Homogeneous vs. Inhomogeneous Broadening Mechanisms

Good to see you again. In the last lesson, we made the inversion criterion precise: an optical transition provides gain only when the upper-state population per degenerate substate exceeds the corresponding lower-state population. That population difference determines the sign of the interaction; this lesson adds the frequency dependence. A gain medium is not equally effective at every optical frequency, because its transitions have finite spectral width.

We will distinguish two fundamentally different reasons for that width. In homogeneous broadening, every active particle has the same spectral response. In inhomogeneous broadening, different members of the ensemble have different resonance frequencies, and the measured line is their aggregate. The distinction will let you identify whether lifetime, collisions, phonons, thermal motion, or static disorder is likely to control a given laser transition.

---

A linewidth is an ensemble observation

For an ideal transition between perfectly stationary states, energy conservation would give one sharp resonance frequency,

$${\nu }_0=\frac{E_2-E_1}{h}.$$

Real transitions have a frequency-dependent response described by a normalized line-shape function $g(\nu )$:

$${\int }_{-\infty }^{\infty }g(\nu )d\nu =1.$$

The quantity $g(\nu )$ specifies how probable emission, absorption, or stimulated emission is near frequency $\nu$. Later, it will enter directly into the frequency-dependent stimulated-emission cross section and gain coefficient.

The crucial question is not merely “why is the measured line broad?” It is:

When a narrowband laser probes the medium at a chosen frequency, does it address essentially all active particles, or only a particular subgroup?

That is the operational distinction between homogeneous and inhomogeneous broadening.

Homogeneous broadening

A medium is homogeneously broadened when every active atom, molecule, or ion has the same line center and the same individual line shape. Individual particles need not emit identical photons in repeated events; rather, they are statistically equivalent. A narrow probe detuned from line center therefore encounters the same response function from every member of the ensemble.

For a homogeneous ensemble,

$$g_{\mathrm{ens}}(\nu )=g_{\mathrm{h}}(\nu -{\nu }_0),$$

where $g_{\mathrm{h}}$ is the individual-particle profile.

The common physical theme is time-dependent dephasing: each emitter’s optical polarization loses phase coherence over a finite time, and the finite coherence time produces a finite spectral width.

Inhomogeneous broadening

A medium is inhomogeneously broadened when different particles have different resonance frequencies. Each individual particle may have a narrow homogeneous profile, but its center frequency ${\nu }_c$ depends on some particle-specific parameter: its velocity, local strain, local electric field, isotope, or nearby environment.

If the distribution of center frequencies is $p({\nu }_c)$, then the observed ensemble profile is

$$g_{\mathrm{ens}}(\nu )={\int }_{-\infty }^{\infty }p({\nu }_c)g_{\mathrm{h}}(\nu -{\nu }_c)d{\nu }_c.$$

Thus, a laser tuned near one frequency mostly interacts with those particles whose own ${\nu }_c$ lies nearby. Tuning the laser to a different part of the line selects a different subensemble.

The following distinction is worth making automatic:

Question	Homogeneous medium	Inhomogeneous medium
Do all particles share one resonance center?	Yes	No; centers are distributed
What does a narrowband probe address?	The whole ensemble	Mainly a resonant frequency class
Typical origin	Dynamic phase interruption or rapid energy-level modulation	Persistent particle-to-particle differences
Common idealized shape	Lorentzian	Gaussian, if the center-frequency distribution is Gaussian

Read the following short section now. It establishes the distinction, connects it to saturation, and derives thermal Doppler broadening as the standard inhomogeneous case.

---

Homogeneous mechanisms: finite coherence shared by all emitters

Natural, or lifetime, broadening

An excited state has a finite lifetime. If its radiating dipole field has the approximate time dependence

$$E(t)\propto e^{-t/T_2}{e}^{-i2\pi {\nu }_{0}t},$$

then its Fourier transform has a Lorentzian spectral profile. In a standard convention,

$$g_{\mathrm{L}}(\nu )=\frac{1}{\pi }\frac{\Delta {\nu }_{\mathrm{L}}/2}{(\nu -{\nu }_0{)}^2+(\Delta {\nu }_{\mathrm{L}}/2{)}^2}.$$

Here $\Delta {\nu }_{\mathrm{L}}$ is the full width at half maximum (FWHM). For exponential decay of optical coherence,

$$\Delta {\nu }_{\mathrm{L}}=\frac{1}{\pi T_2}.$$

This is the Fourier-transform version of the time-frequency uncertainty principle: a dipole that remains phase coherent only briefly cannot define an extremely precise frequency.

Natural broadening is homogeneous because all otherwise identical atoms have the same finite level lifetimes. It is always present, though it is often much smaller than other widths.

