Davor Krajnović's Recent Research Highlights

Are kinematically distinct cores really decoupled components: the case of NGC5813

(appeared on the main page in June 2015.)

One of the first targets with MUSE, still during the early stages of commissioning of the instrument on Paranal, was an elliptical galaxy NGC5813. Why? Because it intrigued me for a long time... In 2003, we published a paper, led by Eric Emsellem and the SAURON team, showing integral-field kinematics of a representative sample of early-type galaxies. There were many famous object in that sample, and this one had another distinction: NGC5813 was the first known object to have a "kinematically distinct core", discovered by Geroge Efstathiou, Richard Ellis and David Carter in 1980. The SAURON data were very good, but they didn't reveal all secrets of NGC5813. The structure of the NGC5813 remained a mystery for good 35 years. The new MUSE data, however, allow us to start understanding the hidden nature of NGC5813 and provide a good topic for An Astro Highlight based on this paper.

Kinematically distinct cores, or kinematically decoupled cores, or kinematically peculiar cores, or most commonly only referred to as KDCs, are exciting features on the mean velocity maps of galaxies. As you can see from the first panel on the plot below, which shows the map of the mean velocities of stars in NGC5813 (essentially a map of ordered motions of stars), the centre of the galaxy, its core, seems to rotate, while the rest of the galaxy's body does not seem to have any ordered rotation. Why is that? How do you make structures like that? Are they stable and long lived?

The mean velocity and the velocity dispersion maps of NGC5813 as seen by MUSE, showing the famous KDC, but also a very specific structure on the velocity dispersion map (two peaks along the major axis). This figure is based on Fig 2. from a recently published paper in MNRAS (Krajnović et al. 2015). NGC5813 is the first know KDC, and the MUSE data, constraining dynamical models, allowed us to understand its internal structure. It is not a decoupled component, separated from the rest of the galaxy's body. Instead, it is made of two components of stars, each following a family of orbits with opposite net angular momentum: they are counter-rotating. This counter-rotation together with a specific distribution of stars within the components defines the structure we see on the kinematics maps. North is up, East is left. A specialist talk about KDCs in general, which I gave at the Macquarie University, Sydney, in February 2015, is available here (27Mb).

These are some of the question astronomers are trying to answer since the discovery of KDCs. There are many types of KDCs (velocity maps of a number of KDC you can see on the figure at the bottom of the page): some have similar velocity maps as NGC5813, where the core seems to rotate and the rest of the body doesn't, but there are even more spectacular cases, such as NGC4365, where the rotation axes of the core and the body are misaligned by about 90 degrees, or NGC5322 where the core is counter-rotating with respect to the rest of the body (180 degrees misalignment). Sometimes, but not very often, the misalignments is even different from the 90 and 180 degrees (can you spot such galaxy on the figure below?). When astronomers started discovering more and more KDCs in 1990s, they became the topic of the day as they look like kinematic proofs that galaxies indeed grow by interactions. KDCs are the smoking guns of mergers, but what kind of mergers and how are they actually created is still a bit of an uncharted territory (although there are plenty of ideas out there).

KDC are often found in large and massive elliptical galaxies. These objects are essentially balls, or even rugby balls, of stars embedded in dark matter halos and hot gas. Some have active galactic nuclei and large scale jets, which seem to stir the halos of hot gas, rising bubbles of gas around the central galaxy (all this is true for NGC5813). KDC are also found in smaller galaxies, but these seem to be different from the KDCs in big galaxies: they are made of young stars and they are likely to disappear as their stellar population ages, as it was elegantly shown by Richard McDermind and collaborators in this paper.

