53. Dark Matter in elliptical galaxies?

(Guest post by Dr. Jörg Dabringhausen, Charles University in Prague, Dec. 18th 2020)

The hypothesis of dark matter in galaxies was originally brought up by observations. Zwicky (1933) first found out that galaxies were usually moving too fast to stay in the observed galaxy clusters, if the luminous matter was all there is in galaxies. With “luminous matter”, essentially all stars were meant. Stars are understood well in terms of how much mass in a star leads to a certain light strength, or luminosity. But if the light emitted by the galaxies in a galaxy cluster is translated to a stellar population similar to the stellar population of the Milky Way, the stellar population would not have enough mass by a factor of a couple hundreds to keep the galaxies bound to the cluster. Thus, the galaxy clusters would have dispersed billions of years ago, and today we would be surrounded by a uniform distribution of galaxies. But that is not what we see: galaxies are still in galaxy clusters today.

But the problem was not only with galaxy clusters. Rubin & Ford (1970) found out, that the Andromeda Galaxy rotates so fast, that its stars would disperse if only the standard gravity would keep them together. And the Anromeda galaxy turned out be the rule rather than the exception; all spiral galaxies that were studied later on showed similar trends (for example Rubin et al. 1980). So, not only galaxy clusters would disperse, but also the (spiral) galaxies themselves. It is like the riders (that is the stars) on a merry-go-round (that is the galaxy). Forces keep the riders on circles around the merry-go-round, and if the forces for some reason become weaker or cease to exist (for example because the link between the rider and the merry-go-round breaks), the riders would move away from it. But again, this is against our observations: There are large spiral galaxies everywhere around us (including our Milky Way), and the stars in them move on stable orbits.

In general, the problem of missing mass in galaxies is nowadays omnipresent. It arises because there are different ways to estimate masses in astronomy. One such way is to make educated guesses about the age and the composition of the stellar population of a galaxy, and calculate from there how much units of mass it should have per unit of luminosity. Astronomers call this a stellar mass estimate. Another way is to measure the radius of a galaxy and how fast stars move on average in it, then make some educated guesses about the dynamics of the galaxy, and calculate the ratio of mass to light from there. Astronomers call this a dynamical mass estimate. Ideally, stellar and dynamical mass would agree for the same galaxy, because the galaxy only has one real mass (within uncertainties, of course). In practice however, the dynamical mass is usually larger than the stellar mass, and the factor ranges from slightly above one to 10000 or so. Apparently, the error lies somewhere in the guesswork leading to the two different mass estimates. Astronomers tried to solve the problem of the missing visible matter in two general ways: Either by adding more matter, so that the matter in total would produce the observed gravitational force, or by changing the laws of gravity themselves and saying that the visible matter is all the matter there is in galaxies.

Adding more matter is mathematically the simpler solution, which is also why many people favoured it at first. The gravitational force is then linear in the critical range of values, that is weak to moderate gravity. This means that if there is twice the matter, there is also twice the gravitational force, independent of the total amount of matter there is. Note that from this point of view, the type of matter does not matter, as long as it is invisible, or nearly so. Also the Earth is near invisible next to the Sun, even though they both consist basically of the same kind of matter (that is atoms, not something exotic). It is only a matter of temperature that makes the Sun brighter than the Earth. Indeed, there was a theory that the missing matter are earth-like bodies (that is free-floating planets and brown dwarfs), until the needed quantity of those bodies was observationally excluded. More and more alternatives for the additional matter were excluded as well, so that we are today at the Lambda-Cold-Dark-Matter Model (LCDM-model) for this class of models. However, the LCDM-model requires exotic dark matter beyond the standard model of particles. But this kind of matter has not been discovered yet, including in the largest accelerators like CERN. Nevertheless, this first group of physicists still believes the LCDM-model to be true in general (even though there are some changes to be made) and therefore they continue to search for the so far still hypothetic dark-matter particle.

The second group of physicists rather correct the law of gravity than adding a hypothetic particle beyond the standard model of particle physics. It is like whichever way you go, you have to expand a theory which has been extremely succesful so far: you either have to give up the standard model of particle physics in order to save the LCDM-model, or have to have to give up general relativity, with Newtonian gravity as its limiting case for weak and moderate gravity. This new theory of gravity is, unlike Newtonian gravity, not linear in the critial range. This means that twice the matter does not necessarily mean twice the gravity when the gravitational force is weak enough. This has a funny consequence, which is in contrast to our daily-life experience, namely that the same amount of matter suddently looks like it becomes more gravitating when you spread it out thinly enough. Lüghausen et al. (2015) therefore called it “phantom dark matter”, because this dark matter is a mirage that disappears when the real matter is put close enough together. (Of course, inside the Solar system, the matter must be on average dense enough for the gravitational force to be linear – otherwise we would not be able to send spaceships with high precision to other planets using Newtonian gravity.) This second set of theories leads to Modified Newtonian Dynamics or Milgromian Dynamics (MOND).

Here, I will concentrate on the “missing” matter of elliptical galaxies – “missing” in the sense that there is usually less matter if seen from a stellar perspective than if seen from a dynamical perspective on the same galaxy. Are there alternatives to adding exotic dark matter to the visible matter, and thus supportive to the second group of physicists?

First of all, let’s start with the question of what an elliptical galaxy is. A very short answer would be that they are more or less like the spiral galaxies, but without the disks that contain the spirals. So, only the central bulge is there, and hence, they are called ellipitical because of their elliptical shape. That central bulge can however be very massive, and the most massive elliptical galaxies are even more massive than the most massive spiral galaxies (bulge and disk of the spirals together)!

Going a bit more to the details of elliptical galaxies, they show however some diversity in their mass and radius. I will distingish them into three different kinds of objects, namely ultra-compact dwarf galaxies (UCDs), conventional elliptical galaxies (Es) and dwarf spheroidal galaxies (dSphs), and discuss the invisible matter in each of them. We will see that the invisible matter is just a mirage in some of them, while others contain really some more matter than originally accounted for, but not the exotic dark matter predicted by the LCDM-model.

