Dwarf galaxies in LCDB : distribution around hosts, dust emission and internal structure

(1)

DWARF GALAXIES IN ΛCDM:

DISTRIBUTION AROUND HOSTS, DUST EMISSION AND INTERNAL STRUCTURE

Memoria de tesis doctoral presentada por Isabel Mar´ıa Eugenia Santos Santos para optar al t´ıtulo de Doctor en Astrof´ısica

Supervisado por

Prof. Dra. Rosa Dom´ınguez Tenreiro & Dr. Christopher Brook

Noviembre 2018

(2)

A mis padres

(3)

This thesis has received support from MINECO/FEDER (Spain) AYA2012-31101 and AYA2015- 63810-P grants. I acknowledge funding from the European Union’s Horizon 2020 research and in- novation programme under the Marie Sklodowska-Curie grant agreement No. 734374 (LACEGAL- RISE) for a secondment at the Astrophysics group of Univ. Andr´es Bello (Santiago, Chile). Also funding from the Univ. Aut´onoma de Madrid for a stay at the Leibniz Institut fur Astrophysik Potsdam to obtain the international mention for my PhD. Thank you to Dra. P. Tissera and Dr. N.

Libeskind for hosting me at UNAB and AIP, respectively.

(4)

Chapter 1 General introduction

Dwarf galaxies are faint, low-metallicity, and small galaxies. They can be orbiting other more massive galaxies as satellites or be isolated in the field. Typically, they are defined as those objects with stellar masses below 10⁹ M (Bullock & Boylan-Kolchin 2017). In order to differentiate them from globular clusters, it is additionally required that they present high mass-to-light ratios, meaning they are embedded in a dark matter halo. In this sense, it is observed that their dynamical masses are extremely large as compared to the masses inferred from light only, which implies dwarf galaxies are dark matter-dominated systems.

With the recent advent of powerful telescopes it has been revealed that these low-luminosity galaxies represent the most numerous type of galaxy in the Universe, and more particularly, of our local neighborhood, i.e. the Local Group. The Local Group is a group of galaxies dominated by two massive spirals (the Milky Way, MW, and Andromeda, M31), separated ∼ 770 kpc, and isolated from other massive galaxies out to at least ∼3 Mpc. Thanks to the combination of updated galaxy redshift surveys and measurements of peculiar velocities we now have a better understanding of its location in the Universe from a large-scale, cosmic web point of view (“cosmography”):

it is located inside the Virgo supercluster, along a filament that connects the Virgo cluster with the less massive Fornax cluster, and has a mostly flattened distribution due to a pushing of matter from above by the Local Void (Courtois et al. 2013). In addition to the MW and M31, it consists of ∼ 80 dwarf galaxies (between satellites and field dwarfs), and several unclassified objects that remain as dwarf galaxy candidates. This census¹ is continuously increasing thanks to newer observational surveys with ever deeper magnitude limits. Indeed, apart from the Small and Large Magellanic clouds that are two exceptionally bright MW satellites, most remaining dwarf galaxies are ultra-faint dwarfs.

Such a nearby sample of dwarf galaxies has allowed their detailed study from both integrated property measurements and measurements of individual spatially-resolved stars. Beyond the Local Group, spectro-photometric analysis of faint galaxies is still very complicated, therefore the Local Group represents the best laboratory for studying dwarf galaxy properties, as well as their implications in galaxy formation and cosmological models.

The formation and evolution mechanisms of dwarf galaxies still remain quite uncertain (Mateo 1998). This is mainly due to their very diverse properties, which make it difficult to decipher clear

1The most updated census and observed properties inventory of Local Group dwarf galaxies can be found athttp:

//www.astro.uvic.ca/˜alan/Nearby_Dwarf_Database_files/NearbyGalaxies.dat, by A. McConnachie.

9

(9)

evolutionary tracks. Indeed, contrary to more massive spiral and elliptical galaxies, dwarf galaxies in the local Universe show a wide variability of observed properties regarding morphologies, gas and stellar mass content, sizes, star-formation histories, metallicities, dust content, rotational velocities, and different wavelength emission (luminosity and colors), among others. Some corre- lations between structural, kinematical, and population properties do exist, but overall, dwarfs do not follow the extrapolations of well-known galaxy scaling relations at low masses, which at high masses show little scatter, covering instead a wide parameter space.

For example, low-luminosity galaxies are observed to present a range of different morphologies. This has lead to a classification including dwarf spheroidals (dSph), irregulars (dIrr), el- lipticals (dE), spirals (dS), transition-types (dIrr/dSph), blue compact dwarf galaxies (BCD) or ultra-faint dwarfs (UFD). These different types yield different star formation histories (SFHs), as evidenced by synthetic color-magnitude analyses of their resolved stars. Spheroidals, which are typically devoid of gas, present no recent star formation and are composed mainly of an old stellar population. In turn, irregulars commonly retain neutral hydrogen and show an extended star formation activity throughout their entire life-time (Gallart et al. 1996;Mateo 1998;Tolstoy et al.

2009;Dolphin et al. 2005).

This variety of SFHs and stellar populations is reflected as well in a diversity of infrared (IR) spectral energy distribution (SED) shapes, in contrast to high-mass spirals that always show common patterns. Indeed, the dust emissions from low-metallicity galaxies considerably differ from that of more metal-rich galaxies. They harbour warmer dust with higher gas-to-dust ratios, and present some characteristic features thats can be summarized as follows (R´emy-Ruyer et al. 2015):

1. An excess of emission in the submm (∼500 µm), causing a flattening of the submm/far-IR slope;

2. Broadening of the IR peak of the SED, implying the presence of a warmer dust component; 3.

Less polycyclic aromatic hydrocarbon emission lines. In order to fit the data, observers add new ad hoc extra dust components to their modelling, with whatever properties provide the right match.

In particular they introduce a modified black-body to recover point 1.- However, the possibility exists that the differences above can be explained in the framework of a model for dust emission and absorption taking into account the components of the interstellar medium or gas phase (Silva et al. 1998).

Low-mass galaxies also show a wide variety of rotation curves shapes, as compared to more massive rotationally-supported galaxies (de Blok et al. 2008; Oh et al. 2015). While the latter rise sharply at inner radii and then reach a flat, constant rotational velocity at outer radii (Rubin et al. 1978), dwarf galaxies with rotating gas generally produce rotation curves that rise slowly and sometimes do not reach a final constant velocity at their last measured point (Swaters et al.

2009;Oh et al. 2011). Furthermore, this diversity in rotation curve shapes occurs for dwarfs with the same baryonic mass (Oman et al. 2015), which reproduces as a large scatter in the low-mass end of the baryonic Tully-Fisher relation (Bradford et al. 2016). Rotational velocities at different radii are measured from the mean azimuthal rotation of the HI gas, which is assumed to follow the gravitational potential created by the total mass enclosed at that radius. Hence these observations reveal dwarfs have a different internal structure than massive spirals, and in particular, suggest that mass in the centers of dwarf galaxies is distributed more extendedly.

Another puzzle concerning dwarf galaxies observed properties is their spatial distribution.

Satellite galaxies in the Local Group show a very anisotropic spatial distribution around their hosts (Pawlowski 2018). In particular, a high fraction of MW satellites, as well as globular clusters and streams, are aligned in a plane that is approximately perpendicular to the Galactic disk (Pawlowski

(10)

11 et al. 2012). Estimates of their angular momentum vectors, derived by combining proper motion data, radial velocities, and distances, evidence in addition a high degree of co-orbitation of satellites within the plane they define (Fritz et al. 2018). The lack of statistics is a caveat when assessing this matter, but similar claims for planes of satellites have recently been made in M31 (Ibata et al.

