Astronomical Science

Disentangling the Kinematics and Stellar Populations of Counter-rotating Stellar Discs in

Lodovico Coccato1 Different scenarios have been proposed to simultaneously, allowing the different for- Lorenzo Morelli2, 3 explain the formation of these peculiar ob­ mation scenarios to be disentangled. Alessandro Pizzella2, 3 jects. The internal origin scenario (Evans & Enrico Maria Corsini2, 3 Collett, 1994) describes the dissolution of Lucio Maria Buson3 a triaxial potential or a bar. In this process, Results of the spectroscopic observations Elena Dalla Bontà2, 3 the stars, moving on box orbits, escape from the confining azimuthal potential well We started an observational campaign and move onto tube orbits. During this with the integral field unit (IFU) of the 1 ESO process, half of the box-orbit stars are VIMOS spectrograph at the VLT aimed at 2 Dipartimento di Fisica e Astronomia scattered onto clockwise-streaming tube determining the most efficient mecha- “G. Galilei”, Università degli Studi di orbits, the other half onto counterclockwise nisms in building large-scale counter- Padova, Italy ones. In this way, two identical counter- rotating stellar discs, like those observed 3 INAF–Osservatorio Astronomico di rotating stellar discs can be built. In the in NGC 4550. For this study, we devel- Padova, Italy gas accretion scenario, a disc oped a new spectroscopic decomposition acquires gas from extragalactic reservoirs technique that fits a galaxy spectrum and onto retrograde orbits (e.g., Pizzella et al., separates the contributions of two counter- Spectroscopic VIMOS/IFU observations 2004). If the amount of acquired gas is rotating stellar components (Coccato et are presented for three galaxies known sufficiently large, the new gas drives away al., 2011). At each position on the sky, the to host two stellar counter-rotating discs the pre-existing gas and it settles onto code builds two synthetic templates (one of comparable sizes. For the first time a counter-rotating disc. New stars are then for each stellar component) as a l­inear both the kinematics and stellar popu­ born from the acquired gas, forming the combination of spectra from an in­put stel- lation properties of the two counter- counter-rotating stellar disc. In the binary lar library, and convolves them with two rotating discs in the observed galaxies merger scenario (e.g., Crocker et al., 2009), Gaussian line-of-sight velocity distribu- were separated and measured. The two disc galaxies with opposite spin direc- tions with different kinematics. Gaussian ­secondary, less massive, stellar compo- tions collide and form two coplanar coun- functions are also added to the convolved nent rotates in the same direction as the ter-rotating stellar discs, depending on the synthetic templates to account for ionised ionised gas and is on average younger geometry of the encounter. gas emission lines (Hγ, Hβ, [O iii] and [N i]). and less metal-rich than the main galaxy disc. These results support the scenario These different formation mechanisms are The spectroscopic decomposition code of gas accretion followed by star for­ expected to leave differing signatures in returns the spectra of two best-fit syn- mation as the origin for large counter- the properties of the stellar populations of thetic stellar templates and ionised gas rotating stellar discs in galaxies, and set the counter-rotating component. In par- emission (Figure 1), along with the best-­ an upper limit of 44% to those formed ticular, age is a key element in differentiat- fitting parameters of luminosity fraction, by binary galaxy mergers. ing among these scenarios. The internal velocity and velocity dispersion. The line origin predicts the same mass, luminosity, strengths of the Lick indices of the two chemical composition and age for both counter-rotating components are measu­ Counter-rotating galaxies counter-rotating stellar components. Gas red on the two best-fit synthetic templates. acquisition followed by star formation We then compare the indices Hβ, Mgb, Counter-rotating galaxies are those that predicts younger ages for the counter- Fe5270, and Fe5335 to the predictions of host two components that rotate in oppo- rotating stellar component in all cases, and stellar population models to infer the stel- site directions to each other. These pecu- it allows for different metallicity and lar age, metallicity ([Z/H]), and abundance liar objects have been observed in all mor- α-enhancement between the two discs. ratio of α-elements ([α/Fe]) of the two phological types, from ellipticals to spirals. Direct acquisition of stars through mergers counter-rotating discs. We also computed They are classed by the nature (stars vs. also allows for different metallicity and the mass-to-light ratios and the mass frac- stars, stars vs. gas, gas vs. gas) and size α-enhancement, but the younger compo- tion of the two counter-rotating compo- (counter-rotating cores, rings, discs) of nent will be determined by the difference nents from the luminosity fraction and the the counter-rotating components (see in age between the host galaxy and inferred stellar population parameters. Bertola & Corsini [1999] for a review). In the merged system. Roughly, one would The estimated mass is used to determine this work, we investigate the peculiar class expect younger counter-rotating stars in which component is the main (i.e., the of counter-rotating galaxies with two coun- ~ 50% of the cases. most massive) and which is the secondary ter-rotating stellar discs of compa­rable (i.e., the least massive). size. The prototype of this class of ob­jects A proper spectroscopic decomposition is the famous E7/S0 galaxy NGC 4550, that separates the relative contribution of We started this project (programme IDs: whose counter-rotating nature was first the two counter-rotating stellar compo- 383.B-0632 and 087.B-0853A) by observ- discovered by Rubin et al. (1992). To date, nents to the observed galaxy spectrum is ing with VIMOS/IFU three galaxies that there are only a few known counter-rotating therefore needed. In this way, the kine­ were known to host counter-rotating stel- galaxies similar to NGC 4550, but the matics and the stellar populations of both lar discs of comparable size: NGC 3593 ­census will increase as a result of the new the stellar components can be measured (Bertola et al., 1996), NGC 4550 (Rubin et two-dimensional spectroscopic surveys. al., 1992) and NGC 5719 (Vergani et al.,