1) UCDs

UCDs (Figures 1 and 2) stand a little apart from the other elliptical galaxies, and some doubt that some of them really are galaxies, and not just very massive star clusters. The reason lies in their compactness, which makes them look much like very massive globular clusters. However, their compactness also places them deeply in the Newtonian regime, so there is literally no room for the phantom dark matter of MOND. Yet, it was claimed that they may contain dark matter (see for example by Drinkwater et al 2004 and Hasegan et al. 2005).

Figure 1: A “family picture” of elliptical galaxies. The two bright objects near the center and at the upper right corner, respectively, are the bright “conventional” elliptical galaxies NGC 1404 and NGC 1399. Slightly above NGC 1404 is a UCD, and near the lower border of the image is a small conventional elliptical galaxy. Not all the spots on the image are galaxies. There are stars too, and even bright ones. They can be recocnized by the spikes around them. However, the described galaxies are all members of the Fornax Galaxy Cluster. They are thus at approximately at the same distance, and are consequently to scale to each other. Image credit: Michael Hilker.
Figure 2: This is probably what a UCD would look like if it was a member of the Milky Way. Shown here is actually Omega Centauri, the brigtest globular cluster of the Milky Way. However, it has sometimes been doubted that Omega Centauri really is a globular cluster, and not a UCD, because of some peculiarities of Omega Centauri. For instance, Omega Centauri shows clear evidence for stellar populations of different ages, as opposed to other, less massive globular clusters. Image credit: ESO.

The reason for that is that at the turn of the millenium, it was popular among atronomers that the stellar initial mass function (IMF) is universal (see for example Kroupa 2001). What this means is that all stellar systems formed with a fixed ratio of massive stars to light stars, and only the age of the stars and their chemical composition may change from stellar system to stellar system. This is not to say that people back then were unaware of the influence that, for example, different temperatures and chemical composition had on the process of star formation. Rather, they were looking for different IMFs, but did not find supportable evidence for them in resolved stellar populations. However, when modeling a UCD (or any other kind of stellar system) with the universal IMF, there is maximum ratio between stellar mass and stellar light that can be reached for any reasonable stellar ages and chemical compositions. Nevertheless, there are many UCDs above that limit, and Dabringhausen et al. (2008) showed that this is not just a statistical uncertainty. So, there must be a reason for this unseen mass, and the exotic dark matter that comes with the LCDM-model was a proposition.

However, Murray (2009) voiced serious doubts that the LCDM-model could accomodate enough exotic dark matter inside the tiny radii of UCDs. This is even though the dark-matter halos around the galaxies can be very massive in the LCDM-model. However, the LCDM-model then also predicts that the halos would be very extended, and thus the density (that is mass per volume) of the dark-matter halo would be very thin. So, the total mass of the dark-matter halo may be gigantic, but the fraction of its mass inside a UCD would be tiny because of the small radius of the UCD, and this tiny amount of dark matter inside the UCD would not influence the internal dynamics of the UCD much. Thus, in short, it is not the exotic dark matter of the LCDM-model that increases the mass of the UCDs. It is then likely “conventional” matter, for example from a different IMF. Thus, the word “universal” IMF is then misleading because the IMF is in fact not universal, but “standard” IMF or “canonical” IMF are pretty good replacements. After all, this IMF pretty much seems to be the standard in our immediate surroundings (in an astromical sense); that is regions whose mixture of chemical elements is like that of the Sun and which do not form so many stars at present.

In UCDs, the conditions under which star formation took place were probably far away from those we know to produce the standard IMF. Thus, Dabringhausen et al. (2009) proposed that the UCDs may have formed with an IMF that had a different shape than the standard IMF, namely one that formed more massive stars. (IMFs that have more massive stars than they should have according to the standard IMF are called “top-heavy”.) These massive stars are known to be short-lived, and after they have burned all their nuclear fuel, they leave remnants which produce little or no light compared to their mass. These remnants exist of course in any aged stellar population, but if the IMF had more massive stars once, it has more stellar remnants now. The stellar remnants thus increase the ratio between mass and light, and make a UCD “darker”. Dabringhausen et al. (2012) also tried an alternative way to detect those additional stellar remnants by looking for systems, where a stellar remnant accretes matter from a companion star. Those stellar systems become distinctive X-ray sources, and are thus countable. They compared the numbers they found in UCDs to the numbers they found in globular clusters (that is stellar systems more or less like UCDs, but less massive), and they found more X-ray sources in UCDs than they expected. This as well could indicate that there are more high-mass stars per low-mass stars in UCDs. Based also on their works, Marks et al. (2012) proposed an IMF that changes with the mass of the stellar system (that is from globular clusters to UCDs) and with the chemical composition. Thus, they gave up the notion of the universal IMF, but explained changes in the ratio between mass and light in UCDs with changes in their IMFs.

Another way to increase the mass of UCDs, but not their emission of light, are central massive black holes. In a black hole so much mass is kept, that nothing that comes too close to it can escape it, not even light. Black holes are a prediction of general relativity and known to exist. For example, very massive stars become black holes when all their nuclear fuel is burned, and the pressure from stellar radiation no longer opposes the pull of gravity. Or, as another example, there is a massive black hole at the center of the Milky Way, and many other galaxies as well, even though it is less clear than for massive stars how those came to be. (This year’s Nobel Prize for physics was about the detection of this central black hole.) But if massive black holes are common at the centers of galaxies, why can’t UCDs have them as well? However, a massive central black hole is easy to overlook at the distance of known UCDs. That is because at the distance of UCDs, the stars look like they are almost located at a single point in space, whereas the mass of the central massive black hole is precisely located a this single point. Thus, if seen from Earth, there is not much difference in the distribution of matter, while the central massive black hole would still add its mass to the mass of the stellar population. Therefore, only by careful observations with the telescopes with the best optical resolution, one has a chance to detect them. Nevertheless, massive central black holes were indeed proposed as a solution for the problem of the missing mass in UCDs; for example by Mieske et al. (2013) and Janz et al. (2015). Seth et al. (2014) then observationally confirmed a massive central black hole in a UCD for the first time. Later, massive black holes were also discovered in other UCDs, see for example Afanasiev et al. (2018).

