Local Field Corrections and the Clausius-Mossotti Relation

Every constitutive relation in electromagnetism — $\mathbf{P}=\varepsilon_0\chi\mathbf{E}$ , $\mathbf{D}=\varepsilon_0\varepsilon_r\mathbf{E}$ — is written in terms of the macroscopic field, obtained by averaging Maxwell's equations over a volume large compared with the atomic spacing. But no atom ever feels that average. An atom sits at a point, and what drives its electrons is the applied field plus the field of every other dipole, minus its own. In a dilute gas those neighbour fields are negligible. In a liquid, glass, or crystal the neighbours are close, all polarized the same way, and their combined contribution at any site is a real, sizeable surplus. That surplus is the local field correction, and forgetting it gives wrong refractive indices, misplaced spectral lines, and radiative rates off by factors of three. This article takes two inputs and returns one master relation and one universal factor. From the electric-dipole and polarizability articles it inherits the response $\mathbf{p}=\alpha\mathbf{E}_{\mathrm{loc}}$ — a dipole proportional to whatever field the atom sees — and from electrostatics it borrows the field inside a spherical cavity. Lorentz closed the loop self-consistently: the polarization sets the local field, the local field sets each dipole, and the dipoles sum back to the polarization. That loop yields the Clausius–Mossotti relation, its optical twin the Lorentz–Lorenz equation, and the local-field factor $L=(n^2+2)/3$ this chapter hands forward to spontaneous emission, the Einstein coefficients, and radiative lifetimes.

Key takeaways

An atom in a dense medium is driven not by the macroscopic field $\mathbf{E}$ but by a stronger local field $\mathbf{E}_{\mathrm{loc}}$ , because the surrounding polarized matter adds its own field at the site.
For an isotropic medium or cubic crystal, Lorentz's cavity construction gives $\mathbf{E}_{\mathrm{loc}}=\mathbf{E}+\mathbf{P}/(3\varepsilon_0)$ — the cavity's surface charge supplies $+\mathbf{P}/3\varepsilon_0$ while the discrete inner neighbours sum to zero.
Feeding this into $\mathbf{P}=N\alpha\,\mathbf{E}_{\mathrm{loc}}$ gives the Clausius–Mossotti relation $\frac{\varepsilon_r-1}{\varepsilon_r+2}=\frac{N\alpha}{3\varepsilon_0}$ , bridging microscopic polarizability and bulk permittivity.
At optical frequencies with $\varepsilon_r=n^2$ it becomes the Lorentz–Lorenz equation, defining the nearly density-independent molar refractivity $R_m$ .
The same factor $L=(n^2+2)/3$ renormalizes radiative properties: spontaneous emission speeds up by $\approx n\,L^2$ , resonances red-shift, and oscillator strengths need correcting before comparison with vacuum.
Key idea: an embedded atom responds to $\mathbf{E}_{\mathrm{loc}}=\mathbf{E}+\mathbf{P}/(3\varepsilon_0)$ , not to $\mathbf{E}$ ; self-consistency with $\mathbf{P}=N\alpha\mathbf{E}_{\mathrm{loc}}$ gives Clausius–Mossotti / Lorentz–Lorenz and a local-field factor $L=(\varepsilon_r+2)/3=(n^2+2)/3$ that rescales emission, absorption, and line positions.

The core insight

An embedded atom responds to the local field $\mathbf{E}_{\mathrm{loc}}=\mathbf{E}+\mathbf{P}/(3\varepsilon_0)$ , not to the macroscopic $\mathbf{E}$ ; enforcing self-consistency between this and $\mathbf{P}=N\alpha\mathbf{E}_{\mathrm{loc}}$ produces the Clausius–Mossotti / Lorentz–Lorenz relation and a local-field factor $L=(\varepsilon_r+2)/3=(n^2+2)/3$ that renormalizes emission, absorption, and line positions.

