The evolution of the sizes and angular momentum content of galaxies in the COLIBRE simulations

Abstract

We analyse the sizes and specific angular momentum content of galaxies in the Colibre cosmological hydrodynamical simulations spanning two orders of magnitude in mass resolution. We compare the predicted size-mass and angular momentum-mass relations to a broad range of observational measurements spanning redshifts $z=0$ to $4$. At $z=0$, Colibre reproduces observed size-mass relations over the sampled mass range $10^8 \lesssim M_\star/{\rm M_\odot}\lesssim 10^{11.5}$, and for multiple size definitions, including two- and three-dimensional stellar half-mass radii, half-light radii across several wavelengths, as well as alternative measures such as baryonic half-mass radii and characteristic radii defined by stellar surface density thresholds. The simulations also recover the observed segregation of galaxies in the size-mass plane by morphological type and star formation rate, and reproduce the distinct, approximately parallel sequences followed by star-forming discs and quenched spheroids in the stellar specific angular momentum-mass plane. The angular momentum content of star-forming Colibre galaxies match that of observed systems out to $z\approx 1.5$. At higher redshifts, massive galaxies ($ 10^{9.5}\lesssim M_\star/{\rm M_\odot}\lesssim 10^{11}$) in the simulations are somewhat smaller than observed, and the separation between star-forming and passive populations in the size-mass plane is reduced relative to observations, while at lower masses the agreement remains good. This apparent discrepancy may reflect the effects of dust attenuation, which is neglected in our analysis and may preferentially obscure the central regions of observed systems. Overall, our findings highlight the close connection between galaxy size, angular momentum, and morphology over cosmic time.

Figures (19)