Naturally, also a mixture of non-standard IMFs and central massive black holes is possible to explain why UCDs are so massive for their light. However, what is important here is that there are less far-fetched alternatives to exotic dark matter in UCDs.

2.) Conventional elliptical galaxies

The conventional elliptical galaxies are not only usually more massive than the UCDs, but also far more extended. What I mean with “conventional” is that they were among the first galaxies to be identified as galaxies – this was in the 1920ies, when people like Hubble first discovered that some “nebulae” are not just gas clouds inside the Milky Way, but distant stellar islands just like the Milky Way. It is unclear what mass exactly is required for an elliptical galaxy in order to be coventional, perhaps 108 Solar masses or so. This unclearity is because there is an extension of elliptical galaxies to even lower masses, which are however not (compact, star-cluster-like) UCDs, but (extended, galaxy-like) dwarf Spheroidal galaxies (dSphs). However, there are some specialities on dSphs about dark matter and its seeming existence, and therefore I will treat them in an own section. What I will not do, though, is to distinguish the elliptical galaxies into dwarf elliptical galaxies and elliptical galaxies proper, because this distinction in merely historical in my eyes (see also Ferguson & Binggeli 1994 about this). The most massive of all galaxies (about 1012 Solar masses) are conventional elliptical galaxies, too.

So, how much exotic dark matter do elliptical galaxies contain, if any? Cappellari et al. (2006), for instance, found out that the conventional elliptical galaxies they observed had on average 30 percent too much mass for the IMF they assumed. They suggested that the missing mass could be the dark matter predicted by the LCDM-model. However, for this finding, they also assumed that the standard IMF is universal for all star-forming regions. Tortora et al. (2014) later tried to fix this without exotic dark matter, but MOND. They also failed with a universal IMF, but not if the IMF was changing with the mass of the galaxy. So, the real question is: Can the IMF change with galaxy mass or is the standard IMF also the universal IMF?

For answering this question, let’s look at star clusters, which are the building blocks of galaxies. Could a star cluster have a star more massive than the cluster itself? Of course not. Actually, Weidner et al. (2010) found out that the mass of the most massive star of a star cluster is much lower still. An impressive example of this was observed by Hsu et al (2012): They compared a large cluster of some mass with several adjacent small star clusters with the same mass in total. All the other parameters like age, chemical composition, and so on are the same, just how the total mass of the stars is bundled is different. However, the massive star cluster has heavier stars than the several small star clusters. This would not be a problem by itself, if the overall star formation was the same in all galaxies; that is when all galaxies form the same number of light star clusters per massive star cluster. But this is not the case. Weidner et al. (2004) found that the mass of the most massive cluster that can form in a galaxy depends on its star formation rate; that is how many stars form in a galaxy per time unit. Low-mass elliptical galaxies have low star formation rates and massive elliptical galaxies have high star formation rates. Thus, low-mass conventional elliptical galaxies have a lack of massive stars. This already is an argument against a universal IMF in all star clusters and in all galaxies.

The galaxies with the highest star formation rates (that is also the most massive galaxies) produce also star clusters in the mass range globular clusters and UCDs. Now, lets assume that these most massive star clusters are in fact UCDs and that these UCDs have IMFs with more massive stars per low-mass stars than “normal” star clusters (see the section about UCDs). Then the real IMF deviates from the once-thought universal IMF not only in low-mass star clusters (by not having any massive stars), but also in high-mass star clusters (by having too many massive stars). Now, remember what we have said about IMFs with more massive stars than the standard IMF: when they grow old, they produce less light per unit mass than the standard IMF. Or when a certain amout of light is observed, a stellar population with more massive stars and a certain age must have more mass to produce it. The stellar populations of elliptical galaxies are usually that old that the massive stars (which are short-lived) have already evolved into dark stellar remnants, and only the light stars continue to shine. So, if the IMF behaves with the star formation rate of the galaxies like it is assumed nowadays (see for example Kroupa & Weidner 2003 or Fontanot et al 2017), then the low-mass elliptical galaxies have a little less mass than assumed with the standard IMF for their light, and the massive elliptical have a little more mass than assumed with the standard IMF. This goes up to about twice the mass for the most massive conventional elliptical galaxies, and the point where the mass estimate is equal to that for the standard IMF is at approximately 109 Solar masses. Thus, for most conventional elliptical galaxies, the mass estimates are above the mass estimates for the standard IMF, and the “missing” mass is about the mass detected by Cappellari based on the standard IMF. (See also Dabringhausen et al. 2016 if you want to follow the brightness of elliptical galaxies with their mass, and Dabringhausen 2019 if you wish to go deeper on elliptical galaxies and non-standard IMFs). Thus, again like with UCDs, there is an alternative, more down-to-earth explanation for the excess mass of those elliptical galaxies.

3.) Dwarf speroidal galaxies (dSphs)

Dwarf spheriodal galaxies (dSphs, Figure 3) are in a way the low mass extension to “conventional” elliptical galaxies, because in a plot of their radius against their mass, they continue the line established by the conventional elliptical galaxies to lower masses. However, the brightest ones are in light and mass like UCDs, but way more extended than UCDs. In other words, there is a gap in radius between dSphs and UCDs (see Gilmore et. al 2007), in contrast to conventional elliptical galaxies and dSphs.

Figure 3: The Fornax Dwarf Galaxy. This is arguably the largest dwarf speroidal galaxy around the Milky Way. Image credit: ESO / Digitized Sky Survey 2.

If it is true that dSphs are in fact very low-mass conventional elliptical galaxies, then we would expect them to be about 20 percent or so lighter than expected based on their light with a standard IMF. But in fact, they are way more massive. Just in order get a feeling for the numbers we are dealing with: Let’s say the standard IMF would predict a ratio of mass to light of 2 for a dSph, the ratio for the corrected IMF would then give 1.5, but the measured value is 2000 (all numbers are in Solar units). So, how can we be wrong to a factor up to approximately 1000 (even though in many cases less)?