Be careful to distinguish a population lifetime $T_1$ from a coherence time $T_2$. With no additional pure dephasing, a radiatively decaying upper state gives approximately

$$T_2=2T_1.$$

Collisions and lattice fluctuations can make $T_2$ much shorter than this lifetime-limited value without necessarily depopulating the upper state quickly.

Collision, impact, or pressure broadening

In a gas, collisions perturb the phase accumulated between two energy levels. An elastic collision may leave the atom in the same electronic state yet change the phase of its radiating polarization. Inelastic processes can additionally shorten the relevant level lifetime. Both effects reduce the effective coherence time.

At higher gas density, collisions occur more often, so pressure broadening usually increases approximately with pressure. The individual emitters are all subject to the same collision statistics; a collision does not establish a permanently distinct resonance class for one atom. Consequently, impact broadening is normally homogeneous and approximately Lorentzian.

The basic competition in a gas is therefore:

• At low pressure, collisions are relatively rare and Doppler broadening often dominates.
• At high pressure or high density, collisions can dominate and the line behaves more homogeneously.

Phonon broadening in crystals

A crystalline host is not perfectly rigid. Lattice vibrations, quantized as phonons, modulate the local environment and therefore the transition energies of dopant ions. When that modulation is rapid and statistically similar across equivalent sites, it is analogous to a continual sequence of phase-perturbing collisions. The resulting broadening is homogeneous.

This mechanism is especially important for transitions whose electronic orbitals couple strongly to lattice vibrations. Its temperature dependence is often pronounced because phonon populations increase with temperature.

---

Inhomogeneous mechanisms: a distribution of resonance centers

Doppler broadening in gases

Suppose a gas has thermal velocities $v_z$ along the optical axis. For $|v_z|\ll c$, an atom sees or emits light with an approximate shift

$$\nu \simeq {\nu }_0\left(1+\frac{v_z}{c}\right).$$

Atoms moving toward the beam are shifted one way in frequency; atoms moving away are shifted the other way. Since an equilibrium gas has a Maxwell-Boltzmann distribution of line-of-sight velocities, it has a Gaussian distribution of resonance frequencies.

The Doppler FWHM is

$$\Delta {\nu }_{\mathrm{D}}={\nu }_0\sqrt{\frac{8k_{\mathrm{B}}T\ln 2}{m{c}^2}}.$$

It increases with temperature and decreases with particle mass. At a fixed temperature, an optical transition has a larger Doppler width than a microwave transition because $\Delta {\nu }_{\mathrm{D}}$ scales with ${\nu }_0$.

Doppler broadening is inhomogeneous under the usual dilute-gas condition: over the optical coherence time, an atom’s velocity remains sufficiently well defined that it belongs to a particular Doppler-shifted class. A narrow laser frequency then interacts most strongly with atoms having the corresponding $v_z$.

This is closely related to a familiar spectroscopy point: a common bulk velocity shifts an entire line but does not broaden it. Broadening requires a distribution of line-of-sight velocities. Thermal motion, turbulence, unresolved velocity gradients, and multiple kinematic components can all create such a distribution, although in laser media the thermal Doppler distribution is the canonical case.

Static site variations in solids and glasses

In a doped glass or imperfect crystal, ions nominally of the same species can occupy locally different environments. Site-to-site variations in strain, nearby charges, local electric fields, composition, or crystal-field shifts give each ion a slightly different transition energy.

Those local differences persist for long compared with an optical coherence time. The ensemble is then a superposition of many resonance frequencies: inhomogeneous broadening.

Rare-earth ions in glasses are an important example. Their optically active electrons can be relatively shielded from rapid lattice vibrations, limiting homogeneous phonon broadening, while static variations of the local host environment produce a substantial distribution of transition centers. This is why Nd-doped glasses can exhibit important inhomogeneous broadening even though Nd:YAG at ordinary temperatures is often treated as predominantly homogeneously broadened.

The visual hallmark appears when a narrow intense field saturates an inhomogeneous gain medium.

---

Saturation as a practical test

Line shape alone provides a useful clue, but it is not a definitive classification. A Gaussian-looking line suggests a distribution of resonance centers, while extended Lorentzian wings suggest homogeneous dephasing; nevertheless, multiple mechanisms coexist and their convolution can resemble either simple shape over a limited measurement range.

A more decisive conceptual test is narrowband saturation.

In a homogeneously broadened medium, a narrow intense field at one frequency changes the shared population and coherence response of essentially every active emitter. The gain or absorption is therefore reduced across the whole homogeneous line, not just at the pump frequency.

In an inhomogeneously broadened medium, that field predominantly saturates the resonant frequency class. Atoms whose centers lie well away from the field frequency remain largely unsaturated. A weak probe scanned across the line then finds a narrow reduction centered on the saturating laser frequency: spectral hole burning.

