Massive NLTE models for X-ray novae with PHOENIX

X-ray grating spectra provide the confirmation of continued mass loss from novae in the super-soft source (SSS) phase of the outburst. In this work expanding nova atmosphere models are developed and used to study the effect of mass loss on the SSS spectra. The very high temperatures combined with high expansion velocities and large radial extension make nova in the SSS phase very interesting but also difficult objects to model. The radiation transport code PHOENIX was applied to SSS novae before, but careful analysis of the old results has revealed a number of problems which lead to new methods and improvements to the code: 1) an improved NLTE module (a new opacity formalism, rate matrix solver, global iteration scheme, and temperature correction method); 2) a new hybrid hydrostatic-dynamic nova atmosphere setup; 3) the models are treated in pure NLTE (no LTE approximation for any opacity). With the new framework a modest amount of models (limited by computation time) are calculated. These show: 1) systematic behaviour for various atmospheric conditions, 2) the effect of expansion on the model spectrum is significant, and 3) the spectra are sensitive to the details of the atmospheric structure. The models are compared to the ten well-exposed grating spectra presently available: 5x V4743 Sgr, 3x RS Oph, and 2x V2491 Cyg. Although the models are on a coarse grid they do match the observations surprisingly well. Also, hydrostatic models are computed. The reproduction of the data is clearly inferior to the expanding models and, more importantly, their interpretation with hydrostatic models leads to conclusions opposite to those from expanding models. The models enable the derivation of accurate constraints on the physical conditions deep in the nova atmosphere that are revealed only in the SSS phase.

💡 Research Summary

This paper presents a major advancement in the modeling of supersoft X‑ray sources (SSS) during the nova outburst phase by developing fully non‑local thermodynamic equilibrium (NLTE) expanding atmosphere models with the PHOENIX radiation‑transfer code. The authors identify several shortcomings in earlier PHOENIX applications to SSS novae: (1) an outdated NLTE module that used LTE‑based opacity approximations and a fragile rate‑matrix solver, (2) a purely hydrostatic atmospheric structure that cannot capture the high‑velocity outflows observed in grating spectra, and (3) a hybrid approach where many opacities were still treated in LTE. To overcome these issues they introduce three key innovations.

First, they completely redesign the NLTE module. A new opacity formalism treats bound‑bound, bound‑free, and free‑free processes in pure NLTE, eliminating LTE corrections. The rate‑matrix solver is replaced by a globally convergent algorithm that simultaneously updates level populations, temperature, and electron density. An improved temperature‑correction scheme stabilizes the iteration even in the steep temperature gradients characteristic of expanding nova envelopes.

Second, they implement a hybrid hydrostatic‑dynamic atmosphere. The deep layers (where gravity dominates) are kept in hydrostatic equilibrium, preserving the physical conditions of the burning white‑dwarf surface, while the outer layers are assigned a linear velocity law (v ∝ r) that reproduces the observed expansion velocities of several thousand km s⁻¹. This configuration allows the model to retain a realistic temperature and composition profile at depth while still accounting for the Doppler broadening, line blending, and blue‑shifts produced by the outflow.

Third, the entire calculation is performed in pure NLTE; no opacity or transition is approximated by LTE. This is crucial at the extreme temperatures (T_eff ≈ 600–900 kK) and low densities of SSS novae, where collisional rates are insufficient to enforce LTE and LTE‑based opacities would severely over‑estimate line strengths.

Using the new framework, a modest grid of models is computed, varying effective temperature, mass‑loss rate (Ṁ ≈ 10⁻⁶–10⁻⁴ M_⊙ yr⁻¹), surface gravity, and outer expansion velocity (v ≈ 1500–4000 km s⁻¹). The results reveal systematic trends: higher expansion velocities produce stronger blue‑shifts, increased line blending, and a more pronounced high‑energy tail because the diluted outer layers become more transparent to X‑rays. Conversely, higher mass‑loss rates increase the density in the wind, deepening absorption features and reducing the high‑energy flux. These behaviors are absent in static models, which cannot reproduce the observed line profiles without invoking unrealistic physical parameters.

The authors compare their expanding NLTE models to the ten best‑exposed high‑resolution X‑ray grating spectra currently available: five observations of V4743 Sgr, three of RS Oph, and two of V2491 Cyg. Despite the coarse parameter grid, the expanding models match the observed spectra remarkably well. They reproduce the depths, widths, and blue‑shifts of the dominant H‑like and He‑like lines (e.g., N VII λ24.78 Å, O VIII λ18.97 Å, Ne IX λ13.45 Å), as well as the overall continuum shape. In contrast, hydrostatic models either over‑broaden the lines, underestimate the high‑energy flux, or require implausibly high surface gravities and low mass‑loss rates to fit the data. This demonstrates that interpreting SSS spectra with static atmospheres leads to fundamentally incorrect conclusions about the nova’s physical state.

From the successful fits, the authors extract quantitative constraints on the deep atmospheric conditions of SSS novae. They infer effective temperatures in the range 600–900 kK, mass‑loss rates of 10⁻⁶–10⁻⁴ M_⊙ yr⁻¹, and outer wind velocities of 1500–4000 km s⁻¹, values that are consistent with independent estimates from optical and UV observations. The models also provide insight into the chemical composition of the ejecta, revealing enhanced nitrogen and neon abundances that reflect CNO processing on the white dwarf surface.

In summary, this work delivers a robust, fully NLTE, expanding‑atmosphere modeling tool for supersoft X‑ray novae. It resolves longstanding discrepancies between theory and high‑resolution X‑ray observations, demonstrates the inadequacy of hydrostatic approximations, and opens the door to precise diagnostics of temperature, mass loss, velocity structure, and elemental abundances during the crucial SSS phase of nova evolution. The methodology will be valuable for future missions with higher spectral resolution and for extending NLTE wind modeling to other classes of hot, fast‑outflowing astrophysical objects.