Two-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars

arXiv:2604.07317 · astro-ph.SR · Submitted 2026-04-08 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: We're looking at the paper: "Multi-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars".

Jocelyn: Here's the gist of the paper. Multi-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars presents an approximate NLTE procedure that accounts for scattering in computing energy and Planck mean opacities from a multitude of spectral lines to better capture heating and cooling in the outflowing parts of time-dependent radiation-hydrodynamical simulations The net result is a multi-component wind structure not only in density and velocity, but also in temperature The Need for an Approximate NLTE Formalism Previous multi-dimensional radiationhydrodynamics (RHD) simulations of unified atmospheres and winds of Wolf-Rayet (WR) stars, O-type stars, and luminous blue variable (LBV) stars assumed that the energy- and Planck-mean opacities are equal to the flux-mean opacity This assumption neglects important scattering effects by effectively forcing LTE when computing the thermal exchange between gas and radiation To improve this, a fast but physically motivated NLTE treatment for calculating line opacities in the outflowing parts of multi-D atmosphere and wind simulations is necessary The authors developed an approximate NLTE (aNLTE, henceforth) formalism that accounts for scattering when computing the energy- and Planck-mean line opacities in the outflowing parts of the atmosphere This formalism involves introducing opacity multipliers for a grid of density, gas, and radiation temperature, and a dilution factor for different Sobolev-like optical depths These parameters are then tabulated for various densities, gas and radiation temperatures, and dilution factors to be interpolated during the RHD simulations to obtain the line opacity parameters based on the local conditions within the grid Modeling Line Heating and Cooling The net heating and cooling rate is given by a formula derived from considering interactions with spectral lines in an accelerating medium For a two-level model atom, the line source function can be written as (e.g., Hubeny & Mihalas 2015, page 458) S l = epsilon lBl(Tgas) + (1 − epsilon l)J¯ϕ The line heating and cooling rates are then given by q˙ = 4παlϵl(1 − χl)WB˜l(Trad) − Bl(Tgas) This indicates that the line heating and cooling depend on the temperature imbalance between the diluted stellar radiation field and the local gas temperature, which is modulated by the efficiency of photon thermalization within the resonance zone Results of Shock Heating in Outflows The simulated structures originating from the iron-opacity bump below the stellar atmosphere at ∼200 kK propagate outwards and give rise to a structured line-driven outflow High-density structures in the wind move much more slowly than their low-density counterparts Due to their high velocity dispersion, upon interaction, they produce localized shock fronts with the gas temperature exceeding the photon temperature For a typical O-type supergiant star with a relatively dense wind, about 35% of the gas in the outer wind parts of the simulation has temperatures exceeding the local frequency-integrated radiation temperature, with 4% heated to above the effective temperature Comparison with Stationary Equilibrium The comparison between time-dependent simulations and simplified radiative equilibrium models shows that while plane-parallel approximations fail at larger radii, including spherical correction terms yields good agreement with the median values of the 1D simulation The resulting gas (Tgas,sc) and radiation (Trad,sc) temperatures show good agreement with the median values of the 1D simulation at larger distances Observational Implications The added EUV and X-ray emissions are generally attributed to embedded wind shocks that locally heat the plasma to high temperatures In practice, spectral modeling often adjusts shock strengths and volume filling factors so that the resulting emissivities yield approximately the canonical scaling value LX ∼ 10−7Lbol Limitations and Future Improvements A natural extension would be to include bound-free and free-free continuum opacities and to generalize the radiation field to a number of frequency bins Work is currently underway in our group to replace the fitting routine with a machine-learning approach that predicts the line opacity multiplier directly, thereby avoiding fit-functions and associated inconsistencies like these In our simulations, extensive structures in the wind arise even with Sobolev-based line-driving because of the turbulent wind-launching region stemming from instabilities associated with the iron-bump However, it remains both methodologically and computationally very challenging to include the LDI in a multi-D model covering also the sub-surface iron opacity bump region In this work, we have developed and implemented a formalism to treat aNLTE effects and scattering in a multi-D, time-dependent, RHD unified atmosphere and wind model of a hot, massive O-type star Our new treatment of the energy balance here leads to local pockets in the wind where the gas temperature significantly exceeds the local radiation temperature <ref:2604.07317#pg13> The simulations presented here thus represent first steps toward self-consistent multi-dimensional O-type star wind models with realistic multi-component density, velocity, and temperature structures <ref:2604.07317#pg13> Acknowledgements The computational resources used for this work were provided by Vlaams Supercomputer Centrum (VSC), funded by the Research Foundation-Flanders (FWO) and the Flemish Government DD, JS, NM, CVdS acknowledge support from the European Research Council (ERC) under the Horizon Europe grant agreement 101044048, from the Belgian Research Foundation Flanders (FWO) Odysseus program under grant number G0H9218N, and from KU Leuven C1 grant BRAVE C16/23/009 Data analysis and visualization in this project relied on Numpy (Harris et al.