2013) and in nearby galactic systems like Centaurus A (Tully et al. 2015; M¨uller et al. 2018), strengthening the presumption that these are not unique occurrences.

A way of better understanding the formation and evolution of dwarf galaxies is by contrasting observational data with predictions from theoretical models of galaxy formation within a standard cosmological context, i.e.,Λ Cold Dark Matter (ΛCDM).

This ΛCDM model is able to explain observational data (e.g., Big Bang nucleosynthesis, the Cosmic Microwave Background and its anisotropies, baryon acoustic oscillations, gravitational lensing). Also, numerical N-body simulations within this framework containing only DM particles (DM-only) have resulted very successful in reproducing the observed large scale structure (LSS) in the Universe (Klypin et al. 2016). The general consensus is that the LSS we see today forms from infinitesimal matter density perturbations arising after the Big Bang that grow via gravitational in- stabilities. The advanced, highly non-lineal phases of gravitational instability are described by the Zeldovich Approximation (Zel’Dovich 1970) extended to the Adhesion Model (see e.g. Gurbatov et al. 1989; Kofman et al. 1990; Gurbatov et al. 2012), whose predictions are now confirmed in detail by numerical simulations (Cautun et al. 2014). At a given scale, walls surrounding voids, filaments and nodes (i.e., the Cosmic Web (CW) elements also known as caustics, seeBond et al.

1996) successively emerge in multi-streaming regions at different spatial locations, and they per- colate giving rise to the CW. Then these elements vanish as mass elements flow through the voids towards the walls, through the walls towards the filaments, and finally from these to nodes. The increasing density of nodes can reach the threshold allowing them to decouple from the general expansion of the Universe, turn around and collapse (involving a CW patch) giving rise to virial- ized halos, as the spherical collapse model states (Gunn & Gott 1972;Padmanabhan 1993). We see that theories, models and simulations agree on ΛCDM as a viable framework for different cosmological observations including large scale structures. However, several problems arise when comparingΛCDM predictions for small scales with the observed dwarf galaxy Universe.

The first of these problems concerns the number and mass function of substructure. While DM- only simulations predict the existence of thousands of subhalos around a central galaxy, less than 100 satellite galaxies are observed orbiting the MW and M31. This has been called the ’missing satellites problem’ (Bullock 2010). Furthermore, observed satellites seem to live in less massive DM subhalos than expected, assuming a standard “abundance matching” one-to-one correspon- dence between the predicted halo mass function and the observed luminosity function. A linear function of DM subhalo masses down to convergence limit is expected inΛCDM simulations, but stellar velocity dispersion measurements of observed satellites reveal that they live in very low- mass subhalos, leaving the more massive ones unoccupied (Walker et al. 2009). This problem remains even when considering the future discovery of new ultra-faint dwarfs by surveys with deeper magnitude limits, since it is considered unlikely that there are hundreds of hidden luminous objects within the virial radius of our Galaxy. It is partly alleviated however by invoking mechanisms that may surpress star formation below a certain mass limit, like reionization UV feedback in the early universe (z . 6) that heats up the intergalactic medium and prevents consequent gas collapse (Bullock et al. 2000). In this case it would be expected that some DM halos are unable

(11)

to form luminous galaxies and stay dark. But this solution does not account for the most massive DM subhalos, which should be unaffected by reionization. This conflict is known as the “Too big to fail” problem (Boylan-Kolchin et al. 2011).

Another strong prediction of ΛCDM is the internal structure of halos. DM-only simulations have revealed an almost universal DM halo density profile, which becomes increasingly steep (or cuspy) in the inner regions (Navarro et al. 1997). This result is at odds with the observed diversity of rotation curve shapes in dwarf galaxies (Oman et al. 2015), which suggests that the mass density profile of observed galaxies is more often shallow (or cored) in the center. Since dwarfs are DM- dominated systems, this implies that the DM halo is less dense than expected by theory, issue which is referred to as the “Cusp-core problem” (Moore 1994).

With respect to satellite spatial anisotropy, theory (e.g. the Zeldovich Approximation and the Adhesion model, see above) predicts, and simulations show, that the flow of mass and its consequent collapse at high redshift occurs following preferential directions given by the velocity field shear tensor (Libeskind et al. 2012). In this way, subhalos are accreted onto their host halos mainly following the filaments of the CW. These preferential directions lead to the appearance of planes of satellites, but the statistics of systems in simulations with satellites as significantly aligned as those observed has been so far very low (Pawlowski 2018). Co-orbitation of aligned satellites has been as well very rarely found (Buck et al. 2016). This issue known as the “Planes of satellites problem” does not contradict any prediction of theΛCDM model, but remains without a physical or cosmological explanation.

The above discrepancies may occur because of comparing a DM-only Universe with real galaxies, which are composed additionally of gas and stars. Indeed, considering the effects of baryonic physics on dark matter halos leads to some reconciliation with the observational data. In the last decades, new high-resolution cosmological simulations have been developed that incor- porate hydrodynamics equations (i.e. accurate physics) and phenomenological prescriptions at sub-resolution scales to model some aspects of baryonic physics. These include detailed recipes for gas cooling, star formation, supernovae (SNe) and stellar feedback (ejection of thermal and mechanical energy), and chemical enrichment of gas particles, among others. As a consequence, simulations have been able to produce populations of galaxies that have global properties similar to observed galaxy populations in many respects.

Dwarf galaxies have been shown to be specially susceptible to the presence of a baryonic disk galaxy in the center of the DM halo, and to subgrid physics effects. On one side, the creation of an axisymmetric gravitational potential on top of the spherical/triaxial one produced by the halo generates a gravitational torque pattern acting on the satellites (which is null in the polar and planar regions, seeDanovich et al. 2015). Tidal stripping effects on the satellites, either from the halo or the disk, can strip off baryonic and dark mass from their outer parts, deform them generating tidal tails, or even trigger their total disruption (Mart´ınez-Delgado et al. 2008;Brooks & Zolotov 2014).

Furthermore, the collisional behaviour of gaseous particles implies that satellites can absorb and dissipate energy. Therefore, environmental interactions among satellites or between a satellite and the disk galaxy (or even the halo gas) can easily induce hydrodynamical and thermodynamical effects like shock waves or ram pressure stripping, or induce changes in their immediate orbital angular momentum directions (G´omez-Flechoso 2001;D’Onghia et al. 2010).

With respect to subgrid recipes modelling internal physical processes, simulations have shown that very strong outflows of gas produced by SNe explosions can produce a sudden change of

(12)

13 the gravitational potential and the thermodynamical gas state, expanding the central dark matter content of low mass halos and yielding shallower inner DM density profiles (Governato et al.

2010; Teyssier et al. 2013; Di Cintio et al. 2014; O˜norbe et al. 2015). Furthermore, this change can be irreversible if the process ocurrs repeatedly and at short timescales (Read & Gilmore 2005;

Pontzen & Governato 2012). These baryonic processes represent possible mechanisms to lower dwarf galaxy central masses yielding less compact inner mass density distributions that can be reconciled with the slowly-rising rotation curves observed (Read et al. 2016).