• Figure 1: Median stellar surface density profiles, $\Sigma_\star(R)$, for $z=0$ central galaxies of mass $M_\star = 10^9\, {\rm M_\odot}$ and $10^{11}\, {\rm M_\odot}$. Different coloured curves correspond to different simulations. Thick segments mark the radial range used for profile fitting and error bars indicate the $16^{\rm th}$ to $84^{\rm th}$ percentile scatter in $\Sigma_\star(R)$ for individual galaxies in L200m6. Grey lines show the mean of the best-fitting Sérsic profiles obtained separately for each simulation and mass bin; shaded grey regions span the range corresponding to individual fits to each of the median $\Sigma_\star(R)$ profiles. Coloured circles show median $R_{\star,50}$ values measured directly from the particle data, and upward arrows indicate the values inferred from the best-fitting Sérsic profiles. Coloured squares correspond to the radius $R_1$ at which $\Sigma_\star=1\,{\rm M_\odot\, pc^{-2}}$. Downward arrows mark the $z=0$ gravitational softening lengths, $\epsilon_{\rm prop}$, for each simulation, which exceed the minimum radii where profiles from runs with different resolutions converge.
• Figure 2: Virtual Hubble Space Telescope observations of a $z=0$ star-forming disc galaxy (left panel; $M_\star=8.8\times 10^{10}\,{\rm M_\odot}$, SFR=$2.5\,{\rm M_\odot\, yr^{-1}}$) and a passive elliptical galaxy (right panel; $M_\star=1.5\times 10^{11}\,{\rm M_\odot}$, SFR=$0.2\,{\rm M_\odot\, yr^{-1}}$). Both galaxies are viewed along a line of sight parallel to the total stellar angular momentum vector. The circles in each panel highlight three measures of galaxy size used in this paper (see Section \ref{['SSSecSizes']} for details): (1) the 2D half-stellar-mass radius, $R_{\star,50}$ (solid white circle); (2) the 2D half-baryonic-mass radius, $R_{\rm bar,50}$ (dashed light blue circle); and (3) the radius $R_1$ at which the stellar surface density profile first drops below $\Sigma_\star = 1\,{\rm M_\odot\,pc^{-2}}$ (dot–dashed yellow circle). The image scale is indicated by the horizontal dashed line in the lower-right corner, which spans $15\,{\rm kpc}$.
• Figure 3: Three-dimensional stellar half-mass radius ($r_{\star,50}$; upper panels) and stellar specific angular momentum ($j_\star$; lower panels) as a function of stellar mass at $z=0$, 1, 2, and 3 (left to right). Median relations are constructed by combining galaxies from different runs above a resolution-dependent mass limit. For m5 resolution (L025m5 at $z=0$, and L050m5 at $z>0$) we include all galaxies with $M_\star \ge 100\, m_{\rm bar}$; for L200m6 and L400m7, we include galaxies with $M_\star \ge 500\, m_{\rm bar}$ and $\geq 1000\, m_{\rm bar}$, respectively. The line segments above the upper-left panel indicate the stellar-mass range where each simulation contributes the highest number of galaxies. Connected circles show relations for central galaxies, while dashed lines show the entire galaxy population (centrals plus satellites). Shaded regions denote the $16^{\rm th}$–$84^{\rm th}$ percentile scatter for the full distribution of centrals. Blue lines show the median relations for galaxies within 0.5 dex of the redshift-dependent star-formation main sequence defined by eq. \ref{['eqSFR']} (star-forming systems), while red lines show the medians for galaxies at least 0.5 dex below the main sequence (passive systems). Only bins that contain at least 25 galaxies are plotted. The grey line segment in the lower panels shows the $j_\star\propto M_\star^{2/3}$ scaling predicted by the virial theorem. For $z\lesssim 2$ and $10^9\lesssim M_\star/{\rm M_\odot} \lesssim 10^{11}$, star-forming central galaxies are larger (by $\approx0.1$–0.3 dex) and have higher $j_\star$ (by $\approx0.2$–0.5 dex) than passive galaxies. Conversely, massive ($M_\star \gtrsim 10^{11}\,{\rm M_\odot}$) and low-mass ($M_\star \lesssim 10^9\,{\rm M_\odot}$) star-forming and passive have similar sizes and angular momenta at all redshifts. Including satellite galaxies does not affect the trends, which can be seen by comparing the dashed lines to the connected circles.
• Figure 4: The upper panel shows mass-weighted size estimates: the two-dimensional half-mass radii $R_{\star,50}$ from xGASS Hardwick2022 and GAMA Bellstedt2024, the radius $R_1$ at which the stellar surface density equals $\Sigma_\star = 1\,{\rm M_\odot\,pc^{-2}}$Trujillo2020, and the baryonic half-mass radii of SPARC galaxies Zichen2025. Also shown are the median $z=0$ two-dimensional half-mass radii of colibre galaxies in L025m5, L200m6, and L400m7 (coloured lines). The latter, shown for all galaxies (centrals plus satellites), were computed directly from the stellar particle data and were used to calibrate colibre's sub-resolution physics models; these relations should only be compared with the observed $R_{\star,50}-M_\star$ relations. The lower panel shows light-weighted sizes from the GAMA Survey, including the $r$-band two-dimensional half-light radii $R_{r,50}$Casura2022, and the radii enclosing 50 and 90 per cent of the $u$- and $z$-band light. Error bars, where present, correspond to the $16^{\rm th}$ to $84^{\rm th}$ percentile scatter.
• Figure 5: Size–mass relations for $z=0$ central galaxies in the colibre simulations compared with recent observations. The top panels show median 2D stellar half-mass radii, $R^{\rm Ser}_{\star,50}$, compared with observational measurements from Bellstedt2024. The bottom panels show the median 2D $r$-band half-light radii, $R^{\rm Ser}_{r,50}$, and are compared with the measurements of Casura2022 (connected circles) and Shen2003 (dot-dashed lines). In both cases, galaxy sizes were measured from single-component Sérsic fits to the stellar mass surface density or $r$-band surface brightness profiles, as described in Section \ref{['SSecAnalysis']}. Results from L025m5, L200m6, and L400m7 are shown using dot-dashed, solid, and dashed lines, respectively, with thin line segments indicating mass bins that contain fewer than 10 galaxies. The shaded regions highlight the $16^{\rm th}$ to $84^{\rm th}$ percentile scatter for L200m6. Observed data are plotted as connected symbols, with error bars indicating the $16^{\rm th}$–$84^{\rm th}$ percentile range. The left column shows results for all central galaxies, regardless of morphological type. The middle and right columns show late- and early-type central galaxies, respectively, for which morphological classification is based on Sérsic index, $n_{\rm S}$: in the top (bottom) panels, galaxies with $n_{\mathrm{S}} \leq 1.5$ (2.5) are classified as late-type, and those with $n_{\mathrm{S}} \geq 2.5$ as early-type. Overall, colibre reproduces the size--mass relations for both early- and late-type galaxies, and achieves good convergence across resolution levels.