Absorption Cross Section

The absorption cross section is the meeting point of three chapters. From quantum mechanics it inherits the transition dipole moment and Fermi's golden rule, fixing how strongly a transition couples to light. From the Einstein rate equations it inherits the bookkeeping of the $B$ coefficient and detailed balance. And from statistical and collisional physics it inherits the line-shape function that spreads that coupling over frequency. The cross section fuses all three into a single number with units of area — the effective target an atom presents to a beam. Multiply intensity by $\sigma$ for the power one atom removes; divide by $\hbar\omega$ for the photon rate. Two facts make this deceptively simple quantity remarkable. First, on resonance a lifetime-broadened two-level atom reaches $\sigma_0=3\lambda^2/2\pi$ — an area set by the wavelength, not the electron cloud, and for visible light some ten million times larger than the atom itself: correctly tuned, a single atom is a giant to a photon. Second, however the line is broadened — by finite lifetime, collisions, or Doppler motion — the area under $\sigma(\omega)$ never changes, nailed to the oscillator strength by the Thomas–Reiche–Kuhn sum rule; broadening only redistributes it, so a hotter, wider line has a lower peak. Built from all three of its parents — which turn out to agree — the cross section then runs out to the Beer–Lambert law a spectrometer reads.

Key takeaways

The absorption cross section $\sigma(\omega)$ is the effective area an atom shows to a beam: absorbed power is $\sigma I$ , photon absorption rate $\sigma I/\hbar\omega$ .
It factorizes as $\sigma(\omega)=\sigma_{\mathrm{int}}\,g(\omega)$ — an integrated strength times a normalized line shape carrying all the frequency dependence.
The area $\int\sigma\,\mathrm{d}\omega=(\pi e^2/2\varepsilon_0 m c)\,f=2\pi^2 r_e c\,f$ is fixed by the oscillator strength alone, independent of how the line is broadened — the Thomas–Reiche–Kuhn sum rule.
On resonance a purely lifetime-broadened two-level atom reaches $\sigma_0=3\lambda^2/2\pi$ , of order the wavelength squared and vastly larger than the atomic size.
The same $\sigma$ links the microscopic dipole moment and Einstein $B$ coefficient to the macroscopic Beer–Lambert coefficient $\alpha=n\sigma$ a spectrometer measures.
Key idea: write $\sigma(\omega)=\sigma_{\mathrm{int}}\,g(\omega)$ — the area $\sigma_{\mathrm{int}}=\pi e^2 f/(2\varepsilon_0 m c)$ is fixed by the oscillator strength and broadening-independent, while $g(\omega)$ merely redistributes it; on resonance a natural-width two-level atom reaches $\sigma_0=3\lambda^2/2\pi$ .

The core insight

Write $\sigma(\omega)=\sigma_{\mathrm{int}}\,g(\omega)$ : the integrated area $\sigma_{\mathrm{int}}=\pi e^2 f/(2\varepsilon_0 m c)$ is fixed by the oscillator strength and is independent of the broadening mechanism, while the line shape $g(\omega)$ only redistributes that fixed area — so on resonance a natural-width two-level atom reaches $\sigma_0=3\lambda^2/2\pi$ , an area set by the wavelength alone.

Where this lives today

The cross section is the working datum of every quantitative absorption measurement. A tunable-diode-laser trace-gas sensor flown to map methane and CO $_2$ reports a column density only because the molecular $\sigma(\omega)$ for each line is tabulated in databases like HITRAN; the instrument measures $\tau_{\mathrm{opt}}=n\sigma L$ and divides out $\sigma$ to get $n$ . Cavity ring-down spectroscopy pushes this to extremes, folding a laser through a cell tens of thousands of times so that cross sections of $10^{-23}$ cm $^2$ and column densities of parts per trillion become readable from the decay of a leaking optical cavity. At the cold, single-particle frontier the resonant value $\sigma_0=3\lambda^2/2\pi$ is a design parameter, not a curiosity. Because a single atom's cross section rivals a diffraction-limited focal spot of area $\sim\lambda^2$ , a tightly focused beam can be measurably extinguished by one atom or molecule — the basis of single-molecule extinction microscopy and of photon-atom coupling in cavity and waveguide QED. The same $\sigma_0$ and its saturation intensity set the scattering-force ceiling ( $\gamma/2$ ) behind every magneto-optical trap and atomic clock; and in astrophysics the oscillator strengths inside $\sigma$ turn stellar line depths into elemental abundances through the curve of growth — the cross section quietly converting starlight into chemistry.