The intensity of the final influence of baryonic processes on dark matter, however, is subject to simulation details. In particular, different simulations are run with different codes to compute the gravitational and hydrodynamical forces. Therefore accurate comparisons of simulation outputs with observational data are needed in order to constrain galaxy formation models, reproduce realistic dwarf galaxies and eventually better understand their formation and evolutionary paths.

This thesis is divided in 5 chapters, each aiming to address a specific problem concerning dwarf galaxies and their properties within aΛCDM context. The methodology has consisted in confronting new observational data from low-mass galaxies with predictions from state-of-the-art zoom-in cosmological hydrodynamical simulations of galaxy formation, in order to

i) test their respective galaxy formation models within theΛCDM cosmological context, and ii) give a physical explanation to some of the diverse observed properties of dwarf galaxies.

To this aim I have made use of simulations from the MaGICC project (Brook et al. 2012;Stinson et al. 2013), the CLUES project (Gottloeber et al. 2010; Yepes et al. 2014), the NIHAO project (Wang et al. 2015), the PDEVA suite (Serna et al. 2003; Dom´enech-Moral et al. 2012), and the Aquarius project (Springel et al. 2008). MaGICC is a suite of isolated galaxies spanning a wide range of masses from dwarfs to massive spiral galaxies, run with the GASOLINE SPH code including detailed stellar feedback prescriptions at sub-resolution. The CLUES simulation is a simulation of the Local Group. By means of constrained initial conditions, MW and M31 analogs are simulated, embedded in the correct large scale environment. The hydrodynamical version used in this thesis has been run with the GASOLINE SPH code as well. PDEVA is an SPH code which primary concern is that conservation laws hold accurately. In particular I have studied galaxy PDEVA-5004, a MW-like disk galaxy hosting a swarm of satellite galaxies. Finally the Aquarius project provides the initial conditions for a set of MW-like galaxies. I have studied galaxy Aquarius-C, run with the hydrodynamical code by P.Tissera that includes a detailed chemical enrichment scheme.

In chapter 2 I test the angular momentum content of galaxies from the MaGICC and CLUES simulations, spanning a wide range of masses and velocities, by means of the mass discrepancy–

acceleration relation. This relation links the total mass enclosed at a given radius of the galaxy with the gravitational acceleration exerted by only the baryonic component, therefore representing an instructive way to display the radial mass distribution of a population of disk galaxies. I show –for the first time with cosmological simulations–, that the observed relation is reproduced within ΛCDM when taking into account the complicated galactic-scale baryonic cycle. We further discuss implications for the obtained scatter.

Chapter 3 explores the low-mass end of the baryonic Tully-Fisher relation with the MaGICC suite of simulated galaxies ranging from massive spirals to very low-mass dwarfs. An accurate comparison to data is done by carefully imitating observational methods and criteria (in particular concerning HI gas) to derive the characteristic rotational velocity of galaxies. Significantly diverg-

(13)

ing relations at low masses arise depending on the method used, but overall consistency is found between corresponding observed and simulated relations.

In chapter 4 I address the ’cusp-core problem’ by studying the diversity of rotation curve shapes found for the suite of NIHAO simulated galaxies. These simulations include a feedback model that produces the expansion of dark matter halos in low mass galaxies, in a mass-dependent fashion.

By comparing the velocity at the inner and outermost parts of their rotation curves with recent high-quality observational data, I systematically show that halo expansion is necessary to recover the scatter in rotation curve shapes observed at each stellar mass. The match found is not perfect though, missing a type of compact, star-forming dwarf galaxy. I discuss why this type of galaxy may not be formed in NIHAO, with regards to stellar feedback processes and baryon-driven expansion effects.

Chapter 5 covers dust emission and IR SED shapes in dwarf galaxies. I use the highly successful GRASIL-3D radiative transfer code to compute the IR SEDs of simulated star-forming dwarf galaxies from the CLUES project, and compare the results to recently observed IR lumi- nosities of both high-mass and low-mass galaxies. Since this is the first time the code is applied to dwarf galaxies, I adapt it to account for very low-metallicity galaxy specificities. I show that GRASIL-3D’s sophisticated two-component dust model naturally reproduces the diversity of observed dwarf SED shapes as well as their particular features. By an accurate quantification of the results, I provide a physical explanation for the observed features within the context of molecular clouds and cirrus as a two-phase model for dust. Furthermore, CLUES dwarf galaxies are shown to match a wide range of observed dwarf galaxy property relations concerning molecular and neutral hydrogen gas content, star formation rate, and metallicity.

In Chapter 6 I tackle the ’planes of satellites’ issue in two simulations of MW-like galaxies:

PDEVA-5004 and Aquarius-C. I search for planar configurations of satellites first from a positional, and then an orbital angular momentum, point of view. I show that following the first approach it is possible to find very high-quality planes but these are not kinematically-coherent structures.

The second approach however reveals groups of co-orbiting satellites that share a common orbital plane, and define long-lasting planar structures in space. As work in progress, I take a quick look into the origin of these co-orbiting groups of satellites, finding hints for a strong link between their common kinematics and mass collapse of the local cosmic web at high redshift.

Finally, chapter 7 provides a summary of the thesis and the specific conclusions reached.

Each chapter has its own introduction, methods, results, consclusions, and references. Chapters 2, 3, 4, and 5, have been published in high-impact international journals. At the beginning of the respective chapter the authors and journal publication numbers are duly referenced.

Bibliography

Bond J. R., Kofman L., Pogosyan D., 1996, Nature, 380, 603

Boylan-Kolchin M., Bullock J. S., Kaplinghat M., 2011, MNRAS, 415, L40 Bradford J. D., Geha M. C., van den Bosch F. C., 2016, ApJ, 832, 11

Brook C. B., Stinson G., Gibson B. K., Wadsley J., Quinn T., 2012, MNRAS, 424, 1275 Brooks A. M., Zolotov A., 2014, ApJ, 786, 87

Buck T., Dutton A. A., Macci`o A. V., 2016, MNRAS, 460, 4348 Bullock J. S., 2010, ArXiv e-prints

(14)

BIBLIOGRAPHY 15 Bullock J. S., Boylan-Kolchin M., 2017, Annual Review of Astronomy and Astrophysics, 55, 343 Bullock J. S., Kravtsov A. V., Weinberg D. H., 2000, ApJ, 539, 517

Cautun M., van de Weygaert R., Jones B. J. T., Frenk C. S., 2014, MNRAS, 441, 2923 Courtois H. M., Pomar`ede D., Tully R. B., Hoffman Y., Courtois D., 2013, AJ, 146, 69 Danovich M., Dekel A., Hahn O., Ceverino D., Primack J., 2015, MNRAS, 449, 2087

de Blok W. J. G., Walter F., Brinks E., Trachternach C., Oh S.-H., Kennicutt, Jr. R. C., 2008, AJ, 136, 2648

Di Cintio A., Brook C. B., Dutton A. A., Macci`o A. V., Stinson G. S., Knebe A., 2014, MNRAS, 441, 2986

Dolphin A. E., Weisz D. R., Skillman E. D., Holtzman J. A., 2005, ArXiv e-prints, astro

Dom´enech-Moral M., Mart´ınez-Serrano F. J., Dom´ınguez-Tenreiro R., Serna A., 2012, MNRAS, 421, 2510