On the second panel of the figure above you can see what makes NGC5813 very special indeed. It shows the map of the velocity dispersion, a measure of random motions, and it is very unusual. Typically the velocity dispersion maps of early-type galaxies (including galaxies with KDCs) have a peak (highest values) in the centre of the galaxy, there the gravitational potential is the deepest. From that point outwards, velocity dispersion decreases pretty much steadily, following the distribution of the light. But the velocity dispersion map in NGC5813 is very different: it has a central peak, but this is then followed by a drop of some 100 km/s, followed by a rise of about 40-50 km/s and a subsequent drop. While general drops and rises in velocity dispersion are known to occur, typically in galaxies with small central stellar discs or bars, the structure of the velocity dispersion map in NGC5813 is very specific. The rise happens only on two sides of the nucleus (and not everywhere around it), along the major axis, and it coincides with the end of the KDC (of the region that exhibits clear ordered rotation). What could be the cause for that?

As I mentioned earlier, NGC5813 intrigued us for quite some time. The drop in the velocity dispersion was known since the discovery of the KDC, and the SAURON observations revealed part of its structure. But unfortunately the field-of-view of SAURON was too small, and the final mosaic of three SAURON pointings didn't cover the full structure (we didn't really know what to expect). The MUSE, however, did it in one shot. It revealed this special structure, which has been actually seen before, but in very different types of galaxies.

A small percentage of early-type galaxies, something like 7%, typically less massive and luminous and quite flat, disc-dominated systems, show what we call a double-sigma profile on their velocity dispersion maps: drop in the centre, followed by two peaks in the velocity dispersion along the major axis. Their velocity maps can be diverse: from very nice counter-rotating KDCs (180 degrees difference in angle between the central and the outer part of the body) to very messy velocity maps, with no ordered rotation. The most famous of these galaxies is NGC4550, and there is a very reasonable explanation for its structure (see this as well): it is made of two discs of stars, which rotate in opposite directions. The two peaks in the velocity dispersion are then the consequence of the counter-rotation. Imagine that in the same spectrum you are detecting one population of stars that go in one direction (quite fast) and another population that goes in the opposite direction (as fast). This will result in the mean observed velocity close to zero (the relative velocities cancels out), but the spread in velocities (the velocity dispersion) will be large. Depending on the type of these two counter-rotating discs, how many stars they have and how are they distributed, one can get a plethora of structures on the velocity and the velocity dispersion maps.

Once we saw the shape of the velocity dispersion maps of NGC5813, we had a good hunch that that NGC5813 could also be built of two populations of stars, which rotate in oposite directions. But NGC5813 is not a flat, disc-like galaxy, and the situation is a bit more complicated. Still with the help of dynamical models, we found that indeed, the galaxy can be described as having two components of stars. They couter-rotate, but not very fast; neither of them is a disc (as in a typical double-sigma galaxy). The distribution of stars is such that one component is dominant in the centre: within the KDC one component consists of 70% of the stars - that is why we see the rotation. Outside of the KDC, the components become more similar, contributing about 50-50% in mass. This 50-50 ratio also explains why there is no visible rotation: stars in both components steadily rotate, but the net motion gets canceled and we don't see it.

What does that mean for formation of the KDCs? Well, it seems KDC in NGC5813 should not be considered as a separate entity, divorced from the rest of the galaxy. It is not a ball of stars that has different kinematics from the rest of stars. It is not "decoupled". It is a result of mixing of different orbital families of stars. There are still many ways you can form this structures, such as accretion of gas from cosmological cold flows, turbulent dics of gas rich galaxies in the early universe (stars in NGC5813 are almost as old as the Universe), mergers of similar-in-mass galaxies.... Plenty of possibilities, and only by understanding the structure of other KDCs in a similar way as here, we can hope to understand their formation.

NGC5813 is a wonderful system. It has a complex kinematics which can be explained with a relatively simple, but very specific dynamical model. It also has very complex multi-phase gas structure, linked to the active galactic nucleus and its jet. But that might be another highlight, in the future.

The mean velocity maps of galaxies with KDCs from ATLAS3D Paper II (Krajnović et al. 2011). All velocity maps were obtained with the SAURON instrument. One before the last in the 2nd row is the SAURON velocity map of NGC5813, also published in 2004. North is up, East is left.

