🔬 optics

Spectra of laser diodes

This paper serves as an introductory supplement to a semiconductor laser diode textbook, focusing specifically on the theoretical aspects of their noise properties and spectral characteristics.

Original authors: Bjarne Tromborg, Palle Jeppesen

Published 2026-04-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Bjarne Tromborg, Palle Jeppesen

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

Imagine a laser diode not as a complex piece of high-tech hardware, but as a tiny, high-speed factory inside a microscopic chip. This factory's job is to produce light (photons) by smashing together tiny particles called electrons and holes.

This paper, written by two physicists (one of whom has sadly passed away, and the other is honoring him), is essentially a manual on how to predict the "mood swings" and "noise" of this factory.

Here is the breakdown of the paper's concepts using everyday analogies:

1. The Factory Floor: Carriers and Photons

Think of the laser as a factory with two main groups of workers:

The Carriers (Electrons & Holes): These are the raw materials. You pump them into the factory via a "current" (like a conveyor belt bringing in supplies).
The Photons (Light): These are the finished products.

The paper uses Rate Equations to describe the factory. It's like a balance sheet:

Input: How fast are we bringing in raw materials?
Output: How fast are we making products?
Waste: How many materials get lost or break down before becoming products?

The goal of the laser is to reach a "steady state" where the factory is humming along perfectly, producing a steady stream of light.

2. The "Threshold": The Tipping Point

Every factory has a point where it starts working efficiently.

Below Threshold: If you turn on the power just a little, the factory is chaotic. The workers are bumping into each other randomly, producing a dim, messy glow (like a regular lightbulb). This is Spontaneous Emission.
Above Threshold: Once you push the power past a specific "tipping point" (the Threshold Current), something magical happens. The workers start coordinating. One worker bumps into another, and they both release a product in perfect sync. This is Stimulated Emission.
The Result: The factory suddenly floods with light. The paper explains that once you cross this line, the number of raw materials (electrons) gets "clamped" (stuck at a specific level) because the factory is so efficient at turning them into light that it can't hold any extra.

3. The "Noise": Why the Light Flickers

Even when the factory is running perfectly, it's not silent. There is always background noise. The paper focuses on three types of noise, which are like different kinds of chaos in the factory:

Intensity Noise (RIN): Imagine the conveyor belt speed fluctuating slightly. Sometimes the factory produces a burst of light, sometimes a dip. This is the "flicker" in the brightness. The paper calculates exactly how much the light will flicker based on how hard you are pushing the factory.
Frequency Noise (Phase Noise): Imagine the factory is a metronome ticking perfectly. Frequency noise is like the metronome getting slightly faster or slower every few ticks. In a laser, this means the color of the light wobbles slightly. This is crucial for high-speed internet because if the "color" wobbles too much, the data gets garbled.
Relaxation Oscillations: This is the most interesting part. Imagine you suddenly turn up the power to the factory. The factory doesn't just jump to the new speed; it overshoots, then undershoots, then overshoots again, like a car suspension bouncing after hitting a bump. The paper shows that the light intensity will "ring" or vibrate at a specific frequency (the relaxation frequency) before settling down.

4. The "Linewidth Enhancement Factor" (The Alpha Factor)

This is a specific number (denoted as $\alpha$ ) that the paper highlights.

The Analogy: Imagine a factory manager who is very emotional. When the factory gets too busy (too many electrons), the manager gets stressed and changes the color of the lights, not just the brightness.
In physics terms, a change in the number of electrons changes the refractive index (which changes the color/frequency of the light).
The paper explains that this "emotional manager" (the $\alpha$ factor) makes the laser's light much "fuzzier" (wider linewidth) than it would be otherwise. If $\alpha$ is high, the light is less pure.

5. The "Langevin Noise Functions": The Random Walkers

To calculate these noises, the authors use something called Langevin Noise.

The Analogy: Imagine the factory floor is crowded with people walking randomly. Sometimes two people bump into each other by pure chance (spontaneous emission). This random bumping is the source of the noise.
The paper treats these random bumps as mathematical "noise sources" that are added to the equations. It's like adding a "static" layer to a radio signal to see how the radio distorts.