D’Onghia E., Springel V., Hernquist L., Keres D., 2010, ApJ, 709, 1138

Fritz T. K., Battaglia G., Pawlowski M. S., Kallivayalil N., van der Marel R., Sohn T. S., Brook C., Besla G., 2018, ArXiv e-prints, arXiv:1805.00908

Gallart C., Aparicio A., Bertelli G., Chiosi C., 1996, AJ, 112, 2596

G´omez-Flechoso M. A., 2001, in Highlights of Spanish astrophysics II, Zamorano J., Gorgas J., Gallego J., eds., Vol. 4, p. 141

Gottloeber S., Hoffman Y., Yepes G., 2010, arXiv.1005.2687 Governato F. et al., 2010, Nature, 463, 203

Gunn J. E., Gott, III J. R., 1972, ApJ, 176, 1

Gurbatov S. N., Saichev A. I., Shandarin S. F., 1989, MNRAS, 236, 385 Gurbatov S. N., Saichev A. I., Shandarin S. F., 2012, Physics Uspekhi, 55, 223 Ibata R. A. et al., 2013, Nature, 493, 62

Klypin A., Yepes G., Gottl¨ober S., Prada F., Heß S., 2016, MNRAS, 457, 4340 Kofman L., Pogosian D., Shandarin S., 1990, MNRAS, 242, 200

Libeskind N. I., Hoffman Y., Knebe A., Steinmetz M., Gottl¨ober S., Metuki O., Yepes G., 2012, MNRAS, 421, L137

Mart´ınez-Delgado D., Pe˜narrubia J., Gabany R. J., Trujillo I., Majewski S. R., Pohlen M., 2008, ApJ, 689, 184

Mateo M. L., 1998, Annual Review of Astronomy and Astrophysics, 36, 435 Moore B., 1994, Nature, 370, 629

M¨uller O., Pawlowski M. S., Jerjen H., Lelli F., 2018, Science, 359, 534 Navarro J. F., Frenk C. S., White S. D. M., 1997, ApJ, 490, 493

O˜norbe J., Boylan-Kolchin M., Bullock J. S., Hopkins P. F., Kereˇs D., Faucher-Gigu`ere C.-A., Quataert E., Murray N., 2015, MNRAS, 454, 2092

Oh S.-H., de Blok W. J. G., Brinks E., Walter F., Kennicutt, Jr. R. C., 2011, AJ, 141, 193 Oh S.-H. et al., 2015, AJ, 149, 180

Oman K. A. et al., 2015, MNRAS, 452, 3650

Padmanabhan T., 1993, Structure Formation in the Universe Pawlowski M. S., 2018, Modern Physics Letters A, 33, 1830004

Pawlowski M. S., Pflamm-Altenburg J., Kroupa P., 2012, MNRAS, 423, 1109 Pontzen A., Governato F., 2012, MNRAS, 421, 3464

Read J. I., Gilmore G., 2005, MNRAS, 356, 107

Read J. I., Iorio G., Agertz O., Fraternali F., 2016, MNRAS, 462, 3628

(15)

R´emy-Ruyer A. et al., 2015, A&A, 582, A121

Rubin V. C., Ford, W. K. J., Thonnard N., 1978, ApJ, 225, L107 Serna A., Dom´ınguez-Tenreiro R., S´aiz A., 2003, ApJ, 597, 878 Silva L., Granato G. L., Bressan A., Danese L., 1998, ApJ, 509, 103 Springel V. et al., 2008, MNRAS, 391, 1685

Stinson G. S., Brook C., Macci`o A. V., Wadsley J., Quinn T. R., Couchman H. M. P., 2013, MNRAS, 428, 129

Swaters R. A., Sancisi R., van Albada T. S., van der Hulst J. M., 2009, A&A, 493, 871 Teyssier R., Pontzen A., Dubois Y., Read J. I., 2013, MNRAS, 429, 3068

Tolstoy E., Hill V., Tosi M., 2009, Annual Review of Astronomy and Astrophysics, 47, 371 Tully R. B., Libeskind N. I., Karachentsev I. D., Karachentseva V. E., Rizzi L., Shaya E. J., 2015,

ApJ, 802, L25

Walker M. G., Mateo M., Olszewski E. W., Pe˜narrubia J., Wyn Evans N., Gilmore G., 2009, ApJ, 704, 1274

Wang L., Dutton A. A., Stinson G. S., Macci`o A. V., Penzo C., Kang X., Keller B. W., Wadsley J., 2015, MNRAS, 454, 83

Yepes G., Gottl¨ober S., Hoffman Y., 2014, New Astron. Rev., 58, 1 Zel’Dovich Y. B., 1970, A&A, 500, 13

(16)

Chapter 2 The distribution of mass components in simulated disc galaxies

This chapter is a copy of

”The distribution of mass components in simulated disc galaxies”, by I. Santos-Santos, C. Brook, G. Stinson, A. Di Cintio, J. Wadsley,

R. Dom´ınguez-Tenreiro, S. Gott¨ober & G. Yepes;

published in MNRAS 455, 476-483 (2015).

2.1 Introduction

Within a ΛCDM context, the angular momentum of disc galaxies originates from tidal torques imparted by surrounding structures in the expanding Universe, prior to proto-galactic collapse (Peebles 1969;Barnes & Efstathiou 1987). Assuming that gas gains a similar amount of angular momentum as the dark matter, and that angular momentum is substantially retained as the gas cools to the centres of dark matter halos (Fall & Efstathiou 1980), then the gas will settle into a disc, fragment and form stars.

Simulations have shown that angular momentum acquisition is more complicated than this picture, involving a complex web structure (e.g. Pichon et al. 2011; Dom´ınguez-Tenreiro et al.

2015). Indeed, there has been significant progress over the past years in our ability to simulate the processes of disc formation within a cosmological context. Without an efficient feedback scheme, angular momentum is lost to dynamical friction during the mergers of overly dense sub-structures (e.g.Navarro & Steinmetz 2000;Maller & Dekel 2002;Piontek & Steinmetz 2011).

Progress was made by implementing increasingly effective recipes for feedback from supernovae (Thacker & Couchman 2001;Stinson et al. 2006) and the inclusion of other forms of feedback from massive stars (Stinson et al. 2013;Hopkins et al. 2014). The benchmark for assessing this progress has primarily been the ability to match the Tully-Fisher relation (e.g.Governato et al.

2004;Dom´enech-Moral et al. 2012), with recent simulations succeeding at this, and in particular matching the Baryonic Tully Fisher relation (BTFR), for galaxies over a range of masses (Brook et al. 2012b;Aumer et al. 2013).