Where this lives today

The local-field factor is a working correction in every laboratory that measures how bright an emitter is. When a rare-earth ion such as Er³⁺ or Nd³⁺ is doped into a glass fibre or crystal, its measured radiative lifetime is shorter than the free-ion value by roughly $n\,L^2$ , and the Judd–Ofelt analysis at the heart of laser-gain engineering must divide that factor out to recover an intrinsic transition strength. The same arithmetic governs colloidal quantum dots, nitrogen-vacancy centres in diamond ( $n=2.42$ , where the boost nears an order of magnitude), and single molecules: comparing brightness across hosts, or against a Purcell cavity, is meaningless until the background enhancement is removed. On the materials side, Clausius–Mossotti generalizes into the Maxwell Garnett and Bruggeman effective-medium rules for nanoparticle composites and metamaterials, where local-field factors raised to high powers make engineered nonlinear susceptibilities enormous near a resonance. Refractometers, meanwhile, still lean on the density-independence of $R_m$ to read sugar, salinity, and protein from a single index measurement.

The physical picture

Put a slab of dielectric in a uniform applied field. Every molecule becomes a little dipole, and at any one molecule the near fields of all the others pile up. Part of that pile-up is already accounted for — the depolarizing field of the sample's outer surfaces is what turns the applied field into the smaller macroscopic $\mathbf{E}$ . What remains is a local surplus from the immediate neighbourhood, invisible to any smooth average. Lorentz isolated it with a spherical cavity carved around the reference atom, large enough that everything outside is smooth polarized continuum, small enough that its field is uniform across the atom. The field at the centre splits into three pieces: the macroscopic $\mathbf{E}$ from far away, the field $\mathbf{E}_{\mathrm{cav}}$ of the bound charge on the cavity wall, and the field $\mathbf{E}_{\mathrm{near}}$ of the discrete dipoles that lay inside the sphere. The wall charge is the interesting part: a uniform $\mathbf{P}$ deposits $\sigma_b=\mathbf{P}\cdot\hat{\mathbf{n}}$ on the boundary, whose field integrates to a clean $+\mathbf{P}/3\varepsilon_0$ along $\mathbf{P}$ . The inner dipoles cancel exactly for a cubic lattice or isotropic arrangement. So the whole correction collapses to a single geometric third — which is why gases, liquids, and cubic crystals share one functional law.

Figure

The Lorentz cavity — the reference atom at the centre sees the macroscopic field E plus the field of bound charge on the cavity wall, σ_b = P·n̂ = P cosθ, which integrates to E_cav = P/3ε₀ along P. Discrete neighbours inside cancel for cubic or isotropic symmetry, so E_loc = E + P/3ε₀

Quantum-optically the enhancement cuts both ways. The field that drives an absorbing atom is boosted by $L$ , and so is the field it radiates when it emits, so lifetimes shorten inside a dense host. An emitter in glass of index $n=1.5$ radiates about three times faster than in vacuum; read that rate as if the atom were in vacuum and you badly overestimate its dipole moment. The boost climbs steeply with host index — as $n\,L^2=n(n^2+2)^2/9$ — so the host is a real experimental knob, which the second figure plots with the values a spectroscopist actually meets.

Figure

The spontaneous-emission enhancement A_medium/A_vac = n L² = n(n²+2)²/9 in the virtual-cavity model, versus host refractive index n. Marked: water n = 1.33 gives ×2.10, glass n = 1.50 gives ×3.01, diamond n = 2.42 gives ×16.6 — one factor n from the photon density of states, L² from the enhanced field at the emitter

The curve carries a warning too. The Clausius–Mossotti susceptibility blows up as $N\alpha/3\varepsilon_0\to1$ , where the induced dipoles reinforce one another so strongly that the medium would polarize spontaneously. That "polarization catastrophe" is a caricature of ferroelectricity and a reminder that this is mean-field theory: superb in the weakly-coupled regime, untrustworthy near an instability.

The core equations

The starting point is the field the atom actually experiences. For an isotropic medium or cubic lattice, Lorentz's decomposition — macroscopic, cavity-wall, and vanishing near-neighbour field — collapses to one correction proportional to the polarization.

