Star Formation Rates Based on Hydrogen Alpha May Be Flawed

The Light That Lied

For decades, astronomers have treated hydrogen alpha light like a reliable witness. When a young, massive star ignites, it floods its surroundings with ultraviolet radiation. That radiation strips electrons from hydrogen atoms. When the electrons snap back into place, they emit a specific shade of red light: hydrogen alpha. The brighter the glow, the more stars are being born. Simple. Clean. Universal.

But a team of researchers led by Sandro Tacchella has been running simulations that suggest the witness has been lying. Not maliciously. Not obviously. But systematically, and in ways that have quietly distorted our understanding of how galaxies grow.

The problem is not that hydrogen alpha fails to trace star formation. It does. The problem is that it does not trace only star formation. And the corrections astronomers have been applying may be missing nearly half the story.

What H Alpha Actually Measures

The standard logic is elegant. Hydrogen alpha emission comes from gas that has been ionized by young, hot stars. Those stars are short lived, on the order of 10 million years. So the hydrogen alpha glow should track recent star formation with high fidelity. Measure the light, apply a dust correction, and you get the star formation rate.

Tacchella and his colleagues (Tacchella et al., 2022) put this logic to a rigorous test. They built high resolution simulations of Milky Way like and Large Magellanic Cloud like galaxies, then ran full radiative transfer calculations. That means they tracked every photon as it scattered off dust grains, got absorbed by atoms, or escaped into space. They included non equilibrium thermochemistry and dust evolution. They did not simplify the physics.

What they found should unsettle anyone who has used hydrogen alpha as a star formation rate indicator.

The simulations show that only about 57 percent of the ionizing photons produced by young stars actually go into producing hydrogen line emission. The rest are lost to three processes. Dust absorbs about 28 percent of the ionizing radiation before it can ionize any hydrogen at all. Helium atoms capture about 9 percent. And roughly 6 percent of the ionizing photons simply escape the galaxy entirely.

The authors found that these losses are not small corrections. They are major factors that have been folded into empirical calibrations without being fully understood. The standard dust correction via the Balmer decrement works reasonably well, recovering the intrinsic hydrogen alpha luminosity within about 25 percent. But the deeper issue is that the intrinsic luminosity itself is not what astronomers think it is.

The Collisional Confusion

There is another complication. Hydrogen alpha can be produced without any stars being involved at all.

When gas is heated to temperatures above about 10,000 Kelvin, collisions between electrons and hydrogen atoms can excite the atoms directly. The atoms then emit hydrogen alpha as they relax. This is collisional excitation, and it has nothing to do with star formation.

Tacchella et al. (2022) found that collisional excitation contributes between 5 and 10 percent of the total hydrogen alpha emission in their simulations. That fraction is remarkably stable across different regions of the galaxy and at different heights above the galactic plane. It does not spike in star forming regions. It is just always there, a background hum of false signal.

For a galaxy forming stars at a moderate rate, that 5 to 10 percent contamination might be tolerable. But for galaxies with strong shocks or active galactic nuclei, where the gas can be heated to much higher temperatures, the collisional contribution could dominate. The hydrogen alpha light would still be bright. The star formation rate inferred from it would be wildly wrong.

The Scattering Surprise

Here is where the story gets stranger. Hydrogen alpha photons do not travel in straight lines from their point of origin to our telescopes. They bounce.

Tacchella et al. (2022) found that scattering by dust grains boosts the observed hydrogen alpha luminosity by about 40 percent. A photon emitted in a dense star forming region might scatter multiple times before escaping the galaxy. Each scatter changes its direction. Some photons that would have been headed away from us get redirected toward us. Some that would have escaped cleanly get absorbed.

This scattering has a subtle effect on how we interpret hydrogen alpha images. The light we see from a particular pixel in a galaxy image did not necessarily originate in that pixel. It could have come from a star forming region tens or hundreds of parsecs away, scattered into our line of sight by intervening dust. The spatial correspondence between hydrogen alpha emission and actual star formation is fuzzier than we thought.