Yet rotation curves of observed galaxies provide significantly more information regarding the angular momentum of galaxies than is contained within the BTFR, allowing more stringent con-

17

(17)

able2.1:Propertiesofthesimulatedgalaxiesorderedbyhalomass.MaGICCgalaxieshavea”g”asprefix,whileCLUESgalaxiesvea”C”.Diskscalelengthshandcentralsurfacebrightnessesµ0arederivedfromexponentialfitstothesurfacebrightnessprofileinIband.NameMhalo(M)M∗(M)MHI(M)h(kpc)µ0(magas −1)Vmax(kms −1)Vflat(kms −1) g15784MW1.49×10 125.67×10 101.96×10 103.2319.09222.10222.10g21647MW8.24×10 112.51×10 105.62×10 91.3017.39189.79163.84g1536MW7.06×10 112.36×10 106.78×10 93.4620.54175.95175.95g5664MW5.39×10 112.74×10 104.19×10 92.3419.51196.66151.40g7124MW4.47×10 116.30×10 93.49×10 92.7920.60120.14120.14g15807Irr2.82×10 111.46×10 104.68×10 91.9418.68141.21141.21g15784Irr1.70×10 114.26×10 92.70×10 92.2720.30106.90106.90g22437Irr1.10×10 117.44×10 81.08×10 91.8821.3375.4075.40g21647Irr9.65×10 101.98×10 83.68×10 81.7522.7760.8560.85g1536Irr8.04×10 104.46×10 84.39×10 81.7021.7567.1667.16g5664Irr5.87×10 102.36×10 82.56×10 81.6622.2859.5059.50g7124Irr5.23×10 101.32×10 82.30×10 81.1621.6152.7752.77C17.23×10 111.45×10 103.86×10 91.5619.56168.83127.10C25.31×10 111.11×10 106.32×10 81.8320.40123.59123.59C32.67×10 115.08×10 92.79×10 92.2521.22119.75119.75C41.87×10 114.18×10 99.77×10 71.4520.07101.21101.21C51.51×10 114.54×10 92.42×10 91.3520.30116.63116.63C61.29×10 112.08×10 92.74×10 91.5321.29101.66101.66C71.18×10 111.57×10 99.10×10 81.4820.4488.8972.92C81.21×10 111.57×10 96.34×10 81.0320.1985.2285.22C98.04×10 101.10×10 91.05×10 81.5522.6470.8570.85C106.44×10 103.78×10 86.33×10 70.8621.5553.3553.35

(18)

2.1. INTRODUCTION 19 straints on galaxy formation models which have not previously been applied to simulated galaxies.

High resolution observations of HI velocities, combined with studies of the gas and stellar mass distributions, provide detailed information on how the different mass components are radially distributed in galaxies with a wide range of rotational velocities Vr(e.g.Begeman et al. 1991;Sanders

& Verheijen 1998;de Blok et al. 2001;Gentile et al. 2004;Kuzio de Naray et al. 2006;Oh et al.

2015).

Differences between the mass implied by measured rotational velocities, and the baryonic mass observed in the form of gas and stars, is usually attributed to dark matter (Rubin & Ford 1970), an assumption which our simulations embrace. In this study we aim to determine whether galaxies simulated in a ΛCDM universe can reproduce the detailed radial mass distribution of observed galaxies.

An instructive way to display the radial mass distribution of a population of disc galaxies is to plot the mass discrepancy-acceleration relation (Sanders & McGaugh 2002;McGaugh 2004).

Mass discrepancy, D, is defined as the ratio of the square of the measured rotation velocity, and the square of the rotation velocity that can be attributed to baryons, D≡(Vr/Vb)². The acceleration is defined at each radius, r, by the baryonic contribution to the centripetal acceleration, gb≡Vb2/r.

Despite a large variety in rotation curve shapes (e.g. Zwaan et al. 1995; Tully & Verheijen 1997; Swaters et al. 2009), disc galaxies with a wide range of Vr show a remarkably tight D-gb

relation (McGaugh 2004,2014). Galaxies that present the same mass discrepancy at various radii all experience, at those radii, the same gravitational radial force as contributed by their baryons; it is as though the rotation velocity attributed to dark matter depends only on the distribution of baryonic mass. Indeed, this tight relation has been interpreted as being causal (Sanders & McGaugh 2002;

McGaugh 2014), and therefore evidence for modified Newtonian dynamics, MOND (Milgrom 1983). This empirical result has no a priori explanation in a ΛCDM cosmology, but a study of semi-analytic models (van den Bosch & Dalcanton 2000) found that galaxies tuned to match the Tully-Fisher relation reveal a characteristic acceleration.

Here, we explore the D-gb relation in a suite of 22 disk galaxies simulated within a ΛCDM context, which vary in their virial, stellar and baryonic masses (M_halo, M∗, M_b), star formation histories, disk scale lengths, central surface brightnesses and circular velocity curve shapes. As we show, the suite of galaxies tightly match the empirical M∗-M_halo(Guo et al. 2010;Moster et al.

2010), M_b-M_halo(Papastergis et al. 2012) and BTFRs (McGaugh 2005).

It is important to emphasise that these simulated galaxies were not tuned to reproduce the BTFR, with free parameters tuned to match the stellar to halo mass relation at one halo mass (Brook et al. 2012b;Stinson et al. 2013), and then fixed in simulations of different mass halos. The simulations have previously been shown to match a wide range of scaling relations including the Tully-Fisher, luminosity-size, mass-metallicity relations, and HI mass to r band luminosity ratio as a function of R band magnitude, at z=0 (Brook et al. 2012b). Further, the simulations match the evolution of the stellar mass-halo mass relation (Stinson et al. 2013;Kannan et al. 2013), as derived by abundance matching (Moster et al. 2013) and a range of relations at high redshift (Obreja et al.

2014). The simulations also expel sufficient metals to match local observations (Prochaska et al.

2011;Tumlinson et al. 2011) of OVI in the circum-galactic medium (Stinson et al. 2012;Brook et al. 2012b).

The paper is organized as follows. Section 2 presents the simulations used, describing their initial conditions and baryonic modelling. The circular velocity curves, BTFR, M∗-Mhalo Mb- M_halo relations and plots of D versus g and radius are shown in Section 3. Residuals around the

(19)

0 5 10 15 20 200

4060 10080 120140

160 g15807_Irr

0 5 10 15 20

0 20 40 60 80 100

120 g15784_Irr

0 5 10 15 20 25 30 35 40 0

50 100 150 200

250 g15784_MW

0 5 10 15 20

100 20 3040 5060

70 g1536_Irr

0 5 10 15 20 25 30 35 0

50 100 150

Vcirc (km s−1)

g1536_MW

0 2 4 6 8 10 12 14 16 0

10 20 30 40 50

60 g5664_Irr

0 5 10 15 20

0 50 100 150

200 g5664_MW

0 2 4 6 8 10 12

0 10 20 30 40 50

g7124_Irr

0 5 10 15 20 25

0 2040 6080 100

120140 g7124_MW

0 2 4 6 8 10 12 14 16 R (kpc) 0

10 20 30 40 50

60 g21647_Irr

0 2 4 6 8 10 12 14

0 50 100 150

200 g21647_MW

0 2 4 6 8 10 12 14 16 100

2030 4050 6070

80 g22437_Irr

Figure 2.1: The circular velocity curves of the 12 MaGICC disk galaxies. Different symbols represent the velocity values due to different mass components (triangles: cold gas; stars: stars;

squares: all baryons; circles: total). Simulations reproduce the variety of observed rotation curves.

Furthermore, like in observations, the features present in the baryonic curves are reflected in the total one.

D-g relation are also shown. Finally, Section 4 discusses the implications of our results.

2.2 The Simulations

Two sets of simulated galaxies have been used, with slightly different input physics, as we will explain. In first place, we use 12 galaxies from the MaGICC (Making Galaxies in a Cosmological Context) project (Brook et al. 2012a;Stinson et al. 2013). These are zoomed-in regions of a total cosmological volume of side 68 Mpc. The resolution varies depending on the “type” of simulated galaxy, labelled MilkyWay or Irregular. The former have m_star=4.0×10⁴M, m_gas=5.7×10⁴M, m_dm=1.1×10⁶M and a gravitational softening length of =312pc (for all particle types), while the latter are more highly resolved with mstar=4.3×10³M, mgas=7.1×10³M, mdm=1.4×10⁵M

and =156pc. The initial power spectrum of density fluctuations is derived from the McMaster Unbiased Galaxy Simulations (MUGS) (Stinson et al. 2010) which use aΛCDM cosmology with WMAP3 parameters, i.e. H0 = 73 km s⁻¹ Mpc⁻¹, Ωm = 0.24, ΩΛ = 0.76, Ωbaryon = 0.04 and σ₈ = 0.76.