6. Measuring the Chaos

The paper also discusses how to measure this noise in the real world.

The Interferometer: To measure the "wobble" in the light's color (phase noise), scientists use a device called a Michelson Interferometer.
The Analogy: It's like taking a runner (the light beam), splitting them into two paths, and making one path run a little longer than the other. When they meet again, you see if they are still running in step (coherent) or if they are out of sync. The paper explains how the "fringe visibility" (how clear the pattern is) tells you how much the light is wobbling.

Summary: Why does this matter?

This paper is a mathematical blueprint for understanding the imperfections of laser diodes.

For Engineers: It tells them how to design lasers that are stable enough for high-speed fiber-optic internet. If the noise is too high, your video call will pixelate.
For Scientists: It connects the microscopic quantum world (electrons bumping) to the macroscopic world (the light beam you see).

In short, the paper takes a messy, chaotic quantum factory and writes a precise rulebook for how to predict its jittery behavior, ensuring that the lasers in our phones and internet cables work reliably.

1. Problem Statement

The performance of optical communication systems is heavily dependent on the noise characteristics of the light source. While various light sources exist, semiconductor laser diodes remain the dominant technology. However, a comprehensive theoretical framework that connects the fundamental physical mechanisms of semiconductor lasers (carrier dynamics, spontaneous emission, and waveguide modes) to their observable noise spectra (Relative Intensity Noise, Frequency Noise, and Field Power Spectral Density) is often scattered across different textbooks and specialized papers.

The specific problems addressed in this paper include:

Deriving a unified mathematical model for the static and dynamic behavior of a generic semiconductor laser diode.
Calculating the noise spectra (field power, relative intensity, and frequency) that characterize laser performance.
Explaining the physical origins of spectral features such as the Lorentzian linewidth, relaxation oscillations, sidemodes, and asymmetry in the spectrum.
Bridging the gap between simple rate equations and complex field equations to account for spontaneous emission noise (Langevin noise) and partition noise.

2. Methodology

The authors employ a multi-layered theoretical approach, progressing from macroscopic rate equations to microscopic field equations:

Rate Equation Model: The foundation is a system of coupled rate equations for the carrier number ( $N$ ) and photon number ( $P$ ). These equations include terms for carrier injection, recombination (spontaneous and stimulated), and photon loss. The model uses standard parameters (e.g., transparency carrier number $N_{tr}$ , photon lifetime $\tau_p$ , gain rate $a$ ).
Small-Signal Analysis: The authors linearize the rate equations around a steady-state operating point to derive transfer functions. This allows the calculation of the modulation response and the identification of the relaxation resonance frequency ( $\Omega_R$ ).
Waveguide and Field Theory: To determine spectral properties, the authors solve the wave equation for the electric field in the laser waveguide. They introduce a loop gain model ( $G(\omega, N)$ ) to describe the laser oscillator, considering boundary conditions at the facets and the interaction between forward and backward traveling waves.
Langevin Noise Formalism: To account for noise, the authors introduce Langevin noise functions ( $F(t)$ , $F_N(t)$ , $F_\phi(t)$ ) representing spontaneous emission, carrier injection fluctuations, and phase noise. They derive the autocorrelation and cross-correlation properties of these noise sources (diffusion coefficients).
Perturbation Expansion: The solutions to the stochastic rate equations are expanded in powers of a noise parameter. This allows the calculation of second-order effects, such as the noise-induced shift in the lasing frequency and the modification of the average photon number.
Spectral Derivation: The authors derive the Power Spectral Density (PSD) for:
- Intensity Noise (RIN): Using the transfer matrix and noise correlation strengths.
- Frequency Noise: Derived from the phase rate equation, linking phase fluctuations to carrier density fluctuations via the linewidth enhancement factor ( $\alpha$ ).
- Field Spectrum: Calculated via the autocorrelation of the field envelope, incorporating the variance of the phase difference $\langle (\Delta \phi)^2 \rangle$ .

