forming galaxies. We use Yun et al. (2001) to convert the 1.49 GHz luminosity of Mrk 739W to an SFR of 3.1 M_☉ yr⁻¹.

We detect two hard X-ray sources coincident with the eastern optical nucleus (Mrk 739E) and western nucleus (Mrk 739W). Both sources show hard X-ray spectra extending out to 10 keV (Figure 4). In the 2–10 keV band, we find an FWHM of 048 ± 0.05 (280 pc) for Mrk 739E and 051 ± 0.07 (295 pc) for Mrk 739W. The Chandra spectra of Mrk 739E is well fit (χ²_ν = 1.4) by a power law (F∝E^{−Γ + 2}) with photon index Γ = 2.1 ± 0.1 and N_H = 1.5 ± 0.2 × 10²¹ cm⁻² consistent with a Seyfert 1. Mrk 739W is well fit (C-stat/dof = 0.8) by a harder power law with more absorption and a photon index Γ = 1.0 ± 0.2 and N_H = 4.6 ± 0.1 × 10²¹ cm⁻². While positive residuals do exist at the location of the neutral 6.4 keV iron Kα line in Mrk 739W, there are too few counts to confirm its existence. The 2–10 keV absorption-corrected luminosities are L_2–10 keV = 1.1 × 10⁴³ and 1.0 × 10⁴² erg s⁻¹ for Mrk 739E and Mrk 739W, respectively. An archival XMM-Newton 2009 observation of Mrk 739 is unable to resolve the emission to either source, but shows L_2–10 keV = 1.0 × 10⁴³ and Γ = 1.92 ± 0.02 consistent with the Chandra spectra of Mrk 0739E.

Zoom In Zoom Out Reset image size

Timing analysis is a critical part of identifying AGNs since nearly all show variability. Bins of 1.5 ks were chosen to ensure >20 counts per bin (Figure 4). We find statistically significant variations in the fluxes of both sources. There is a factor of 2.5 change in flux for Mrk 739W and 0.3 for Mrk 739E during the ≈4 hr observation.

Mrk 739 was detected by Swift BAT with L_{14–195 keV} = 2.4 ± 0.5 × 10⁴³ erg s⁻¹, Γ = 2.6 ± 0.4, and a factor of two variability over five years (Figure 4). Because of the variability and much steeper photon index of Mrk 739W, it is difficult to identify the source of the L_{14–195 keV} luminosity (Mrk 739E, Mrk 739W, or both).

The hard X-ray (>2 keV) band provides one of the best tools for finding AGNs since it is less affected by contamination and absorption and can only be produced in large amounts by AGNs. Our discussion of this binary AGN will be limited to Mrk 739W since Mrk 739E was already known to be an AGN based on its BLR (Netzer et al. 1987).

In Mrk 739W, the hard X-ray emission is point-like at the location of one of the bulge components in the galaxy consistent with an AGN. Energetic phenomena related to vigorous star formation such as OB stars, X-ray binaries, and SN shocks produce hard X-rays. However, in star-forming regions the dominant X-ray emission is from point-like ultraluminous X-ray sources (ULXs; Bertram et al. 2007). ULXs are by definition not located at the centers of galaxies where the central supermassive black hole resides.

The luminous hard X-ray emission, hard photon index, and time variability of Mrk 739W also provide little support for the hypothesis that this source is a ULX. In a study of 154 ULXs, the average luminosity is much lower (L_{0.5–8 keV} = 2.2 ± 4.5 × 10³⁹ erg s⁻¹). The most luminous ULX ever detected, ESO 243-49 HLX-1 (Farrell et al. 2009), has a 0.2–10 keV luminosity of 10⁴² erg s⁻¹, however, its hard X-ray (2–10 keV) luminosity is only 4 × 10⁴⁰ erg s⁻¹ because it has a soft photon index of 3.4. Therefore, if Mrk 739W is a ULX, it is the most luminous ULX in the hard X-rays (L_2–10 keV) by over an order of magnitude. The average ULX power-law index is also Γ = 1.97 ± 0.11, which is significantly softer than Mrk 739W (Γ = 1.0 ± 0.2). Finally, no ULXs have shown such high amplitude variability over the short timescale of hours with this level of variability only seen on timescales of days to weeks (Gladstone 2010).

The measured SFRs in Mrk 739W provide an additional constraint as to whether the hard X-ray emission could be from star formation. An SFR greater than 200 M_☉ yr⁻¹ would be required to generate the observed hard X-ray luminosity based on the relationship between SFR and X-ray emission (Ranalli et al. 2003). The predicted SFR in Mrk 739W is 0.3, 0.6, and 3.1 M_☉ yr⁻¹ from the Hα, UV luminosity, and 1.4 GHz emission. The predicted SFR from FIR emission of the combined system is 6.9 M_☉ yr⁻¹. All of these rates are significantly lower than the 200 M_☉ yr⁻¹ needed to generate the observed hard X-ray luminosity. In addition, it is likely that much of this star formation would be extended and resolved in Chandra.

