The SAMI Galaxy Survey: embedded discs and radial trends in outer dynamical support across the Hubble sequence

Due to longer dynamical timescales, the outskirts of galaxies are thought to retain and reflect recent accretion events (e.g. Hoffman et al., 2010; Wu et al., 2014). In contrast, deep potential wells efficiently funnel gas into the centres of galaxies enabling star formation. This has led to the expectation that the centres of early-type galaxies are primarily populated by stars formed in-situ, while the outskirts are dominated by accreted stars from past merger events. The observation of massive compact galaxies at high redshift (e.g. Daddi et al., 2005; Trujillo et al., 2006; van Dokkum et al., 2008) combined with negative metallicity gradients and a tight age-[Mg/Fe] relationship at low redshift (e.g. Martin-Navarro et al., 2018) suggest that the bulk of the inner in-situ component formed early and rapidly through dissipational processes. The outskirts form over a more extended period in discs (e.g. Driver et al., 2013) or are accreted through mergers with less massive companions (e.g. Oser et al., 2012; Huang et al., 2013). The so-called “two-phase galaxy formation” scenario begins with an early and abrupt phase of intense in-situ star formation leading to a dense proto-galaxy at high redshift followed by a more progressive phase of predominantly minor merging which mainly build up the outer stellar component (e.g. Zhao et al., 2003; Oser et al., 2010, 2012).

A radial transition in the luminosity profiles and stellar kinematics between the inner in-situ and outer accreted stellar components of galaxies is thus expected (e.g. Cooper et al., 2013; Foster et al., 2016; Pulsoni et al., 2017; Rodriguez-Gomez et al., 2017). The transition radius may depend on mass (e.g. Cooper et al., 2013; Pulsoni et al., 2017). Although other processes, such as stellar migration could blur the transition zone (e.g. El-Badry et al., 2016). Traditional integral field unit (IFU) studies have typically focused on the inner effective radius ($R_{e}$) of galaxies (e.g. Emsellem et al., 2007; Cappellari et al., 2011) and been unable to probe this transition. Notable progress has been achieved using the wide-field Mitchell Spectrograph (Raskutti et al., 2014; Boardman et al., 2017) and the use of wide-field multiple slit spectrographs as “proxy” IFUs (e.g. Proctor et al., 2009; Arnold et al., 2011). These studies have pushed 2D stellar kinematics in nearby galaxies out to $\sim 3R_{e}$, but sample sizes have been limited because they are observationally expensive.

Using sparsely sampled stellar kinematic data in the outskirts of nearby galaxies in the SAGES Legacy Unifying Globulars and GalaxieS (SLUGGS) Survey (Brodie et al., 2014), Proctor et al. (2009) showed that although NGC 821 and NGC 2768 have comparable spin (the so-called $\lambda$ parameter, e.g. Emsellem et al. 2007) at one effective radius ($R_{e}$), the two differ markedly beyond $1R_{e}$. Using planetary nebulae as kinematic tracers, Coccato et al. (2009) and Pulsoni et al. (2017) also identified massive early-type galaxies with a marked decrease in rotational support beyond $1R_{e}$. Arnold et al. (2011) found a significant fraction of elliptical galaxies with decreasing rotational support with radius. They interpreted this feature as an “embedded disc”, i.e. a rotationally supported disc-like structure embedded within a larger scale slowly or non-rotating structure. Using the SAURON spectrograph, Weijmans et al. (2009) confirmed the decreasing rotational support of the lenticular galaxy (S0) NGC 821 at large radii. However, Raskutti et al. (2014) and Boardman et al. (2017) did not find significant kinematic transitions in a sample of galaxies observed with the Mitchell Spectrograph. This discrepancy was attributed to differences in the sample selection of the respective studies.

Embedded discs and the bespoke two-phase kinematic transition have sometimes been used interchangeably to describe the same observed phenomenon (e.g. Arnold et al., 2014; Bellstedt et al., 2017). One can think of various formation pathways for embedded discs: e.g. major mergers (Querejeta et al., 2015a), gas dissipation/accretion (e.g. White & Rees, 1978) or the two-phase galaxy formation (Oser et al., 2012). On the one hand, embedded discs are only identifiable by their radial decline in rotation, while on the other hand the two-phase galaxy formation scenario could accomodate a range of radial kinematic transitions given the diversity of possible merger orbital configurations and histories (e.g. Naab et al., 2014) allowed. As such, embedded discs may be special cases of two-phase kinematic transitions.