Vera: 2020), Scipy (Virtanen et al.

Jocelyn: 2020), and Matplotlib (Hunter 2007) References The references listed support the methodology and context of the study, including works by Hubeny & Mihalas, Puls et al., and Sundqvist et al.

Vera: The paper relies on a large atomic database consisting of more than four million spectral lines from hydrogen up to zinc The fitting formula for the line opacity multiplier is based on the distribution function concepts of Gayley (1995) and Poniatowski et al.

Paper summary: Vera: So, to wrap up, the paper "Multi-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars" develops and implements a formalism to treat approximate NLTE effects and scattering in a multi-dimensional, time-dependent radiation hydrodynamics unified atmosphere and wind model of a hot, massive O-type star.

Jocelyn: The main result they show is that their new treatment of the energy balance leads to local pockets in the wind where the gas temperature significantly exceeds the local radiation temperature.

Subrahmanyan: These simulations represent first steps toward self-consistent multi-dimensional O-type star wind models with realistic multi-component density, velocity, and temperature structures.

Vera: So, what does this mean for us observing these stars? It means we can start thinking about the observed EUV and X-ray emissions being caused by those embedded wind shocks that locally heat the plasma to high temperatures.

Jocelyn: In practice, spectral modeling often has to adjust shock strengths and volume filling factors so that the resulting emissivities yield approximately the canonical scaling value of LX equals ten to the negative seven Lbol <ref:2604.07317#pg1>.

Subrahmanyan: The authors also pointed out some limitations for future work. They mentioned that a natural extension would be to include bound-free and free-free continuum opacities and to generalize the radiation field to a number of frequency bins.

Vera: And they also noted that it remains both methodologically and computationally very challenging to include the line driving effect in a multi-dimensional model covering also the sub-surface iron opacity bump region.

Jocelyn: It sounds like they’ve done some really important work moving from simpler LTE assumptions to a more accurate, scattering-aware treatment for these complex stellar winds.

Conclusion: Vera: So, we're looking at this paper titled "Multi-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars." Basically, they took a complex problem of modeling how hot stars lose mass and figured out a way to do it in three dimensions over time.

Jocelyn: Yeah, what that means is they've built a model that tracks the density and velocity of the wind as it moves outward, not just one snapshot in time.

Subrahmanyan: Exactly. They developed this approximate NLTE method to better handle how light scatters within the outflowing gas, which is a big hurdle when you try to get the physics right in these winds.

Vera: And their main finding is that this new approach shows local pockets where the gas temperature gets much hotter than the radiation temperature.

Jocelyn: That's what I mean by it causing those intense EUV and X-ray emissions we see from these stars; it’s not a smooth transition anymore.

Subrahmanyan: From a theoretical side, this is a step toward making these self-consistent models for massive O-type stars, where the density and temperature structures are all linked together dynamically.

Vera: But they also laid out some limitations, saying that their current method still struggles to fully include things like the iron opacity bump in that dense region near the star.

Jocelyn: So, while they've made a huge step forward in capturing the wind structure, there's still a lot of work left to incorporate those smaller but important physical details.

Subrahmanyan: That leads us to thinking about what observations would actually tell us if these dynamic structures are real and how they fit into the larger picture of stellar evolution.