The authors found that this scattering effect is particularly important for the diffuse ionized gas that fills the space between star forming regions. That diffuse glow, long assumed to be powered by ionizing photons leaking out of star forming regions, turns out to be significantly boosted by scattered light. The true escape fraction of ionizing photons from star forming regions may be even lower than the 6 percent they calculated, because some of the light we attribute to escaping radiation is actually scattered.

How the Balmer Decrement Fools Us

Astronomers have a standard trick for correcting hydrogen alpha for dust extinction. They measure the ratio of hydrogen alpha to hydrogen beta, another Balmer line. Dust absorbs more strongly at shorter wavelengths, so hydrogen beta gets more extincted than hydrogen alpha. The observed ratio tells you how much dust is in the way, and you can correct the hydrogen alpha measurement accordingly.

The simulations show that this trick works, but only up to a point. Tacchella et al. (2022) found that the intrinsic hydrogen alpha luminosity can be recovered within 25 percent using the Balmer decrement correction. That is not bad for an empirical correction applied to complex astrophysical systems.

But here is the catch. The dust attenuation law, the mathematical relationship that describes how extinction varies with wavelength, is not universal. The authors found that the attenuation law depends on the amount of attenuation itself. Galaxies with more dust have different attenuation properties than galaxies with less dust. This is true both on spatially resolved scales, within individual galaxies, and on integrated scales, comparing one galaxy to another.

This means that applying a single dust correction recipe to all galaxies introduces systematic errors. Low dust galaxies get overcorrected. High dust galaxies get undercorrected. The errors are not random. They are correlated with the very property astronomers are trying to measure.

Why This Matters for High Redshift Galaxies

The stakes are highest for the most distant galaxies. At high redshift, hydrogen alpha is redshifted into the infrared, where the James Webb Space Telescope can observe it. JWST was designed in part to measure star formation rates in the early universe using hydrogen alpha. The Tacchella et al. (2022) simulations suggest that those measurements carry hidden uncertainties.

At high redshift, galaxies are generally smaller, denser, and more dust rich than local galaxies. The escape fraction of ionizing photons might be different. The collisional excitation contribution might be different. The scattering properties of the dust might be different. Every one of these factors changes the relationship between hydrogen alpha luminosity and star formation rate.

The authors built their simulations for local galaxy analogs, not for high redshift systems. They explicitly state that future applications to cosmological zoom in simulations will be needed to interpret JWST observations. Until those simulations are done, every star formation rate derived from hydrogen alpha in the early universe carries an unknown systematic error.

What the Diffuse Gas Is Telling Us

One of the most interesting results from the Tacchella et al. (2022) study concerns the diffuse ionized gas that extends far above the galactic plane. For decades, astronomers have debated what powers this extraplanar emission. The leading theory is that ionizing photons leak out of the galactic disk and ionize gas in the halo.

The simulations confirm this picture, but with a twist. The extraplanar hydrogen alpha emission is indeed powered by Lyman continuum photons escaping the disk. But the authors found that the morphology of this emission does not trace the star formation rate in any simple way. It traces the geometry of the interstellar medium, the distribution of dust, and the clumpiness of the gas.

A galaxy with a clumpy, porous interstellar medium will leak more ionizing photons and produce brighter extraplanar hydrogen alpha emission than a galaxy with the same star formation rate but a smoother gas distribution. If you measure the total hydrogen alpha emission from such a galaxy, including the diffuse halo, you will infer a higher star formation rate than is actually present.

This is not just a correction factor. It is a fundamental limitation on what hydrogen alpha can tell us about individual galaxies. The emission depends on the three dimensional structure of the interstellar medium, which we cannot observe directly.