This is where MOND finally kicks in, because the visible matter in dSphs is actually thin enough, in contrast to UCDs and Es. MOND can rise the ratio of the mass of a dSph over its light from values of a few (that is a stellar population in Newtonian dynamics) to values up to about 100. This fits the dynamical values of many dSphs, which would contain plenty of “dark” matter in Newtonian dynamics. Thus, in MOND, their dark matter is actually phantom dark matter – it would disappear if the matter was denser. Or, in other words, the difference between stellar and dynamical mass estimates disappears for those dSphs, and all is well. The precise value for a given dSph depends on which value the mass-to-light ratio of the stellar population would have according Newtonian dynamics and on how many stars are distributed over which volume, that is the density of visible matter. Estimates for the mass-to-light ratios in Newtonian and MONDian dynamics for a number of dSphs are for example given in Dabringhausen el al. (2016).

But it is also visible in Dabringhausen el al. (2016) that even MONDian dynamics cannot explain the mass-to-light ratios of the few dSphs, which have a mass-to-light ratio far beyond 100. So, have we finally found a failure of MOND? Not necessarily. So far, we have implicitly always assumed that the galaxies are in virial equilibrium. What this means is for instance the absence of tides because of other distracting souces of gravity. The tides on Earth are the best-known example, even though Earth is dense enough to be near tidal equilibrium, given the gravitational forces from the Moon and the Sun. We only see them so well because because in this case, the tides are happening right under our noses. Ultimately, there are tides on Earth because the Earth is an extended body. Thus, the gravitational force from the Moon pull on the near side of the Earth a bit stronger than on the far side, and the Earth is being stretched a bit by the tides. There are ebb and flow of the oceans on Earth, because the Earth also rotates, while the tides are always directed towards the Moon. There of course also other sources of gravity on Earth which cause tides (the Sun for instance), but the Moon is the strongest.

Also UCDs and conventional elliptical galaxies are dense enough to be nearly unaffected by neighboring galaxies, which are the potential reason for tides in them. But the internal gravity is comparatively weak on the thin matter of dSphs, so that they are easy to stretch by outside forces of other galaxies. Thus, the tidal forces form gigantic tidal “waves” consisting of stars. Every encounter with another galaxy pulls on the galaxy, because the gravitational force is stronger on the near side of the encounter than on the far side. This heats the galaxy up, meaning that the galaxy is being pulled out of virial equilibrium by the encounter and that the average velocities of the stars get faster with enconters. Finally, the tidal forces from encounters with other galaxies make the galaxy break apart.

Now, what would an observer from Earth see? The observer could for example see a dSph that has been heated up by a recent encounter with another galaxy, and is thus out of virial equilibrium. Or the dSph has found its virial equilibrium again, but at the cost of stars which have left the dSph, and are now moving faster or slower than the stars which are still bound to the galaxy. But the observer could be ignorant of this fact, and assume that all the stars (s)he sees are bound to the galaxy. Or the dSph has dissolved already completely, but the stars still move all along on similar orbits, even though they are not bound to each other any more. The radius in which the stars are is then just much larger than it would be, if the stars were bound to each other. If the observer then wrongly assumes the dSph to be in virial equilibrium, all these effects increase the dynamical mass estimate (not the real mass!) (s)he makes for the mass of the galaxy. And those effects could indeed raise the dynamical mass estimate by the required factor. For a discussion of tidal heating of dSphs under Newtonian gravity, see for example Kroupa (1997). McGaugh and Wolf (2010) made a similar study with MOND. Notably, they found for observed dSphs surrounding the Milky Way that if a dSph is more susceptible to tidal forces, it is also more likely to be outside virial equilibrium for MOND. For an interesting theoretical discussion of how a dissolving star cluster in a tidal field could be mistaken for a much more massive (but evidently not more luminous) dSph, see Dominguez et al. (2016).

However, the dSphs which are out of virial equilibrium far enough to increase the dynamically estimated mass-to-light ratio by a few or more compared to the real mass could just be a few dSphs out of a larger sample. For the majority, the effect would simply be too weak now, although their time to dissolve will also come. In other words, this scenario is highly improbable if gravity was Newtonian, because then all dSphs around the Milky Way must be in dissolution. However, if gravity is MONDian, only a few would be near their dissolution, while most would be in or near virial equilibrium – see Dabringhausen el al. (2016).

Figure 4: The Antennae Galaxies (NGC 4038 and NGC 4039), the “poster child” for interacting galaxies with matter trown out as tidal tails. Image credit: Daniel Verschatse.
Figure 5: Another example for interacting galaxies and tidal tails, the Tadpole Galaxy. Despite its name being singular, there is actually two galaxies instead of one at the distance of the Tadpole Galaxy. The smaller galaxy is however currently projected on to the disk of the larger galaxy. It can be seen as blueish blob to the upper right of the larger galaxy. There are also star-forming regions in the tidal tail, which may become tidal dwarf galaxies, if they do not fall back on the primordial galaxies. Image credit: Hubble Space Telescope.

