Origins of the Evil Eye: M64's Stellar Halo Reveals the Recent Accretion of an SMC-mass Satellite

Merging has long been thought to be an important mode of evolution for galaxies like the Milky Way (MW; e.g., Toomre & Toomre 1972; Barnes & Hernquist 1996), complemented by "secular" evolution of the disk (e.g., Kormendy & Kennicutt 2004) and direct gas accretion from the intergalactic medium (IGM; e.g., Tumlinson et al. 2017). Disentangling these processes to understand which present-day galaxy properties are a result of the merger history is a challenging task as the impact of mergers (especially minor ones) on an MW-mass galaxy's disk can be subtle. The nearby galaxy M64 (also NGC 4826) is a prime example. Long nicknamed the "Evil Eye" galaxy due to the presence of an unusual, dusty inner gas disk (Figure 1, top left), M64 is unique in the Local Volume in hosting an outer h i disk that counter-rotates (Braun et al. 1992) with respect to its stellar disk (see Figure 1, top right). The collision between these two disks, with its high relative velocity, is thought to have produced an enhanced star formation rate at their interface, which is likely responsible for M64's infamous visible dust lanes (Braun et al. 1992). This idea is supported by much more recent, larger studies of galaxies with dual counter-rotating gas disks (Cao et al. 2022).

Zoom In Zoom Out Reset image size

Since its discovery, the prevailing assumption has been that M64's counter-rotating gas disk was accreted during a recent merger or interaction (e.g., Rubin 1994). Yet, puzzlingly, M64's outer disk shows little sign of a recent interaction in either stellar kinematics (Rix et al. 1995) or deep ground-based imaging (Watkins et al. 2016; see also Figure 1). This prompted the early idea that M64 may have accreted its counter-rotating gas directly from the IGM (Rix et al. 1995). However, more recent speculation has suggested a "flyby" interaction within the last ∼1 Gyr with the diffuse, gas-rich dwarf galaxy Coma P may be responsible (Brunker et al. 2019). There has been evidence for M64 having a stellar halo, as first uncovered by Kang et al. (2020), who found a distinct, metal-poor stellar halo population in a single Hubble Space Telescope field along M64's major axis. Yet, with only a single field, it is completely ambiguous whether this population was accreted along with M64's counter-rotating gas, during, for example, an interaction with Coma P or an as-yet unknown progenitor or if it was deposited by more ancient accretions.

The approach of resolving bright red giant branch (RGB) stars in nearby galaxies with wide-field, ground-based imagers is ideally suited to this problem (e.g., Bailin et al. 2011; Ibata et al. 2014; Okamoto et al. 2015; Smercina et al. 2020, and others). The debris from any interaction between M64 and a galaxy massive enough to account for its counter-rotating gas, either as a flyby or a full disruption, should be easily visible as an overdensity of point sources with colors and magnitudes consistent with a population of RGB stars at the distance of M64.

In this Letter, we present Subaru Hyper Suprime-Cam (HSC) observations of M64's stellar halo, resolving its stellar populations in three pointings and reaching radii of ∼115 kpc. Throughout this Letter, we assume a distance to M64 of 4.4 Mpc (Tully et al. 2013) and a total stellar mass of M_⋆,Gal = 2.8 × 10¹⁰ M_⊙ (Querejeta et al. 2015). We find evidence for a spectacular shell feature and other tidal structures indicative of an ongoing, late-stage radial merger. Benchmarking against existing Hubble Space Telescope (HST) photometry, we estimate the stellar mass of the progenitor galaxy to be M_⋆ ≃ 5 × 10⁸ M_⊙, with a metallicity of [M/H] ≃−1—very similar to the mass and metallicity of the Small Magellanic Cloud (SMC). The mass of M64's counter-rotating outer gas disk, M_gas,CR ≃ 5 × 10⁸ M_⊙ (Walter et al. 2008), is also comparable to the gas mass of the SMC (Dickey et al. 2000), suggesting that the likeliest origin of M64's unique counter-rotating disk was a recent merger with a gas-rich satellite very similar to the SMC.

Subaru HSC observations were carried out in Subaru HSC's Queue Mode during the 2019A semester, through the Gemini–Subaru exchange program (PI: Smercina, 2019A-0283). Our survey of M64's stellar halo consists of three 15 HSC pointings: two northern fields (Fields 1 and 2) with imaging in g and r bands and a third southern field (Field 3) with imaging in g, r, and i bands, for a total of 5.2 hr of exposure time. These three fields cover an effective area equivalent to a circle of radius ∼100 kpc, centered on M64.

