EDP Sciences
Press Release
Free Access
Issue
A&A
Volume 579, July 2015
Article Number L3
Number of page(s) 4
Section Letters
DOI https://doi.org/10.1051/0004-6361/201526282
Published online 25 June 2015

© ESO, 2015

1. Introduction

According to the current theory of hierarchical formation and evolution of structures, accretion events are believed to have an important role in the cosmological build up of stellar halos in elliptical galaxies (De Lucia & Blaizot 2007), and they are responsible for their growth at relatively low redshifts (z< 2; Oser et al. 2010). In dense environments accretion is even more dramatic, such that close to the dynamical centre of the cluster, central cluster galaxies are expected to have the majority of their stars accreted (Laporte et al. 2013; Cooper et al. 2014).

On the observational front, the outer stellar envelopes are observed to increase in mass by a factor of ~4 since z = 2 (van Dokkum et al. 2010). The presence of an accreted component is usually identified as an excess of light over the extrapolation of the galaxy’s inner profile (Zibetti et al. 2005; Gonzalez et al. 2007; D’Souza et al. 2014), or by high Sersic indexes (n> 4; Kormendy et al. 2009). Observations of blue colour gradients from the centres of early-type galaxies towards their outskirts (Peletier et al. 1990; Liu et al. 2005; Rudick et al. 2010), mainly attributed to a gradient in metallicty (Tamura et al. 2000; Loubser & Sánchez-Blázquez 2012; Montes et al. 2014), are also in agreement with a change in stellar properties driven by the accretion of smaller systems. Records of accretion events are also revealed in the form of spatially extended low-surface brightness features in the outskirts of stellar halos (Mihos et al. 2005; van Dokkum et al. 2014; Duc et al. 2015). These “substructures” are not in a phase-mixed equilibrium in the host galaxy potential and, therefore, can in principle be traced as kinematic features in velocity phase-space.

The galaxy M 87 is one of the nearest central objects (at an adopted distance of 14.5 Mpc; Ciardullo et al. 2002; Longobardi et al. 2015), close to the dynamical centre of the Virgo cluster (Binggeli et al. 1987). It is considered a type-cD galaxy (Weil et al. 1997; Kormendy et al. 2009) with an extended stellar envelope that reaches ~150 kpc in radius. A blue colour gradient towards its outskirts (Rudick et al. 2010) has been interpreted as age and metallicity gradients (Liu et al. 2005; Montes et al. 2014), consistent with a late build-up of its halo. The orbital properties of globular clusters (GCs; Agnello et al. 2014) and ultra compact dwarfs (Zhang et al. 2015) also favour accretion onto the halo. This galaxy has been the target of several deep imaging surveys (Mihos et al. 2005; Rudick et al. 2010), and its close proximity has made it possible to identify hundreds of planetary nebulas (PNs) and to measure their line-of-sight velocities (LOSVs; Doherty et al. 2009; Longobardi et al. 2015).

In this letter we use the synergy between PN kinematics and deep imaging to identify an on-going accretion event in the outer halo of M 87, which we find to be a non-negligible perturbation of the galaxy properties at the distances where it is traced.

thumbnail Fig. 1

Top left: projected phase space, VLOS vs. major axis distance (R), for all spectroscopically confirmed PNs (black asterisks) in the halo of M 87. Red lines separate the elliptical annuli analysed to isolate cold components associated to the substructure. Top-right: histograms of the LOSVD in the three elliptical annuli. In each panel, the blue lines show the best-fit model computed as a combination of three Gaussians. Black dashed lines show the relative contribution of each component to the LOSVD, with parameters for the cold components given in the plot. Bottom left: as in the top left panel, however, the green circles and magenta diamonds show the PNs associated with the cold secondary peaks in the LOSVD. Orange squares show a kinematically selected GC substructure from Romanowsky et al. (2012). Bottom right: probability that a PN is drawn from the halo component (dark grey area) or from the chevron (green, magenta areas). Stars represent PN probabilities at their measured VLOS.

Open with DEXTER

2. Kinematic evidence for an accretion event in M 87

We acquired kinematic data from the FLAMES/VLT spectroscopic survey for a large sample of PNs in the outer regions of M 87 (Longobardi et al. 2013, 2015). The total sample consists of 254 objects classified as M 87 halo PNs and 44 intracluster PNs, for which we have obtained LOSVs with an estimated median velocity accuracy of 4.2 km s-1. We concentrate here on the M 87 halo PNs which cover a range of radii from ~15−150 kpc.

