SDSS J085431.18+173730.5: THE FIRST COMPACT ELLIPTICAL GALAXY HOSTING AN ACTIVE NUCLEUS

We primarily made use of the SDSS-III (Ahn et al. 2012) imaging data and catalog to explore the parameter space in searching for compact objects that we are interested in. We present the optical images of cE_AGN in Figure 1, a $g-r-i$ color composite taken from the SDSS skyserver at the top and a Canada–France–Hawaii Telescope (CFHT) i-band image at the bottom. We find that the SDSS image quality is rather poor with an average seeing in r-band of ∼1''. The SDSS measured half-light radius of cE_AGN is 198. Therefore, to avoid any uncertainty in the derived structural parameters due to the resolution limit, we searched for higher-quality images in other publicly available archival imaging databases. We found higher-quality imaging data in the CFHT³ archive, with an average seeing of 06, which was observed in the i-band.

Zoom In Zoom Out Reset image size

We acquired the pre-processed, level 2 calibration fits files that include dark current subtraction, flat normalization, and the magnitude zero-point supplemented in the header. The observations were made as a part of the CFHT Very Wide survey. With the seeing of 06 and a pixel scale of 018/pix, CFHT data provide significantly higher image quality compared to the SDSS observations. However, note that the CFHT MegaCam images are known to have severe issues with sky background subtraction due to the mosaic pattern of multiple CCD chips. Thanks to its compactness, our object extends entirely in one chip, thus sky background subtraction becomes straightforward.

As we can see in Figure 1, a background (z = 0.105) galaxy is located next to cE_AGN. We removed the background contamination by subtracting its light from the fits image using the IRAF⁴ ellipse and bmodel tasks. An unsharp masking technique has been applied to examine the presence of any substructural features such as spiral arm/bar or inner disk in cE_AGN. Finally, we used galfit (Peng et al. 2002) to carry out a 2D multicomponent structural analysis where we modeled the light distribution with several different possible combinations, i.e., one with a simple Sérsic function and the other with a double Sérsic function with or without a central Gaussian point source. For the double Sérsic function, we fixed the outer component to be an exponential, i.e., n = 1. A PSF image, required by galfit, was prepared by stacking the stars found in the same field of view. Foreground and background sources were masked by providing a mask image that we prepared manually.

With visual inspection of these residual images, we find that the use of a double Sérsic function provides a better fit; see the top right panel of Figure 2. The inner component has a Sérsic index n = 1.4 and an effective radius of 158. The outer component has an effective radius of 49. The flux ratio between the inner and outer components, i.e., bulge to disk ratio (B/T), is 2.36.

Zoom In Zoom Out Reset image size

Within the spatial resolution limit of the CFHT image, the presence of the central nucleus is not obvious; however, the addition of a central Gaussian component in the galfit modeling slightly improves the fit in the inner region. The Gaussian component light contribution is less than 1/50 of the bulge. A careful inspection of the unsharp-masked image and the model-subtracted residual images seems to indicate the presence of an inner bar/disk-like structure. The ellipticity profile along the major axis shows a mild change, and the maximum ellipticity coincides with the maximum change in position angle.

The global photometric parameters (e.g., half-light radius and total luminosity) are derived using a non-parametric approach. We follow a similar procedure as described in Janz & Lisker (2008) where the fluxes are summed up within a Petrosian aperture. The K-band luminosity is derived from the 2MASS⁵ archival images. The bulge luminosity is derived using the inner and outer component flux ratio. Since the detailed decomposition has been done in a CFHT i-band image, here we assume that the stellar population in cE_AGN is homogeneous. The results of photometry and structure modeling are listed in Table 1.