Kinematically distinct core in M87

(appeared on the main page on September 2014.)

Here is a highlight of my recent work based on the MUSE Science Verification data obtained at the end of June 2014 (past highlights can be found here). It comes from a paper by Eric Emsellem (ESO), Marc Sarzi (University of Hertfordshire) and myself (Emsellem, Krajnović & Sarzi, 2014, MNRAS, accepted). It spawned press releases in Germany and the UK, and it featured as a Picture of the Week by ESO.

Multi Unit Spectroscopic Explorer, or MUSE for short, is a next generation instrument on ESO's Very Large Telescope situated on Cerro Paranal in Chile. It was installed at the beginning of 2014 on UT4 and we were commissioning it throughout the spring. The first light and some of the commissioning data can be seen on this ESO press release. The standard procedure with the new instruments, before they are offered to the astronomical community, is to demonstrate their scientific capabilities. This is done during the Science Verification (SV) observations. The idea is that astronomers devise programmes which will test the instrument and all the tools that are required to use it, and bring some exciting science. Once MUSE was successfully commissioned, ESO announced a call for the SV proposals, selected a bunch of sound looking ideas and started observing in June (the SV run was about 2 weeks long, split between late June and mid August). The catch with the SV data is that they become immediately public and anyone in the world can use them.

We observed one of the biggest nearest galaxies. Its name is Messier 87 (M87 for short), or NGC4486 (if you prefer the New General Catalogue of Nebulae and Clusters of Stars of John Dreyer), and it is in the centre of the Virgo Cluster of galaxies, some 54 million light-years away. Galaxies like M87 are fascinating from many aspects. They are the largest galaxies, reside in the centres of galaxy clusters (bottoms of their gravitational wells), consists of old stars, have very massive black holes in their centres, which are typically very active, shooting out plasma jets and steering the inter-cluster medium very far beyond the extent of their host galaxies. Galaxies like M87 are often considered the end products of galaxy evolution, but the question still is: how have they become what they are now?

This is a copy of Fig. 2 from Emsellem, Krajnovic & Sarzi (2014). The first panel shows the SAURON velocity map, similar to the one presented in Emsellem et al. (2004), but now binned to a much higher signal-to-noise ratio (of 300, instead of 60 in that paper), which suggested that there is indeed some rotation in the central parts. The second panel presents the MUSE velocity map, and the third panel is a kinemetry reconstruction of the features on the middle panel, highlighting the KDC and the prolate-like rotation of the outer body. The color-bar on the right describes the rotation pattern: red are those regions that are receding from and blue are those that are approaching the observer. Green are the regions with no-net motion. The recession velocity of M87 was subtracted. A dashed magenta line marks the photometric major-axis of M87 (at large radii). The central 2" were masked as these kinematics are significantly influenced by the AGN (having very broad and strong emission-lines). North is up, East is left.

The first time I saw a map of stellar velocities of M87, the one coming from SAURON integral-field spectrograph observations, but not as good as the one shown on the left most panel on the figure, I was puzzled: why are stars not moving? (Have a look at the collage at the bottom of the page of stellar velocity maps typical of the galaxies in the nearby universe). The answer is that they are moving, actually very fast (often faster than a million km/h), but they do not move in a coherent way. Basically every star is moving in a direction different from that of the other stars and the mean motion observed at each point (what is plotted in the figure above) is very low. There was, however, something odd in the SAURON velocity map of M87 and we decided to have a second look at the centre of this galaxy, with an instrument of better characteristics. MUSE has a larger field-of-view than SAURON, covering about 1 arcmin x 1 arcmin on the sky. It also has a beter resolution, both in terms of probing smaller spatial scales (one square pixel is 0.2 arcseconds on a side), as well as in the resolving power of the spectrograph. Basically, with MUSE one can distinguish better than with SAURON between the various velocities of the stars. The improvement can be seen on the middle panel of the figure below.