The physical picture

Think of each atom as a tiny antenna tuned to its resonance $\omega_0$ . A resonant antenna intercepts power not just from its physical footprint but from a region of order one wavelength across — which is why $\sigma_0\sim\lambda^2$ and not the atomic radius squared. The cross section is the effective disk that, absorbing every photon crossing it, would remove power at the atom's true rate: on resonance that disk is enormous; a few linewidths off resonance the tuning fails and it collapses.

Figure

A cross section is an area — and a resonant atom's is astonishingly large. Photons crossing the amber disk of area σ₀ are absorbed the rest sail past. For the sodium D line σ₀ = 3λ²/2π ≈ 1.66×10⁻¹³ m² dwarfs the geometric πa₀² ≈ 8.8×10⁻²¹ m² by ≈ 2×10⁷

The second idea is a conservation law. The area under $\sigma(\omega)$ counts how many electrons participate and how strongly they are bound — the oscillator strength — and no broadening can change it. Finite lifetime, collisions, or the Doppler smear of thermal motion each widens the line, but only by spreading a fixed area more thinly, so the peak drops in exact proportion — a cold sharp line and a hot broad line of the same transition remove the same total photons integrated over frequency. The figure below is the whole story on one plot: two wildly different shapes, identical enclosed area.

Figure

Broadening conserves area not height. A narrow natural Lorentzian of peak σ₀ = 1.66×10⁻¹³ m² and a Doppler Gaussian ≈ 145× broader enclose the same area σ_int = π r_e c f so the hot line's peak collapses to ≈ 1.1×10⁻¹⁵ m²

Finally, the single-atom cross section is the microscopic seed of the macroscopic Beer–Lambert law: multiply $\sigma$ by the number density $n$ and a beam decays as $I(z)=I_0e^{-n\sigma z}$ . Measuring a vapour's transmission therefore measures $\sigma$ directly — the bridge from lab bench back to dipole matrix element.

The core equations

The starting point is operational and needs no quantum mechanics: $\sigma$ is defined as the proportionality between the power an atom absorbs and the intensity driving it. Read as an area, it is the geometric target whose interception rate equals the true absorption rate.

P_{\mathrm{abs}}=\sigma(\omega)\,I,\qquad R_{\mathrm{abs}}=\frac{\sigma(\omega)\,I}{\hbar\omega} \qquad (1)

where: $P_{\mathrm{abs}}$ — power absorbed by one atom (W); $\sigma(\omega)$ — cross section (m $^2$ ); $I$ — incident intensity (W·m $^{-2}$ ); $R_{\mathrm{abs}}$ — photon absorption rate (s $^{-1}$ ); $\hbar\omega$ — photon energy (J). Reading: area times energy flux is a power; dividing by $\hbar\omega$ gives a count rate. Because absorption is resonant, all the frequency dependence lives in a normalized line shape and the strength in a single number out front. This is the organizing move of the subject: separate how much from where.

\sigma(\omega)=\sigma_{\mathrm{int}}\,g(\omega),\qquad \int g(\omega)\,\mathrm{d}\omega=1 \qquad (2)

where: $\sigma_{\mathrm{int}}$ — frequency-integrated cross section (m $^2\cdot$ s $^{-1}$ ); $g(\omega)$ — normalized line shape (units of time), Lorentzian for lifetime or pressure broadening, Gaussian for Doppler, Voigt for both. Reading: since $g$ integrates to one, $\int\sigma\,\mathrm{d}\omega=\sigma_{\mathrm{int}}$ — the peak depends on shape, the area does not. That integrated strength is dictated by the oscillator strength $f$ , the transition measured against one classical electron oscillator. The result is a pure combination of fundamental constants times $f$ , independent of broadening.

\sigma_{\mathrm{int}}=\int\sigma(\omega)\,\mathrm{d}\omega=\frac{\pi e^2}{2\varepsilon_0 m c}\,f=2\pi^2 r_e c\,f \qquad (3)

where: $f$ — absorption oscillator strength (dimensionless, $\le 1$ per transition); $e,m,\varepsilon_0,c$ — elementary charge, electron mass, vacuum permittivity, speed of light; $r_e=e^2/4\pi\varepsilon_0 m c^2=2.818\times10^{-15}$ m — classical electron radius. Reading: the constant $2\pi^2 r_e c=1.67\times10^{-5}$ m $^2\cdot$ s $^{-1}$ (equivalently $\int\sigma\,\mathrm{d}\nu=\pi r_e c\,f=2.65\times10^{-2}f$ cm $^2\cdot$ Hz) is one classical electron's absorbing power, scaled by $f$ . The same $\sigma$ can be read off the Einstein $B$ coefficient, the rate-equation language of absorption. Equating the microscopic photon-absorption rate with the Einstein rate ties $\sigma$ to $B_{12}$ .