What This Does Not Prove

This study does not mean hydrogen alpha is useless. It remains one of the most powerful tools for measuring star formation in galaxies. The Tacchella et al. (2022) simulations confirm that the hydrogen alpha surface brightness profiles of their simulated galaxies match observations of real galaxies very well. The basic picture is correct.

What the study does is quantify the uncertainties. It tells us that the conversion from hydrogen alpha luminosity to star formation rate is not a simple one to one mapping. It depends on dust content, gas geometry, helium abundance, escape fraction, and collisional excitation. All of these factors vary from galaxy to galaxy and from region to region within a galaxy.

The study also does not tell us exactly how to correct for these effects in real observations. The simulations are for specific galaxy types, not a representative sample of the universe. Applying the exact correction factors from this study to all galaxies would be a mistake. The value of the study is in identifying the processes that matter and showing that they are not negligible.

What the Field Gets Wrong

The standard approach in extragalactic astronomy has been to calibrate the hydrogen alpha star formation rate indicator empirically, by comparing hydrogen alpha luminosity to other star formation tracers like far infrared emission or ultraviolet continuum. These calibrations work statistically, across large samples of galaxies. But they paper over the physical complexity.

The Tacchella et al. (2022) study shows that the empirical calibrations are hiding real physics. The fact that a calibration works on average does not mean it works for any particular galaxy. A galaxy with unusually high dust content, or an unusual escape fraction, or a strong contribution from collisional excitation, will have its star formation rate systematically misestimated.

This matters most for the galaxies that astronomers find most interesting: extreme starbursts, low metallicity dwarfs, and high redshift galaxies. These are precisely the systems where the standard calibrations are least reliable.

The Path Forward

The solution is not to abandon hydrogen alpha. It is to stop treating it as a direct readout of star formation and start treating it as what it is: a complex signal that encodes information about star formation, dust, gas physics, and galaxy structure simultaneously.

Tacchella et al. (2022) show that multi wavelength approaches are essential. Combining hydrogen alpha with ultraviolet continuum, far infrared dust emission, and other recombination lines like hydrogen beta and Paschen alpha can break the degeneracies. Each tracer has its own biases. Used together, they can constrain the physical parameters that distort the hydrogen alpha signal.

The simulations also point toward better dust corrections. The Balmer decrement works, but only if you account for the fact that the attenuation law depends on the amount of attenuation. This is a solvable problem. It just requires more careful modeling than the standard approach of applying a single attenuation curve to all galaxies.

What This Actually Means

• ▸The standard conversion from hydrogen alpha luminosity to star formation rate assumes that all ionizing photons from young stars go into producing hydrogen line emission. The simulations show that only about 57 percent do. The rest are absorbed by dust, captured by helium, or escape the galaxy entirely. Every star formation rate derived from hydrogen alpha should be treated as an upper limit, not a measurement.

• ▸Collisional excitation contributes 5 to 10 percent of hydrogen alpha emission in normal galaxies, and potentially much more in galaxies with shocks or active nuclei. If you see hydrogen alpha emission from a region that does not contain young stars, do not assume it is powered by escaping radiation. It might be powered by hot gas.

• ▸Scattering boosts the observed hydrogen alpha luminosity by about 40 percent, and it smears out the spatial correspondence between emission and star formation. High resolution hydrogen alpha images do not show where stars are forming. They show where stars are forming plus a scattered halo of light from elsewhere.

• ▸The Balmer decrement dust correction works within 25 percent on average, but the dust attenuation law varies with the amount of dust. Applying a single correction to all galaxies introduces systematic errors that correlate with dust content. Low dust galaxies are overcorrected. High dust galaxies are undercorrected.

• ▸The most exciting targets for JWST, high redshift galaxies, are also the most uncertain. Until cosmological simulations with full radiative transfer are run for high redshift conditions, every star formation rate derived from hydrogen alpha in the early universe carries unknown systematic errors. The numbers are not wrong. They are just less precise than advertised.