Saturation observation	Interpretation
Gain or absorption decreases across essentially the full line	Predominantly homogeneous broadening
A narrow depleted region appears within a much broader envelope	Predominantly inhomogeneous broadening
A hole is visible but broadened, shallow, or partly filled	Both mechanisms are relevant, or frequency classes mix during the measurement

This distinction matters later for laser oscillation. In a homogeneously broadened gain medium, one lasing mode strongly saturates the gain available to competing modes. In an inhomogeneously broadened medium, separated modes can draw gain from different frequency classes and can therefore coexist more readily.

The following segment from Debdeep Jena’s Cornell lecture develops the physical mechanisms in a complementary way. Watch it after you have the ensemble-versus-subensemble distinction firmly in mind.

---

Mixed broadening is the normal case

“Homogeneous” and “inhomogeneous” are best understood as dominant descriptions, not mutually exclusive material labels. Every real line has a natural homogeneous width. A gas can have both thermal Doppler and collisional broadening. A solid can have phonon-induced homogeneous width alongside static site-to-site variation.

For a thermal gas with both Doppler and Lorentzian homogeneous broadening, the ensemble profile is a Voigt profile: a convolution of a Gaussian and a Lorentzian,

$$g_{\mathrm{V}}(\nu )=g_{\mathrm{G}}(\nu )\ast g_{\mathrm{L}}(\nu ).$$

The dominant contribution can depend on what part of the profile matters:

• The Gaussian Doppler component commonly controls the central width when Doppler broadening is large.
• Lorentzian contributions decay more slowly with detuning and can dominate far in the wings.
• A statement such as “Doppler dominated” usually refers to the observed FWHM or the saturation behavior, so always ask what observable and spectral range are under discussion.

---

Diagnosing the dominant mechanism from a physical situation

A reliable classification process has three stages.

1. Identify whether the perturbation is dynamic or persistent.
   Rapid phase randomization that affects all emitters statistically alike indicates homogeneous broadening. A persistent emitter-specific shift indicates inhomogeneous broadening.

2. Identify the physical source.
   In a gas, compare thermal velocity spread with collision frequency. In a solid, distinguish dynamic phonon coupling from static disorder, strain, or local-field variations.

3. Compare estimated linewidths or use a saturation measurement.
   The largest contribution near the feature of interest generally controls the observed classification. If numerical widths are comparable, call the line mixed and describe both contributions rather than forcing a binary label.

Here are representative classifications.

Physical situation	Dominant mechanism	Classification	Reasoning
An isolated atom or dilute atomic vapor with negligible collisions	Natural lifetime width	Homogeneous	Every atom has the same finite lifetime and coherence decay.
A low-pressure He-Ne discharge	Thermal Doppler width, often much larger than natural and pressure width	Inhomogeneous	Different neon atoms have different axial velocities and thus different Doppler-shifted centers.
A dense gas cell whose pressure is increased substantially	Collisional or pressure broadening	Homogeneous	Frequent collisions interrupt phase evolution for every emitter.
A transition-metal ion in a vibrating crystal host at room temperature	Phonon dephasing	Homogeneous	Dynamic lattice vibrations modulate equivalent dopant ions in statistically similar ways.
Rare-earth ions in a disordered glass or a strained crystal	Static site distribution	Inhomogeneous	Local crystal fields and strain differ from site to site, shifting line centers.
A plasma-emission line with a distribution of line-of-sight thermal or turbulent velocities	Doppler-type velocity distribution	Inhomogeneous	Each emitter has a shifted center in the observation frame; a single bulk velocity would only shift the line.

A final caution: broadening mechanisms can change regime as conditions change. Raising gas pressure, for example, both increases collision rates and can modify the importance of velocity classes. The correct statement is therefore conditional: under the stated density, temperature, host material, and timescale, this mechanism dominates.

---

Takeaways

The central distinction is microscopic:

$$\boxed{\text{Homogeneous: every emitter has the same spectral response.}}$$ $$\boxed{\text{Inhomogeneous: different emitters have different resonance centers.}}$$

Natural lifetime broadening, collision broadening, and dynamic phonon broadening are standard homogeneous mechanisms. Doppler broadening and static local-environment variations are standard inhomogeneous mechanisms.

A Lorentzian profile often signals homogeneous dephasing, while a Gaussian profile often signals a distribution of centers, but a measured profile may be Voigt-like and mixed. The cleanest conceptual diagnostic is saturation: homogeneous saturation reduces the full line, whereas inhomogeneous saturation creates a spectral hole by depleting only a resonant subensemble.

Next, we will use the normalized line-shape function quantitatively to calculate a stimulated-emission cross section from transition parameters.