The data were reduced using the most up-to-date version of the HSC optical imaging pipeline (Bosch et al. 2018), which produces final magnitudes in the Sloan Digital Sky Survey (SDSS) filter system. Sources were detected separately in each available filter, and photometry was performed on the coadded stack of all available images. For each source, a clipped mean of deblended pixels is measured in an annulus that spans from 7 to 15 times the 68% enclosed light radius of the point spread function. This source-specific background measurement, normalized to the effective aperture area, is then subtracted from the source fluxes. All magnitudes were then corrected for Galactic extinction using the maps of Schlafly & Finkbeiner (2011). ¹³ Seeing conditions were excellent, yielding measured FWHM between 04 and 075 for the brightest sources in all three fields.

While the initial survey design provided three-filter coverage to equal depth across all three fields, poor weather on Maunakea in the 2019A semester substantially reduced the available observing time in the required good/excellent seeing conditions. The survey was, therefore, only partially completed, and Subaru does not carry over programs to subsequent semesters. Given this, throughout the paper, we base most quantitative inferences of M64's halo properties on our completed Field 3 observations due to the immense benefit of color–color information in selecting stellar sources from its three-band photometry (see Smercina et al. 2017, 2020 for further details). For Field 3, we conducted a suite of artificial star tests (ASTs) to characterize the photometric quality and completeness of the data. Artificial stars were drawn uniformly from the g − i versus i color–magnitude space, with 22.6 < i < 28.6 and 0 < g − i < 3.5. r-band magnitudes were then calculated using the g − r/r − i stellar locus (see Section 3). Artificial stars were then inserted uniformly across the field in 10 separate runs, with 87,800 stars per run, for a total of 878,000 artificial stars. Following Smercina et al. (2020), we consider an artificial star "recovered" by the pipeline if the detected source is within 03 of the input position and within 1 mag of the input magnitude in each filter. From these ASTs, we estimate general 50% source-completeness limits of g₀ = 27.44, r₀ = 26.69, and i₀ = 25.91 for Field 3.

In ground-based resolved star surveys at the photometric depths probed here, the separation of stars at the same line-of-sight distance as the host galaxy and more distant background galaxies, many of which appear as point sources in seeing-limited observations, is a primary impediment. Due to the variable coverage and depth across our survey footprint, we adopt two different sets of selection criteria for star–galaxy separation and identification of RGB candidates in M64's halo. For the initial analysis that includes the shallower Fields 1 and 2, we adopt a less stringent selection: sources <15 in size and lying within our visually identified RGB selection region in the g − r versus r color–magnitude diagram (CMD). We show a Hess CMD of all sources <15 in Figure 2 (top left). For Field 3, and our quantitative analyses of M64's halo, we adopt a stringent selection; to pass, sources must be <075 in size and must be no further from the g − r/r − i color locus (Figure 2, top center; see also Smercina et al. 2020) than the quadrature sum of their g − r and r − i uncertainties. RGB candidate stars are then selected in the g − i versus i CMD (Figure 2, top right).

Zoom In Zoom Out Reset image size

In Figure 2 (bottom) we show a spatial map of RGB star candidates selected using our less stringent criteria for all three fields. We have applied nearest-neighbor filtering to the map, such that only sources with a seventh-nearest neighbor within 40'' are displayed, to reduce the presence of the numerous background sources that contaminate this selection. ¹⁵ Remarkably, a very prominent substructure is visible in M64's southern halo, ∼30° counterclockwise from its major axis, extending out ∼45 kpc to a bright shell. On the opposing (northern) side, there is a much more diffuse but similarly shaped plume, which extends out to >100 kpc. These structures are characteristic of a radial merger, where stars follow a plunging orbit (e.g., Johnston et al. 2008; Amorisco 2015; Hendel & Johnston 2015). The difference in morphology between the northern and southern features suggests that the denser southern shell is the younger structure, with a still-distinct pileup of stars near the apocenter (similar to the simulations of Cooper et al. 2011).

Figure 3 shows a density map of selected RGB stars in all three fields, with color mapped to three bins of inferred photometric metallicity. The metallicity of each star was estimated using a grid of PARSEC isochrones (Bressan et al. 2012), with age of 10 Gyr and metallicities ranging from [M/H] = −2 to 0 in steps of Δ[M/H] = 0.05 dex. Visually, the southern shell and northern complement are reasonably high metallicity—[M/H] > −1.2. There does not appear to be a clear radial gradient in metallicity.

Zoom In Zoom Out Reset image size

In this section, we explore the properties of M64's radial merger remnant and its probable progenitor in greater detail. First, using existing HST data, we show that our Subaru HSC observations are probing accreted stars in M64's halo, with little contamination from the disk. We then estimate the mass and metallicity of the southern shell feature observed in Field 3. We use this insight to estimate the initial properties of the disrupted progenitor dwarf galaxy and discuss the results in the context of other nearby systems.