The Messenger 151 – March 2013 33 Astronomical Science Coccato L. et al., Counter-rotating Stellar Discs in Galaxies

Figure 1. Fit of the galaxy spec- Table 1. Luminosity-weighted values for the stellar   '` ,F( %D( trum (black) in a spatial bin. The population parameters of the stellar discs. best-fitting model (green) is the   sum of the spectra of the ionised gas component (magenta) and Age [Gyr] [Z/H] [α/Fe] TW   the two stellar components (blue NGC 3593 Ck and red). The differences in the Main: 3.6 ± 0.6 −0.04 ± 0.03 0.09 ± 0.02   position of absorption line features Secondary: 2.0 ± 0.5 −0.15 ± 0.07 0.18 ± 0.03 and in the Hβ equivalent widths NGC 4550 -NQL@KHRD   between the two stellar compo- Main: 6.9 ± 0.6 −0.01 ± 0.03 0.20 ± 0.02 nents (indicating different kinemat- Secondary: 6.5 ± 0.5 −0.13 ± 0.04 0.28 ± 0.02   ics and stellar population content) NGC 5719 are clearly evident.   Main: 4.0 ± 0.9 0.08 ± 0.02 0.10 ± 0.02      Secondary: 1.3 ± 0.2 0.3 ± 0.02 0.14 ± 0.02 6@UDKDMFSGÄ

2007). The combination of the collecting are younger, more metal-poor, and have the kinematics and the Lick indices of power of an 8-metre-class telescope, the higher abundances of α-elements than the fitted components. The measured large wavelength coverage (4150–6200 Å) the main stellar discs. kine ­matics are used to construct the two- and the high spectral resolution (full width dimensional velocity fields, where the half maximum [FWHM] of 2.0 Å) with the Table 1 summarises the mean properties structure of the adopted spatial binning is HR-blue grism makes VIMOS on the VLT of the two counter-rotating stellar popula- still visible, and the counter-rotation the best-suited integral field unit for this tions, and Figures 2, 3 and 4 illustrate our between the two stellar components is kind of study. results. In these figures, the extension of remarkable (lower panels). The Lick indi- the VIMOS/IFU field of view is plotted over ces are shown in the diagnostic plots In all the sample galaxies, we confirm the the galaxy image (upper-left panels). The (upper right panels) with the predictions of presence of two stellar discs of compara- collected spectra are then spatially binned single stellar population models (Thomas ble size, and one ionised gas component. over the field of view with the Voronoi et al., 2011). We use Hβ and the combined The secondary stellar component and the ­binning technique to increase the signal- [MgFe]: = √(Mgb · [0.82 · Fe5270 + 0.28 · ionised gas component counter-rotate to-noise ratio (see Coccato et al. [2013] for Fe5335]) index as indicator of stellar age with respect to the main stellar disc. More- further details). The spectral decomposi- and metallicity. We also take into account over, in all the observed galaxies, these tion technique is then applied on the the variation in the abundance of counter-rotating secondary components spectrum of each spatial bin to measure α-elements and metallicity by measuring

 -&"  Figure 2. Results of the   FD&XQ :_%D< spectroscopic decom- FD&XQ %D<l  :_ position of NGC 3593.   %D<  FD &XQ  :_ The top left panel shows the galaxy image with FD&XQ  B  %D<  the location of the _ Ä  : RD

Ä VIMOS observed field of

QB :9'<    `  FD&XQ D view. The bottom panels ' "@ % :9'<   :9'<l  FD&XQ   show the two-dimen- #$

