[Paper Review] Monte Carlo Simulation of Laser Diodes Sub-Poissonian Light Generation

Laurent Chusseau, Jacques Arnaud|HAL (Le Centre pour la Communication Scientifique Directe)|May 17, 2001

Semiconductor Quantum Structures and Devices13 references3 citations

TL;DR

This paper presents a Monte Carlo simulation of laser diodes that tracks individual electron and hole level occupancies and photon counts in the optical cavity, enabling accurate modeling of sub-Poissonian light generation. The key finding is that poor thermal contact at high power levels narrows the frequency band where photo-current spectral density remains below shot noise, revealing new noise effects like spectral-hole burning and gain fluctuations.

ABSTRACT

When laser diodes are driven by high-impedance electrical sources the variance of the number of photo-detection events counted over large time durations is less than the average number of events (sub-Poissonian light). The paper presents a Monte Carlo simulation that keeps track of each level occupancy (0 or 1) in the conduction and valence bands, and of the number of light quanta in the optical cavity. When there is good electron-lattice thermal contact the electron and hole temperatures remain equal to that of the lattice. In that case, elementary laser-diode noise theory results are accurately reproduced by the simulation. But when the thermal contact is poor (or, almost equivalently, at high power levels) new effects occur (spectral-hole burning, temperature fluctuations, statistical fluctuations of the optical gain) that are difficult to handle theoretically. Our numerical simulation shows that the frequency domain over which the photo-current spectral density is below the shot-noise level becomes narrower as the optical power increases.

Motivation & Objective

• To model sub-Poissonian light generation in laser diodes using a microscopic, stochastic approach.
• To investigate the impact of thermal contact quality on photodetection noise beyond standard linearized theories.
• To simulate individual quantum transitions and photon statistics in mesoscopic laser structures with few quantum dots.
• To validate analytical models of photo-current spectral density under varying pumping and temperature conditions.
• To explore the emergence of non-equilibrium effects such as spectral-hole burning and gain fluctuations in high-power laser diodes.

Proposed method

• A discrete-event Monte Carlo simulation tracks the time evolution of electron and hole occupancies in conduction and valence bands, with each level either occupied (1) or unoccupied (0).
• The simulation models all microscopic processes: electron-hole recombination, stimulated emission, absorption, Auger transitions, and thermalization with the lattice.
• Photon number in the optical cavity is updated based on stimulated emission and loss rates, with detection modeled via Poissonian counting of photo-events.
• The system uses a simplified model with evenly spaced energy levels (spacing ε) and assumes Fermi-Dirac statistics for level occupancies under thermal equilibrium.
• Noise spectral density is computed from the power spectrum of the detected photo-current, with analytical results derived from linearized birth-death processes.
• The model incorporates a normalized pumping rate $ J^* = J heta / k_B T $, and the spectral density depends on the loss parameter $ \alpha $ and relaxation frequency $ f_r $.

Experimental results

Research questions

• RQ1 How does poor thermal contact affect the frequency bandwidth over which photo-current spectral density remains below the shot-noise level in laser diodes?
• RQ2 To what extent do spectral-hole burning and optical gain fluctuations emerge in high-power laser diodes under non-equilibrium conditions?
• RQ3 Can a Monte Carlo simulation accurately reproduce standard laser noise theory when thermal contact is good and temperature equilibration is maintained?
• RQ4 How do individual level occupancies and photon statistics influence the emergence of sub-Poissonian light in mesoscopic semiconductor lasers?
• RQ5 What are the limitations of linearized analytical models when thermalization is poor or when the number of working levels is small?

Key findings

• When thermal contact is good, the Monte Carlo simulation reproduces standard laser noise theory with high accuracy, validating the model’s consistency with established physics.
• At high optical power levels with poor thermal contact, the frequency band where spectral density is below shot noise becomes narrower, indicating increased noise in the low-frequency regime.
• The relaxation frequency $ f_r \approx 0.074 $ GHz for typical parameters $ \alpha = 0.5 $, $ J^* \approx 0.58 $, which governs the bandwidth of sub-Poissonian behavior.
• Spectral-hole burning and gain fluctuations emerge as significant noise sources when thermal equilibration is impaired, challenging standard theoretical treatments.
• The analytical formula for the photo-current spectral density, derived from linearized birth-death processes, matches the simulation results closely under good thermalization.
• The simulation demonstrates that the population inversion factor $ n_p = \frac{(1+\alpha)^2}{4\alpha} $ and differential gain $ \gamma = \frac{\epsilon}{k_B T} \frac{1-\alpha^2}{2\alpha} $ are critical in determining noise characteristics.