There is also another argument against dark matter in dSphs. Galaxies are usually not by themselves, but surrounded by other galaxies. Together, these galaxies form gravitationally bound galaxy clusters. But how do these galaxy clusters form? According to the LCDM-model, this happens by the infall of galaxies from all directions. They can come, the dSphs included, with any amount of exotic dark matter into a galaxy cluster. We will call those galaxies “primordial galaxies” from now on, because there is also another way to form galaxies that look like dSphs to an observer. This other way is through close encounters of already existing galaxies. In such encounters, matter is pulled away from the existing galaxies by gravity though tides (Figures 4 and 5), and new small galaxies can form from this matter. We know that this process happens. Otherwise, the elongated streaks of matter of, for instance, the Antennae Galaxies and the Tadpole Galaxy would be difficult to explain. Simulations of interacting galaxies, which are set up to reproduce situations like in the Antennae Galaxies, show also those streaks of matter like the ones observed (see for example Bournaud & Duc 2006 or Wetzstein et al. 2007). They are called tidal tails for obvious reasons. The Tadpole Galaxy even has a new small star-forming regions in its tidal tail, which may become dSphs. If aged enough, these dwarf galaxies may be difficult to distinguish from primordial galaxies of the same mass, though (see Dabringhausen & Kroupa 2013). However, in the following, we call galaxies of tidal origin “tidal dwarf galaxies”, in order to distinguish them from primordial galaxies. The tidal dwarf galaxies cannot contain the exotic dark matter of the LCDM-model, even if their progenitor galaxies did. The reason is that all matter that ends up in a tidal dwarf galaxy, whether visible or not, must have occupied similar regions of space with similar velocities also before the encounter of the existing galaxies. The total amount of the exotic dark matter may be huge, but most dark matter had other velocities and other locations, and therefore does not qualify to be bound to the tidal dwarf galaxy. After all, simulations of galaxy encounters by, for example, Barnes & Hernquist (1992) show that most visible matter that is to become a tidal dwarf galaxy comes from the disks of spiral galaxies. This visible matter does not only form a thin disk, as opposed to the presumed dark matter halo, but it also moves with the same velocity in the same direction, again in contrast to the presumed dark matter halo. Also, the tidal dwarf galaxies that form in an encounter of galaxies can only move in the plane of the encounter (because of the conservation of angular momentum). Thus, there is an easy way to distinguish the dSphs in the LCDM-model: those which move in a plane and those which cannot be assigned to a plane. Those in a plane are very likely tidal dwarf galaxies and cannot have any exotic dark matter. Those, however, which cannot be assigned to a plane might also be primordial and can thus contain dark matter (see for example Kroupa et al 2010). Now, what do observations tell us about the pattern of motion of the dSphs? In the Milky Way, it was shown by Lynden-Bell (1976) and by Kroupa et al. (2005) that the then known dSphs are most likely arranged in a plane. Later, additional objects and also velocities were added, but the long-lasting disk of Satellites was always confirmed (see for example Pawlowski et al. 2012 and Pawlowski & Kroupa 2020). This was according to some proponents of the LCDM-model just an exception, while other, they said more normal galaxies would have dSphs with random motions around them. However, it was shown then that also the Andromeda Galaxy has a disk of dSphs around it (for example Ibata et al 2013), and Centaurus A as well (Mueller et al 2018). In short, disks of satellites around major galaxies are more the rule than the exception, see for example Ibata et al (2014) for an attempt of a census. Thus, galaxies in these planes must manage their high dynamical mass-to-light ratios without exotic dark matter, despite numerous claims to the contrary from the LCDM-community. If MOND is the correct description of gravitation, then the large gravitating (phantom) masses of the satellite galaxies, as opposed to their small masses in stars, is beautifully resolved.

Conclusion

I have discussed the reasons for “dark” matter in elliptical galaxies, which comes ultimately from the comparison of different mass estimates. Also, some assumptions which were used for the lack of better knowledge have been proven wrong by now. This concerns the theory of a universal IMF in all star-forming regions, which was leading to a mismatch between the mass estimates from stellar populations and from the dynamics in UCDs and conventional elliptical galaxies. If the “one-size-fits-all” IMF is replaced by a more elaborate picture of the IMF, those differences disappear easily without using exotic dark matter or MOND. For dSphs, the situation is different. They cannot have exotic dark matter because it could not bind to them, but neither can their extreme mass-to-light ratios be explained with different stellar populations. Here, MOND and tidal fields offer an answer. Thus, adding more exotic dark matter to all galaxies until their dynamics is fitted might appear the simpler solution on first sight, but it is not necessarily the correct one. The seemingly more complicated solution without exotic dark matter stands a better test result here.


In The Dark Matter Crisis by Joerg Dabringhausen. A listing of contents of all contributions is available here.

43. Pavel Kroupa on ” The vast polar structures around the Milky Way and Andromeda “

In case you, like me, have missed Pavel Kroups’s recent talk at the Joint Astronomical Colloquium in Heidelberg, you now have the opportunity to watch a movie of the event and download the slides. The movie is quite long (more than an hour), but it is worth watching it to the end. While the talk is titled “The vast polar structures around the Milky Way and Andromeda”, Pavel talks about much more, starting with tidal dwarf galaxies and ending with a discussion of indications for an alternative model of gravity.

This presentation is very similar and in most parts identical to Pavel’s presentations held at Monterey at the conference “Probes of Dark Matter on Galaxy Scales” and in Durham at the “Ripples in the Cosmos” conference. The latter talk resulted in quite a discussion on Peter Coles’ (aka Telescoper) blog “In the Dark”, following his criticism of Pavel’s talk as being “poorly argued and full of grossly exaggerated claims”. The video of a very similar presentation now offers everybody the opportunity to develop their own opinion on the issue. Given the numerous questions Pavel got during his talk and afterwards, people must have thought that it was worth the effort to argue with him, in contrast to Peter’s opinion.

 

See the overview of topics in The Dark Matter Crisis.

38. Are there two types of dwarf galaxies in the universe?

Dwarf galaxies, that is galaxies less massive than a few billion solar masses, are expected to be formed through two processes. They might either be the luminous components of small dark matter halos, formed early in the universe when gas fell into the potential well of those halos. These dwarf galaxies are called primordial dwarf galaxies (PDGs) and are expected to be dominated by their dark matter content.

The other formation mechanism is a process observed even in the present-day universe. When two major disk galaxies collide, the gas and the stars in the disks are expelled by tidal forces induced by the encounter to large distances. An example for a very prominent structure that has been created through tidal interactions between disk galaxies is the ‘tail’ that extends to the upper right corner in the figure below. Within this tidal debris, new objects of dwarf galaxy mass form. This is why dwarf galaxies of this second type are called tidal dwarf galaxies, or TDGs.