We see a notable chevron (or “V” shape) structure in the projected phase-space of the PN sample, as shown in Fig. 1 (top left). To isolate this kinematical substructure we utilise a three-component Gaussian mixture model to identify high-density, narrow features on top of a broader distribution. We note that there is not enough data to statistically favour this model over simpler models with BIC or AIC (Bayesian/Akaike Information Criteria; Liddle 2007); however, the substructure is visually indicated and we confirm it with photometry in Sect. 4. A brief description of the technique is given in the following paragraph; for more details see Pedregosa et al. (2011).

A Gaussian mixture model (GMM) is a probabilistic model, which assumes that a distribution of points can be described as a linear combination of K independent Gaussian probability density functions (PDFs), or components, expressed by (1)where, x is a data vector (here the LOSVs), Pk is the mixture weight that satisfies the conditions 0 ≤ Pk ≤ 1, and , and p(x | μk,σk) are the individual Gaussian PDFs, with mean μk, and dispersion σk. The GMM classifier implements the expectation-maximization (EM) algorithm, i.e. an iterative process that continuously updates the PDF parameters until convergence is reached. At the end of the EM procedure, the posterior probabilities, γk(x) for a data value to belong to each of the k Gaussian components are returned. These are described by (2)To apply the GMM to our LOSV distribution (LOVSD), we bin the PN M 87 halo sample into seven elliptical annuli, or stripes in phase space, covering the entire PN velocity phase space. The LOSVD in each annulus is analysed as a combination of three Gaussians, where the centres (μk), widths (σk), and weights (Pk) are treated as free parameters in the EM algorithm, and have uncertainties , , and , with S = [ ∑ nγk(xn) ] (MacKay 2003).

thumbnail Fig. 2

Spatial and colour distribution associated with the kinematic substructure identified in the phase space of the M 87 halo PNs. Left panel: V-band image of a 1.6×1.6 deg2 centred on M 87 from Mihos et al. (in prep.). Full circles, and diamonds indicate the spatial position of the M 87 halo PNs in the chevron substructure. Magenta and green colours indicate PN LOSVs above and below VLOS = 1254 km s-1, the LOSV at the end of the chevron. Central panel: unsharp masked image of M 87 median binned to enhance faint structures. The crown-shaped substructure is visible at distance of 800′′1200′′(~60−90 kpc) along the major axis, NW of M 87. Details are given in Sect. 3. The blue line measures 90 kpc. Right panel: (BV) colour image of M 87 from Mihos et al. (in prep.) with chevron PNs overplotted (white dots). The dashed ellipse indicates the isophote at a major axis distance of 1200′′. The crown is found in a region where the (BV) colour is on average 0.8, bluer than on the minor axis.

Open with DEXTER

We find cold components in three out of seven elliptical bins, for which we show the histogram of the data, along with the best-fit GMM and reduced χ2 in Fig. 1 (top-right panel). We also plot (bottom-left panel) the LOSV phase space for the 254 PNs in the halo of M 87: black crosses represent PNs of the smooth halo LOSVD of M 87, while magenta diamonds and green dots are PNs that have a higher probability (see Eq. (2)) of belonging to the chevron. Finally, we show the probability that a given PN is drawn from each of the components as a function of its velocity (bottom-right panel). The GMM assigned a total of 54 PNs to the chevron substructure, which covers 700 (~50 kpc) for major axis distances 500″ <R< 1200″. The separation, ΔV, between the two peaks of the cold components becomes smaller at larger distances. For the three elliptical bins, it is ΔV = 654.5±40.1, 246.4±20.5, and 153.3±20.6 km s-1, respectively. At R ~ 1200″ (~90 kpc) the width of the chevron goes to zero with LOSVs close to the galaxy’s systemic velocity (Vsys = 1275 km s-1; Longobardi et al. 2015)1. The PNs on the arms of the chevron are seen on both the northern and southern sides of the galaxy, as shown in Fig. 2 (see Sect. 3). The broad Gaussian, with average mean velocity ~1290 km s-1 and dispersion ~320 km s-1 in the three bins, traces the M 87 halo.

The search for kinematic features in the phase space of GCs has resulted in the discovery of a similar chevron-like structure (Romanowsky et al. 2012), shown in Fig. 1 as orange squares. Though the morphology in the phase space is similar, it differs in a number of physical properties: the width goes to zero at RGC ~ 1500″ with VLOS,GC = 1307 km s-1, versus R ~ 1200″ and VLOS ~ 1250±21 km s-1 for the PNs. Moreover, the 27 chevron GCs show a very different spatial distribution with the highest density of points on the NE photometric minor axis (Romanowsky et al. 2012; D’Abrusco et al. 2015), and few GCs near the crown substructure traced by the PNs.

