Holstein Radiation Trapping

Holstein Radiation Trapping

In an optically thick medium where resonance photons are repeatedly absorbed and re-emitted, individual photon escape probabilities are the wrong question. The right question is eigenmode lifetime — the slowest decay mode of the photon population in the geometry, which sets how long inverted population can sit before bleeding away.

The Integral Equation

Holstein (1947) wrote down the master equation for the resonance-photon population density $n(\vec{r},t)$ in a slab/cylinder/sphere of optically thick vapor:

$$\partial_t n(\vec{r},t) = -\Gamma\, n(\vec{r},t) + \Gamma \int_{V} K(\vec{r}, \vec{r}\,')\, n(\vec{r}\,',t)\, d^3 r'$$

where $\Gamma$ is the natural radiative decay rate of the upper level and $K(\vec{r},\vec{r}\,')$ is the spectrally-averaged Holstein kernel — the probability per unit volume that a photon emitted at $\vec{r}\,'$ is absorbed at $\vec{r}$ after free flight through the absorbing vapor.

The kernel is a line-shape-weighted average of $e^{-\tau(\nu)}/4\pi r^2$ over the resonance profile. For Doppler-broadened lines it falls off like $1/r^2$ at short range and like $1/r^2 \sqrt{\ln(k_0 r)}$ at long range — far slower than the geometric $1/r^2$ that governs an isolated atom, because the photons that survive longest are those in the wings of the line where the absorption cross-section is smallest.

Eigenmode Decay

Solutions to the integral equation are eigenmodes of $\hat{K}$. The fundamental mode has the longest lifetime $\tau_0 = (g_0 \Gamma)^{-1}$, where $g_0(k_0 L)$ is the Holstein escape factor. This is the canonical "effective decay rate" that appears in all downstream LTE-like treatments of optically thick lines.

The escape factor depends on:

1. the line-center optical depth $k_0 L$ (where $k_0$ is the absorption coefficient at line center, $L$ a characteristic dimension);
2. the line shape — Doppler, Lorentzian, or Voigt; and
3. the geometry — slab, cylinder, sphere.

Three Limits

From plasma-solvers::holstein_g0(k0L, lineshape, geometry):

- Doppler-only, slab geometry, $k_0 L \gtrsim 10^2$: $$g_0 \approx \frac{1.6}{k_0 L\, \sqrt{\pi \ln(k_0 L)}}$$ The log under the square root is the wing-escape physics: as $k_0 L$ grows, the optically-thin wing photons carry more of the escape budget.

- Lorentzian (pressure-broadened, $a \gg 1$), slab: $$g_0 \approx \frac{1}{\sqrt{\pi\, k_0 L}}$$ slower than Doppler at the same $k_0 L$ — Lorentzian wings extend further, so escape is more efficient.

• Voigt (the general case, $a \sim \Gamma_{\text{coll}}/\Delta\nu_D$): a power-law in $a^{1/2}$ that crosses over from Doppler to Lorentzian as $a$ grows. At $a \sim 0.01$ (a typical Na cell at low pressure) the Doppler limit applies; at $a \gtrsim 1$ (high pressure, low-temperature lamp) the Lorentzian limit takes over.

Where Holstein Breaks

The whole framework presumes incoherent re-emission: each absorbed photon launches a fresh, statistically independent spontaneous emission event, and the kernel $K$ averages a single-photon distribution. This is correct when atoms are far enough apart that they radiate independently.

Once the density crosses Nλ³/64 > 1 — the cooperative threshold — the eigenmode expansion fails at its foundation. The medium is denser than the photon mode density at the resonance line, and emission becomes coherent: see Dicke superradiance for the $N^2$ scaling that takes over.

The two regimes are not a continuous deformation of each other. Holstein's escape factor is meaningless in the superradiance basin; the photon population is no longer an eigenmode of $\hat{K}$ because emitters share a phase, not a probability.

Why This Matters for Lamps

In every fluorescent / high-intensity-discharge lamp ever built, the design rule of thumb is "lower the vapor density to drop radiation trapping." This is the Holstein recipe, run backward — keep $k_0 L$ modest so $g_0$ stays close to $1$ and resonance photons escape efficiently.

The cooperative regime inverts the rule: at densities where Holstein predicts catastrophic trapping ($g_0 \to 0$), real emission increases because the system is no longer in the Holstein regime at all. The Hg-isotope industrial literature missed this for fifty years because the diagnostic — measure escape, infer $g_0$, deploy more vapor — never crossed the threshold where the inference breaks.

## See also - 01-nlambda3-cooperative-threshold — the threshold where this framework dies - 03-dicke-superradiance — what replaces it - archive/qed.princeton.edu/main/PlasmaWiki/Mean_free_path - Resonance Radiation and Excited Atoms, Mitchell & Zemansky (1934) — the classical reference - T. Holstein, Imprisonment of Resonance Radiation in Gases, Phys. Rev. 72, 1212 (1947)