In Bellstedt et al. (2017), most embedded discs were found in either elliptical (E) galaxies or galaxies with uncertain morphological types, suggesting that the visual morphological classification of galaxies with embedded discs might be more representative of their global rather than central (within $1R_{e}$) dynamical structure. Past studies have confirmed the link between the photometric properties of galaxies and their kinematic properties (e.g. Krajnović et al., 2013; Cortese et al., 2016). However, fast and slow rotator kinematic morphologies based on the kinematics within $1R_{e}$ don’t necessarily correspond to the lenticular and elliptical visual morphological types (e.g. Emsellem et al., 2007, 2011; Cappellari, 2016b).

While fast and slow rotator kinematic morphologies are based on a single kinematic measure, other studies have suggested that higher order kinematic moments are also a useful probe of a galaxy’s dynamical state and better reflect their formation (Naab et al., 2014; van de Sande et al., 2017b). Similarly, Kalinova et al. (2017) compared the shape of radial kinematic profiles with visual morphologies. In every case, some correspondence between visual and kinematic morphology was found, but all concluded that visual morphologies do not map directly onto kinematic morphologies.

Arnold et al. (2011, their fig. 11) reported a trend between the global morphology of galaxies and the local spin gradient using SLUGGS data. They found that lenticular galaxies had steeper gradients than ellipticals. However, Raskutti et al. (2014) did not find a relationship between morphology and the spin gradient in a sample of 33 massive elliptical galaxies observed with the Mitchell Spectrograph, even when combined with the SLUGGS data. A reanalysis using kinemetry by Foster et al. (2016) subsequently and independently confirmed the trend for the SLUGGS sample.

In this work, we define a sample of 384 galaxies from the Sydney AAO Multi-object Integral Field (SAMI) Galaxy Survey with available large-scale stellar kinematics. We study the balance of dynamical support between rotation and random/pressure support. Although recent works have focused on the local (e.g. Wu et al., 2014; Bellstedt et al., 2017) or central (within $R_{e}$) spin parameter $\lambda$ (Cappellari et al., 2007; Emsellem et al., 2007) over the classic and more intuitive $V/\sigma$ measurement (e.g. Binney, 1978), our work mainly shows the latter as both lead to qualitatively identical conclusions. Our emphasis is on large-scale $V/\sigma$ radial profiles, their slope and relation to morphological types. Special attention is given to contentious issues from the literature based on smaller samples already outlined. This study uses a wider morphological baseline (ellipticals to late-type spirals) and appreciably larger sample than past studies of large scale stellar kinematics, although we note that the visual morphological classifications (Cortese et al., 2016) are somewhat less certain owing to lower image quality and higher distances.

We assume a $\Lambda$CDM cosmology with $\Omega_{\rm m}=0.3$, $\Omega_{\lambda}$ $=$ $0.7$ and $H_{0}=70$ km s^-1 Mpc^-1.

The spectroscopic data used in this work were obtained using the SAMI instrument (Croom et al., 2012) as part of the SAMI Galaxy Survey. The SAMI instrument is a multiple integral field spectrograph capable of simultaneously obtaining spatially resolved spectroscopy of multiple targets using hexabundles (Bland-Hawthorn et al., 2011; Bryant et al., 2014). Each IFU hexabundle of 61 fibres has a 73% filling factor with a 15 arcsec diameter. The SAMI instrument has 13 such hexabundles and 26 individual sky fibres over a 1 degree field-of-view. These are connected to the AAOmega spectrograph (Sharp et al., 2006) on the 3.9m Anglo-Australian Telescope. The observational campaign is now complete and the final SAMI Galaxy Survey contains $\sim 3000$ low redshift ($0.004\leq z\leq 0.095$) galaxies. The early, first and second public data releases are described in Allen et al. (2015), Green et al. (2018) and Scott et al. (in preparation), respectively. Details of the SAMI target selection for the Galaxy And Mass Assembly (GAMA, Driver et al. 2011) fields and cluster sample can be found in Bryant et al. (2015) and Owers et al. (2017), respectively.