6 l FD&XQ :9'<  sional kinematics of the :9'<l :9'<  :9'<l  :9'<  FD&XQ main, secondary and  :9'<   l ionised gas compo- JOB :9'<l  nents, respectively. The   top-right panels show    l ll        the measured equivalent 61 @QBRDB ,F%DÄ ,FAÄ width of the Lick indices; ,@HMBNLONMDMS 2DBNMC@QXBNLONMDMS (NMHRDCF@R grids with the predictions  -&" -  -&"  -&" from single stellar popu- lation models are also shown. The crosses on    the top-right side of the panels show the mean B B B error bars on the meas- RD RD RD

QB  QB  QB  ured indices. Each spa- tial bin returns two sets 8@ 8@ 8@ of Lick indices, one for l l l each stellar component: red symbols represent OB the main stellar compo- l l l nent, whereas blue sym- l l    l l    l l    bols represent the sec- 7@QBRDB 7@QBRDB 7@QBRDB ondary counter-rotating l l    l l    l l    component. 5JLR 5JLR 5JLR

34 The Messenger 151 – March 2013 -&"  Figure 3. As Figure 2,   _%D< FD&XQ but for NGC 4550. FD&XQ

  %D<l  FD &XQ  _ %D<  FD&XQ _  B   Ä RD Ä

 %D<  QB  ` FD&XQ _ ' :9'< %D "@  :9'<l  FD&XQ  :9'<  #$  6 FD&XQ l :9'<l :9'<  :9'<l  :9'<  FD&XQ  :9'<   l :9'<l  JOB

     l ll        61 @QBRDB :,F%D<Ä ,FAÄ ,@HMBNLONMDMS 2DBNMC@QXBNLONMDMS (NMHRDCF@R

 -&"  -&"  -&"

 -   B B B RD RD RD

QB  QB  QB  8@ 8@ 8@

l l l

OB l l l l l    l l    l l    7@QBRDB 7@QBRDB 7@QBRDB

l l    l l    l l    5JLR 5JLR 5JLR

-&"   Figure 4. As Figure 2,   :_%D< FD&XQ but for NGC 5719. %D<l FD&XQ :_   %D<  FD &XQ  :_ FD&XQ  B  %D<   :_ Ä RD

Ä :9'<  QB 

 ` FD&XQ ' "@ %D :9'<    :9'<l  FD&XQ  #$ :9'<  6 l FD&XQ :9'<l :9'<  :9'<l  :9'<  FD&XQ  :9'<   l JOB :9'<l 

     l ll            61 @QBRDB ,F%DÄ ,FAÄ ,@HMBNLONMDMS 2DBNMC@QXBNLONMDMS (NMHRDCF@R  -&"- &"- &" -

   B B B RD RD RD

QB  QB  QB  8@ 8@ 8@

l l l

OB l l l l l    l l    l l    7@QBRDB 7@QBRDB 7@QBRDB

l l    l l    l l    5JLR 5JLR 5JLR

The Messenger 151 – March 2013 35 Astronomical Science Coccato L. et al., Counter-rotating Stellar Discs in Galaxies

the magnesium Mgb index and the mean &@K@WHDRVHSGK@QFDlRB@KDBNTMSDQlQNS@SHMFRSDKK@QCHRBR Figure 5. Schematic iron index = (Fe5270 + Fe5335)/2. representation of the probability PE of observ- (. - ing the secondary stellar 3 NGC 3593. An isolated, highly inclined, disc younger than the

S0/a spiral at a distance of 7 Mpc. It is .1, main galaxy disc, as a characterised by a patchy spiral dust pat- l,EQ@BSHNMNEBNTMSDQlQNS@SHMF ,EQ@BSHNMNEBNTMSDQlQNS@SHMF function of the fraction F@K@WHDROQNCTBDCAXF@R@BBQDSHNM F@K@WHDROQNCTBDCAXAHM@QXLDQFDQR M of galaxies formed tern in the centre. The two counter-rotat- through the binary gal- ing stellar discs have different scale axy merger scenario. lengths, but the same intrinsic flattening.

The secondary component dominates the 2% RDBNMC@QX RDBNMC@QX OQHL@QX BNLONMDMSXNTMFDQ BNLONMDMSXNTMFDQ BNLONMDMSXNTMFDQ innermost 500 pc. The ages of the two (.- %Q@BSHNMl, %Q@BSHNM, %Q@BSHNM, components date the accretion event to 3 between 2.0 and 3.6 Gyr ago, i.e., 1.6 ± /QNA@AHKHSX/$NEjMCHMF@BNTMSDQlQNS@SHMFF@K@WXVHSG 0.8 Gyr after the formation of the main RDBNMC@QXBNLONMDMSXNTMFDQSG@MSGDL@HMBNLONMDMS .!2$15 / ,l, ,l, stellar disc (­Coccato et al., 2013). $