Institute of Astronomy, KU Leuven · Centre for mathematical Plasma Astrophysics, Department of Mathematics, KU Leuven · Royal Observatory of Belgium · Zentrum für Astronomie der Universität Heidelberg · Universität Heidelberg, Interdisziplinäres Zentrum für Wissenschaftliches Rechnen

astro-ph.SR

Submitted: 2026-04-08

Updated: 2026-11-15

Journal ref: A&A 714, A418 (2026)

DOI: 10.1051/0004-6361/202660303

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 80/100

The gist: Multi-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars presents an approximate NLTE procedure that accounts for scattering in computing energy…

Terminology

Summary

Multi-dimensional, time-dependent approximate NLTE unified model atmospheres with winds for hot, massive stars presents an approximate NLTE procedure that accounts for scattering in computing energy and Planck mean opacities from a multitude of spectral lines to better capture heating and cooling in the outflowing parts of time-dependent radiation-hydrodynamical simulations The net result is a multi-component wind structure not only in density and velocity, but also in temperature

The Need for an Approximate NLTE Formalism

Previous multi-dimensional radiationhydrodynamics (RHD) simulations of unified atmospheres and winds of Wolf-Rayet (WR) stars, O-type stars, and luminous blue variable (LBV) stars assumed that the energy- and Planck-mean opacities are equal to the flux-mean opacity This assumption neglects important scattering effects by effectively forcing LTE when computing the thermal exchange between gas and radiation To improve this, a fast but physically motivated NLTE treatment for calculating line opacities in the outflowing parts of multi-D atmosphere and wind simulations is necessary The authors developed an approximate NLTE (aNLTE, henceforth) formalism that accounts for scattering when computing the energy- and Planck-mean line opacities in the outflowing parts of the atmosphere This formalism involves introducing opacity multipliers for a grid of density, gas, and radiation temperature, and a dilution factor for different Sobolev-like optical depths These parameters are then tabulated for various densities, gas and radiation temperatures, and dilution factors to be interpolated during the RHD simulations to obtain the line opacity parameters based on the local conditions within the grid

Modeling Line Heating and Cooling

The net heating and cooling rate is given by a formula derived from considering interactions with spectral lines in an accelerating medium For a two-level model atom, the line source function can be written as (e.g., Hubeny & Mihalas 2015, page 458) S l = epsilon lBl(Tgas) + (1 − epsilon l)J¯ϕ The line heating and cooling rates are then given by q˙ = 4παlϵl(1 − χl)WB˜l(Trad) − Bl(Tgas) This indicates that the line heating and cooling depend on the temperature imbalance between the diluted stellar radiation field and the local gas temperature, which is modulated by the efficiency of photon thermalization within the resonance zone

Results of Shock Heating in Outflows

The simulated structures originating from the iron-opacity bump below the stellar atmosphere at ∼200 kK propagate outwards and give rise to a structured line-driven outflow High-density structures in the wind move much more slowly than their low-density counterparts Due to their high velocity dispersion, upon interaction, they produce localized shock fronts with the gas temperature exceeding the photon temperature For a typical O-type supergiant star with a relatively dense wind, about 35% of the gas in the outer wind parts of the simulation has temperatures exceeding the local frequency-integrated radiation temperature, with 4% heated to above the effective temperature

Comparison with Stationary Equilibrium

The comparison between time-dependent simulations and simplified radiative equilibrium models shows that while plane-parallel approximations fail at larger radii, including spherical correction terms yields good agreement with the median values of the 1D simulation The resulting gas (Tgas,sc) and radiation (Trad,sc) temperatures show good agreement with the median values of the 1D simulation at larger distances

Observational Implications

The added EUV and X-ray emissions are generally attributed to embedded wind shocks that locally heat the plasma to high temperatures In practice, spectral modeling often adjusts shock strengths and volume filling factors so that the resulting emissivities yield approximately the canonical scaling value LX ∼ 10−7Lbol

Limitations and Future Improvements

A natural extension would be to include bound-free and free-free continuum opacities and to generalize the radiation field to a number of frequency bins Work is currently underway in our group to replace the fitting routine with a machine-learning approach that predicts the line opacity multiplier directly, thereby avoiding fit-functions and associated inconsistencies like these In our simulations, extensive structures in the wind arise even with Sobolev-based line-driving because of the turbulent wind-launching region stemming from instabilities associated with the iron-bump However, it remains both methodologically and computationally very challenging to include the LDI in a multi-D model covering also the sub-surface iron opacity bump region

Summary and Conclusion

In this work, we have developed and implemented a formalism to treat aNLTE effects and scattering in a multi-D, time-dependent, RHD unified atmosphere and wind model of a hot, massive O-type star Our new treatment of the energy balance here leads to local pockets in the wind where the gas temperature significantly exceeds the local radiation temperature <ref:2604.07317#pg13> The simulations presented here thus represent first steps toward self-consistent multi-dimensional O-type star wind models with realistic multi-component density, velocity, and temperature structures <ref:2604. p. 0

Where the paper says it

The net result is a multi-component wind structure not only in density and velocity, but also in temperature. This likely has important consequences for the formation and interpretation of observed O-type star wind spectra.

