🔭 astrophysics

Full non-LTE multi-level radiative transfer I. An atom with three bound infinitely sharp levels

This paper presents and validates a new numerical scheme for solving the full non-LTE multi-level radiative transfer problem by simultaneously iterating radiative transfer, kinetic equilibrium, and Boltzmann equations to account for deviations in both atomic populations and particle velocity distributions from equilibrium.

Original authors: Tristan Lagache, Frédéric Paletou, Malali Sampoorna

Published 2026-02-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tristan Lagache, Frédéric Paletou, Malali Sampoorna

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Cosmic Traffic Jam

Imagine a star's atmosphere not as a static cloud of gas, but as a massive, chaotic highway filled with two types of travelers:

Atoms (the cars).
Photons (the light beams or headlights).

For decades, astronomers have tried to understand how light travels through this highway to create the spectra (rainbows of light) we see from stars. They used a standard model called Non-LTE (Non-Local Thermodynamic Equilibrium).

The Old Model (The "Average Driver" Assumption):
In the old model, scientists assumed that while the number of cars (atoms) and the brightness of the headlights (light) might be chaotic, the speed of the cars was perfectly predictable. They assumed every atom moved at a "standard" speed, like a crowd of people walking at a perfectly uniform pace. They ignored the fact that some cars might be speeding up, slowing down, or swerving because of collisions or the pressure of the headlights hitting them.

The New Model (Full Non-LTE):
This paper introduces a new, much more realistic way to solve the problem. It's called Full Non-LTE (FNLTE). Instead of assuming everyone walks at the same speed, this new model calculates the exact speed and direction of every single atom as it interacts with the light. It asks: "If a photon hits a fast-moving atom, how does that change the atom's speed? And if that atom changes speed, how does that change the light it emits next?"

It's like upgrading a traffic simulation from "average speed" to a video game where every car has its own unique speed, and the traffic flow changes based on how the cars bump into each other.

The Experiment: The Three-Lane Highway

To test this new, super-complex math, the authors didn't try to simulate the entire universe immediately. They built a miniature model:

The Atom: A simple atom with only three energy levels (like a car with only three gears: Low, Medium, High).
The Condition: They assumed the "gears" were perfectly sharp (no fuzzy edges), making the math slightly easier to start with.

They wanted to see if their new "Full Speed" model could reproduce the results of the old "Average Speed" model when they forced the old rules, and then see what new things happened when they let the atoms move freely.

The Method: A Dance of Iterations

Solving this is incredibly hard because everything depends on everything else. It's a "chicken and egg" problem:

You need to know the atoms' speeds to know what light they emit.
But you need to know the light to know how the atoms' speeds change.

The Solution (The Iterative Dance):
The authors created a computer algorithm that acts like a dance partner who keeps adjusting their steps:

Guess: Start with a guess (using the old, simpler method).
Calculate Light: Based on that guess, calculate the light.
Update Speeds: Based on that light, calculate how the atoms' speeds change.
Update Light Again: Based on the new speeds, recalculate the light.
Repeat: Do this over and over (75 times in their test) until the speeds and the light stop changing and settle into a perfect, self-consistent rhythm.

The Results: Why the Old Model Was "Lying"

When they compared their new "Full Speed" results with the old "Average Speed" approximations, they found something interesting:

1. The "Overshoot" Mystery:
In the old models (specifically a method called XRD), the light intensity at the very surface of the star would sometimes spike to impossible levels (an "overshoot"). It was like a traffic jam suddenly causing cars to fly into the sky.

The Cause: The old model assumed that while atoms emit light based on their weird, changing speeds, they absorb light as if they were all walking at a normal, average pace. This was a contradiction.
The Fix: In the new FNLTE model, the atoms absorb light based on their actual weird speeds. When they did this, the "flying cars" (the overshoot) disappeared. The light profile became smooth and realistic.

2. The Speed Distribution:
The most exciting new result is that they can now see the speed distribution of the atoms.

Deep inside the star (where it's hot and crowded), the atoms behave normally, moving at standard speeds (Maxwellian distribution).
But near the surface (where the star is thin), the atoms start to behave strangely. The "fast" atoms become more common, and the "slow" ones become rare. The "traffic" is no longer uniform; it's skewed.

The Takeaway

Think of this paper as the difference between looking at a crowd from a helicopter (seeing only the average density) versus walking through the crowd with a stopwatch (seeing exactly how fast each person is running and how they bump into each other).

The Old Way: Good enough for a quick glance, but it missed the subtle, chaotic details that happen at the edges of the star.
The New Way: It captures the full chaos. It shows that atoms near the surface of a star aren't just sitting there; they are dancing to a different rhythm than the atoms deep inside.

Why does this matter?
Because stars are the building blocks of the universe. If we want to understand exactly what stars are made of, how old they are, or how they evolve, we need to understand the light they send us. This new method ensures that when we decode that light, we aren't making assumptions about how the atoms are moving. We are actually calculating their movement, leading to a clearer, more accurate picture of the cosmos.

In short: The authors built a super-accurate simulator for starlight that finally accounts for the fact that atoms are messy, fast-moving particles, not just static dots. And it turns out, the old way of looking at them was slightly "off" near the surface of the star.

1. Problem Statement

The standard non-local thermodynamic equilibrium (non-LTE) radiative transfer problem typically accounts for deviations in the radiation field and atomic level populations from their equilibrium distributions. However, it relies on the assumption that the velocity distribution functions (VDFs) of massive particles (atoms) remain Maxwellian.