4.1. Mass and Metallicity of the Southern Shell

In Figure 4 we show the map of high-quality RGB stars, color coded by inferred photometric metallicity from our grid of PARSEC isochrones. We choose a selection region for the southern shell that encompasses all visibly structured debris (beginning at ∼10 kpc). We show this selection region overlaid on the RGB star map in Figure 4 (left), as well as the CMD for all good-quality stars within (top right). We estimate an average metallicity of [M/H] = −1 for this shell region.

Zoom In Zoom Out Reset image size

To estimate the mass, we first choose a region of identical 0.12 deg² area at the edge of Field 3 (∼100 kpc from M64) to estimate the background. We show the CMD of high-quality stars in this background region in Figure 4 (bottom right), to confirm that there is no visible RGB. Next, using the results of the ASTs (see Section 2), we estimate the total completeness using the stringent stellar selection within the shell region and correct the background-subtracted counts to a bulk number of expected RGB stars in the shell region. Last, using a [M/H] = −1 isochrone (with a Kroupa initial mass function and 10 Gyr age), we convert these RGB counts to an estimated total stellar mass in the southern shell of M_⋆,shell = 1.80 ± 0.54 × 10⁸ M_⊙. Following past works (e.g., Harmsen et al. 2017), we adopt a 30% uncertainty on all star-count and isochrone-derived stellar masses. ¹⁷

An important consideration for this analysis is that it has been well documented that radial mergers of the kind M64 is experiencing may pull in situ (i.e., disk) stars from the central galaxy during pericenter passages (e.g., Gómez et al. 2017; Laporte et al. 2018; Sanderson et al. 2018), complicating the interpretation of visible debris. To investigate this possibility in M64, we use archival HST observations from the PHANGS-HST survey (Lee et al. 2022). The CMD of stars detected in this single Advanced Camera for Surveys (ACS) field in M64's outer disk is shown in Figure 5 (top left). As discovered by Kang et al. (2020), M64's outer disk exhibits two distinct stellar populations: a metal-rich population, described well by a stellar isochrone with [M/H] = −0.3, and a metal-poor population described well by a stellar isochrone with [M/H] = −1.

Zoom In Zoom Out Reset image size

We convert the PHANGS-HST photometry to the SDSS g/i filters and "degrade" it to the Subaru HSC depth achieved in Field 3, using g- and i-band completeness functions (shown in Figure 5, top right) derived from our ASTs. We show the predicted CMD of these converted HST sources in Figure 5 (second panel from the left), including predicted detections (black) and nondetections (gray) at the depth of Field 3. The metal-rich disk population, visible in the HST CMD, is largely invisible at the depth of our Subaru observations due to the depth of the g-band observations and the red colors of RGB stars. The HST-identified disk stars are sufficiently red that they fall faintwards of our g-band limit, even at the tip of the RGB, while the metal-poor halo population is well recovered by our RGB selection region. We also show an observed CMD of detected Subaru sources in the same region as the HST observations (Figure 5, third panel from left), which matches the predicted Subaru-depth HST CMD well.

As final evidence that our Subaru observations are purely probing the accreted stars belonging to the shell progenitor, we compare radial mass density profiles derived from the metal-poor HST stars, metal-rich HST stars, and Subaru stars (Figure 5, bottom). For each of the HST populations (selected from the regions shown in Figure 5, top left), we calculate their densities in 0.5 kpc bins. As expected (e.g., Kang et al. 2020), the disk and halo populations exhibit very different profiles, with the disk following a steep profile such as observed by Watkins et al. (2016; shown in Figure 5, gray squares), and the halo stars following a shallower power-law profile. For the Subaru profile, we calculated the density in 1 kpc bins for radii <25 kpc and 1.5 kpc bins >25 kpc. We then background-subtracted and converted counts to stellar mass in each bin, following the method described above in this section. We used the ASTs to calculate completeness and correct counts separately for each radial bin. The RGB star-count-derived Subaru density profile matches the HST halo profile almost exactly and is distinct from the steeper profile of the metal-rich disk population.

In summary, the estimated stellar mass of 1.80 × 10⁸ M_⊙ within the shell appears to be entirely composed of stars accreted from the progenitor satellite.

4.2. Total Mass of the Progenitor

In this Letter, we have presented a first-ever view of the resolved stellar populations in the stellar halo of M64, using Subaru HSC. Detected RGB stars in M64's halo reveal a spectacular ongoing radial merger, including a diffuse northern plume and a dense southern shell. We estimate a stellar mass in the southern shell of M_⋆,shell = 1.80 ± 0.54 × 10⁸ M_⊙, with an average metallicity of [M/H] = −1. Using existing HST observations, which separate M64's metal-rich disk and metal-poor accreted populations, we determine that the stars detected in our Subaru survey are entirely consistent with the accreted population, with little in situ contamination from the disk. Scaling from the mass of the southern shell, we estimate a nearly equal stellar mass in the northern plume of M_⋆,plume = 1.42 ± 0.71 × 10⁸ M_⊙. Using the profile of inferred accreted stars in M64's inner disk, from existing HST observations, we estimate that another 2.13 ± 0.64 × 10⁸ M_⊙ of accreted stellar mass exists in M64's interior, giving a total stellar mass of the progenitor of M_⋆,prog ≃ 5 × 10⁸ M_⊙.