The AAOmega spectrograph is set up with the 580V and 1000R gratings. This setup yields a wavelength coverage in the blue and red arms of 3750-5750 Å and 6300-7400 Å with a median spectral resolution of $R\sim 1809$ and $R\sim 4310$, respectively. The SAMI data are reduced by the 2dFdr pipeline. The pipeline subtracts the bias frames and applies flat fielding, cosmic ray removal, wavelength calibration using CuAr arc frames and sky subtraction. Following the technique of Sharp & Birchall (2010), a spectrum is extracted for each fibre. For detail of the spectral data reduction steps in 2dFdr, see Hopkins et al. (2013). Spectra are then flux calibrated using a primary standard star observed on the same night as the observations. Flux scaling and telluric absorption corrections are performed using a secondary standard star observed simultaneously with each field. To deal with gaps between fibres within the hexabundles and ensure continuous sampling, each field is typically observed with 7 dithers. For each galaxy, datacubes are reconstructed using a minimum of 6 dithers. Flux and covariance are carefully propagated onto a grid as described in Sharp et al. (2015). For full details of the SAMI data reduction, see Allen et al. (2015) and Sharp et al. (2015).

The line-of-sight velocity distribution (LOSVD) for each spaxel is parametrised with a Gaussian using the penalised pixel-fitting (pPXF, Cappellari & Emsellem 2004; Cappellari 2016a) algorithm. The velocity moments (i.e. the recession velocity, $V$, and velocity dispersion, $\sigma$) are measured by fitting template spectra for each spaxel. As described in van de Sande et al. (2017b), spectra within elliptical annuli are first combined into a high quality spectrum before fitting in order to determine the best template to subsequently use for fitting the LOSVD on corresponding member spaxels. This two-step fitting process mitigates uncertainties associated with possible template mismatch issues, especially in the lower signal-to-noise spectra. For each spaxel, the LOSVD ($V$, $\sigma$) is estimated through minimising the residuals between the observed spectrum and the corresponding pre-determined template broadened through convolution with a Gaussian.

We use the Multi-Gaussian Expansion technique (MGE, Emsellem et al., 1994) and the code from Cappellari (2002) to measure the light-weighted global photometric ellipticity, position angle and effective radius ($R_{e}$) of each galaxy. The MGE analysis is performed on Sloan Digital Sky Survey (SDSS) or VLT Survey Telescope (VST) 400 arcsec $r$-band cutout images. We first process each image with SExtractor and PSFEx (Bertin, 2011) to mask neighboring objects and build a model of the point spread function. We use the regularised version of MGE, which allows the fit to reflect the underlying light distribution rather than asymmetries such as bars, spiral arms or other non-axisymmetric features (Scott et al., 2013). More detail on the MGE fitting can be found in D’Eugenio et al. (in prep).

Stellar masses are estimated from the $i$-band absolute magnitude and $(g-i)$ colours following Taylor et al. (2011, equation 8). Whenever available (i.e. for 98 percent of the galaxies in our sample), visual morphologies are taken from Cortese et al. (2016). Two percent of galaxies either have uncertain morphological types or have not yet been visually classified.

Our sample selection is as follows. Galaxies from the internal team release version v0.9.1 with no spaxels satisfying $V_{\rm error}<30$ km s^-1 and $\sigma_{\rm error}<0.1\sigma+25$ km s^-1 are discarded. From the remaining parent sample of 845 galaxies with reliable stellar kinematics, we identify a sample of galaxies with measured stellar kinematics ($V$ and $\sigma$) out to at least 1.5$R_{e}$. The outermost radius probed must have a minimum filling factor of 85 percent. There are 408 SAMI galaxies satisfying this selection. A further 24 galaxies are removed from the sample due to short profiles (see Section 3 for more detail). This yields a final sample of 384 galaxies. Fig. 1 shows how the final large scale stellar kinematics sample compares with the parent sample. The selection is biased in favour of earlier morphological types and higher Sérsic index values because stellar kinematics are more readily measured in systems where the stellar light dominates the flux budget (rather than emission from ionised gas). By construction, galaxies with effective radii larger than two thirds that of the field of view of an individual hexabundle will not make it into our sample. Fig. 1 suggests that this angular size limit has prevented galaxies with $R_{e}\gtrsim 5$ kpc from entering the final sample.

In order to look for possible kinematic transitions in the outskirts of SAMI galaxies, we measure radial variations in the ratio of ordered to random stellar motion. For this purpose and to avoid weighting excessively against the fainter outskirts, we compute $V/\sigma(R)$ locally following e.g. Wu et al. (2014):

where $V_{i}$ and $\sigma_{i}$ are the measured recession velocity and velocity dispersion of the $i^{\rm th}$ spaxel in the elliptical annulus ($A$) with circularised radius $R$.