The MUSE velocity map of M87 (middle panel above) is somewhat larger than the previous SAURON map. Notably it adds information in the top-left and bottom-right corners, or, in astro terms, in the North-East and South-West regions. This part, which was not covered by SAURON, is actually very important, as it shows that stars in M87 do coherently rotate about the center of the galaxy (evidence of this could also be seen in this paper). Another unexpected finding brought up by MUSE is that the stars in the very centre also rotate coherently, but their rotation is distinct from the large scale rotation (seen in the corners). It is distinct in two senses: the two rotations are not spatially connected and are around different axes, with some 140 degrees difference between them. The axis of the outer rotation is actually close to be aligned with the major axis of the galaxy (dashed magenta line above), the main symmetry axis which traces the orientation of the galaxy. There are very few galaxies that show this kind of rotation, often described as being "prolate-like", as a body with a prolate symmetry is expected to exhibit such a rotation.

Similar structures are know to exist, typically in massive elliptical galaxies. They are called "kinematically distinct cores" (KDCs). What makes this kinematic configuration even more surprising is that neither of the axis of rotation coincides with the minor axis of the galaxy. That kind of configuration is rarely found, and it very strongly suggest that the body of M87 is of a triaxial shape. As a matter of fact, now that we have detected the coherent rotation, albeit low, we can have some hope in determining the actual shape of M87. Before, with essentially zero net rotation, many models of very different shapes could be made to describe M87 (its light distribution and no rotation), but to hardly any one could attach a decent significance. Now, we have a possibility to change this.

To summarise, MUSE observations of M87 revelad that this central galaxy in the Virgo cluster has a very complex kinematic structure, comprising of an outer prolate-like rotation and a central KDC. The two rotation patterns seem distinct and with a mutual orientation of about 140 degrees. How does one build a galaxy like this? M87 is a product of a long history of galaxy mergers, some of which were between galaxies of smilar masses and most of which were between a big galaxy (M87) and many small satellites. By knowing the actual shape and kinematics, by knowing the chemistry and age of the stars, but also the properties of globular clusters and surrounding satellite galaxies, we may start deciphering the colourful history of M87.

And MUSE is a really impressive instrument!

ATLAS3D Paper 23. Angular momentum and nuclear surface brightness profiles

(appeared on the main page on 01.09.2013.)

Here is a highlight of my recent work based on ATLAS3D data. These two diagrams are from Krajnović et al. 2013 (MNRAS, 433, 2812K) and with colours and symbols show the nuclear shape of the radial surface brightness profiles in the λR - ε space of nearby early-type galaxies. (λR and ε respectively measure the specific angular momentum of stars and the apparent shape of galaxies within the half-light radius). The shape of the light profiles can be related to the formation of galaxies. They either continue increasing at the last resolution point (typically about 10 pc for galaxies in this study) or they turn downwards at some point and stay flat until the last observable point (so, it is critical to use Hubble Space Telesope observations). The nuclear regions in galaxies where the light profile turns down and stays flat are called "cores". The other type of light profiles, those that keep increasing at the smallest spatial scales we can probe, are sometimes called "core-less" (or power-laws). The idea is that cores are made by massive black holes which are bound in a bianry system and kick far out all stars that cross their paths. These black holes are brought together in a merger of two galaxies of simular sizes. However, if there is a lot of gas present during the merger (i.e. brought by one of the galaxies), even if cores are created by the massive black hole binary, new stars will be made from the gas and the light profile will be core-less.