\sigma(\omega)=\frac{\hbar\omega\,B_{12}}{c}\,g(\omega) \qquad (4)

where: $B_{12}$ — Einstein absorption coefficient (energy-density-per-angular-frequency convention). Reading: $B_{12}$ carries the strength, $c$ converts energy density to flux, $\hbar\omega$ restores the per-photon energy — the same transition in different books. Descending further, $\sigma$ follows from the transition dipole moment through Fermi's golden rule, with a factor $1/3$ from averaging over random atomic orientation. This is the most microscopic form; everything above repackages it.

\sigma(\omega)=\frac{\pi\omega}{3\varepsilon_0\hbar c}\,|\mathbf{d}_{ba}|^2\,g(\omega) \qquad (5)

where: $\mathbf{d}_{ba}$ — transition dipole moment between lower state $a$ and upper state $b$ ; $\hbar$ — reduced Planck constant. Reading: coupling enters as $|\mathbf d_{ba}|^2$ (a rate goes as amplitude squared), the $1/3$ is the orientation average $\langle(\hat\varepsilon\cdot\mathbf d_{ba})^2\rangle=|\mathbf d_{ba}|^2/3$ , and $g(\omega)$ sets the frequency. Evaluate this on resonance for an atom broadened only by spontaneous emission and every material constant cancels. What survives is a purely geometric peak — one of the most striking results in light–matter physics.

\sigma_0=\frac{3\lambda^2}{2\pi}=\frac{6\pi c^2}{\omega_0^2} \qquad (6)

where: $\lambda=2\pi c/\omega_0$ — resonant wavelength; $\omega_0$ — transition angular frequency. Reading: the resonant cross section knows only the wavelength — not the dipole moment, not $\gamma$ — since peak and width both scale with the same matrix element and cancel. The line shape producing that peak, for lifetime or collision broadening, is the Lorentzian. Its peak value $2/\pi\gamma$ is what makes equation (6) come out geometric.

g_L(\omega)=\frac{\gamma/2\pi}{(\omega-\omega_0)^2+(\gamma/2)^2},\qquad g_L(\omega_0)=\frac{2}{\pi\gamma} \qquad (7)

where: $\gamma$ — full width at half maximum (rad·s $^{-1}$ ); for natural broadening $\gamma=A_{21}=1/\tau$ ; $\tau$ — upper-state radiative lifetime. Reading: a narrower line (small $\gamma$ ) has a taller peak $2/\pi\gamma$ , conserving the unit area. Finally, the macroscopic payoff. Density times cross section is the Beer–Lambert coefficient, and a cell's transmission is an exponential in the optical depth $n\sigma L$ .

\alpha(\omega)=n\,\sigma(\omega),\qquad I(z)=I_0\,e^{-\alpha z} \qquad (8)

where: $\alpha$ — absorption coefficient (m $^{-1}$ ); $n$ — number density of absorbers (m $^{-3}$ ); $z$ — propagation distance. Reading: the dimensionless optical depth $\tau_{\mathrm{opt}}=n\sigma L$ sets transparency versus opacity.

Derivation AThe resonant cross section is purely geometric

The goal is to prove the flagship result $\sigma_0=3\lambda^2/2\pi$ from three ingredients — the dipole cross section (5), the Lorentzian peak (7), and the Wigner–Weisskopf spontaneous-emission rate — showing that when the width comes only from radiative decay, the dipole moment and the linewidth cancel completely, leaving an area fixed by the wavelength alone — which is why a resonant atom is a giant to a photon however weakly it is bound.

Step 1. Evaluate the dipole cross section on resonance: $\sigma(\omega_0)=\dfrac{\pi\omega_0}{3\varepsilon_0\hbar c}\,|\mathbf d_{ba}|^2\,g_L(\omega_0)$ . Why: line centre is where $\sigma$ peaks, and the golden-rule prefactor and line-shape peak are the only frequency-dependent pieces.

Step 2. Insert the Lorentzian peak $g_L(\omega_0)=2/\pi\gamma$ . Why: for a lifetime-broadened line the whole area is squeezed into a Lorentzian of width $\gamma$ whose height goes as $1/\gamma$ — the seed of the cancellation.

Step 3. Bring in the spontaneous-emission rate, $\gamma=A_{21}=\dfrac{\omega_0^3|\mathbf d_{ba}|^2}{3\pi\varepsilon_0\hbar c^3}$ . Why: for pure natural broadening the width is not independent — it is set by the same matrix element that sets the strength, which makes them cancel.

Step 4. Solve for the matrix element, $|\mathbf d_{ba}|^2=\dfrac{3\pi\varepsilon_0\hbar c^3\gamma}{\omega_0^3}$ , and substitute into Step 1. Why: trading $|\mathbf d_{ba}|^2$ for an expression in $\gamma$ cancels both the dipole moment and the width in one stroke.