The second set of galaxies is from a single simulation with initial conditions from the CLUES project (Constrained Local UniversE Simulations,Gottloeber et al. 2010;Yepes et al. 2014). Again the zoom-in technique is used, this time together with observational data (masses of nearby X- ray clusters and peculiar velocities obtained from catalogs) imposed as constraints on the initial

(20)

2.2. THE SIMULATIONS 21 conditions, in order to simulate a cosmological volume that is representative of our local universe.

The Hoffman-Ribak algorithm, using the observational data mentioned above, is used to constrain scales down to ≈ 5h⁻¹Mpc. This way structures like the Virgo cluster, Coma cluster and Great attractor, are always reproduced by the simulations.

On smaller scales, the distribution of structure is essentially random, and several dark matter- only realizations are run until a Local Group analogue (a Milky Way-M31 like binary group) is found. Then this Local Group region is re-simulated with baryons and at a higher resolution. In this case, the re-simulation includes 4096³ effective particles in a spherical volume of 2h⁻¹Mpc around the Local Group. The mass resolution of particles is m_star=1.3×10⁴M, m_gas=1.8×10⁴M

and m_dm=2.9×10⁵M, and the gravitational softening lengths are _bar=223pc between baryons and

dm=486pc between dark matter particles. This CLUES simulation also follows a WMAP3 cosmological model.

The CLUES simulation used is not one of the previously published CLUES simulations, which were evolved using the PMTree-SPH MPI code GADGET2, but rather part of a new set with the same initial conditions but different physics prescriptions for star formation and feedback, as explained below.

All simulations in this study are evolved using the parallel N-body+SPH tree-code GASOLINE (Wadsley et al. 2004), which includes gas hydrodynamics and cooling, star formation, energy feedback and metal enrichment to model structure formation. We describe here the most important implementations (for details seeGovernato et al. 2010andStinson et al. 2013).

When gas gets cold and dense, stars are formed according to a Schmidt law with star formation rate ∝ ρ^1.5. Stars feed energy and metals to the surrounding interstellar medium. Energy feedback by supernovae is implemented by means of the blastwave formalism (Stinson et al. 2006) where

_SN× 10⁵¹erg of thermal energy is released. The amount of metals deposited from SNe explosions is computed from a Chabrier IMF, and they diffuse between gas particles as described in Shen et al. (2010). GASOLINE also accounts for the effect of a uniform background radiation field on the ionization and excitation state of the gas. In the case of the MaGICC simulations, metal-line cooling (Shen et al. 2010) and early stellar feedback from massive stars (Stinson et al. 2013) prior to their explosion as SNe are also included.

The CLUES simulations follow the physics used inGovernato et al.(2010) andGuedes et al.

(2011), which formed realistic dwarf and Milky Way galaxies respectively. These runs do not include metal line cooling, nor do they include early stellar feedback. As argued inFeldmann &

Mayer(2015), gas cooling is also affected by UV and soft X-ray emission from nearby massive stars (e.g. Cantalupo 2010;Kannan et al. 2014); it is not clear whether adding metal line cooling, without their potential counterparts, such as radiative ionization by local sources, results in a better model.

Further, in these types of simulations, feedback recipes are not well constrained, but are basi- cally tuned to balance whatever cooling rate is included, in order to match the constraints imposed;

in these simulations, constraints come from the stellar to halo mass relation. The lower cooling rates of the CLUES set of simulations is compensated by the lack of early stellar feedback, result- ing in our two sets of simulations following very similar trends in their structural properties, as we will see below. One could argue that sub grid local feedback processes are included in the CLUES simulation by adjusting the cooling function.

We emphasise that the two sets of simulations share the same implementations of SNe feedback and star formation. Yet there are differences in cooling and feedback, which will result in differ-

(21)

0 2 4 6 8 10 12 14 200

406080 100120 140160 180

Vcirc (km s−1)

1

5 10 15 20

2

5 10 15 20 25 30 3

2 4 6 8 10 12 14 4

5 10 15 20

5

0 2 4 6 8 10 12 14 0

20 40 60 80 100

120 6

2 4 6 8 10 12 14 16 18 7

1 2 3 4 5 6 7 8 9

R (kpc)

8

5 10 15 20

9

1 2 3 4 5 6 7 8 9 10

Figure 2.2: The circular velocity curves of the 10 galaxies selected from the CLUES simulation.

Triangles: cold gas; stars: stars; squares: all baryons; circles: total.

ences in the amount of gas cycling through central regions of the galaxies. We will comment on some systematic differences between the two sets of simulations, in terms of the relations explored in this paper.

Halos in both simulations have been identified using Amiga’s Halo Finder (AHF; Knollmann

& Knebe 2009), where their masses are defined as the mass inside a sphere containing∆vir '350 times the cosmic background matter density at redshift z=0.

The analysis of the simulation data was largely performed using the open source PYNBODY package (Pontzen et al. 2013).

The properties of the simulated galaxies are presented in Table 3.1. The MaGICC simulations are 12 disk galaxies, separated into two sub-sets labelled as Milky-Way (MW) and irregular (Irr) type galaxies, although they all are disc galaxies with stellar masses ranging from 1×10⁸- 5×10¹⁰M. From the CLUES simulation we have selected the halos that satisfy the following conditions: (i) Not a sub-halo, (ii) M_halo>4×10¹⁰M. These integrate a sample of 10 well resolved isolated galaxies. Since this is a Local Group simulation, the three most massive galaxies are loose analogues of the Milky Way, M31 and M33, and the rest are isolated dwarf galaxies.

2.3 Results

We emphasise the MaGICC set of simulations by showing each individual galaxy in colour in all plots. This is because these galaxies have been thoroughly explored in the literature, as noted in the introduction. The CLUES simulations have not been as extensively analysed in other contexts, and are shown as grey dots. Considering that our results emphasise the ability of the suite of simulations to match various relations, and that one may expect any differences in the two sets of simulations to increase any scatter found around the relations we explore, we feel that it is justified to include all galaxies in the derived results. Thus, our fits include all simulated galaxies.

Nothing in our conclusions changes if only MaGICC galaxies are included, although the number and diversity of galaxies would be less.

(22)

2.3. RESULTS 23

2.3.1 Circular velocity curves

Figures2.1 and2.2show the gaseous (triangles), stellar (stars), baryonic (squares) and total (circles) circular velocity curves of the MaGICC and CLUES simulated galaxies, respectively. These are measured at radii ranging from 0.7kpc to 10×h where h is the disc scale length (see Table 3.1). These circular velocities are calculated using the gravitational potential along the midplane of the aligned simulated disc. We checked that our results are not significantly changed by assuming a spherical potential and averaging the mass within spherical shells (i.e., the classical GM/r potential).

The simulated galaxies reach a flat value of circular velocity which persists to large radii, and lack the strong peak at small radii that not so long ago was ubiquitous in simulations due to overcooling. A couple of galaxies, g5664 MW and C1, do have significant bulges, which is reflected in the heightened inner region of their circular velocity curves.

