The Extended Baryonic Halo of NGC 3923

Abstract

Galaxy halos and their globular cluster systems build up over time by the accretion of small satellites. We can learn about this process in detail by observing systems with ongoing accretion events and comparing the data with simulations. Elliptical shell galaxies are systems that are thought to be due to ongoing or recent minor mergers. We present preliminary results of an investigation of the baryonic halo—light profile, globular clusters, and shells/streams—of the shell galaxy NGC 3923 from deep Dark Energy Camera (DECam) g and i-band imaging. We present the 2D and radial distributions of the globular cluster candidates out to a projected radius of about 185 kpc, or $\sim 37R_e$, making this one of the most extended cluster systems studied. The total number of clusters implies a halo mass of ${\mathcal{M}}_h\sim 3\times {10}^{13}$ M${}_{\odot }$. Previous studies had identified between 22 and 42 shells, making NGC 3923 the system with the largest number of shells. We identify 23 strong shells and 11 that are uncertain. Future work will measure the halo mass and mass profile from the radial distributions of the shell, N-body models, and line-of-sight velocity distribution (LOSVD) measurements of the shells using the Multi Unit Spectroscopic Explorer (MUSE).

1. Introduction

Galaxies at high redshift are much more compact at a given mass than in the local universe, implying that galaxies and their halos grow with time due to the accretion of mostly lower-mass galaxies [1,2,3]. Therefore, we can learn about the details of this process by observing nearby systems that are in the process of merging or accreting. Merger remnants and galaxies with tidal streams are good candidates. Elliptical shell galaxies are other promising environments for studying the build-up of halos. Shell galaxies are surrounded by interleaved “umbrellas” or shells of stellar material that match the appearance of structures created during the accretion of low-mass satellites on near-radial orbits in simulations (Figure 1) [4].

In this contribution, we present preliminary results of a study of the globular cluster system and shells of the elliptical shell galaxy NGC 3923. NGC 3923 is one of the most studied shell galaxies because it has the largest number of detected shells. Different studies have identified between 22 and 42 shells [5,6]. Therefore, new, deeper data will help determine the true number of shells and look for additional shells at larger radii. Mergers will also bring in new globular clusters (GCs) so the distribution and total number of GCs can be used to estimate the halo mass. We adopt a distance to NGC 3923 of $21.3\pm 1.4$ Mpc (${(m-M)}_0=31.64\pm 0.14$) [7]. Using a total apparent magnitude of $V_T^0=9.69$ [8] results in an absolute magnitude of $M_V=-21.95$.

2. Observations and Reduction

NGC 3923 was observed on 22–23 April 2015 using Dark Energy Camera (DECam) on the Blanco 4-m telescope at the Cerro Tololo Inter-American Observatory (CTIO) [10]. The conditions were photometric with seeing of about 1 arcsec. Total integration times of 10,400 sec in $g^{\prime }$-band and 10,800 sec in $i^{\prime }$-band were obtained. A Fermat spiral dither pattern was used to maximize uniformity across the gaps between the 62 detectors. Sky subtraction is critical since we are searching for low-surface brightness features on scales of a degree on the sky. Therefore, exposures alternated between the NGC 3923 field and five different sky fields (Figure 2).

These observations were done as part of the larger Neighborhood Watch survey of the baryonic structures within about 20 Mpc using DECam $u^{\prime }{g}^{\prime }{i}^{\prime }$ and VIRCAM J and $K_s$ imaging. The goals are to use globular clusters, dwarf galaxies, and shells to study stellar populations and the formation of structures in the local universe. Targets include the Fornax Cluster and the CenA, Sombrero, NGC 2997, NGC 6744, and NGC 3115 groups.

The sky-subtraction algorithms used by the DECam community pipeline [11] are not designed for extended objects and over-subtract the sky in the vicinity of large, bright galaxies. Therefore, we reduced the data using the THELI reduction software [12] that used the images of the sky fields to create the background model. This eliminated the problem of sky over-subtraction.

3. Results

3.1. The Globular Cluster System

At at distance of 21.3 Mpc, GCs are unresolved from the ground in 1 arcsec seeing. Therefore, we identified candidate GCs by selecting objects consistent with being point sources and then doing statistical background subtraction. The candidate GCs from [13]—taken in much better image quality using GMOS-S —were used as a “training” set to guide our selection. Source detection and photometry were done using SExtractor and PSFEx [14,15]. Principal component analysis (PCA) was applied to the photometry parameters in order to select point sources (Figure 3a). The eigenvectors are dominated by the FLUX_RADIUS and SPREAD_MODEL parameters. We then applied cuts in color-magnitude space as shown in Figure 3b. Finally, we fit a plane to the source density of objects more than 30 arcmin from NGC 3923, and statistically (randomly) subtracted this background.

