Central dark matter content of dwarf galaxies
Kyle Oman, 29 May 2020
Dwarf galaxies typically have very low stellar and gas densities, such that even in their centres the local dark matter density exceeds that of ordinary matter. The dark matter is therefore the primary driver of the dynamics of such systems, making them ideal laboratories to investigate the detailed internal dark matter distribution of galaxies. There is a long-standing debate over whether the central density of dwarfs is compatible with galaxy formation in the fiducial ΛCDM cosmology. While making a measurement of the central density is conceptually straightforward, in practice there are many possible pitfalls and caveats which must be understood in detail before drawing conclusions about cosmology.
Research project
In this project we investigate what possible combinations of cosmological and galaxy formation models are compatible with the apparent diversity in dwarf galaxy kinematics, while also taking into account the extent to which systematic errors in the kinematic modelling of galaxies may compromise the inference of the dark matter density in galaxies. This is supported by detailed comparisons with numerical simulations, where simulated galaxies are ‘observed’ with a virtual radio telescope to put their analysis on even footing with that of real galaxies.
The diversity of dwarf galaxy kinematics
The basic measurement of dark matter density in late-type dwarf galaxies is tied to the rotation curve &ndash the speed at which the gas disk of the galaxy rotates at each radius. The more mass that is contained within a given radius, the faster the gas must orbit, otherwise it would collapse to the centre of the galaxy. The rotation curve therefore corresponds to a measurement of the total mass distribution within the galaxy. The rotation speed can be measured from the Doppler shift of spectral lines emitted in different locations across the galactic disc &ndash the 21-cm radio emission line of neutral hydrogen gas is often used. To get from a rotation curve to a measurement of the dark matter distribution, the contribution due to ordinary matter must be removed. The density of stars is tied to the density of light, so optical observations give a handle on the stellar distribution, while gas density is instead often estimated based on the intensity of the 21-cm line. Once the distribution of ordinary matter is subtracted from the total matter distribution, the remainder is the dark matter distribution.
Following the above process, the central dark matter density of many dwarf galaxies has been measured in an observational and modelling effort spanning several decades. The picture that emerges is one where otherwise broadly similar dwarfs can have very different central densities, to an extent much greater than that expected based on the properties of dark matter halos in the fiducial ΛCDM cosmology. In an earlier part of this project, we arrived at a re-statement of this discrepancy in which the number of modelling steps and required assumptions are minimized as much as possible, which we published in 2015.
Possible explanations for this discrepancy fall into four broad categories:
- The cosmological model may need to be modified. For instance, instead of the fiducial cold collisionless fluid, the dark matter may have a self-scattering interaction (SIDM).
- The gravitational force law may need to be modified. As one example, the MOND model can explain some features of galaxy rotation curves.
- The galaxy formation model may need to be taken into account. Violent gas flows driven by supernova explosions can couple gravitationally to the dark matter and re-distribute it within a galaxy, but there is currently no consensus on the extent to which this mechanism operates in real galaxies.
- The kinematic models used to extract rotation curves from astronomical measurements may need to be improved. The rotation curve may not be a faithful tracer of the underlying mass distribution.
We recently published an overview of the merits and drawbacks of each of these possible solutions, aided by a new diagnostic shown in Fig. 1. We found that none of them currently offer a fully satisfactory explanation for the diversity of dwark kinematics, although it seems clear that at least some of the diversity must be attributed to systematic errors in the modelling of observations.
 |
|---|
| Figure 1. Loosely speaking, ηrot is a measure of the central dark matter density, with higher density systems having higher values of ηrot, while ηbar is a measure of the degree to which dark matter is dynamically important in the central region of each system, with ‘dark matter-dominated’ systems on the left. Dwarf galaxies (log(Vmax)~1.8, orange) exhibit an interesting anti-correlation between these two metrics, which turns out to be a diagnostic able to discriminate between different scenarios seeking to explain the diversity in dwarf galaxy kinematics. (Taken from Santos-Santos et al. 2019.)** |
Synthetic observations of simulations
In a 2019 paper, we used our MARTINI software package to ‘observe’ galaxies from the APOSTLE suite of galaxy formation simulations. An example is shown in Fig. 1. The syntetic observations capture the details of how 21-cm line emission would be produced by the galaxies simulated, and how this would be received by a radio telescope, such as the Karl G. Janksy Very Large Array. The resulting data files can be manipulated and analyzed in exactly the same fashion as those for real galaxies. By using the same 3DBAROLO kinematic modelling software used in recent studies of local dwarf galaxies, we were able to assess to what extent the rotation curves we measured were faithful representations of the ‘true’ rotation curves, which we can easily measure directly from the raw APOSTLE simulation outputs.
  |
|---|
| Figure 2. The left panel shows the gas density map of a simulated galaxy with a strongly warped disc. This galaxy was ‘observed’ (at one orientation) using the MARTINI tool. The right panel shows a visualization of the observation. Conceptually, a radio telescope does not record ‘images’, but rather an emission intensity as a function of both position and frequency, combining an image and a spectrum into one. The right image shows a visualization of this intrinsically 3D data, with two spatial axes corresponding to (RA, Dec), and a frequency axis. The surface is an iso-intensity surface and rotates between the intensity map projection (roughly ellipsoidal) and the major-axis position + frequency projection (roughly diagonal band). The strong warp seen on the left leads to many of the large ‘bumps’ in the right panel. |
We found that non-circular motion &ndash gas which orbits on an elongated, approximately elliptical path, rather than a circle &ndash can cause significant errors in the rotation curve measurement, even when the amplitude of the non-circular component of the orbits appears to be small.
Revisiting measurements of local dwarfs
We are now prepared, in terms of methodology and technical development, to undertake a critical re-assessment of 21-cm survey data sets. We will search for the subtle signatures of asymmetries and kinematic irregularities which can significantly impact the reliability of the rotation curve measurement. This will be supported by comparison with next-generation numerical simulations, such as those which will make up the forthcoming Colibri and Lyra suites. The ultimate aim is to establish the (in)compatibility of observed dwarf galaxy kinematics with galaxy formation in the fiducial cosmology &ndash is a warm or self-interacting particle required to reconcile theory with measurement?