The differences in the physical modelling and initial conditions used in both simulations are not readily visible, with galaxies of similar masses reaching similar maximum, flat velocities (see for example g7124 MW & C3, g15807 Irr & C1 or g5664 Irr & C10). However, there does appear to be a tendency for the MaGICC galaxies to have more slowly rising rotation curves than the CLUES simulations. The broad range of observed rotation curve shapes has been noted for some time (e.g.Zwaan et al. 1995;Swaters et al. 2009), with recent attempts to quantify this range (Oman et al. 2015;Brook 2015) and compare with cosmological models. The relatively small number of galaxies of each different suite used in this study means it is difficult to compare with observations in a quantitative manner. Whether a larger suite of cosmological simulations can match the range of observed rotation curve shapes will be explored in a later study (Santos-Santos et al. in prep).

For this study, we note that there may be slight systematic differences between the MaGICC and CLUES rotation curves shapes, which we will explore in terms of the mass discrepancy relation, and will be seen to be relatively minor.

Visually, one can appreciate that overall, the simulations do produce diversity in rotation curve shapes, and that the baryon contribution increases with increasing mass. One can also see that features from the baryonic components are often reflected in the total-components curves. This is known as ”Renzo’s Rule” (Sancisi 2004;McGaugh 2014), and has long been observed in real galaxies. In particular, these bumps and features are noticeable in galaxies g15807 Irr, g15784 Irr, g1536 MW, g5664 MW of Figure 2.1, and C1, C5, C6, C7 & C8 of Figure 2.2. These results represent evidence that the different mass components affect each other throughout the disc region as they co-evolve within aΛCDM Universe.

2.3.2 Baryonic and Halo Masses

In the left panel of Figure2.3 the stellar-to-halo mass relation of the simulations is plotted, along with the empirical relation as determined byGuo et al. (2010) andMoster et al. (2010), whilst in the right panel of Figure2.3the baryon-to-halo mass relation is plotted, along with the empirical relation as determined byPapastergis et al.(2012). The baryonic mass is defined as the sum of the mass coming from stars and cold gas particles, where the latter is estimated as a multiple of the atomic HI gas mass Mg = ηM^HI, with η= 4/3 (following e.g. McGaugh 2012). The Saha equation is solved to determine an ionization equilibrium and the HI mass. This remains an approximation since an accurate model of HI mass would require full radiative transfer. In particular self shielding

(23)

10

¹¹

10

¹²

M

halo

(M

_¯

)

10

⁸

10

⁹

10

¹⁰

10

¹¹

M

star

(M

¯

)

Guo10 Moster10

10

¹¹

10

¹²

M

halo

(M

_¯

)

10

⁹

10

¹⁰

10

¹¹

M

bar

(M

¯

)

Papastergis12

Figure 2.3: The stellar-to-halo mass (left panel) and baryon-to-halo mass (right panel) relations, with MaGICC galaxies in blue and CLUES galaxies in red. Also shown are the empirical stellar- halo mass relations from Guo et al. 2010 (green line) and Moster et al. 2010 (black line), and the baryon-to-halo mass relation from Papastergis 2012 (cyan line).

from the UV background is not included in our model and may affect our derived HI masses, while photo-ionization of HI from the galaxy itself is also excluded.

As stated in Section 3.2, the MaGICC simulations were tuned to match the stellar mass- halo mass relation at one galaxy mass (in particular, to match the stellar to halo mass of galaxy g15784 Irr), and shown to then match the relation over a range of masses (Brook et al. 2012b;

Obreja et al. 2014, see also the Nihao simulations,Wang et al. 2015, which use very similar imple- mentation of physics, and note that other models are also able to match the relation, e.g. Munshi et al. 2013;Schaye et al. 2015). The CLUES simulations were also calibrated to match the relation.

So, although in some sense it is not surprising that the simulations match the relation to which they were tuned, they actually match the relation over a far wider mass range than the one on which the parameter search was performed.

We show that the simulations also match the empirical M_b-M_halo relation, implying that they also have the same total amount of cold gas as observed galaxies at z= 0. As far as we know, this is the first time that simulations have been shown to match this important empirical relation, and emphasise that it was not a direct result of a parameter search.

2.3.3 The Baryonic Tully-Fisher relation

The maximum velocity found in each simulated galaxy, Vmax, is a good approximation of the flat velocity, V_flat, in most cases. For the cases mentioned above in which a couple of MW type galaxies have significant bulges, we show different values of Vmaxand V_flat in Table3.1.

In Figure 2.4we plot the BTFR using Vflat, with the MaGICC and CLUES sets of simulations shown as blue and red dots, respectively. In the case of C7, the galaxy is about to undergo a merger, and we use the maximum velocity from the inner 10 kpc as V_flat which is the central galaxy, and

(24)

2.3. RESULTS 25

10

²

V

flat

(km s

⁻¹

) 10

⁹

10

¹⁰

10

¹¹

M

bar

(M

¯

)

log M

b

=3.78log V

flat

+2.16 V

max

McGaugh&Schombert15

Figure 2.4: The baryonic Tully-Fisher relation: total baryonic mass M_bar(stars+ cold gas) plotted against rotation velocity Vflat. Blue points are from the MagiCC suite while the red points are the galaxies from the CLUES simulation. The dashed line shows the linear fit to the simulated data, with slope=3.78. The small green points show Vmax rather than V_flat, which results in a slightly flatter relation, with slope=3.49 (see text for details). The dotted line is the observational relation using measurements in the V band given inMcGaugh & Schombert 2015with slope=3.92.

use the baryonic mass from within this same radius. The fit to the V_flat BTFR is logMb = 3.78logVflat+ 2.16

The scatter is very small, with the galaxy that is furthest from the fit being C7, the one which has a very close companion galaxy with which it is dynamically interacting.

If we simply use V_max in each case, the relation is slightly flatter, and can be seen as small green dots in Figure2.4, with fit

logM_b= 3.49logVmax+ 2.67 and similar scatter.

These fits are consistent with the observational fits found in the literature (seeMcGaugh 2012, for a summary), as is the trend for a flatter relation when using V_maxrather than V_flat.

2.3.4 Mass discrepancy

“Mass discrepancy” refers to the difference between the total mass and the baryonic mass enclosed at a certain radius, which can be inferred from the rotation curve of a galaxy. The value of the mass discrepancy D is calculated as the squared ratio of the observed velocity to that due to the observed baryons D=(V^r/Vb)².

Figure2.5shows mass discrepancy plotted against radius, each point representing a point along the rotation curves (MaGICC galaxies are colored points, CLUES are small gray dots). In the upper

(25)

10

⁰

10

¹

10

²

(V /V

∗

)

2

10

⁰

R (kpc) 10

¹

10

⁰

10

¹

(V /V

b

)

2

Figure 2.5: Mass discrepancy versus radius (stars, top panel; all baryons, lower panel). Data points for each MaGICC galaxy are represented in a different color according to figure 2.1. Data points for all CLUES galaxies are small dots in gray color (as in figure 2.2). As occurs with observed galaxies, the lower the surface brightness of the galaxy the higher the mass discrepancy encountered: mass discrepancy does not hold a correlation with radius. Smaller values of D are found when all baryons are taken into account as expected.