07317#pg13>

Acknowledgements

The computational resources used for this work were provided by Vlaams Supercomputer Centrum (VSC), funded by the Research Foundation-Flanders (FWO) and the Flemish Government DD, JS, NM, CVdS acknowledge support from the European Research Council (ERC) under the Horizon Europe grant agreement 101044048, from the Belgian Research Foundation Flanders (FWO) Odysseus program under grant number G0H9218N, and from KU Leuven C1 grant BRAVE C16/23/009 Data analysis and visualization in this project relied on Numpy (Harris et al. 2020), Scipy (Virtanen et al. p. 1

Where the paper says it

In state-of-the-art 1D model atmosphere codes such as CMFGEN (Hillier & Miller 1998), PoWR (Hamann & Gräfener 2004), and FASTWIND (Puls et al. 2005, 2020), the radiative transfer and the statistical equilibrium conditions are solved in full non-local thermodynamic equilibrium (NLTE).

2020), and Matplotlib (Hunter 2007)

References

The references listed support the methodology and context of the study, including works by Hubeny & Mihalas, Puls et al., and Sundqvist et al. p. 1

Where the paper says it

The atmospheric model is then iterated toward convergence by coupling to the NLTE equations, assuming radiative equilibrium1 , possibly also involving an iterative solution of the stationary equation of motion (Sander et al. 2017; Sundqvist et al.

The paper relies on a large atomic database consisting of more than four million spectral lines from hydrogen up to zinc The fitting formula for the line opacity multiplier is based on the distribution function concepts of Gayley (1995) and Poniatowski et al. p. 3
Where the paper says it

The solid lines provide the fits to the opacity multiplier using the fitting formalism introduced by Gayley (1995). Calculation and tabulation of line mean opacities To compute the Sobolev-based line-mean opacities, we use the Munich Atomic Database (Pauldrach et al.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI can perform self-consistent multi-dimensional O-type star wind models by separating flux-, energy-, and Planck-mean opacities to better describe radiative acceleration and net heating and cooling of gas in optically thin outflowing regions.

  2. The improved system can derive an aNLTE mean opacity description accounting for scattering in a medium with a significant velocity gradient by precomputing line opacity multipliers for a grid of density, radiation temperature, gas-to-radiation temperature ratio, and dilution factor.

  3. The AI can use the derived line opacity parameters to compute on-the-fly the line contributions to kappa E, kappa B, and kappa F while running multi-D O-type star models using MPIAMRVAC.

  4. The system can generate realistic multi-component density, velocity, and temperature structures in O-type stars by allowing gas temperatures to exceed radiation temperatures at shock fronts due to the relaxation of the equality kappa E = kappa B = kappa F.

  5. The AI can better predict the occurrence of high-ionization wind lines and general complexity observed in O-type star spectra by simulating shock heating that is physically motivated, rather than relying on ad hoc fitting parameters.

Abstract

Multi-dimensional unified model atmospheres with winds of massive stars have so far been studied under the assumption of equal flux, Planck, and energy weighted mean opacities, which effectively means these models have been in local thermodynamic equilibrium (LTE). Although LTE may be a valid approximation in deeper atmospheric layers, it breaks down in the extended outflowing parts. As such, the opacities governing the heating and cooling of the gas are neither the same nor equal to flux-mean opacity in those regions. We present an approximate NLTE procedure that accounts for scattering in the computation of energy and Planck-mean opacity from a multitude of spectral lines in an accelerating medium. The formalism evaluates the opacities using Sobolev escape probabilities and effective thermalization parameters from a line database consisting of 4 million spectral lines. RHD simulations are calculated as before with a hybrid opacity scheme combining Rosseland means with line opacities in an accelerating medium. Due to their high velocity dispersion, upon interaction, they produce localized shock fronts with the gas temperature exceeding the photon temperature. Due to improved treatment of heating and cooling in outflowing parts, the radiation and gas temperatures in the wind are no longer the same, as was the case in previous multi-dimensional simulations. Instead, gas gets heated at shock fronts, but due to strong radiative cooling remains localized. The net result is a multi-component wind structure not only in density and velocity, but also in temperature. This likely has important consequences for the formation and interpretation of observed O-type star wind spectra.

Sources

Related papers