JWST/NIRSpec Balmer-line Measurements of Star Formation and Dust Attenuation at z~3-6

TL;DR

JWST/NIRSpec CEERS enables direct Balmer-line measurements out to $z\sim6.5$, allowing nebular attenuation from the Balmer decrement and dust-corrected SFRs from $L_{H\alpha}$. The analysis applies metallicity- and binary-star–aware conversions to convert $L_{H\alpha}$ into SFR($H\alpha$) and maps the SFR–$M_*$ relation in three redshift bins, finding general agreement with the evolving main sequence and no strong redshift evolution in attenuation at fixed $M_*$ up to $z\sim6.5$. These results imply a relatively stable dust-mass distribution in star-forming galaxies across cosmic time and highlight the need for larger, well-calibrated samples to fully interpret dust attenuation indicators at $z>3$ and to improve reionization-era inferences.

Abstract

We present an analysis of the star-formation rates (SFRs) and dust attenuation properties of star-forming galaxies at $2.7\leq z<6.5$ drawn from the Cosmic Evolution Early Release Science (CEERS) Survey. Our analysis is based on {\it JWST}/NIRSpec Micro-Shutter Assembly (MSA) $R\sim1000$ spectroscopic observations covering approximately $1-5$$μ$m. Our primary rest-frame optical spectroscopic measurements are H$α$/H$β$ Balmer decrements, which we use as an indicator of nebular dust attenuation. In turn, we use Balmer decrements to obtain dust-corrected H$α$-based SFRs (i.e., SFR(H$α$)). We construct the relationship between SFR(H$α$) and stellar mass ($M_*$) in three bins of redshift ($2.7\leq z< 4.0$, $4.0\leq z< 5.0$, and $5.0\leq z<6.5$), which represents the first time the star-forming main sequence has been traced at these redshifts using direct spectroscopic measurements of Balmer emission as a proxy for SFR. In tracing the relationship between SFR(H$α$) and $M_*$ back to such early times ($z>3$), it is essential to use a conversion factor between H$α$ and SFR that accounts for the subsolar metallicity prevalent among distant galaxies. We also use measured Balmer decrements to investigate the relationship between dust attenuation and stellar mass out to $z\sim6$. The lack of significant redshift evolution in attenuation at fixed stellar mass, previously confirmed using Balmer decrements out to $z\sim2.3$, appears to hold out to $z\sim 6.5$. Given the rapidly evolving gas, dust, and metal content of star-forming galaxies at fixed mass, this lack of significant evolution in attenuation provides an ongoing challenge to explain.

Figures (4)

• Figure 1: Left: Redshift distribution of all 113 CEERS galaxies at $2.7 \leq z \leq 6.5$, from which the sample of star-forming galaxies we analyze is drawn. The three redshift bins we delineate are indicated in green ($2.7 \leq z < 4.0$), blue ($4.0 \leq z < 5.0$), and magenta ($5.0 \leq z < 6.5$). Right: Stellar mass distributions for the three redshift samples indicated in the left-hand panel, using the same color coding. For each redshift distribution, the median stellar mass is marked with a vertical dotted line. These median stellar mass values are $\log(M_*/M_{\odot})=$9.59, 9.38, and 8.38, respectively, for the $2.7 \leq z < 4.0$, $4.0 \leq z < 5.0$, and $5.0 \leq z < 6.5$ redshift samples.
• Figure 2: SFR(H$\alpha$) vs. $M_*$. Green, blue, and magenta symbols are used, respectively, for the $2.7\leq z < 4.0$, $4.0 \leq z < 5.0$, and $5.0 \leq z < 6.5$ samples, and galaxies with H$\beta$ upper limits are indicated as SFR(H$\alpha$) lower limits (i.e., due to the lower limit on the Balmer decrement). The median error bar for each sample is shown in the lower-right corner of the plot in its designated color. Along with CEERS data points, we plot the best-fit relation from speagle2014 (their equation (28)), at the median redshift of each sample ($z=3.30, 4.60$ and $5.65$, respectively, for the $2.7\leq z < 4.0$, $4.0 \leq z < 5.0$, and $5.0 \leq z < 6.5$ samples), and offset by $-0.34$ dex in the $y$-axis to account for different assumptions regarding the conversion between observables and SFR.
• Figure 3: Composite spectra for each of the three redshift bins, where, from bottom to top, we show spectra, respectively, for the $2.7\leq z<4.0$, $4.0\leq z < 5.0$, and $5.0\leq z < 6.5$ samples. In each row, the left set of panels represents the "low-mass bin," while the right side indicates the "high-mass bin," where each redshift sample is divided at the median stellar mass. Each composite spectrum is zoomed in on the regions covering H$\beta$ and [OIII]$\lambda\lambda 4959,5007$, as well as H$\alpha$, [NII]$\lambda\lambda 6548,6583$, and [SII]$\lambda\lambda 6717,6731$. These features are marked and labeled.
• Figure 4: Attenuation vs. $M_*$ based on the Balmer line ratio, H$\alpha$/H$\beta$. In each panel, the background grayscale histogram corresponds to the distribution of local SDSS galaxies. The running median H$\alpha$/H$\beta$ ratio for $z\sim 2.3$ star-forming galaxies in the MOSDEF survey is shown in red shapley2022. In the left panel, we show individual CEERS galaxies color-coded by redshift as in previous plots. On the right, plotted H$\alpha$/H$\beta$ ratios are measured from composite spectra in two bins of stellar mass for each redshift range, as shown in Figure \ref{['fig:hahbplots']}.