It is interesting that Mrk 739W has not been detected as an AGN using optical emission line spectroscopy. Noguchi et al. (2010) found that optical emission line studies are biased against "buried AGNs" that have a small scattering fraction or a small amount of narrow-line region gas. AGNs with a low ratio of [O iii] to hard X-ray luminosity ($L_{[\rm O\,\mathsc{iii}]}$/L_2–10 keV < 0.1) tend to be "buried AGNs." The $L_{[\rm O\,\mathsc{iii}]}$/L_2–10 keV = 0.008, consistent with a "buried AGN" and the lowest ratios found in their study. This finding is also consistent with a recent study that found that merging AGNs selected in the ultra-hard X-rays tend to have low $L_{[\rm O\,\mathsc{iii}]}$/L_{14–195 keV} ratios and are preferentially misclassified using optical line diagnostics (Koss et al. 2010).

For AGNs with low luminosity in the [O iii] line, nebular emission from star formation can overwhelm the AGN signature in optical emission line diagnostics. Schawinski et al. (2010) found that for $L_{[\rm O\,\mathsc{iii}]}$ = 10⁴⁰ erg s⁻¹, nearly 54% of star-forming galaxies with AGNs will be classified as star-forming or composites. The small value of $L_{[\rm O\,\mathsc{iii}]}$ = 7.5 × 10³⁹ erg s⁻¹ in Mrk 739W suggests that star formation is overwhelming the AGN photoionization signature.

CO velocity profiles can provide information on the dynamics of the molecular gas. The ¹²CO 3–2 and 2–1 spectra in Mrk 739 have almost identical shapes. Each spectrum has a narrow profile with FWHM = 94 ± 8 and 98 ± 6 km s⁻¹ for single-Gaussian fits to the CO (2–1) and CO (3–2) profiles, respectively. These profiles are significantly narrower than the CO (2–1) emission from NGC 6240 (Figure 5) and imply a nearly face-on orientation to any disk-like structure in this system.

Zoom In Zoom Out Reset image size

CO measurements also provide estimates of the amount of molecular gas. In Mrk 739, I_CO = 109 ± 33 and 169 ± 51 Jy km s⁻¹ for the 2–1 and 3–2 lines. Following Solomon et al. (1992), L'_CO = 10.9 × 10⁸ and 7.5 × 10⁸ K km/s pc⁻² for the 2–1 and 3–2 lines. Adopting α = 1.5–4 M_☉ (K km pc⁻²)⁻¹ for the conversion from CO luminosity to molecular hydrogen, we find log  M_☉(H₂) = 9.2–9.6, similar to the Milky Way (Sanders et al. 1984).

Measurements of radial velocities provide important insights about the dynamics of the merger. We use the Na i λλ5890, 5896 (Na D) absorption lines from stars and cold gas since narrow emission lines in AGNs often have blueshifts (Bertram et al. 2007). There is an offset of ≈40 km s⁻¹ between the two bulge components (8995 ± 15 km s⁻¹ for Mrk 739E and 8953 ± 15 km s⁻¹ for Mrk 739W). The CO data also show evidence of two components with the peak brightness temperatures similar to the radial velocities in the Na D absorption lines (Figure 5). When fit with Gaussians, the peaks are consistent with the Na D radial velocities (8921 ± 22 and 8980 ± 16 km s⁻¹ in CO 2–1 and 8956 ± 12 and 8993 ± 22 km s⁻¹). High resolution (1'') interferometric CO imaging of this system would provide evidence to confirm this picture.

There is also evidence of outflows in the narrow-line region of Mrk 739E. There is a 192 ± 22 km s⁻¹ blueshift in the [O iii] line and a 153 ± 25 km s⁻¹ in the lower ionization [O i] λ6300 line compared to the Na D absorption. This blueshift is consistent with other nearby QSOs which have an average [O iii] blueshift of −174 km s⁻¹ (Bertram et al. 2007). In Mrk 739W, there is no evidence of outflows in the narrow-line region.

Using Chandra and UV photometry and following Vasudevan et al. (2009), the bolometric luminosity is 1.0 × 10⁴⁵ erg s⁻¹ in Mrk 0739E (Figure 5). The extinction-corrected 2500 Å luminosity is log L_2500 Å = 43.7 ± 0.3. Using Hβ and continuum emission (Vestergaard & Peterson 2006), the black hole mass is log M_BH = 7.04 ± 0.4, giving an Eddington ratio of λ_Edd = 0.71, the highest among all the Swift BAT selected AGNs (Vasudevan et al. 2010). Our estimates are consistent with Ho et al. (2008) who find λ_Edd = 0.78 using only optical spectra and the same method to determine black hole mass. Uncertainties in intrinsic dust reddening as well as the inclination angle and spectral hardening parameter in the accretion disk model can lower the Eddington ratio at most 58% to λ_Edd = 0.30.

In Mrk 739W, the bolometric luminosity is 2×10⁴³ erg s⁻¹ using only the hard X-ray data with a bolometric correction factor of 22 from Vasudevan et al. (2009).