\mathbf{E}_{\mathrm{loc}} = \mathbf{E} + \frac{\mathbf{P}}{3\varepsilon_0} \qquad (1)

where: $\mathbf{E}_{\mathrm{loc}}$ — field driving a reference atom, excluding its own field; $\mathbf{E}$ — macroscopic (Maxwell-averaged) field; $\mathbf{P}$ — polarization; $\varepsilon_0$ — vacuum permittivity. The correction points along $\mathbf{P}$ : the neighbourhood reinforces the applied field, so a dense medium responds more strongly than a naive count of independent dipoles. Each atom responds linearly to the field it sees; the single-atom polarizability times the number density gives the polarization — the microscopic constitutive relation that closes the self-consistency loop.

\mathbf{P} = N\alpha\,\mathbf{E}_{\mathrm{loc}} \qquad (2)

where: $N$ — number density of polarizable atoms; $\alpha$ — molecular polarizability, $\mathbf{p}=\alpha\mathbf{E}_{\mathrm{loc}}$ , units $\mathrm{C\,m^2\,V^{-1}}$ . The field here is $\mathbf{E}_{\mathrm{loc}}$ , not $\mathbf{E}$ — the whole subtlety lives in that subscript. Eliminating $\mathbf{E}_{\mathrm{loc}}$ between (1) and (2) and using $\mathbf{P}=\varepsilon_0(\varepsilon_r-1)\mathbf{E}$ gives the bridge from microscopic to bulk, derived below.

\frac{\varepsilon_r-1}{\varepsilon_r+2} = \frac{N\alpha}{3\varepsilon_0} \qquad (3)

where: $\varepsilon_r$ — relative permittivity, $\mathbf{D}=\varepsilon_0\varepsilon_r\mathbf{E}$ . The local field's signature is the $\varepsilon_r+2$ denominator: drop the correction and you get the naive $\varepsilon_r-1=N\alpha/\varepsilon_0$ , valid only at lowest order in density. At optical frequencies $\varepsilon_r=n^2$ , so (3) becomes the Lorentz–Lorenz equation. Dividing by mass density and multiplying by molar mass isolates the molar refractivity, nearly independent of temperature, pressure, and phase.

\frac{n^2-1}{n^2+2} = \frac{N\alpha}{3\varepsilon_0}, \qquad R_m=\frac{n^2-1}{n^2+2}\,\frac{M}{\rho}=\frac{N_A\,\alpha}{3\varepsilon_0} \qquad (4)

where: $n$ — refractive index, $n=\sqrt{\varepsilon_r}$ ; $R_m$ — molar refractivity, additive over atoms and bonds; $M$ — molar mass; $\rho$ — density; $N_A$ — Avogadro constant. Since $R_m$ depends only on $\alpha$ , it barely changes under compression or phase change. When $\alpha(\omega)$ is that of a driven bound electron it has resonant poles, and the local-field denominator shifts them: a single Lorentz oscillator in (3) moves the pole of $\varepsilon_r$ below the bare atomic resonance.

\omega_r^2 = \omega_0^2 - \frac{N e^2 f}{3 m_e \varepsilon_0} \qquad (5)

where: $\omega_r$ — collective resonance in the medium; $\omega_0$ — bare isolated-atom resonance; $f$ — oscillator strength; $e,m_e$ — electron charge and mass. A red-shift, linear in $N$ and $f$ , as neighbours soften the restoring force. Finally, the low-frequency ratio $E_{\mathrm{loc}}/E$ is the local-field factor itself, reappearing wherever a field couples to an embedded atom.

L \equiv \frac{\varepsilon_r+2}{3} = \frac{n^2+2}{3}, \qquad \frac{A_{\mathrm{medium}}}{A_{\mathrm{vac}}} = n\,L^2 = n\left(\frac{n^2+2}{3}\right)^2 \qquad (6)

where: $L$ — local-field factor; $A$ — Einstein spontaneous-emission coefficient, $\tau=1/A$ . In the virtual-cavity model the rate carries a factor $n$ from the photon density of states and $L^2$ from the field at the emitter; absorption picks up the same $L^2$ .