This progenitor satellite was, therefore, likely remarkably similar to the SMC in both stellar mass and metallicity. This is important for understanding the origins of M64's counter-rotating gas disk as its accreted gas component is very similar in mass (≃5 × 10⁸ M_⊙; Walter et al. 2008) to the gas content of the SMC (≃4 × 10⁸ M_⊙; Dickey et al. 2000). The morphology of the debris implies that the disrupting satellite experienced a last pericenter passage likely <1 Gyr ago (e.g., Johnston et al. 2008; Cooper et al. 2011), and the merger has been ongoing for the last several gigayears. This means that any flyby interaction between M64 and the distant Coma P (Brunker et al. 2019) would have occurred during M64's merger with the current progenitor. We therefore assert that the shell progenitor is the more likely donor of M64's counter-rotating gas as it is much more massive (though as noted by Brunker et al. 2019, Coma P is surprisingly gas rich). This makes sense in the visual context of M64's halo: the plume–shell structure suggests that the satellite's last pericenter passage occurred from northwest to southeast, which is counter to the rotational direction of M64's stars, at least in projection.

The likely origins of the Evil Eye's counter-rotating gas are at last revealed, after decades of curiosity. We suggest that its outer gas disk was accreted recently during a 52:1 merger with an SMC-mass galaxy and is now colliding with an existing inner gas disk, fueling a burst of star formation at the disk–disk interface and driving the visible dust lanes from which it earns its name. Future observational and theoretical studies will help test this idea. Resolved spectroscopic studies of the kinematics and abundance gradients of stars across the debris field, enabled by upcoming wide-field integral field unit instruments, would provide definitive evidence that the structures were accreted from the same progenitor. Moreover, studies using hydrodynamic simulations of similar galaxies in the Illustris and EAGLE suites have begun to explore the phenomenon of kinematically misaligned gas and stars (e.g., Starkenburg et al. 2019; Khoperskov et al. 2021; Lu et al. 2021; Casanueva et al. 2022). Results from these simulations may help inform the possible connection between M64's counter-rotating gas and the evolution of its recent merger. For example, it is currently unclear when the gas would have been accreted in our presented scenario.

While a fascinating case study of how mergers shape and drive the properties of nearby galaxies, M64's unique set of properties make it an exciting target for understanding stellar halos generally. Connecting the counter-rotating gas to a radial merger event would place a strong observational prior on the satellite's orbital properties, mainly that it must have been partially retrograde. Such a reliable prior on a halo progenitor's orbital properties has only been achieved for the MW (e.g., Helmi et al. 2018, which has a very different merger history) and otherwise requires model-dependent assumptions (e.g., Hammer et al. 2018, for M31). Efforts to reproduce M64's halo properties in models and galaxy formation simulations therefore has the potential to contribute more broadly to a better understanding of the buildup of stellar halos around galaxies.

A.S. was supported by NASA through grant No. GO-14610 from the Space Telescope Science Institute, which is operated by AURA, Inc., under NASA contract NAS 5-26555. E.F.B. was partly supported by the National Science Foundation through grant 2007065 and by the WFIRST Infrared Nearby Galaxies Survey (WINGS) collaboration through NASA grant NNG16PJ28C through subcontract from the University of Washington. A.M. gratefully acknowledges support by the ANID BASAL project FB210003, FONDECYT Regular grant 1212046, and funding from the Max Planck Society through a "PartnerGroup" grant.

Based on observations utilizing Pan-STARRS1 Survey. The Pan-STARRS1 Surveys (PS1) and the PS1 public science archive have been made possible through contributions by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, the Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation grant No. AST-1238877, the University of Maryland, Eotvos Lorand University (ELTE), the Los Alamos National Laboratory, and the Gordon and Betty Moore Foundation.

Based on observations obtained at the Subaru Observatory, which is operated by the National Astronomical Observatory of Japan, via the Gemini/Subaru Time Exchange Program. We thank the Subaru support staff—particularly Tsuyoshi Terai, Sakurako Okamoto, and Tom Winegar—for invaluable help in preparing the observations and navigating the Subaru Queue Mode.

The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.

Software: HSC Pipeline (Bosch et al. 2018), Matplotlib (Hunter 2007), NumPy (Oliphant 2006; Van Der Walt et al. 2011), Astropy (Astropy Collaboration et al. 2018), SciPy (Virtanen et al. 2020), SAOImage DS9 (Smithsonian Astrophysical Observatory 2000).