Ultra-diffuse galaxies in the EAGLE simulation

TL;DR

This work uses the high-resolution EAGLE Recal-L025N0752 simulation to investigate ultra-diffuse galaxies (UDGs) and test competing formation scenarios in field and satellite environments. By identifying 181 UDGs and comparing them to 529 normal dwarfs in a matched magnitude range, the study shows UDGs are not a distinct population but a dwarf subset with extended stellar distributions. Field UDGs arise from higher angular momentum in star-forming gas and recent outer-star formation—likely facilitated by galactic fountains—rather than from high halo spins or SN-driven stellar expansion; satellite UDGs predominantly form via tidal interactions, with about 60% pre-infall. The results align with observations of UDG abundances and scaling relations, providing a multi-channel formation framework and concrete predictions about gas recycling and environmental effects in shaping UDGs.

Abstract

We use the highest-resolution EAGLE simulation, Recal-L025N0752, to study the properties and formation of ultra-diffuse galaxies (UDGs). We identify 181 UDGs and find their properties closely match observations. The total masses of EAGLE UDGs range from ${\sim}5\times 10^{8}~M_{\odot}$ to ${\sim}2\times 10^{11}~M_{\odot}$, indicating that they are dwarf galaxies rather than failed $L_\star$ galaxies. EAGLE UDGs are not a distinct population, but rather a subset of dwarf galaxies, as their properties generally form a continuous distribution with those of normal dwarf galaxies. Unlike the situations in previous studies, the extended sizes of field UDGs in EAGLE are not driven by high halos spin or by supernova-induced stellar expansion, but instead largely arise from high spins in their star-forming gas and thus the newly formed stars at large radii. This might be attributed to galactic fountains, by which star-forming gas are launched to large halo-centric distances and acquire additional angular momentum through interactions with the circumgalactic medium. For satellite UDGs, ${\sim} 60 \%$ of them were already UDGs before falling into the host galaxy, while the remaining ${\sim} 40\%$ were normal galaxies prior to infall and subsequently transformed into UDGs due to tidal effects after infall.

Figures (9)

• Figure 1: Left: The relation between the fitted galaxy effective radius, $r_\mathrm{e}$ and central surface brightness, $\mu_0$. UDGs, lying in the blue shaded area of the main panel, are defined as galaxies with $r_\mathrm{e} \geq 1.5~$pkpc and $\mu_0 \geq 24~\mathrm{mag}~\mathrm{arcsec}^{-2}$, while the rest are defined as normal galaxies (red shaded area), as divided by black dotted lines. The upper and right panels show the histograms of $r_\mathrm{e}$ and $\mu_0$, respectively, and the blue (red) dashed lines show the median values of UDGs (normal galaxies). Using this approach, we manage to identify UDGs with median effective radii and central surface brightnesses ($\left<r_{\rm e}\right> = 2.37$ pkpc and $\left<\mu_0 \right> = 25.12$ mag arcsec$^{-2}$) that reasonably match those seen in observations 2015ApJ...798L..45V. Right: Similar to the left panel, but for the distribution of stellar and total masses. The inset panel displays the properties of field galaxy subsamples selected for comparison (see Section \ref{['sec:field']} for details), with dash-dotted lines representing their mean stellar and total masses. In the full sample, normal galaxies generally have higher total and stellar masses than UDGs; however, in the subsample, this difference is largely reduced, making it more suitable for direct comparison.
• Figure 2: General properties of the complete sample of UDGs and normal galaxies at $z=0$. From left to right, the top panels present the PDFs of dark matter fractions of subfind galaxies ($f_{\rm DM}$), dark matter fractions within the fitted galaxy effective radius ($f_{{\rm DM}, <r_{\rm e}}$), and $g$-band magnitude ($M_{g}$), whereas the lower panels show the PDFs of Sérsic index ($n_{\rm sersic}$), color ($g-r$), and stellar metallicity ($[{\rm Fe}/{\rm H}]$). UDGs and normal galaxies are plotted with blue and red, respectively. The vertical dashed lines mark the median values. Most of the UDGs have a dark matter fraction $\gtrsim$ 90%, suggesting they are not dark matter deficient galaxies, and they generally tend to have a fainter magnitude, a lower Sérsic index, a slightly redder color and a lower stellar metallicity, compared to normal galaxies.
• Figure 3: The mean UDG satellite abundance as a function of host halo mass, i.e., the $N_\mathrm{UDG\ satellite}-M_\mathrm{200,\,host}$ relation. The individual EAGLE host galaxies are plotted using blue dots, and the mean $N_\mathrm{UDG\ satellite}-M_\mathrm{200,\,host}$ relation computed in different mass bins is shown using blue points with $1\sigma$ error bars. For comparison, we also plot the observational data from Munoz2015, vanderBurg2016, Roman2017a, Roman2017b, van_der_Burg2017, Mancera_Pina2019, and Karunakaran2023. The dashed line shows the best-fitting relation from Karunakaran2023. The UDG satellite number from the EAGLE simulation broadly aligns with these observation results at the low mass end.
• Figure 4: The relation between sSFR and relative distance to the host halo. The purple-shaded region in the upper right corner illustrates the criteria for selecting field galaxies (i.e., sSFR $> 10^{-11}~\mathrm{yr}^{-1}$, either as host halos themselves (marked as 'central') or as satellite subgroups in an FOF with $d/R_\mathrm{200,\ host}>1.5$). The yellow-shaded region shows satellite galaxies ($d/R_\mathrm{200,\ host} \leq 1$). Histograms of relative distance and sSFR are shown in the upper and right panels, respectively. This division allows us to distinguish the different UDG formation mechanisms in different environments (i.e., in the field and dense environment).
• Figure 5: PDFs of spin parameters for field galaxies. From left to right, top to bottom, the distributions for dark matter, stars, gas, and star-forming gas (i.e., gas with ${\rm SFR} > 0$) are shown. Blue and red lines represent UDGs and normal galaxies, respectively. The dotted lines show the cumulative distribution functions (CDFs, right $y$-axis). The vertical dashed lines indicate the median spin parameter of each distribution. The KS test results are summarized in the top-right corner of each panel. UDGs have higher spins in star-forming gas and stars than normal galaxies, while the differences in dark matter and gas spins are insignificant.