Derivation AClausius–Mossotti from the Lorentz local field

The goal is to turn three familiar relations into the master bridge $\frac{\varepsilon_r-1}{\varepsilon_r+2}=\frac{N\alpha}{3\varepsilon_0}$ by demanding they hold at once: the geometric local field $\mathbf{E}_{\mathrm{loc}}=\mathbf{E}+\mathbf{P}/3\varepsilon_0$ , the microscopic response $\mathbf{P}=N\alpha\mathbf{E}_{\mathrm{loc}}$ , and the macroscopic definition $\mathbf{P}=\varepsilon_0(\varepsilon_r-1)\mathbf{E}$ . The polarization appears in all three; forcing consistency eliminates $\mathbf{E}_{\mathrm{loc}}$ and $\mathbf{P}$ and leaves a pure relation between the bulk $\varepsilon_r$ and the atomic $\alpha$ . Watch where the tell-tale factor of three lands.

Step 1. Substitute the local field (1) into the response (2): $\mathbf{P}=N\alpha\left(\mathbf{E}+\mathbf{P}/3\varepsilon_0\right)$ . Why: this is self-consistency in one line — the polarization on the left is produced by a local field that itself contains the polarization.

Step 2. Collect terms: $\mathbf{P}\left(1-\frac{N\alpha}{3\varepsilon_0}\right)=N\alpha\,\mathbf{E}$ , so $\mathbf{P}=\dfrac{N\alpha}{1-N\alpha/3\varepsilon_0}\,\mathbf{E}$ . Why: the factor $1/(1-N\alpha/3\varepsilon_0)$ is the local-field enhancement — the medium is more polarizable than the bare $N\alpha$ , and the denominator warns of the catastrophe at $N\alpha/3\varepsilon_0\to1$ .

Step 3. Equate to $\mathbf{P}=\varepsilon_0(\varepsilon_r-1)\mathbf{E}$ , giving $\varepsilon_0(\varepsilon_r-1)=\dfrac{N\alpha}{1-N\alpha/3\varepsilon_0}$ . Why: this is the only place the bulk constant $\varepsilon_r$ enters, converting a microscopic identity into a measurable one.

Step 4. Cross-multiply: $(\varepsilon_r-1)\left(1-\frac{N\alpha}{3\varepsilon_0}\right)=\frac{N\alpha}{\varepsilon_0}$ , which rearranges to $\varepsilon_r-1=\frac{N\alpha}{3\varepsilon_0}(\varepsilon_r+2)$ , the boxed result. Why: the $\varepsilon_r+2$ — not $\varepsilon_r+1$ , not a bare $1$ — is the fingerprint of the one-third cavity factor; omit the local field and you lose it. *Verified: as $N\to0$ it reduces to the dilute limit $\varepsilon_r-1\to N\alpha/\varepsilon_0$ ; $N\alpha/\varepsilon_0$ is dimensionless, so both sides are pure numbers. Cf. Jackson §4.5.*

Go deeper · Derivation B · 1 minThe one-third from bound charge on the cavity wall

Derivation BThe one-third from bound charge on the cavity wall

The goal is to prove that the mysterious $1/3$ in the local field is just the depolarizing factor of a sphere, obtained by integrating the field of the bound surface charge a uniform polarization paints on the cavity wall. A uniform $\mathbf{P}$ leaves no net bulk charge, but wherever a surface interrupts it, $\sigma_b=\mathbf{P}\cdot\hat{\mathbf{n}}$ appears. On the cavity around the reference atom that charge varies as $\cos\theta$ , and its field at the centre is a single angular integral whose value $2/3$ gives the famous third.

Step 1. Take $\mathbf{P}=P\hat{\mathbf{z}}$ and a cavity of radius $a$ ; the wall bound charge is $\sigma_b=\mathbf{P}\cdot\hat{\mathbf{n}}=P\cos\theta$ , with $\theta$ from $\mathbf{P}$ . Why: the charge is positive on the hemisphere opposite $\mathbf{P}$ and negative on the one along it — the source of a field along $\mathbf{P}$ at the centre.