3. Localising the substructure with deep imaging

In Fig. 2 (left), we show the position of the chevron PNs overplotted on 1.6×1.6 deg2V-band image of M 87, with an estimated surface brightness limit of μV = 28.5 mag arcsec-2 (Mihos et al., in prep.). Here we see the large spatial extent of the substructure associated with the chevron PNs. Because of its shape on the image, we refer to it as the “veil of M 87”. Now we are interested whether this feature is also visible in the optical light.

To this end, we constructed an unsharped masked image, which is the difference between the original image and a smoothed image. We utilised the IRAF task fmedian to smooth the original image using a window with a size of 1450″×1450″. This window size was chosen so that it contained the large scale extension of the substructure. Looking at the highest concentration of PNs, in the NW region of M 87, it can be seen to extend over many hundreds of arcseconds (~800″). Thus, the adopted box size is ~1.8 times larger than the long side of the feature.

The results of the unsharped masking can be seen in Fig. 2 (central panel), where the high frequency structures are now clearly visible. A previously unknown debris structure, with a crown-like shape, can be seen on top of M 87 at the NW side. We refer to this as the “crown” of M 87’s veil. It has a characteristic width of ~300″, an extension of ~800″, and is almost perpendicular to M 87’s photometric major axis. Just as the PN spatial distribution showed, the edge of this feature is found at R ~ 1200″(~ 90 kpc).

Over the same major axis distances at which the substructure is located, we also observe variations in the M 87 ellipticity profile. Between 300″ <R< 800″ the ellipticity increases and then flattens to a value of e ~ 0.43 for R> 800″ (Kormendy et al. 2009). In particular, the region at which the gradient flattens reflects the crown-like overdensity.

4. Physical properties of the accreted satellite

4.1. Luminosity and α-parameter

To understand the physical origin of this structure, we compute its total luminosity. We calculate it in a region of ~800″×300″, after the subtraction of the local background, determined using a region photometry method (for more details see Rudick et al. 2010). We find a total luminosity in the range Lcrown = 3.7±0.9×108 L,V. When compared to the luminosity determined in the same region from the original image, L ~ 5.9×108 L,V, it is clear that at these distances the substructure represents a significant fraction, ~60%, of the total light.

In the region of the crown, we count NPN,crown = 12±3 PNs, while we find a total of NPN,chevron = 54±7 PNs associated with the entire chevron (see Sect. 2). When correcting these numbers for incompleteness factors as in Longobardi et al. (2015), they become 19 and 142, respectively. Hence, by scaling Lcrown to the total number of PNs associated with the chevron, we obtain the total luminosity associated with the progenitor of the M 87 veil to be L ~ 2.8±1.0×109 L,V.

The total number of PNs is proportional to the total bolometric luminosity of the parent stellar population, and the proportionality is quantified with the luminosity-specific PN density, or, α-parameter (Buzzoni et al. 2006). Utilising the computed luminosity, and the completeness-corrected estimate for NPN,chevron, we can calculate the α-parameter for the progenitor of the substructure. Considering that the typical probability that NPN,chevron PN belong to the chevron is ~0.7 (Fig. 1), we obtain α2.5 = 1.8±0.7×10-8NPN. Here α2.5 is 2.5 mag down the luminosity function as in Longobardi et al. (2015), and we have assumed a bolometric correction for the V-band of BCV = 0.85 (Buzzoni et al. 2006) and BC = 0.07 for the Sun.

4.2. Colour and mass

In Fig. 2 (right panel) we show the chevron PNs overplotted on the BV colour image of M 87, which combines the V-band data (see Sect. 3) with deep B-band photometry with a surface brightness limit of μB = 29 mag arcsec-2. Close to R ~ 1200″ the colour shows an azimuthal variance such that along the photometric minor axis the measured values are redder (Mihos et al., in prep.). This feature correlates with the spatial number density of the chevron PNs, showing a deficit in number along the photometric minor axis. This suggests that the bluer regions are the result of the accreted material on top the light from M 87’s halo. In particular, the crown structure is measured to have integrated colour (BV) = 0.76±0.05.

