The evolution of galaxy shapes in CANDELS: from prolate to oblate

TL;DR

This study models galaxy shapes in CANDELS, revealing an evolution from prolate to oblate forms over cosmic time, consistent with simulation predictions and aiding targeted galaxy selection.

Contribution

It introduces a method to infer intrinsic galaxy shapes from projected distributions, demonstrating shape evolution from prolate to oblate with redshift and mass.

Findings

Galaxies are rounder at smaller sizes across all masses and redshifts.

Prolate shapes dominate at high redshift and low mass, oblate shapes at low redshift and high mass.

Results align with simulations predicting shape transition due to galaxy core baryon dominance.

Abstract

We model the projected b/a-log a distributions of CANDELS main sequence star-forming galaxies, where a (b) is the semi-major (semi-minor) axis of the galaxy images. We find that smaller-a galaxies are rounder at all stellar masses M and redshifts, so we include a when analyzing b/a distributions. Approximating intrinsic shapes of the galaxies as triaxial ellipsoids and assuming a multivariate normal distribution of galaxy size and two shape parameters, we construct their intrinsic shape and size distributions to obtain the fractions of prolate, oblate and spheroidal galaxies in each redshift and mass bin. We find that galaxies tend to be prolate at low m and high redshifts, and oblate at high M and low redshifts, qualitatively consistent with van der Wel et al. (2014), implying that galaxies tend to evolve from prolate to oblate. These results are consistent with the predictions from simulations (Ceverino et al. 2015, Tomassetti et al. 2016) that the transition from prolate to oblate is caused by a compaction event at a characteristic mass range, making the galaxy center baryon dominated. We give probabilities of a galaxy's being prolate, oblate or spheroidal as a function of its M, redshift, projected b/a and a, which can facilitate target selections of galaxies with specific shapes at hight redshifts. We also give predicted optical depths of galaxies, which are qualitatively consistent with the expected correlation that AV should be higher for edge-on disk galaxies in each log a slice at low redshift and high mass bins.