NGC 4550. An E7/S0 galaxy in the

Cluster, at a distance of 16 Mpc, and it is The most efficient mechanism to build mising ΠE (M). In our case N = T = 3, often indicated as the prototype of galax- counter-rotating stellar discs therefore MMAX < 44% at 1σ confidence ies with counter-rotating stellar discs. It level. has an nearby, NGC 4551, Our results favour the external origin of to the northeast of NGC 4550 at a pro- the counter-rotating components in all Thus, it is fundamental to measure the dif- jected distance of 14 kpc. An interaction in the three observed galaxies, and discard ference in ages of the counter-rotating the past between these two systems the internal origin scenario. As stated components in a larger sample of counter- could have produced the counter-rotating above, in the gas-accretion scenario the rotating galaxies to identify the most effi- stellar disc in NGC 4550, although no stellar component associated with the cient mechanism and derive more precise photometric signatures of the interaction, ­ionised gas is predicted to always be constraints. Large spectroscopic surveys such as tidal tails or gas streams, have younger than the main stellar component. like MaNGA will help to identify other been detected. The two counter-rotating In binary galaxy mergers instead, one potential candidates by recognising the stellar discs in NGC 4550 have the same would expect that the younger stellar kinematic signatures of stellar counter- scale lengths, but slightly different ellip­ component is associated with the ionised rotating discs, such as the two symmetric ticity (εmain = 0.6, εsecond = 0.5) meaning gas only in 50% of the cases. Although peaks in the velocity dispersion map. that they have different scale heights. The we always observe the secondary compo- The MaNGA survey is one of the Next measured ages date the accretion event nent to be younger than the main galaxy Generation Sloan Digital Sky Surveys, and ~ 7 Gyr ago, i.e. less than 1 Gyr after the disc, thus favouring the gas accretion it will collect IFU spectroscopic data for formation of the main stellar disc (Coccato ­scenario, we cannot rule out the binary ~ 10 000 galaxies in the nearby Universe et al., 2013). galaxy merger scenario. within six years. Moreover, the next gener- ation integral field unit MUSE at the VLT NGC 5719. An Sab galaxy at a distance On the other hand, we can infer an upper will represent a significant step forward for of 23 Mpc, and a member of a rich group. limit to the fraction of galaxies with two these studies, because of the better It is currently interacting with the Sbc counter-rotating stellar discs that were ­spectral resolution, wavelength coverage ­galaxy NGC 5713, which is to the west of formed by binary mergers by statistical and field of view with respect to VIMOS. NGC 5719 at a projected distance of arguments. We define M as the fraction 77 kpc. The interaction between the two of galaxies hosting two counter-rotating systems is traced by a bridge of neutral stellar discs of comparable sizes that were References hydrogen (Vergani et al., 2007), which fuels generated by a binary major merger (thus, Bertola, F. & Corsini, E. M. 1999, IAU Symposium, the secondary counter-rotating stellar (1 – M) is the fraction produced by gas 1 8 6 , Galaxy Interactions at Low and High component in NGC 5719. The two s­ tellar accretion), and E as the event of finding ­Redshift, eds. Barnes, J. E. & Sanders, D. B. components in NGC 5719 have similar the stellar component co-rotating with the ( ­Dordrecht: ­Kluwer), 149 luminosity, but the secondary is less mas- gas to be the youngest. The probability Bertola, F. et al. 1996, ApJ, 458, L67 Coccato, L. et al. 2011, MNRAS, 412, L113 sive because of its younger age. Moreover, PE to observe E is: PE = 1 – M/2 (see Fig- Coccato, L. et al. 2013, A&A, 549, 3 the youngest ages are ob­­­served in the ure 5). The probability ΠE (M) of observing Crocker, A. F. et al. 2009, MNRAS, 393, 1255 secondary component at ~ 700 pc from E in exactly N galaxies out of a sample Evans, N. W. & Collett, J. L. 1994, ApJ, 420, L67 the centre, where the H emission lines of T counter-rotating galaxies is given by Pizzella, A. et al. 2004, A&A, 424, 447 β Rubin, V. C., Graham, J. A. & Kenney, J. D. P. 1992, are more intense (Coccato et al., 2011). the first term of the binomial distribution. ApJ, 394, L9 The ages of the two components date the It is therefore possible to compute the Thomas, D., Maraston, C. & Johansson, J. 2011, accretion event between 1.3 and 4.0 Gyr most probable value for the fraction of MNRAS, 412, 2183 ago, i.e. 2.7 ± 0.9 Gyr after the formation of counter-rotating galaxies produced Vergani, D. et al. 2007, A&A, 463, 883 the main stellar disc (­Coccato et al., 2011). by binary mergers, MMAX, simply by maxi­

36 The Messenger 151 – March 2013