Thus, TDGs form from the baryonic material in the galactic disks of the progenitor galaxies, but can they also contain dark matter? Even in a disk galaxy with a massive dark matter halo, the vast majority of the dark matter would be located outside the galaxy’s disks. Of the small amount of dark matter within the disk, only a tiny fraction would furthermore be moving in the same direction and would have the same velocity as the stars and the gas in the disks. The vast majority of the dark matter would therefore have different initial conditions regarding its location and motion than the gas and the stars. But during a galaxy collision, only material with similar initial conditions is thrown on similar trajectories by the tidal forces and has a chance of becoming bound to the gravitational field of a forming TDG. The vast majority of the dark matter, having different initial conditions, will therefore be thrown onto different trajectories. While the dark matter on such different trajectories may be able to cross the shallow gravitational field of a TDG, it would do so at a high relative velocity. Therefore, this dark matter cannot become bound to the TDG. As an analogy for an encounter between a TDG and a chunk of dark matter, consider two spaceships orbiting a planet. Even if they orbit the planet at the same altitude, they can only rendezvous if they follow each other on the same orbit. For all other possible choices of orbits (say one is flying to the south and the other is flying to the west), the spaceships would fly past each other quickly if they do not crash.

In summary, it is one of the major characteristics of TDGs that they cannot contain much dark matter, even if their progenitor galaxies did (e.g Bournaud 2010).

TDG

Credit: NASA, H. Ford (JHU), G. Illingworth (UCSC/LO), M.Clampin (STScI), G. Hartig (STScI), the ACS Science Team, and ESA

If the standard model of cold dark matter is correct, there should be a co-existence of these two types of dwarf galaxies in the universe: dark-matter dominated PDGs and TDGs without significant dark matter content. This is the Dual Dwarf Galaxy Theorem (Kroupa 2012).As they would have very different compositions, the two types should fall into two easily distinguishable groups. The natural question to ask in order to test this prediction is:

Are there really two distinct populations of dwarf galaxies in the universe?

This is investigated in the article “Dwarf elliptical galaxies as ancient tidal dwarf galaxies” by Dabringhausen & Kroupa (2013). The principle of their study is simple: they just had to compare the observed properties of old dwarf galaxies with known tidal dwarf galaxies. For the comparison, they use two properties, which are easy to determine observationally. These properties are:

  • The stellar mass, i.e. only the mass in stars, without the mass in gas, dust or dark matter. It can be determined from the luminosity of the system (more stars = brighter object).
  • The projected half-light radius, which is a measure of how extended the system is.

There are extensive catalogs listing these two properties for so-called pressure-supported systems, i.e. systems of stars in which the stars move on chaotic orbits (in contrast to the ordered rotation of  disc galaxies). The following plot shows these data points.

Screenshot from 2020-10-26 11-27-10

                                           Credit: Dabringhausen & Kroupa (2013)

These objects include globular clusters (GCs), ultra-compact dwarf galaxies (UCDs), massive elliptical galaxies (nEs), and dwarf elliptical galaxies (dEs). The first two types of objects (green points) appear to be free of dark matter, while the second two (red points) are generally assumed to sit in dark matter halos. The study of Dabringhausen & Kroupa is particularly interested in the dEs, as these are in the mass- and size-range of observed TDGs, but are generally assumed to be PDGs.

Adding Tidal Dwarf Galaxies

For a meaningful comparison, the properties of these dEs have to be compared with those of known TDGs. To be confident that an object is a TDG, it has to be associated with interacting galaxies (another possibility is to look at numerical simulations of galaxy collisions and extract the properties of TDGs formed in those models). However, this gives rise to a complication: TDGs associated with a pair of interacting galaxies are young, many of them are still forming some stars and such young TDGs can contain a lot of gas. The dEs, in contrast, are old systems without gas. So the observed properties of the young TDGs have to be aged before they can be compared to the dEs. As the TDGs age, they will loose their gas. The paper lists three possible processes:

  1. The gas is converted into stars.
  2. The gas is removed because the feedback of massive stars in the TDG heat it.
  3. The gas can be removed through ram-pressure stripping as the TDG moves through the intergalactic medium.

Because those gas-removal processes happen slowly, their major effect on the TDG properties is an increase of the system’s half-light radius: as (gas) mass is lost, the TDG will be less bound and the distribution of stars will expand. This allowed Dabringhausen & Kroupa (2013) to estimate where aged TDGs would show up in the figure:

Screenshot from 2020-10-26 11-30-25

                                             Credit: Dabringhausen & Kroupa 2013

The TDGs (blue symbols) fit in quite nicely with the dEs. The lower points on the error bars represent the TDG properties as observed, i.e. still young. Their radii are a lower limit: the TDGs cannot shrink as they slowly loose their gas. The upper end of the error bars assumes that most of the TDG’s mass, 75% to be precise, has been lost. This coincides nicely with the upper end of the dE distribution, too. There is in principle no reason why a TDG couldn’t loose even more of its initial mass, but such TDGs are likely to be destroyed very easily (see further below).

So, the TDGs and the dEs populate the same region in the figure. What does this tell us?

Due to their different composition (PDGs being dark matter dominated, TDGs being dark matter free), one would expect to observe two distinguishable groups of dwarf galaxies. The opposite is found: dEs populate only one region in the plot, and the same region is covered by (aged) TDGs. Consequently, this suggests that the observed dEs are in fact old TDGs. But then there is no room for primordial, dark matter-dominated dwarf galaxies.

This finding is also consistent with the expected numbers of TDGs in the universe. Numerical simulations of close encounters between possible progenitor galaxies show that on average one or two long-lived, massive TDGs are created per such encounter (see Bournaud & Duc 2006). By considering the total number of encounters between possible progenitor galaxies until the present day, Okazaki & Taniguchi (2000) found that such a rate of TDG-production would already be enough to account for all dEs in the Universe.