Table 1.  Global Properties of SDSS J085431.18+173730.5

Galaxy	Mk	Mr	Re	n	$\langle {\mu }_{r}\rangle$	g − r	z	Age	log(Z/Z$\odot$)	${M}_{\bullet }$	M*
	(mag)	(mag)	(pc)		(mag arcsec−2)	(mag)		(Gyr)	(dex)	log(M${}_{\odot }$)	log(M${}_{\odot }$)
cE_AGN	−20.03	−18.08	490	1.4	19.36	0.83	0.0140	9.95 ± 0.4	−0.12 ± 0.01	6.3	9.2
CGCG 036-042 (P14)	−20.78	−18.21	559	1.7	19.64	0.77	0.0068	7.15 ± 1.2	−0.18 ± 0.07	⋯	9.5

Note. The global photometric parameters—total brightness and half-light radius—are derived using Petrosian photometry where the galactic extinction is corrected using Schlafly & Finkbeiner (2011). We analyzed the SDSS-optical spectroscopy to derive stellar population parameters and BH mass; see the text.

Download table as:  ASCIITypeset image

The SDSS spectroscopic observation of cE_AGN was done as part of the BOSS⁶ survey. The redshift measured by the SDSS pipeline is z = 0.0140. We exploit the optical spectrum of cE_AGN twofold. First, we aim to derive the properties of the underlying old stellar population using the information that is available in absorption features. Second, we measure the mass of the central BH by analyzing the stellar continuum subtracted emission lines. For this, we first fitted the observed galaxy spectrum in a wavelength range from 4000 to 7000 Å with the simple stellar population (SSP) models from Vazdekis et al. (2010). In doing so, prominent emission-line regions were masked. For this purpose, we used a publicly available full spectrum fitting tool ULySS⁷ (Koleva et al. 2009). An additive continuum of third order was added to account for a non-thermal continuum while matching the galaxy and SSP spectra. The quality of fit is shown in the lower panel of Figure 3, where the observed spectrum matches within 5% of the flux, excluding the masked emission-line regions. The best-fit SSP parameters are age = 9.95 Gyr and log(Z/Z${}_{\odot }$) = −0.12 dex.

Zoom In Zoom Out Reset image size

Finally, we reanalyzed the residual spectrum that mainly contains emission lines (Figure 3, lower panel). The emission lines were de-blended into a narrow and a broad component by fitting multi-Gaussian models. A single Gaussian was used to model the narrow component of Hα, assuming the narrow component has the same profile and redshift as the [N ii] doublet as shown in the inset of Figure 3. However, for the [O iii] doublet (λ 4959 and λ 5007) and the narrow component of Hβ, we adopt double Gaussians to account for the blue wing (Greene & Ho 2005). For the broad components, we use multiple Gaussians, which yield FWHMs of 1550 and 2400 km s⁻¹ for the broad Hβ and Hα components, respectively. Finally, we measured Hα and Hβ narrow and broad emission fluxes from the best-fit narrow and broad component lines, respectively.

Having carefully measured the Hα and Hβ flux from the stellar continuum subtracted spectrum, the central black hole mass is measured by using the calibration provided by Equation (4) in Ho & Kim (2015), where we replace Hβ FWHM with Hα FWHM and L5100 is converted from the L(Hα), using the conversion factor in Greene & Ho (2005). The different normalization factors f are assumed for classical bulge and pseudo-bulge galaxies. With a Sérsic index n = 1.4, cE AGN may not be either a de Vaucouleurs bulge (n  =  4) or an exponential disk (n  =  1). Therefore, we used the scaling factor that was derived from combining both types of galaxies. The derived BH mass is $2.1\times {10}^{6}{M}_{\odot }$. The uncertainty for the BH mass measurement is approximately 0.4 dex, which is mainly due to the uncertainty in the scaling factor.

AGNs are not common in low-mass early-type galaxies (dE or cE classes). According to common knowledge, these galaxies are devoid of gas and have less of a chance of acquiring gas from external sources as massive galaxies do; thus, there is no fuel left to trigger the nuclear activity. Nevertheless, the number of detections of BHs in low-mass galaxies has been increased in recent literature with the advent of large-scale surveys, high-precision instruments, and the sophisticated modeling of the host galaxy and BH kinematics (e.g., Greene & Ho 2004; Seth et al. 2014; den Brok et al. 2015).