Step 2. A ring at angle $\theta$ has area $\mathrm{d}A=2\pi a^2\sin\theta\,\mathrm{d}\theta$ at distance $a$ ; only its $z$ -component survives, contributing $\mathrm{d}E_z=\dfrac{1}{4\pi\varepsilon_0}\dfrac{\sigma_b\,\mathrm{d}A}{a^2}\cos\theta$ . Why: transverse contributions cancel against the diametrically opposite element, reducing the problem to one integral over $\theta$ .

Step 3. Integrate: $E_z=\dfrac{P}{2\varepsilon_0}\displaystyle\int_0^{\pi}\cos^2\theta\sin\theta\,\mathrm{d}\theta=\dfrac{P}{2\varepsilon_0}\cdot\dfrac{2}{3}=\dfrac{P}{3\varepsilon_0}$ . Why: the radius $a$ cancels, so the cavity field is a pure geometric constant — the same third for any sphere size, which is what makes the construction well-defined. *Verified: the angular integral $\int_0^\pi\cos^2\theta\sin\theta\,\mathrm{d}\theta=2/3$ is standard; the result is the sphere's depolarizing factor $1/3$ . Cf. Griffiths §4.4.*

Limiting cases

The single relation (3) interpolates between dilute and dense regimes, collapsing to a familiar law in each corner:

Case	Result	Meaning
Dilute gas	$\dfrac{N\alpha}{3\varepsilon_0}\ll1\Rightarrow\varepsilon_r-1\approx\dfrac{N\alpha}{\varepsilon_0},\ n-1\approx\dfrac{N\alpha}{2\varepsilon_0}$	Local field $\approx$ macroscopic; the linear-in-density law used for atomic vapours.
Optical, non-magnetic	$\dfrac{n^2-1}{n^2+2}=\dfrac{N\alpha}{3\varepsilon_0}$	With $\varepsilon_r=n^2$ , Lorentz–Lorenz; times $M/\rho$ gives the additive molar refractivity $R_m$ .
Polarization catastrophe	$\dfrac{N\alpha}{3\varepsilon_0}\to1\Rightarrow\varepsilon_r\to\infty$	Mean-field divergence flagging a polarization instability; theory unreliable here.
Emission rate	$\dfrac{A_{\mathrm{medium}}}{A_{\mathrm{vac}}}=n\,L^2,\ L=\dfrac{n^2+2}{3}$	Factor $n$ from the photon density of states, $L^2$ from the field at the emitter.
Resonant medium	$\omega_r^2=\omega_0^2-\dfrac{Ne^2f}{3m_e\varepsilon_0}$	The collective resonance red-shifts by a plasma-like term linear in density.

Common trapThe polarizability α and the susceptibility χ are not the same response

It is tempting to write $\varepsilon_r=1+N\alpha/\varepsilon_0$ , treating the polarizability $\alpha$ as the susceptibility $\chi=\varepsilon_r-1$ divided by number density. It is not. The polarizability responds to the local field $\mathbf{E}_{\mathrm{loc}}$ ; the susceptibility responds to the macroscopic field $\mathbf{E}$ . They differ by exactly the local-field factor: $\chi=\dfrac{N\alpha/\varepsilon_0}{1-N\alpha/3\varepsilon_0}$ , so $\varepsilon_r=1+N\alpha/\varepsilon_0$ is only the leading term in density. The signature you dropped is the $3\varepsilon_0$ — not $\varepsilon_0$ — and its $\varepsilon_r+2$ denominator. Using the naive relation for a liquid or solid can be wrong by tens of percent.

Worked examples

1 · The polarizability of a water molecule from its refractive index

Liquid water has $n=1.333$ at the sodium D-line, $\rho=1.00\ \mathrm{g\,cm^{-3}}$ , $M=18.02\ \mathrm{g\,mol^{-1}}$ . Use Lorentz–Lorenz to extract the molecular polarizability and compare with the accepted value.