The black lines in the second plot give another hint at a connection between dEs and TDGs. Because TDGs are formed by colliding galaxies, many of the TDGs will end up as satellite galaxies. When such satellites orbit around a much more massive host galaxy, they will be affected by tidal forces. If the satellite is too extended, its own gravity is not strong enough to keep it bound against the tidal forces of the host. The exact radius depends on the masses of the host and the satellite, as well as the satellite’s orbit. The black lines in the plot give an impression of the tidal radius of satellite galaxies, assuming they orbit at a typical satellite distance of 100 kpc around different host galaxies. For the lowermost line, the host is assumed to be heavy, while the uppermost line corresponds to a rather light host. Above a given line, a satellite of a galaxy with the corresponding mass is not stable anymore, but will be disrupted by tidal forces. So if a TDG loses so much mass that it expands above this line, it will be destroyed and vanish from the plot. Thus, if the dEs are indeed TDGs, the position and slope of the cutoff at large half-light radii is easily explained.

Conclusion

The results of Dabringhausen & Kroupa (2013), if confirmed by future studies, suggest that there is only one type of dwarf galaxies in the Universe. Virtually every galaxy that is classified as an old dwarf galaxy, i.e. a dE, would be an aged TDG which originated from the debris of interacting galaxies. We emphasize also that TDGs have been shown to lie on the baryonic Tully-Fisher Relation (Gentile et al. 2007), which they cannot if this relation is defined by dark matter. These results are very problematic for cold dark-matter based models, which predict that in addition to TDGs a plethora of primordial dwarf galaxies with a completely different composition exists as a second group of dwarf galaxies.  However, the result of Dabringhausen & Kroupa (2013) fits in nicely with the peculiarities of the Milky Way (e.g. Pawlowski et al. 2012) and Andromeda (Ibata et al. 2013) satellite galaxies: they co-orbit within thin planes, which is expected for a population of TDGs. But again this distribution is at odds with the predicted distributions of primordial galaxies.

When it comes to their properties and distribution, tidal dwarf galaxies seem to develop a lead over dark-matter dominated, primordial dwarf galaxies.

By Marcel S. Pawlowski and Pavel Kroupa  (07.03.2013): “Are there two types of dwarf galaxies in the universe?” on SciLogs. See the overview of topics in The Dark Matter Crisis.

36. Andromeda’s satellites behave as expected … if they are tidal dwarf galaxies

Today’s issue of Nature contains a very exciting study by Rodrigo Ibata et al. which might be a game-changer in the research areas of galaxy formation and near-field cosmology. It is titled “A vast, thin plane of corotating dwarf galaxies orbiting the Andromeda galaxy” and already now should be seen as a candidate for the most-exciting paper of 2013.

UPDATE Jan. 4th: The article is now also available on the arXiv.

Pavel Kroupa and I have been waiting for this paper to appear for quite some time. Several months ago we’ve heard the first rumors that Ibata from the University of Strasbourg has detected, with great significance, a plane of satellite galaxies around our neighboring spiral galaxy Andromeda (M31). My curiosity even made me look into available data, which supported what we had heard. Chatting with Rodrigo during a recent N-body meeting in Bonn (after his paper was accepted) finally confirmed these rumors. Seldom have I been looking forward to a paper this curiously, while at the same time being aware of its essential content already.

The Facts

So, what is it all about? Ibata and his collaborators have performed the Pan-Andromeda Archeological Survey (PandAS, lead by Alan McConnachie), an extensive observational campaign of the region around the Andromeda galaxy. This survey has unveiled many of Andromeda’s satellite galaxies and allowed the team to measure the distances to these satellite galaxies in a homogeneous manner (Conn et al. 2012). They then looked at the spacial distribution of the satellite galaxies around their host, motivated by the distribution of satellite galaxies of our own Galaxy. Around the Milky Way, the satellites are distributed and orbit in a thin plane, which we recently termed a vast polar structure (VPOS, Pawlowski et al. 2012a). In fact, the satellite objects are correlated to a degree which is at odds with cosmologically motivated expectations.

Now Ibata et al. find that out of the 27 satellite galaxies in their sample, 15 lie in a common plane. They report that this plane has a thickness of only 13 kpc (40,000 light years), while it has a diameter of at least 400 kpc (1.3 million light years), possibly reaching further out beyond the PAndAS survey region. They can rule out that a chance-alignment is responsible for this configuration with very high confidence, the likelihood that such a well-pronounced structure appears at random is only 0.13 per cent.

An illustration of the Andromeda satellite galaxies which belong to the co-orbiting satellite plane. The top-right vie shows the satellites plane edge-on, as seen from the Milky Way, while the bottom left shows the plane rotated by 90 degrees (the orientations of these two views are indicated in the lower right). The top-left is a optical picture of the Andromeda galaxy. Image Credit: Ibata et al.

But it is not only the existence of this plane which is stunning. The plane is aligned perfectly with the Milky Way, in a way such that we see it edge-on. This fortunate orientation allowed Ibata et al. to also look at a kinematical coherence. We can measure the radial velocities of the satellite galaxies, which lie within the plane due to the planes orientation. This reveals that 13 out of the 15 satellite galaxies in the plane show a common sense of rotation. This, again, is similar to the VPOS around the Milky Way, in which at least 8 satellites orbit in the same sense, while at least one is counter-orbiting in the same plane (Pawlowski 2012). The authors state that including this kinematic information into their analysis increases the significance of the satellite plane to 99.998 per cent. This is just amazing.

Here you can find a very nice video animation illustrating the structure’s orientation with respect to the Milky Way.

Unfortunately the letter itself is behind Nature’s pay-wall, so you can only access it if you have a Nature subscription. I’ll update this blog post if a freely accessible arXiv version becomes available. For the meantime, please be referred to the accompanying press releases. UPDATE Jan. 4th: The arXiv version of the article can be found here.

The Interpretation

In my opinion, the importance of this discovery can not be over-stated, which is in line with Nature publishing a comment on the discovery in the same issue (“Astronomy: Andromeda’s extended disk of dwarfs” by R. Brent Tully) and even making the letter its cover story. The about-the-cover text already hints at the study’s importance:

“Recent studies of the dwarf galaxies of the Milky Way have lead some astronomers to suspect that their orbits are not randomly distributed. This suspicion, which challenges current theories of galaxy formation, is now bolstered by the discovery of a plane of dwarf galaxies corotating as a coherent pancake-like structure around the Andromeda galaxy”

I suppose that due to the restrictive space constraints set by Nature (4 pages, 30 references), the letter is short and does not discuss the study’s implications in extensive detail. In their letter, Ibata et al. mention two broad ideas which might lead to an explanation for the structure’s existence.