This paper addresses the Full non-LTE (FNLTE) problem, which requires the self-consistent calculation of:

The radiation field.
The atomic level populations.
The VDFs of the atoms themselves, which may deviate from the Maxwellian distribution due to radiative and collisional processes.

The authors tackle the significant numerical challenge of solving the kinetic equilibrium equations for VDFs simultaneously with the radiative transfer equations (RTE) and population balance equations. As a first step, they model a three-level atom with infinitely sharp energy levels (zero natural broadening) to isolate the effects of velocity distribution deviations without the added complexity of natural line broadening.

2. Methodology

The authors developed a new numerical scheme based on the kinetic theory of particles and photons (Oxenius 1986; Hubený et al. 1983).

Theoretical Framework

Coupled Equations: The solution requires the simultaneous iteration of:
- Radiative Transfer Equations (RTE): For specific intensities of all allowed transitions.
- Integrated Kinetic Equilibrium Equations (IKEE): For atomic level populations ( $n_i$ ).
- Boltzmann Equations: For the velocity distribution functions ( $f_i$ ) of excited levels.
Atomic Profiles: For infinitely sharp levels, the atomic absorption ( $\alpha$ ) and emission ( $\beta$ ) profiles in the atomic frame are Dirac delta functions. The observer's frame profiles are obtained by convolving these with the relevant VDFs.
Scattering Mechanisms: The formalism naturally includes both resonance scattering (same level up and down) and Raman scattering (cross-redistribution, XRD), where a photon is absorbed in one line and re-emitted in another.

Numerical Strategy

The iterative process is initialized using a standard Multi-Level Approximate $\Lambda$ -iteration (MALI) solution assuming Complete Redistribution (CRD). The subsequent steps involve:

Calculating emission and absorption coefficients based on current VDFs.
Solving the RTE to update photon distributions ( $I_{ij}$ ).
Computing partial scattering integrals ( $J_{ij}(u)$ ) and total scattering integrals ( $J_{ij}$ ).
Updating level populations ( $n_i$ ) by solving the IKEE system.
Crucial Step: Updating the VDFs ( $f_i$ ) by solving the Boltzmann equations at every optical depth and velocity point.
Recomputing observer-frame profiles via convolution with the updated VDFs.
Repeating until self-consistency is reached.

3. Key Contributions

First Multi-Level FNLTE Solution: This is the first successful numerical solution of the FNLTE problem for a multi-level atom (specifically a three-level system). Previous work was limited to two-level atoms.
Self-Consistent VDFs: The method explicitly solves for the deviation of atomic velocities from Maxwellian distributions, rather than assuming them.
Rigorous Treatment of Cross-Redistribution (XRD): The paper provides a complete derivation of emission and absorption profiles that account for Raman scattering without the simplifying assumptions used in previous "approximate" XRD models.
Validation Framework: The authors established a rigorous validation protocol by comparing their FNLTE results against:
- Two-level PRD (Partial Redistribution) solutions (Hummer 1962).
- Standard Multi-level CRD (Complete Redistribution) solutions.
- Approximate XRD models (Sampoorna et al. 2013).

4. Results

The authors applied their code to an isothermal, plane-parallel hydrogen atmosphere model (based on Avrett 1968 parameters).

Validation:
- The code accurately reproduced standard PRD and CRD benchmarks with relative errors $< 0.5\%$ .
- When compared to the approximate XRD method, the FNLTE results showed small discrepancies ( $\sim 0.2-0.4\%$ ) in the source functions, confirming the numerical stability of the new scheme.
New Physical Insights:
- Source Function Anomalies: The approximate XRD method produced a non-physical "overshoot" (source function $> 1$ ) in the $2 \leftrightarrow 3$ transition near the stellar surface at high frequencies.
- Resolution of Overshoot: The full FNLTE approach eliminated this overshoot. The authors identified that the XRD overshoot was caused by a contradiction in the approximate method: it used the non-Maxwellian VDF ( $f_2$ ) to calculate the emission profile but incorrectly assumed a Maxwellian VDF to calculate the absorption profile. The FNLTE method self-consistently calculates both using the actual $f_2$ .
- Velocity Distribution Deviations: The study revealed that while the VDF for the first excited level ( $f_2$ ) remains close to Maxwellian (deviating by only $\sim 0.08\%$ ), the VDF for the second excited level ( $f_3$ ) shows significant deviations from Maxwellian, particularly at high velocities and near the surface. This deviation is directly linked to the non-LTE source function behavior.

5. Significance

Theoretical Advancement: This work moves beyond the "semi-classical" approximations of previous radiative transfer models by fully coupling the kinetic equations of matter and radiation. It demonstrates that assuming Maxwellian velocity distributions can lead to significant errors in the formation of spectral lines, particularly for higher excited states and in the outer layers of stellar atmospheres.
Diagnostic Power: By providing a method to calculate non-Maxwellian VDFs, this approach opens the door to more accurate diagnostics of stellar atmospheres, where velocity distributions are driven by the radiation field itself.
Future Directions: The paper sets the stage for more realistic models including natural line broadening, non-Maxwellian free electrons, and stimulated emission. The authors note that while computationally expensive, the development of efficient operator methods (similar to MALI) for the full FNLTE problem is the next logical step.

In conclusion, this paper establishes the foundational numerical framework for Full non-LTE multi-level radiative transfer, proving that self-consistent treatment of atomic velocity distributions is necessary to resolve inconsistencies in previous approximate models and to accurately describe the physics of spectral line formation in non-equilibrium environments.