Step 1. Form the Lorentz–Lorenz ratio. With $n^2=1.777$ , $\dfrac{n^2-1}{n^2+2}=\dfrac{0.777}{3.777}=\mathbf{0.2057}$ — about a fifth of the way to the polarization catastrophe.

Step 2. Multiply by the molar volume $V_m=M/\rho=1.802\times10^{-5}\ \mathrm{m^3\,mol^{-1}}$ : $R_m=0.2057\times1.802\times10^{-5}=\mathbf{3.71\times10^{-6}\ m^3\,mol^{-1}}=3.71\ \mathrm{cm^3\,mol^{-1}}$ , the tabulated value.

Step 3. Invert $R_m=N_A\alpha/3\varepsilon_0$ : $\alpha=\dfrac{3\varepsilon_0 R_m}{N_A}=\dfrac{3(8.854\times10^{-12})(3.71\times10^{-6})}{6.022\times10^{23}}=\mathbf{1.63\times10^{-40}\ C\,m^2/V}$ .

Step 4. Convert to the polarizability volume tables quote, $\alpha'=\alpha/4\pi\varepsilon_0=1.47\times10^{-30}\ \mathrm{m^3}$ , i.e. α′ ≈ 1.47 Å³ — within a couple of percent of the accepted $1.45\ \mathrm{\text{Å}^3}$ . A visible-light index has recovered a molecular constant.

2 · How much faster does an emitter radiate in glass?

An atom with vacuum radiative lifetime $\tau_{\mathrm{vac}}=10\ \mathrm{ns}$ is embedded in glass of index $n=1.50$ . Estimate its lifetime in the virtual- and real-cavity (Onsager) models.

Step 1. Local-field factor: $L=\dfrac{n^2+2}{3}=\dfrac{2.25+2}{3}=\mathbf{1.417}$ .

Step 2. Virtual cavity: $\dfrac{A_{\mathrm{medium}}}{A_{\mathrm{vac}}}=n\,L^2=1.50\times(1.417)^2=\mathbf{3.01}$ , so $\tau_{\mathrm{medium}}=10/3.01=\mathbf{3.3\ ns}$ — three times shorter than in vacuum.

Step 3. Real cavity: $n\left(\dfrac{3\varepsilon_r}{2\varepsilon_r+1}\right)^2$ with $\varepsilon_r=2.25$ ; the bracket $\dfrac{6.75}{5.5}=1.227$ gives $1.50\times1.227^2=\mathbf{2.26}$ , hence $\tau_{\mathrm{medium}}\approx4.4\ \mathrm{ns}$ .

Step 4. The two models bracket the truth: the lifetime shortens by about 2.3× to 3.0×, so quoting an intrinsic dipole from the raw lifetime without dividing out the factor overstates it similarly. The spread between models is the honest uncertainty.

3 · Is water near the polarization catastrophe?

Using the polarizability from Example 1, find the density at which Clausius–Mossotti diverges and compare it with water's actual density.

Step 1. The catastrophe is $N\alpha/3\varepsilon_0=1$ , i.e. $N_{\mathrm{crit}}=\dfrac{3\varepsilon_0}{\alpha}=\dfrac{3(8.854\times10^{-12})}{1.63\times10^{-40}}=\mathbf{1.62\times10^{29}\ m^{-3}}$ .

Step 2. Water's actual molecular density is $N=\dfrac{\rho N_A}{M}=\mathbf{3.34\times10^{28}\ m^{-3}}$ .

Step 3. The ratio $N/N_{\mathrm{crit}}=\mathbf{0.206}$ — exactly the Lorentz–Lorenz ratio $(n^2-1)/(n^2+2)=0.2057$ from Example 1, as it must be. Water sits at about one-fifth of threshold, well inside the trustworthy regime, which is why $R_m$ is so reliably additive for ordinary liquids.

Physics of the portrait

The local field is where the atom re-enters classical electromagnetism. For most of the nineteenth century the dielectric constant was a bulk number with no owner; connecting it to anything microscopic took a picture of matter as a lattice of polarizable charges and the nerve to ask what field one charge really feels.