  • Either all the satellites in the plane were accreted together, which is unlikely because the very small thickness of the satellite plane restricts the size of an accreted group to less than 14 kpc. Such groups are not observed in the universe.
  • Or the satellites within the plane were formed in place around Andromeda, for example as tidal dwarf galaxies.

Overall, the authors prefer not to make any strong conclusions, instead stating that “the formation of this structure around M31 poses a puzzle”, which is also the prevailing tone of the press release. This is why I would like to share some of my thoughts on the discovery and also highlight some very relevant publications that obviously did not make it into the letter.

Filamentary Accretion?

The letter by Ibata et al, but also the comment by Tully, discusses that the accretion of dwarf galaxies along cosmic filaments might be responsible for the structure. However, there are several reasons why this idea does not work. First of all, the filaments found in cosmological simulations are too thick. They would need to be as thin as the observed structures (< 14 kpc) to have a chance to explain the planes, but their size typically is on the order of 500-1,000 kpc. This is supported by studies like Vera-Ciro et al (2011), who, analyzing the behavior of dark matter particles in cosmological simulations, conclude that

“[…] at later times the cross-section of the filaments becomes larger than the typical size of Milky Way like haloes and, as a result, accretion turns more isotropic […]”.

Consequently, the satellite structure can not be both: of filamentary origin and young, which contradicts the argument in Tully’s comment.

In Pawlowski et al. (2012b) we have also shown that even in case of the VPOS of the Milky Way satellites, a filamentary accretion origin can be ruled out because the coherence of the orbital poles of the sub-halos in cosmological high-resolution simulations is not strong enough to explain the alignment of the MW satellite orbits. The filament might initially lead to a preferred direction of infall, but does not produce a thin, co-rotation plane of sub-halos but a prolate distribution. And now the Andromeda satellite disc is even thinner and more coherent than the VPOS. For more details, please have a look at my blog post on filamentary accretion.

Tidal Dwarf Galaxies

In contrast to the often mentioned accretion along cosmic filaments, the tidal dwarf galaxy scenario is a much more natural explanation for co-orbiting discs of satellite galaxies. In this scenario, two galaxies interact, such that the tidal forces rip out matter from the galactic discs, which form spectacular tidal tails. Within this tidal debris new galaxies (tidal dwarf galaxies or TDGs) form, a process which is observed to happen in the universe and also reproduced by simulations. As the TDGs form from a common tidal tail, they share a common orbital direction and are generally found in a thin plane. Just as it is observed around the Milky Way and now Andromeda.

In fact, this TDG scenario can also explain the existence of counter-orbiting satellites, of which there seem to be two in the Andromeda disc and at least one around the Milky Way (Pawlowski et al. 2011). There is even a study proposing that Andromeda experienced such an galaxy encounter (Hammer et al. 2010), during which TDGs have been formed. These might even be responsible for the VPOS of the Milky Way (Fouquet et al. 2012), in which case the Milky Way should lie within the satellite plane around Andromeda … which is indeed the case. Unfortunately, all these very relevant papers did not make it into the short Nature letter.

All this is also why I have to disagree with a sentence in R. Brent Tully’s discussion of the letter (which of course got picked up by the media …). He states that

“No theorist of galaxy formation would have dared to predict such a situation”.

This is not quite true. I would also argue that the authors of Fouquet et al. (2012) have been expecting such a situation in their tidal dwarf galaxy scenario and that most researchers working on tidal dwarf galaxies would probably predict such an orientation for TDGs. Even I wrote about this in my 2012 paper on the Milky Way VPOS:

“The M31 satellites are preferentially distributed in a structure extending approximately from north to south in Galactic coordinates, just as the MW VPOS extends in the north–south direction. A common direction of the satellite distributions of both galaxies is expected in a tidal scenario that formed both satellite populations together, as TDGs form in a plane defined by the orbit of the interaction.”

There is one major argument against the tidal dwarf galaxy scenario: tidal dwarfs do not contain a significant amount of dark matter, while some of the observed satellite galaxies seem to be completely dark matter dominated. This argument is based on two major assumptions which, however, might both be questioned:

  1. The dwarf galaxies are dynamically relaxed, gravitationally bound systems. If they are not and do not contain dark matter, high mass to light ratios might be derived from their velocity dispersion by mistake (e.g. Kroupa 1997, Klessen & Kroupa 1997).
  2. The underlying gravity law is Newtonian. If the gravity law is modified, e.g. In the low acceleration regime, most satellite galaxies would not need dark matter (e.g. Famaey & McGaugh 2012).

Conclusion

Because of the new study we now know that both satellite galaxy systems for which we have full three-dimensional positions available show strong planar alignments. This coherence is also supported by the available kinematic data: the objects in the VPOS around the Milky Way and in the disc of satellites around Andromeda mostly co-orbit in the same direction.

Such a phase-space coherence is expected if the satellite galaxies were born as tidal dwarf galaxies, but completely at odds with all current cosmological simulations in which the satellites are assumed to be represented by dark matter dominated sub-halos. Therefore, the discovery by Ibata and collaborators, in my opinion, supports the tidal dwarf galaxy scenario and will contribute to a paradigm shift in the field of galaxy formation. We might have to re-consider what we know about near-field cosmology and will have to develop a new understanding of the origins of dwarf satellite galaxies. In the end, this publication might even have an impact on our understanding of the laws of gravity.

The cosmological implications of VPOS-like structures are discussed at length in our paper Kroupa et al. (2010) “Local-Group tests of dark-matter Concordance Cosmology: Towards a new paradigm for structure formation” and in the review by Kroupa (2012) “The dark matter crisis: falsification of the current standard model of cosmology”.

By Marcel Pawlowski and Pavel Kroupa  (03.01.2013): “Andromeda’s satellites behave as expected … if they are tidal dwarf galaxies” on SciLogs. See the overview of topics in  The Dark Matter Crisis.