The (BV) colour is a good estimator of the mass-to-light-ratio, Υ, of the underlying stellar population. Adopting for (BV) = 0.76 (McGaugh & Schombert 2014), the total stellar mass associated with the disrupted galaxy is M = 6.4±2.3×109 M.

From the distribution and velocities of chevron PNs in Fig. 2, a possible interpretation of the satellite orbit could be that it was first disrupted entering M 87 from the south (along the green dots), with the debris then moving up north, turning around in the crown region, and coming back down on both sides across M 87 (the veil, magenta dots). The velocities would then imply that the northern side of M 87 is closer to the observer. The dynamical time for such an orbit is of the order ≲1 Gyr (Weil et al. 1997).

5. Summary and conclusion

In this Letter we have presented kinematic and photometric evidence for an accretion event in the halo of the cD galaxy M 87. This event is traced by PNs whose velocity phase space shows a distinct chevron-like feature, which is a result of the incomplete phase-space mixing of a disrupted galaxy. At major axis distances of R ~ 60−90 kpc, where the width of the chevron goes to zero, a deep optical image shows the presence of a crown-like substructure that contributes ≳60% of the total light in this area.

The “crown of M 87’s veil” is the densest part of the entire substructure, which covers ~50 kpc along the major axis. In this region, a radial variation in M 87’s ellipticity profile is also observed. Looking at the spatial distribution of all the chevron PNs, it traces the azimuthal variation observed in the colour of M 87, showing a deficit in the number of tracers along the photometric minor axis where the galaxy is redder, and a higher fraction where the substructure is strongest and the colour is bluer.

We determined several physical properties of the disrupted satellite: a total luminosity of L = 2.8±1.0×109 L,V, colour (BV) = 0.76±0.05, and total stellar mass of M = 6.4±2.3×109 M. The inferred value for the α-parameter is α = 1.8±0.7×10-8NPN. The similar colours of the accreted satellite and ICL suggest that the cD halo of M 87 is presently growing through the accretion of similar star-forming systems as those that originate the diffuse IC component.

The evidence for on-going accretion in the outer halo of M 87 is consistent with the observed size growth of giant elliptical galaxies and with predictions by theory. The presence of the newly discovered substructure within the halo of M 87 demonstrates that beyond a distance of ~60 kpc its halo is still assembling.


1

The stability of the fitted parameters and the measured distance of the chevron edge were tested with 100 GMM runs for different mock data sets and initialisation values.

Acknowledgments

A.L. is grateful to J. Elliott for helpful comments that improved the manuscript. J.C.M. is supported by NSF grant 1108964.

References

All Figures

thumbnail Fig. 1

Top left: projected phase space, VLOS vs. major axis distance (R), for all spectroscopically confirmed PNs (black asterisks) in the halo of M 87. Red lines separate the elliptical annuli analysed to isolate cold components associated to the substructure. Top-right: histograms of the LOSVD in the three elliptical annuli. In each panel, the blue lines show the best-fit model computed as a combination of three Gaussians. Black dashed lines show the relative contribution of each component to the LOSVD, with parameters for the cold components given in the plot. Bottom left: as in the top left panel, however, the green circles and magenta diamonds show the PNs associated with the cold secondary peaks in the LOSVD. Orange squares show a kinematically selected GC substructure from Romanowsky et al. (2012). Bottom right: probability that a PN is drawn from the halo component (dark grey area) or from the chevron (green, magenta areas). Stars represent PN probabilities at their measured VLOS.

Open with DEXTER
In the text
thumbnail Fig. 2

Spatial and colour distribution associated with the kinematic substructure identified in the phase space of the M 87 halo PNs. Left panel: V-band image of a 1.6×1.6 deg2 centred on M 87 from Mihos et al. (in prep.). Full circles, and diamonds indicate the spatial position of the M 87 halo PNs in the chevron substructure. Magenta and green colours indicate PN LOSVs above and below VLOS = 1254 km s-1, the LOSV at the end of the chevron. Central panel: unsharp masked image of M 87 median binned to enhance faint structures. The crown-shaped substructure is visible at distance of 800′′1200′′(~60−90 kpc) along the major axis, NW of M 87. Details are given in Sect. 3. The blue line measures 90 kpc. Right panel: (BV) colour image of M 87 from Mihos et al. (in prep.) with chevron PNs overplotted (white dots). The dashed ellipse indicates the isophote at a major axis distance of 1200′′. The crown is found in a region where the (BV) colour is on average 0.8, bluer than on the minor axis.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.