Origins · 1 minHendrik Antoon Lorentz and how this was discovered

Hendrik Antoon Lorentz

1853–1928 · electron theory of matter and the optics of dense media

Hendrik Antoon Lorentz built the bridge between Maxwell's continuum electromagnetism and the atomistic picture of matter, and the local field is one of its keystones. Where Maxwell describes smooth averaged fields, Lorentz insisted that matter is made of discrete charged particles — electrons, he called them, before Thomson isolated one — and that the field driving each is not the macroscopic average but the sum over all the others. His cavity construction, treating distant matter as continuum and near matter as discrete dipoles that cancel by symmetry, produced the clean $\mathbf{E}_{\mathrm{loc}}=\mathbf{E}+\mathbf{P}/3\varepsilon_0$ and the index–polarizability relation now called Lorentz–Lorenz — obtained independently by the Dane Ludvig Lorenz, whose near-identical name is a genuine coincidence. The same electron theory let him explain the Zeeman splitting (the 1902 Nobel Prize, shared with Zeeman), derive the Lorentz force and transformation, and lay much of the ground Einstein would build on. Here he is the figure who made the question precise: not "what is the field in the material" but "what is the field at the atom".

History

1850

— Mossotti links a dielectric's response to the polarization of its constituent molecules

1869

— Ludvig Lorenz derives the (n²−1)/(n²+2) refractivity relation from wave optics

1878

— Hendrik Lorentz obtains the same relation from the electron theory of matter

1879

— Clausius recasts it in terms of the permittivity, fixing the Clausius–Mossotti form

1902

— Lorentz shares the Nobel Prize in Physics for the electron theory of matter

Going further

Texts. Hecht, Optics, 5th ed., §3.5 (dispersion, Clausius–Mossotti) · Jackson, Classical Electrodynamics, 3rd ed., §4.5 (polarizability and susceptibility) · Griffiths, Introduction to Electrodynamics, §4.4 (field of a polarized sphere) · Ashcroft and Mermin, Solid State Physics, Ch. 27 · Foot, Atomic Physics, Ch. 7 · Loudon, The Quantum Theory of Light, 3rd ed., Ch. 1 and 6 (emission in a medium) · Saleh and Teich, Fundamentals of Photonics, 3rd ed., Ch. 5 · Demtröder, Laser Spectroscopy, Vol. 1 (line shifts).

Provenance. A verifier should recompute: the Clausius–Mossotti relation by eliminating $\mathbf{E}_{\mathrm{loc}}$ between $\mathbf{E}_{\mathrm{loc}}=\mathbf{E}+\mathbf{P}/3\varepsilon_0$ and $\mathbf{P}=N\alpha\mathbf{E}_{\mathrm{loc}}$ (Jackson §4.5); the cavity integral $\int_0^\pi\cos^2\theta\sin\theta\,\mathrm{d}\theta=2/3$ giving $E_{\mathrm{cav}}=P/3\varepsilon_0$ (Griffiths §4.4); the factor $L=(n^2+2)/3=1.417$ and emission enhancement $nL^2=3.01$ for $n=1.5$ , with real-cavity comparison $2.26$ (Loudon Ch. 6); and the water values $R_m=3.71\ \mathrm{cm^3\,mol^{-1}}$ , $\alpha'=1.47\ \mathrm{\text{Å}^3}$ from $n=1.333$ , $\rho=1.00\ \mathrm{g\,cm^{-3}}$ , $M=18.02\ \mathrm{g\,mol^{-1}}$ , with $N/N_{\mathrm{crit}}=(n^2-1)/(n^2+2)=0.206$ . Constants are CODATA $\varepsilon_0,e,m_e,N_A$ ; other numbers are standard tabulated values.

Prerequisites. Electric Dipole Approximation · Transition Dipole Moment · Oscillator Strength · Resonant Absorption. Leads to. Spontaneous Emission · Einstein A and B Coefficients · Radiative Lifetime and Natural Linewidth · Lorentzian Line Shape.