(26)

2.3. RESULTS 27

0 5 10 15 20 25

(V /V

∗

)

2

10

¹

10

²

10

³

10

⁴

g (km

²

s

⁻²

kpc

⁻¹

)

0 2 4 6 8 10

(V /V

b

)

2

Figure 2.6: Mass discrepancy versus acceleration produced by stars (top panel), and by all baryons–stars + cold gas–(lower panel). The 22 galaxies are shown with colors as in Figures 2.1 and2.2. The dashed lines are the observational D-g relations (equations 8 & 9 ofMcGaugh 2014). An horizontal dashed line is also shown in both panels to emphasize the asymptotic behaviour of the relation to D=1. Binned data is shown as black squares. The errors are the rms deviations from the best 3 degree polynomial fit found for the data in each bin.

(27)

panel mass discrepancy is computed as the squared ratio of the observed velocity to that predicted by the stars, and in the lower panel all baryons (stars plus cold gas) are taken into account.

At the radii where D=1, the rotation of the galaxy can be explained by the contribution of baryons (stars in the upper panel) alone, while the mass discrepancy (need for dark matter) appears when D>1. As with observed galaxies, the mass discrepancy does not appear at the same radius for all galaxies, and increases with radius but not at a constant rate for every system (McGaugh 2014).

Furthermore, galaxies separate readily when mass discrepancy is plotted against radius, with low mass galaxies having a larger dark matter contribution to the mass at any given radius, compared to higher mass galaxies.

We note the most prominent difference between simulations and observations for the relations shown in Figure 2.5 is that (V/V^b)² is higher at low radii for the lowest mass observations, than for the corresponding data from the simulations. As this difference does not appear in the (V/V∗)² case, one may infer that the simulations have an excess of cold gas in the inner regions of low mass galaxies.

Figure 2.6 shows the mass discrepancy-acceleration relation for all the simulated galaxies, where the acceleration is derived from the gravitational potential of the stars (top panel) and baryons (lower panel), g≡V²/r. In this plot, since radius is inversely proportional to the gravitational acceleration, an increase in radius along a rotation curve is read from right to left. One can observe that more massive galaxies reach higher values of g. Note that although individual galaxies inhabit different regions of the plot, as can be readily seen for the MaGICC set which are plotted as different colors, they all follow a single relation.

We divide the data in 10 bins and show in Figure 2.6 the fit to the data in each bin as black squares. Error bars show the standard deviation within each bin. Dashed lines represent the fits found for observational data, using equations 8 & 9 ofMcGaugh(2014).

A slight deviation from the general trend can be seen at g∼10²and D∼2, with some points of the simulated galaxies falling below the D-g relation. This feature is also seen in the observed mass discrepancy–acceleration relations. In the simulations, it is specially evident for MaGICC galaxies g21647 Irr, g7124 Irr, g5664 Irr and g1536 Irr in the “all baryons” relation. These are the galaxies that have the most slowly rising rotation curves, as mentioned in section2.3.1. As discussed above, the stronger feedback scheme present in these simulations removes more baryonic mass from the centre of some of the simulated galaxies, causing the velocity in this inner region to be lower than expected by the one-to-one D-g relation. Better statistics are required to determine whether the deviations from the relation seen in the simulations are more or less prominent than seen in the observations, and may relate to the observed diversity of rotation curve shapes.

Another possible difference between the observations and simulations is found in the baryonic D-g relation, where the observed galaxies extend down to D=1, while the simulated galaxies do not quite reach this asymptotic value. Comparing such fine details of the relations would likely require better matched observational and simulated data, in terms of the distributions of stellar masses and scale-lengths. In particular, the observed sample appears to have more massive disc galaxies, which may be dominated by baryons in the inner region and hence extend down to lower values of D than the simulated sample, which has only 1 galaxy with M∗>3×10¹⁰M.

The intrinsic scatter we find with respect to the fit in each bin, as well as the deviation from the observed relations, is small and decreases as D tends to 1. In Figure 2.7 the histograms of the residuals found around the fits are shown to be narrow, with a very similar spread to that of observed galaxies.

(28)

2.4. CONCLUSIONS 29

4 3 2 1 0 1 2 3 4

∆

_∗

0 20 40 60 80 100 120 140 160 180

N

4 3 2 1 0 1 2 3 4

∆

_b

0 20 40 60 80 100 120 140 160

Figure 2.7: Histograms of the residuals around the fits. Left: only stellar contribution; Right: all baryons.

2.4 Conclusions

In this paper we have moved beyond using the Tully-Fisher relation as a test of the angular momentum content of galaxies simulated in a ΛCDM cosmological context, to include the wealth of information contained within extended rotation curve data. We do this by exploring the mass discrepancy-acceleration relation through the full radial range of the disc in each of 22 simulated galaxies, a suite that spans more than two orders of magnitude in stellar mass, with rotation velocities ranging from 52 to 222 km s⁻¹.

The simulated suite of late type galaxies is shown to match the empirical relation between baryonic mass, which includes their stars and cold gas, and halo mass, as well as the baryonic Tully-Fisher relation.

Despite showing significant diversity in the shapes of their rotation curves, and in the contribution of dark matter to the total mass budget, the simulated galaxies follow a single relation in both the D-g∗ and D-gb plots, with small scatter. The implication is that not only the total amount of angular momentum attained by the simulated galaxies is correct, as shown by the BTFR, but also that their final internal distribution of gas, stars and dark matter at all radii through their discs is similar to that observed in real galaxies.

The acquisition of the angular momentum within the simulations is complicated, as compared to the simple models of angular momentum acquisition within a CDM universe that assume col- lapsing spheres of gas, torqued by large scale structure (Fall & Efstathiou 1980). The baryon cycle within the simulated discs studied here (seeBrook et al. 2014, for details) involve a complex web structure, large scale outflows of low angular momentum gas (Brook et al. 2011), and redistribu- tion of low angular momentum gas through large scale galactic fountains (Brook et al. 2012a).

Further, the dark matter distributions respond to the gas flows in a manner that is dependent on the simulated galaxy mass (Di Cintio et al. 2014b,a).

Yet in many ways, our model remains straight forward once theΛCDM initial conditions are set, with much of the physics involved in driving the complicated galactic scale baryon cycle occurring on ”sub-grid” scales, and accounted for by using relatively simple prescriptions. These

Dwarf galaxies in LCDB : distribution around hosts, dust emission and internal structure

DWARF GALAXIES IN ΛCDM:

DISTRIBUTION AROUND HOSTS, DUST EMISSION AND INTERNAL STRUCTURE

Contents

Chapter 1

General introduction

Bibliography

Chapter 2

The distribution of mass components in simulated disc galaxies

2.1 Introduction

2.2 The Simulations

2.3 Results

2.3.1 Circular velocity curves

2.3.2 Baryonic and Halo Masses

10

10

M

(M

)

10

10

10

10

M

(M

)

Guo10 Moster10

10

10

M

(M

)

10

10

10

M

(M

)

Papastergis12

2.3.3 The Baryonic Tully-Fisher relation

10

V

(km s

) 10

10

10

M

(M

)

log M

=3.78log V

+2.16 V

McGaugh&Schombert15

2.3.4 Mass discrepancy

10

10

10

(V /V

)

10

R (kpc) 10

10

10

(V /V

)

0 5 10 15 20 25

(V /V

)

10

10

10

10

g (km

s

kpc

)

0 2 4 6 8 10

(V /V

)

∆