The Hubble sequence in JWST CEERS from unbiased galaxy morphologies

TL;DR

This study demonstrates that a Hubble-like galaxy morphological sequence exists up to redshift ~4, using a novel method to compare galaxy structures across cosmic time with data from HST and JWST.

Contribution

The paper introduces a new approach to construct comparable galaxy images across redshifts and uses it to analyze morphological evolution up to z~4.

Findings

A continuous galaxy morphological sequence exists up to z~4.

The approach improves separation between early- and late-type galaxies.

Progenitors of low-mass galaxies are mainly star-forming disks at all epochs.

Abstract

Whether the "Hubble sequence" of galaxy morphologies exists up to z~4 is still disputed, and one of the challenges is characterizing galaxy structure consistently across a wide range of redshifts. To enable a fair comparison across cosmic time, we constructed "absolute" images of galaxies spanning 0.15<z<4.5 and 8<log $M_{\star}$<11 from HST CANDELS and JWST CEERS surveys, by matching the effective resolution and surface brightness limit of galaxies, accounting for cosmological dimming and evolution in size and mass-to-light ratio. We measured the structural parameters of 2825 galaxies and used the UMAP technique to study the evolution of the morphological phase space. We find a continuous sequence spanning late-type to early-type galaxies, with no redshift gradient - indicating that a Hubble-like sequence is established by z~4. We show that our approach recovers a cleaner separation between early- and late-type galaxies than visual classifications. By tracing progenitors using empirical mass assembly histories, we find that progenitors of low-mass galaxies are predominantly star-forming disks at all epochs. Progenitors of massive galaxies follow two distinct paths: a stable star-forming disk population with little structural evolution, and an early-type population that builds up rapidly from irregular progenitors and quenches within a few Gyr, consistent with a compaction-driven quenching scenario.