3. Key Contributions

Unified Noise Model: The paper provides a complete derivation linking the microscopic Langevin noise sources to macroscopic observable spectra (RIN, Frequency Noise, and Field PSD) within a single consistent framework.
Derivation of the Linewidth Enhancement Factor ( $\alpha$ ): The paper explicitly demonstrates how the parameter $\alpha$ (relating changes in gain to changes in refractive index) amplifies the phase noise. It shows that the laser linewidth is proportional to $(1+\alpha^2)$ , explaining why semiconductor lasers have significantly broader linewidths than gas lasers.
Detailed Analysis of Relaxation Oscillations: The model predicts the characteristic "peaks" in the RIN and frequency noise spectra at the relaxation resonance frequency ( $f_R$ ). It explains how these peaks broaden and shift with bias current.
Sidemode and Asymmetry Analysis: By extending the analysis to a multimode field equation, the authors explain the existence of sidemodes (satellite peaks) and the asymmetry in the laser spectrum caused by amplitude-phase coupling.
Partition Noise and Shot Noise: The paper rigorously treats partition noise (fluctuations due to photons being transmitted or reflected at the facet) and shot noise, showing how they contribute to the RIN floor and the external field phase noise.
1/f Noise and Measurement Limits: The authors incorporate low-frequency 1/f noise and discuss the practical implications of measurement bandwidth and integration time on the observed variance of phase noise, addressing the divergence issues in theoretical models.

4. Key Results

RIN Spectrum: The Relative Intensity Noise spectrum is dominated by a peak near the relaxation resonance frequency ( $f_R \propto \sqrt{I - I_{th}}$ ). At high frequencies, the spectrum flattens to a shot-noise limit ( $\hbar\omega_0 / \langle P_{out} \rangle$ ).
Frequency Noise: The frequency noise spectrum exhibits a Lorentzian shape at low frequencies (determined by the linewidth) and peaks at the relaxation resonance. The high-frequency limit is determined by spontaneous emission.
Field Power Spectral Density:
- The central peak is Lorentzian with a Full Width at Half Maximum (FWHM) of $\Delta f = \frac{R_{sp}(1+\alpha^2)}{4P_s}$ .
- The spectrum includes satellite peaks (relaxation resonances) at multiples of the relaxation frequency.
- The spectrum is asymmetric due to the coupling between amplitude and phase fluctuations (phase-amplitude coupling).
Linewidth Enhancement: For a typical $\alpha = 5$ , the linewidth is enhanced by a factor of 26 compared to a laser without amplitude-phase coupling.
Noise-Induced Frequency Shift: The analysis confirms that, to the second order, spontaneous emission does not cause a net shift in the lasing frequency ( $\hat{\omega} = 0$ ), although it does increase the average photon number slightly.
Experimental Validation: The theoretical predictions for the variance of phase noise $\langle (\Delta \phi)^2 \rangle$ versus time delay match experimental data obtained via Michelson interferometry, validating the model's accuracy for delays up to several nanoseconds.

5. Significance

Educational Value: The paper serves as a comprehensive "appendix" or advanced chapter for optical communication textbooks, providing a rigorous yet accessible derivation of laser noise theory that is often omitted in standard texts.
Design Implications: The results are critical for the design of high-speed optical communication systems. Understanding the RIN and frequency noise spectra allows engineers to optimize laser bias currents, select appropriate modulation bandwidths, and mitigate noise-induced bit errors.
Fundamental Physics: The work clarifies the interplay between quantum noise (spontaneous emission), carrier dynamics, and optical feedback, offering deep insights into the stability and coherence of semiconductor lasers.
Legacy: As a posthumous work finalized by Palle Jeppesen and Jesper Mørk, it preserves the rigorous mathematical and physical insights of Bjarne Tromborg, a leading figure in semiconductor laser theory, ensuring these foundational concepts remain available for future research in photonics.

In summary, this paper provides a definitive theoretical treatment of semiconductor laser noise, successfully connecting the stochastic nature of spontaneous emission to the deterministic spectral shapes observed in experiments, and highlighting the critical role of the linewidth enhancement factor in limiting laser coherence.