Reading Cepheid chemistry through a star’s pulsating atmosphere

A changing atmosphere, a steadier chemical signal

Classical Cepheids help astronomers measure cosmic distances because their pulsation periods are closely related to their intrinsic brightness. Their chemical composition also affects that relationship, making accurate abundance measurements important for the distance scale. As young stars, Cepheids additionally trace the relatively recent chemical enrichment of the Milky Way.

But reading their chemistry is difficult. As a Cepheid pulsates, its temperature, surface gravity and spectral absorption lines change. An analysis must distinguish those atmospheric changes from differences in the amounts of particular elements.

R. Albarracín and colleagues present a method designed for the high-resolution component of the 4MOST 4MIDABLE-HR survey. Their archival tests suggest that a more detailed treatment of how atoms interact with radiation can reduce artificial changes in inferred composition across a pulsation cycle.

Letting radiation influence the atoms

A common simplifying assumption in stellar spectroscopy is local thermodynamic equilibrium, or LTE. It describes atomic populations using local thermal conditions. Cepheids have warm, low-density, extended atmospheres, where radiation can drive departures from that assumption and alter the absorption lines used to measure abundances.

The researchers instead include non-LTE, or NLTE, line formation for 15 elements, including iron, oxygen and several heavier elements. Carbon and chromium remain treated in LTE.

Their tool uses the Payne neural-network approach, trained on nearly 450,000 synthetic spectra. The network approximates model spectra for different atmospheric conditions and chemical compositions; the pipeline then searches for models that fit the observed absorption lines. Its training therefore comes from spectral calculations, rather than from an unconstrained attempt to infer chemistry directly from observations.

For validation, the team used archival observations from UVES, HARPS and FEROS. They reduced the spectra to a common resolving power of about 20,000, approximating 4MOST-HR. These are tests on existing observations, not results from the forthcoming survey.

The final catalog contains 975 spectra of 415 variable stars, including 351 classical Cepheids. Other targets include Type II Cepheids, anomalous Cepheids and RR Lyrae stars.

Testing the method against stellar pulses

Twenty-three classical Cepheids had at least five accepted spectra, allowing the researchers to compare abundance estimates at different pulsation phases.

For the magnesium, titanium, calcium, silicon and iron measurements examined in their LTE comparison, the NLTE analysis reduced phase-to-phase dispersion by an average of about 0.06 dex. Dex denotes a base-10 logarithmic difference in abundance.

Across the multiphase sample, the mean per-star dispersion in iron abundance was 0.07 dex. Many elemental abundance ratios showed similarly small variations, although cobalt, barium and strontium were less consistent. This matters operationally: 4MOST observations cannot feasibly be scheduled to catch every Cepheid at a particular phase.

The team also checked oxygen measurements using different spectral features. For a selected set of 94 FEROS spectra from 54 Cepheids, oxygen estimates from weaker lines available to 4MOST agreed with estimates from the well-studied 777-nanometer triplet, with a scatter of 0.08 dex. Most of those comparison stars lie near the Sun’s Galactic radius, limiting the range tested.

Mapping chemistry across the disk

Using 335 classical Cepheids with the necessary spectroscopic and photometric information, the researchers traced abundance gradients between roughly 4 and 21 kiloparsecs from the Galactic center.

The overall linear iron gradient was −0.051 ± 0.002 dex per kiloparsec: stars farther outward generally contained less iron relative to hydrogen. Fits allowing a change in slope favored a transition near 15 kiloparsecs, consistent with earlier evidence that the outer-disk gradient becomes shallower.

The exact outer slopes remain poorly constrained. Some individual-element fits have unstable results because of smaller samples and limited spatial coverage. The study also reports the first NLTE radial gradients for cobalt, nickel, yttrium and europium in Cepheids.

An enhanced barium abundance persists even after the NLTE treatment. Its origin remains unresolved; the analysis does not establish a particular additional production mechanism.

Preparation, not a new cosmic expansion measurement

The method still uses one-dimensional, hydrostatic model atmospheres, despite Cepheids’ dynamic behavior. Training coverage is thinner in some low-gravity conditions, and uncertainty tests for the survey’s continuum-normalization procedure remain future work.

The study does not measure a new Hubble constant or recalibrate the Cepheid luminosity relation. Its contribution is a tested chemical-analysis method and a Galactic abundance catalog—work intended to support those later applications without confusing a star’s changing atmosphere with changing chemistry.