Bottom-heavy initial mass functions reveal hidden mass in early galaxies

JWST observations have revealed that massive galaxies formed and evolved far faster than predicted by galaxy formation models, with many having already assembled a large mass in stars $\sim12$ billion years ago [1-7]. However, masses of distant galaxies are highly uncertain, as they assume a distribution of stellar birth masses (the initial mass function [IMF]) similar to that in the Milky Way (MW). Specifically, the contribution from low-mass stars, which make up the bulk of stellar mass, is not directly observed, but inferred based on an extrapolation of the MW IMF. Here, we provide the first robust measurements of the IMF beyond the local Universe. Using ultra-deep spectra of nine massive, quiescent galaxies at $z\sim0.7$ from the ambitious JWST-IMFERNO program, extended to bluer wavelengths with deep spectra from LEGA-C [8], we find that these distant galaxies have excess low-mass stars. In other words, they have more bottom-heavy IMFs than the MW. For the oldest two galaxies, which are direct descendants of JWST’s “impossibly early” galaxies, the bottom-heavy IMFs increase their stellar masses by a factor of $3-4$. These galaxies thus amplify the tension with galaxy formation models.

💡 Research Summary

This paper presents the first robust, direct measurements of the stellar initial mass function (IMF) in massive, quiescent galaxies at intermediate redshift (z ≈ 0.7) using a combination of ultra‑deep JWST‑NIRSpec/Multi‑Object Spectroscopy (the IMFERNO program) and the optical spectra from the LEGA‑C survey. The authors selected nine galaxies from the COSMOS‑Web field based on their quiescent stellar populations, H‑band magnitudes, and availability of high‑quality LEGA‑C data. The JWST spectra cover the rest‑frame near‑infrared (≈ 1.0–1.1 µm) with a median signal‑to‑noise ratio (S/N) > 80 Å⁻¹, while the LEGA‑C spectra provide rest‑frame optical coverage (≈ 3700–8000 Å) with S/N > 15 Å⁻¹. This wavelength baseline enables simultaneous constraints on elemental abundance patterns and IMF‑sensitive absorption features (e.g., Na I, Ca II triplet, FeH) that are otherwise degenerate.

The analysis employs the full‑spectrum fitting code “alf” (absorption line fitter), which models the spectra with flexible stellar population synthesis (SPS) templates. The code simultaneously fits 18 elemental abundances, age, velocity dispersion, and a double‑broken power‑law IMF characterized by two low‑mass slopes (0.1–0.5 M⊙ and 0.5–1 M⊙). A single‑age simple stellar population (SSP) is adopted as the baseline, and a two‑burst star‑formation history is tested but yields consistent results. The key diagnostic is the IMF mismatch parameter α_IMF = (M/L)_free / (M/L)_MW, i.e., the ratio of the mass‑to‑light (M/L) derived when the IMF is free to the M/L assuming a Milky Way (Kroupa) IMF. Values α_IMF > 1 indicate an excess of low‑mass stars (a “bottom‑heavy” IMF).

The results show that several galaxies have α_IMF significantly above unity, with a clear trend of increasing α_IMF with higher stellar velocity dispersion (σ_v) and higher iron abundance (