Ask the tutor: Ask me to walk through why the discrete dipole sum inside the Lorentz sphere vanishes for a cubic lattice but not for a tetragonal one, and what that does to the refractive index.

Why isn't the macroscopic field E good enough — didn't we already average correctly?

Averaging is correct for what it is: $\mathbf{E}$ is the true field smoothed over many atoms, exactly what Maxwell's equations propagate. But an atom is not a smoothed region — it is a point, and there the near field of its neighbours differs sharply from the average. Those neighbour fields are partly counted in $\mathbf{E}$ through the depolarizing field, but a local surplus of order $\mathbf{P}/3\varepsilon_0$ remains. In a dilute gas it is negligible and $\mathbf{E}_{\mathrm{loc}}\approx\mathbf{E}$ ; in condensed matter it is a leading effect, and ignoring it is the commonest quantitative error in relating bulk to molecular optics.

Why does the discrete near-neighbour sum inside the cavity vanish?

Inside the Lorentz sphere the surrounding matter is treated as discrete dipoles whose fields must be summed individually. For a cubic lattice, or a fully random isotropic arrangement, that sum is exactly zero — a symmetry effect: a dipole's field is positive in some directions and negative in others, and on a cubic lattice or averaged over all orientations the contributions cancel term by term. That is why the simple $\mathbf{P}/3\varepsilon_0$ holds for so many materials. In a non-cubic crystal the sum does not vanish, the local field becomes anisotropic, and the scalar relation gives way to a tensor version.

Is the molar refractivity really independent of density, and why does that matter?

Very nearly, and it is one of the most useful facts in optics. Because $R_m=\frac{n^2-1}{n^2+2}\frac{M}{\rho}=\frac{N_A\alpha}{3\varepsilon_0}$ depends only on the intrinsic $\alpha$ , it barely changes when a substance is heated, compressed, or taken from liquid to vapour — even though $n$ changes a lot. That constancy tests the local-field picture and makes $R_m$ additive: a molecule's refractivity is roughly the sum of tabulated atom and bond contributions. Chemists use it to predict the index of an unmeasured compound, and departures flag effects the isotropic derivation misses, such as permanent-dipole orientation.

The local field enhances the field on the atom — so why does emission speed up rather than slow down?

Because it enhances both halves of the interaction, absorption and emission alike. During emission the atomic dipole radiates into the medium, and the same neighbour-polarization that boosts an incoming field boosts the outgoing one, so the atom couples more strongly to the radiation field and decays faster. On top of that local $L^2$ factor is a factor $n$ from the photon density of states; the product $nL^2$ is the total speed-up. This differs from a cavity Purcell effect, which tailors the density of states with mirrors — the local-field boost is present in any bulk dielectric with no cavity at all.

Which cavity model should I trust — virtual or real?

It depends on how the emitter fits into the host. The virtual-cavity (Lorentz) model treats the emitter as one more polarizable point among the host molecules and gives $nL^2$ ; it suits a dopant chemically similar to its surroundings. The real-cavity (Onsager) model carves an actual empty sphere around the emitter and gives $n(3\varepsilon_r/(2\varepsilon_r+1))^2$ ; it suits one that genuinely excludes the host, such as a molecule in solvent. The two differ by about 30 percent at $n=1.5$ — a fair estimate of the theoretical uncertainty, so when precision matters, quote both.

What is the polarization catastrophe, and does any real material reach it?

Clausius–Mossotti contains a factor $1/(1-N\alpha/3\varepsilon_0)$ that diverges when $N\alpha/3\varepsilon_0\to1$ : the induced dipoles reinforce one another so strongly that a tiny fluctuation would grow into a permanent polarization with no applied field. That is the catastrophe, a mean-field caricature of the ferroelectric transition. Ordinary liquids and glasses sit well below it — water is at about one-fifth — so the theory is safe there. Real ferroelectrics do approach the instability, but near it the assumption that every atom feels the same average field breaks down and the relation stops being quantitative. Read the divergence as the boundary where the simple theory ends.