This process resulted in the selection of 500 candidate GCs detected by DECam within a projected radius of 0.5 degree, which corresponds to 185 kpc, about 37 $R_e$ (Figure 4a). This makes the NGC 3923 globular cluster system (GCS) one of the most extended known, comparable to the GCSs of the Milky Way and M31 [16]. This is mainly due to the very wide-field imaging used here. As other galaxies are observed on this scale, we expect that most GCSs of massive galaxies will be found to have similar extents.

In the GCSs of other elliptical galaxies, the “blue” and “red” GCs usually have different radial distributions [13]. We explore the radial distributions of the NGC 3923 GCs by dividing the candidates at $(g^{\prime }-i^{\prime })=0.85$, the minimum between the blue and red color peaks from [13], and plotting the radial distributions in Figure 4b. For this and further analysis, we include 249 GCs from [13] that are detected closer to the center of the galaxy than our current method can find. As found in previous work, the distribution of the blue GCs is more extended than that of the red GCs. The colors of the blue GCs are consistent with the colors of GCs in dwarf galaxies [17,18], and spectroscopic abundances show that the blue clusters are more metal-poor than the red clusters [19]. This is evidence that the two populations are distinct and may have had different formation histories. It is commonly thought that many of the blue GCs may have been accreted from merging dwarf galaxies, but the orbits of outer GCs can be unexpectedly tangential (see [19]), so more kinematic information is needed to test this hypothesis.

The total number of GCs associated with NGC 3923 can be estimated by integrating over the globular cluster luminosity function (GCLF). Following [20], we assume that the GCLF is a Gaussian with a peak at $M_g^0=-7.2$ or $g^0=24.44$. Completeness tests have not been performed yet, so the faint magnitude limit ($g^{\prime }=23.5$) was chosen to be well below the magnitude where incompleteness is significant. Fitting the observed luminosity function gives a reasonable Gaussian sigma of 1.4 mag and 3142 total clusters. This gives a globular cluster specific frequency of $S_N=N_{\mathrm{GC}}{10}^{0.4(M_V+15)}=5.2$, similar to previous results [13].

Harris et al. (2017) [21] has shown that the total number of GCs is proportional to the total halo mass. With this method, the 3142 GCs in NGC 3921 imply a halo mass of ${\mathcal{M}}_h\sim 3.5\times {10}^{13}$ M${}_{\odot }$. This compares to a dynamical mass of ${\mathcal{M}}_{dyn}\sim 2\times {10}^{12}$ M${}_{\odot }$ at 30 kpc [22].

3.2. Shells

The shells around NGC 3923 were identified by eye after removing the underlying galaxy light. The galaxy subtraction was done using a two-Sersic model with Imfit image-fitting code [23]. We also enhanced the shell edges in the inner regions by applying an erosion filter edge detection algorithm [5]. The results are given in Figure 5. We have identified 23 strong shells in common with [5]. Eleven shells are classified as uncertain, and eight shells from [5] are not detected. No new shells have been detected yet, but analysis is ongoing.

The shells from a single minor merger should have an alternating, or interleaved, pattern relative to the center of the galaxy. The pattern for the 23 strong shells that we identified in NGC 3923 is shown in Figure 6. The pattern does not alternate as expected. In future work we will look for multiple patterns and additional shells. The spacing of the shells can then be used to constrain the potential and the enclosed mass [24].

4. Summary and Future Work

In this work we have identified and characterized the GCs and shells in the halo of NGC 3923. A total of 500 GC candidates are detected in deep DECam imaging out to a projected radius of 185 kpc. This implies a total GC population of about 3100 and a halo mass of ${\mathcal{M}}_h\sim 3\times {10}^{13}$ M${}_{\odot }$. In continuing work, we are trying to detect fainter sources and characterize the photometric completeness. Twenty-three interleaved shells are confirmed, and so far we have not detected previously uncatalogued shells. We are currently applying edge-detection and other algorithms that may be useful for detecting, defining, and characterizing the shells.

Future analysis will focus on deriving the mass profile of NGC 3923 out to a projected radius of about 130 kpc. We will use N-body simulations to reproduce the positions and morphologies of the shells and the velocities of satellite galaxies. This will put constraints on the shape and profile of the potential. Measuring the kinematics of the GCs will also be extremely important. Finally, we will be using the Multi Unit Spectroscopic Explorer (MUSE) instrument on the European Southern Observatory Very Large Telescope (ESO VLT) to measure the line-of-sight velocity distribution (LOSVD) of the stars in the low-surface brightness shells themselves ($B\sim 27$ mag arcsec${}^2$). Fitting the width of the LOSVD with radius from the shell edge can give the gravitational acceleration and the enclosed mass at the shell edge [4].

Elliptical shell galaxies are excellent systems for studying the extent of galaxy halos and the processes that form them. They allow us to use the multiple constraints of globular clusters, shells, and satellite companions to constrain models of their structure and formation.