Barth et al. (2004) identified a dwarf Seyfert 1 AGN candidate, POX 52, at redshift z = 0.021. They classified the host galaxy morphology to be a dwarf elliptical and reported a BH mass of 1.5 × 10⁵$\;{M}_{\odot }$. In the last few years, a number of searches has been devised to list dwarf galaxies with central massive BHs (Greene & Ho 2004, 2007; Reines et al. 2013; Moran et al. 2014). In particular, the study of Reines et al. (2013) presents the largest number of nearby dwarf galaxies that exhibit optical spectroscopic signatures of accreting BHs. Visually classifying these BH–host dwarf galaxies located within z < 0.02,⁸ we find that 13 out of 33 can be classified as dE or dS0. Only one, J122342.82+581446.4, has a broad Hα emission similar to that of cE_AGN. Reines et al. (2013) estimate its BH mass to be ∼10⁶$\;{M}_{\odot }$.

The majority of cEs are found in the vicinity of giant galaxies, e.g., M32, and many of them possess morphological features that might have originated from tidal stripping activity (Huxor et al. 2011). With the discovery of an isolated cE by Huxor et al. (2013), an alternative scenario, other than stripping, has gained much attention in recent literature (Dierickx et al. 2014; Paudel et al. 2014). cE_AGN is also located away from the nearest massive galaxy. Recently, Chilingarian & Zolotukhin (2015) proposed a flyby scenario where those isolated compact ellipticals may be tidally stripped systems that ran away from their hosts.

The discovery of a cE with an AGN signature might help us to explore another aspects of compact early-type galaxies. Bulge-dominated early-type galaxies follow a scaling relation between bulge mass and BH mass. Measuring the BH mass allows us to locate the position of cEs in this relation. We find nothing uncommon, as cE_AGN is placed near M32, following the overall relation between BH mass and bulge mass (see Figure 4, right). M32 is a Local Group cE located at the vicinity of M31, which has been an iconoclast example of a cE-type galaxy (Faber 1973). Based on X-ray and radio observations, the presence of nuclear activity has also been reported in M32 (Ho et al. 2003). It has been long argued that M32 is a stripped bulge of a disk galaxy (Faber 1973; Choi et al. 2002; Graham 2002)—but also see Dierickx et al. (2014). As a hypothesis proposed by Chilingarian & Zolotukhin (2015), the possibility that cE_AGN might also be a tidally stripped bulge that ran away from its hosts, perhaps from the NGC 2672 group, cannot be ruled out. NGC 2672 is a relatively massive galaxy group with a viral mass of ∼10¹³M_☉ (Ramella et al. 2002). It is expected that the member galaxies could be spread out as far as a couple of Mpc in projected distance (Sales et al. 2007). However, given the large physical separation between cE_AGN and NGC 2672, the stripping event might have happened either very early in time or at a fast speed with a very eccentric orbit. The small relative velocity to NGC 2672 might be explained by an orbit in the plane of the sky and cE_AGN being close to apogalacticon.

It is remarkable to see that while following the extension of the mass–size relation of massive early-type galaxies, cE_AGN also seems to follow the observed empirical relation between the bulge Sérsic index and the BH mass suggested by Graham & Driver (2007). Interestingly, a detailed decomposition of the light profile by Graham (2002) reveals that M32's light profile resembles an almost perfect bulge—exponential disk system where the bulge component has a Sérsic index of n = 1.5, which is in agreement with the relation between the Sérsic index and the BH mass. Considering the location in the scaling relations, we suggest cE_AGN is a low-luminosity extension of Es. It is, however, not sufficient to rule out the stripping scenario and to support the alternative scenario that assumes a similar evolutionary path as Es, formed possibly via mergers (e.g., Naab et al. 2007). Furthermore, we expect that a detailed study of the underlying stellar population and central kinematics from high-quality two-dimensional spectroscopy may allow us to test the above conclusions, which we plan to do for the coming observing semester.