There are several ways to determine which galaxy has a core or a core-less profiles. One can, for example, fit an outer piece of the light profile with a function (such as the Sersic 1968 profile) and then estimate how much does the nuclear light profile deviate from this outer fit (e.g. Kormendy et al. 2009), in other words shows an "excess" (e.g. core-less) or a "deficit" (core). Anotehr way is to fit a more specific function, such as the so-called core-Sersic function (Graham et al. 2003). In this work, we used the so-called "Nuker-law" (Lauer et al. 1995), which is a double power-law function meant to be applied only to the inner regions of the light profiles (not on the full galaxy). Each method gives somewhat different results (as it measures different things), but, as we show in the paper, they are not very different.

The two diagrams above show the angular momentum versus the ellipticity for 260 ATLAS3D galaxies. On both panels, small open symbols are galaxies with no available HST observations (we do not have data to investigate their nuclear light profiles) and filled small symbols are galaxies for which the classification was not possible (it is uncertain due to significant presence of nuclear dust which masks the central light). Also on both panels, colours of symbols indicate the class of the nuclear profiles: red -- core, blue -- core-less. The green solid line separates fast from slow rotators (divsion of early-type galaxies into slow and fast rotators is described in Emsellem et al. (2007) and Emsellem et al. (2011) papers). Left: Core galaxies are shown with squares and core-less galaxies with circles. The grid of dashed and dotted lines show the region where one can expect to find galaxies that are oblate and axisymmetric, and of certain anisotropy in the distribution of velocities (more details can be found in the paper by Cappellari et al. 2007). Right: Shapes of symbols indicate the kinematic group (Krajnović et al. 2011): group a -- non-rotating galaxies, group b -- featureless non-regularly rotating galaxies, group c -- kinematically distinct cores, group d -- 2σ galaxies made of two counter-rotating discs, and group e - regularly rotating (disc-like) galaxies. Kinematic classification is not provided for galaxies with no HST data. The contours show the distribution of a family of oblate objects with an intrinsic shape of εintr= 0.7 ± 0.2 (from Emsellem et al. (2011)).

Conclusion: fast rotators are typically core-less while slow rotators are cores, but there are significant exceptions to this rule, with an implication that either the accepted formation of cores may not be the only path for their formation, or that there are some additonal core perservations mechanisms at play. Perhaps the most interesting finding is that there are some core-less slow rotator galaxies (blue symbols below the green line), pointing to more complex formation of these galaxies (i.e. formation in gas rich mergers). Unfortunately, a significant number of slow rotators which could be in this regime has not HST imaging. This will be changed soon, as we were awarded HST/WFC3 observations during the Cycle 21 of 12 galaxies in this region, which will allow us to determine the formation paths for these galaxies.

ATLAS3D Paper 17, Fig.5

(appeared on the main page on 17.09.2012.)

Bellow is a highlight of my recent work based on ATLAS3D data. The two diagrams are from Krajnović et al. 2012) and show the relation between λR, a proxy for specific angular momentum of stars within the effective radius (enclosing half of the total galaxy light), and ε, the measure of the apparent shape, for a complete and volume limited sample of nearby early-type galaxies. Galaxies that are above the green line are fast rotators (with high angular momentum) and galaxies below the line are slow rotators (with low angular momentum). In the traditional classification of early-type galaxies, fast rotators are basically S0s, while slow rotators are true ellipticals (for details see Emsellem et al. 2011, MNRAS,414, 888).

Symbols on the left panel are related to the Sérsic index (a measure of the curvature of the light distribution) of the bulge component (galaxies were decomposed into "bulges" and "discs", but this didn't work for all: red symbols are single components fits), while the colours relate to the fraction of disc light in the these galaxies. The right panel is similar, but the symbols are now related to the type of motions of stars found in these galaxies (and explained in Krajnović et al. 2011, MNRAS, 414,2923). Conclusion: a) most of early-type galaxies have a significant fraction of stars in discs, b) Sérsic indices of bulges of early-type galaxies are often small (nb<3), c) fast (ordered) rotation is related to existence of discs, d) slow rotators (ellipticals) actually have large Sérsic indices and often can not be decomposed into two components.

Last modified: 12 Sep 2014