A third hydrogen line helps reveal how dust dims distant galaxies

Measuring how quickly a galaxy forms stars requires accounting for light that never reaches the telescope. Interstellar dust changes the strengths of hydrogen emission lines, which trace radiation from short-lived massive stars. An unsuitable dust correction can therefore distort the inferred rate of recent star formation.

In a preprint posted on November 5, Owen Zafuta, Joseph Havens and colleagues at the University of Kansas use JWST observations to estimate dust corrections individually for 386 star-forming galaxies. Their central finding is substantial diversity: even galaxies at similar redshifts can require very different descriptions of how attenuation changes with wavelength.

Why two lines are not enough

Astronomers commonly compare two hydrogen lines, Hα and Hβ. Under an assumed set of gas conditions, atomic physics predicts their intrinsic brightness ratio. Dust generally suppresses the shorter-wavelength Hβ line more strongly, increasing the observed Hα-to-Hβ ratio, known as the Balmer decrement.

But that ratio alone cannot determine both the overall strength of attenuation and how rapidly it changes with wavelength. Conventional corrections therefore adopt a curve shape in advance, often a Calzetti-like or Small Magellanic Cloud-like prescription.

The researchers instead add a third hydrogen line. This supplies a second independent ratio, allowing them to solve for two parameters: the effective optical depth at a reference wavelength of 5,500 angstroms, and the slope of a power-law attenuation curve. A larger slope means attenuation changes more steeply with wavelength.

Infrared Paschen lines are especially useful because they lie farther in wavelength from Hα and Hβ than additional Balmer lines do. That wider separation helps distinguish the curve’s slope from its overall strength.

From thousands of spectra to 386 solutions

The study uses NIRSpec spectroscopy from the JWST Advanced Deep Extragalactic Survey, retrieved through the DAWN JWST Archive. The parent catalog contains 3,156 galaxies; the analysis focuses on redshifts between 0.5 and 7.

After emission-line requirements and screening for active galactic nuclei, which can contaminate the hydrogen emission, 848 star-forming candidates remained. For each, the team attempted a solution using Hα, Hβ and one eligible additional line. Where several were available, a fixed priority determined which third line entered the fit.

Only 386 candidates produced convergent solutions satisfying the adopted parameter limits. Of the remainder, 345 failed to converge and 117 reached the upper slope boundary. The final sample is therefore a selected subset, not a representative census of all galaxies.

The team also repeatedly perturbed the measured line fluxes according to their uncertainties to assess how those uncertainties affected the fitted parameters. These Monte Carlo calculations propagate observational errors; the galaxy spectra themselves are observations, not simulated galaxies.

Diverse curves, with qualified improvements

The accepted sample has a median attenuation slope of 2.65. Only 17.1% fall between the representative Calzetti-like and SMC-like slopes of 0.7 and 1.3. Fits using the infrared Paβ line provided particularly useful constraints, although measurement quality also affects how tightly any individual solution is pinned down.

To test the corrections, the researchers compared star formation rates inferred from different hydrogen lines, including additional transitions not used to fit each galaxy’s curve. Agreement among the fitted lines is imposed mathematically, so the unused lines provide the more informative check.

The paper reports a median line-to-line scatter of 0.07 dex for its galaxy-specific corrections, compared with 0.09 dex for SMC corrections and 0.11 dex for Calzetti-like corrections. Smaller scatter indicates more consistent estimates, not an independent measurement of the true star formation rate.

Nor does the method improve every reported metric. The main results section gives a larger fraction of factor-of-two disagreements for the galaxy-specific corrections: about 7.5%, versus 6.2% for SMC and 5.5% for Calzetti-like corrections.

What remains uncertain

The calculation assumes intrinsic hydrogen ratios corresponding to an electron temperature of 10,000 kelvin and a specified electron density. Different gas conditions would shift the recovered dust parameters. A single power law also compresses potentially clumpy, uneven obscuration into an effective description rather than uniquely revealing dust geometry.

The higher-redshift subset has steeper recovered slopes and lower optical depths on average. However, it contains only 54 accepted galaxies beyond redshift 3.5, where Paschen coverage largely gives way to less constraining Balmer-only measurements. Changing line availability, measurement quality and selection make genuine evolution difficult to isolate.

The result supports measuring attenuation individually when suitable lines are available—not treating the recovered distribution as a universal dust law or a definitive history of cosmic dust.