To ease comparison with literature, we also measure the local spin parameter:

as adapted from Emsellem et al. (2007) and defined by Bellstedt et al. (2017).

We choose rolling bin sizes of 25 spaxels at 10 spaxels intervals and remove 24 galaxies that have unreliably short profiles with fewer than 7 bins. This brings our final sample to 384 galaxies. The profiles are smoothed using a spline to minimise the effect of stochastic grid sampling.

The measured $V$, $\sigma$, $V/\sigma$ and $\lambda_{\rm loc}$ profiles are shown in Figs. 2 and 3. Velocity and dispersion profiles show a clear trend with stellar mass in Fig. 2 such that more massive galaxies can support higher rotation and dispersion on average. Velocity profiles typically either rise monotonically, rise and fall or plateau with radius. As observed by Falcón-Barroso et al. (2017) for cumulative dispersion profiles, we find a variety of local dispersion profiles. The vast majority ($\sim$96 percent) of dispersion profiles fall monotonically. We do observe a small fraction ($\sim$4 percent) of monotonically rising dispersion profiles or profiles that fall and rise. The most prominent cases are associated with galaxies with unusual kinematic features such as embedded (see Section 4.3) and counter rotating discs (also known as 2$\sigma$ galaxies). In Fig. 3, we have divided the $V/\sigma$ profiles into three velocity dispersion bins chosen to optimise clarity. Most $V/\sigma$ and $\lambda_{\rm loc}$ profiles show a steep rise in the inner parts. Many profiles plateau at increasing radius. In all dispersion bins, we note a large range in $V/\sigma$ and $\lambda_{\rm loc}$ values at a given radius even among galaxies of similar mass. The middle row of Fig. 3 also shows that high mass galaxies on average have higher $(V/\sigma)$ and $\lambda_{\rm loc}$ profiles. This trend with stellar mass is especially noticeable in the intermediate velocity dispersion bin (i.e. middle panel of Fig. 3).

For the sake of simplicity, and to avoid unnecessary redundancy, we present results for $V/\sigma$ only henceforth. Our conclusions do not change qualitatively for $\lambda_{\rm loc}$.

Fig. 4 shows the maximum measured $V/\sigma$ value, $(V/\sigma)_{\rm max}$, as a function of stellar mass ($\log M_{*}/M_{\odot}$). While there is no clear correlation between $(V/\sigma)_{\rm max}$ and stellar mass (when ignoring the colours), the position of galaxies in this space is clearly correlated with visual morphology, albeit with a large range in $(V/\sigma)_{\rm max}$ at fixed morphology. Late-type galaxies (eSp and lSp) typically exhibit higher $(V/\sigma)_{\rm max}$ and tend to have lower masses than early-types (E and S0).

The $V/\sigma$ profiles binned by visual morphology are shown in Fig. 5. The amplitude of the $V/\sigma$ profiles is correlated with the apparent ellipticity of the system such that more apparently flattened systems have $V/\sigma$ profiles that lie above those of rounder ones. Since nearly all galaxies in SAMI are oblate axisymmetric systems (Foster et al., 2017), the apparent ellipticity reflects a combination of the inclination of the system and its intrinsic flattening. Hence, the wide range in measured $V/\sigma$ values is partly explained by the fact that SAMI galaxies are viewed from a range of inclinations. Differences in seeing conditions is a further contributor to this scatter (see e.g. van de Sande et al., 2017b, a). Lenticular and late-type spiral galaxies overall show more rotational support than elliptical galaxies at all radii.

Fig. 6 compares the inner ($0.5R_{e}$) and outer ($1.5R_{e}$) $V/\sigma$ for SAMI galaxies with measured kinematic profiles at both radii (103 galaxies, refer to Fig. 3). A similar trend was shown between 1 and $0.5R_{e}$ in van de Sande et al. (2017a). The data points lie slightly above the one-to-one line at low $V/\sigma$ and increasingly depart from it for higher $V/\sigma$ values. Galaxies with low inner $V/\sigma$ tend to have comparable $V/\sigma$ values in their outskirt, while galaxies with higher inner $V/\sigma$ show progressively steeper positive gradients. A linear regression fit to the data with intercept set to zero yields a slope of $2.09\pm 0.06$, indicating significant departure from a slope of unity. There are very few outlier points in this space, suggesting that dramatic kinematic transitions between 0.5 and 1.5$R_{e}$ are rare in our sample. In Fig. 7, we perform an analysis akin to that presented in Bellstedt et al. (2017, their fig. 12) for $\lambda_{\rm loc}(R)$ and show the $V/\sigma$ gradient as a function of $V/\sigma(1R_{e})$, colour coded by morphology. We confirm a trend with morphology as reported in Bellstedt et al. (2017) that Es tend to have lower $V/\sigma$ values and shallower $V/\sigma$ gradients than S0s. While our sample has few Sp galaxies, they tend to lie below the sequence defined by Es and S0s in Fig. 7 (top panel). Apart from one notable exception, most galaxies are consistent with either a null or positive $V/\sigma$ gradient.

In Fig. 8, we show the $V/\sigma$ gradient between $1.5R_{e}$ and 0.5$R_{e}$ as a function of visual morphology and stellar mass. Despite the large scatter, there is a clear trend between the $V/\sigma$ gradient and morphology such that early-type galaxies have shallower gradients than late-types as found by Arnold et al. (2011) and Foster et al. (2016). The lowest stellar mass bin has slightly lower average $V/\sigma$ gradient than the higher stellar mass bins.

In this work, we explore the outer kinematic properties of galaxies in the SAMI Galaxy Survey. We specifically look at how well visual morphology maps onto the local rotation dominance parameter profiles. We also look for possible kinematic transitions between the in-situ and accreted regime of galaxy formation and embedded discs by identifying radial changes and modulations of the rotation dominance parameter in galaxies.

In this work, we study the balance of rotational to pressure support in the outskirts of 384 SAMI galaxies with stellar kinematics extending to $\geq 1.5R_{e}$. While we mainly show $V/\sigma(R)$ profiles, our results do not change qualitatively for $\lambda_{\rm loc}$. Our conclusions are summarised below.

1. 1.

   We confirm that the structure of galaxies, as probed through their visual morphology, is linked to their dynamical support and we find this to be true at all probed radii. On average, visually discy systems have higher rotational support and steeper $V/\sigma$ gradients than bulge dominated systems. In other words, the visual morphology of galaxies reflects their dynamical state, albeit with large intrinsic scatter and some possible confusion due to inclination.

2. 2.

   We identify three galaxies out of a sample of 384 with convincing embedded discs in the SAMI sample. The fraction of embedded discs in SAMI is roughly an order of magnitude below that of SLUGGS (Arnold et al., 2011; Bellstedt et al., 2017). Once accounting for differences in the distances, spatial sampling and observational effects, the fractions of embedded discs in both surveys are brought into agreement.

The SAMI Galaxy Survey stellar kinematic sample will double in size this year as the recently observed data are analysed and added to the existing sample. In a follow-up paper (Foster et al. in prep), we explore radial tracks in the spin-ellipticity diagram (Graham et al., 2016) to further investigate the link between visual structure and kinematics. The planned multi-object wide-field IFU spectrograph, HECTOR (Bryant et al., 2016) on the Anglo-Australian Telescope is designed to have optimised radial coverage of the stellar kinematics out to at least $2R_{e}$. The associated ambitious galaxy survey is better optimised to probe kinematic transitions and embedded disc fractions.

This research was conducted by the Australian Research Council Centre of Excellence for All Sky Astrophysics in 3 Dimensions (ASTRO 3D), through project number CE170100013. The SAMI Galaxy Survey is based on observations made at the Anglo-Australian Telescope. The Sydney-AAO Multi-object Integral field spectrograph (SAMI) was developed jointly by the University of Sydney and the Australian Astronomical Observatory. The SAMI input catalogue is based on data taken from the Sloan Digital Sky Survey, the GAMA Survey and the VST ATLAS Survey. The SAMI Galaxy Survey website is http://sami-survey.org/.

NS acknowledges support of a University of Sydney Postdoctoral Research Fellowship. JvdS is funded under Bland-Hawthorn’s ARC Laureate Fellowship (FL140100278). M.S.O. acknowledges the funding support from the Australian Research Council through a Future Fellowship (FT140100255). SB acknowledges the funding support from the Australian Research Council through a Future Fellowship (FT140101166). Support for AMM is provided by NASA through Hubble Fellowship grant #HST-HF2-51377 awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS5-26555.