Step 5. Collect: $\sigma(\omega_0)=\dfrac{\pi\omega_0}{3\varepsilon_0\hbar c}\cdot\dfrac{3\pi\varepsilon_0\hbar c^3\gamma}{\omega_0^3}\cdot\dfrac{2}{\pi\gamma}=\dfrac{2\pi c^2}{\omega_0^2}$ — every material constant and $\gamma$ cancels. Why: only $c$ and $\omega_0$ survive, so the result is an area built from the wavelength; the physics has become geometry.

Step 6. Convert with $\omega_0=2\pi c/\lambda$ : $\sigma_0=\dfrac{2\pi c^2}{\omega_0^2}\cdot 3=\dfrac{3\lambda^2}{2\pi}$ , the factor $3$ for light polarized along the dipole (no orientation average, the ideal two-level case). Why: this is Foot's Eq. (7.24); a real degenerate transition multiplies by a statistical-weight ratio, but the wavelength scale is universal.

Go deeper · Derivation B · 1 minThe integrated cross section and the sum rule

Derivation BThe integrated cross section and the sum rule

The goal is to show that the area under $\sigma(\omega)$ is an invariant — untouched by any broadening mechanism — equal to a universal constant times the oscillator strength, then to sum over final states for the Thomas–Reiche–Kuhn rule capping the electrons' total absorbing power. This is the formal statement behind "broadening conserves area, not height," and it lets one tabulated $f$ -value predict absorption in a cold trap, a hot flame, or a stellar atmosphere alike.

Step 1. Integrate the factorized form: $\int\sigma\,\mathrm{d}\omega=\sigma_{\mathrm{int}}\int g\,\mathrm{d}\omega=\sigma_{\mathrm{int}}$ , since $g$ is normalized to unity. Why: the line shape carries all the broadening and integrates to one whether Lorentzian, Gaussian, or Voigt — so the area cannot depend on the mechanism.

Step 2. Bring in the oscillator strength, $f=\dfrac{2m\omega_0}{3\hbar e^2}|\mathbf d_{ba}|^2$ , the ratio of the transition's strength to one classical electron. Why: $f$ is the natural currency because it strips out the transition-specific $|\mathbf d_{ba}|^2$ , leaving a pure number of order one.

Step 3. Substitute into the integrated dipole cross section: $\sigma_{\mathrm{int}}=\dfrac{\pi\omega_0}{3\varepsilon_0\hbar c}|\mathbf d_{ba}|^2=\dfrac{\pi e^2}{2\varepsilon_0 m c}\,f$ . Why: this is the same universal constant $2\pi^2 r_e c=1.67\times10^{-5}$ m $^2\cdot$ s $^{-1}$ for every line — the atom's identity survives only through $f$ and $\omega_0$ .

Step 4. Sum over upper states: $\sum_b f_{ba}=Z_{\mathrm{eff}}$ , so all lines together integrate to $\dfrac{\pi e^2}{2\varepsilon_0 m c}\,Z$ . Why: the Thomas–Reiche–Kuhn sum rule says one electron's oscillator strengths add to one (and to $Z$ for $Z$ electrons), capping the atom's total absorbing power.

Go deeper · Derivation C · 1 minFrom one atom to Beer–Lambert

Derivation CFrom one atom to Beer–Lambert

The goal is to descend from the single-atom cross section to the exponential law a spectrophotometer records, tracking photon flux through a slab of absorbers using nothing but photon conservation. The payoff $\alpha=n\sigma$ turns a bench measurement of transmission and column density into a direct read-out of $\sigma$ — closing the loop from dipole matrix element back to lab data.

Step 1. Convert intensity to a photon flux, $\Phi=I/\hbar\omega$ . Why: absorption is a counting process — atoms remove photons one at a time — so flux, not energy, is the natural variable.

Step 2. A slab $\mathrm{d}z$ holds $n\,\mathrm{d}z$ atoms per unit area, each removing photons at rate $\sigma\Phi$ , so $\mathrm{d}\Phi=-n\sigma\,\Phi\,\mathrm{d}z$ . Why: dilute atoms do not shadow one another, so their capture areas add.

Step 3. Integrate: $\Phi(z)=\Phi_0\,e^{-n\sigma z}$ , hence $\alpha=n\sigma$ and $\tau_{\mathrm{opt}}=n\sigma L$ . Why: constant fractional loss per length is the signature of an exponential; inverting the optical depth $n\sigma L$ recovers $\sigma$ from measured transmission and known column density.

Limiting cases

Case	Result	Meaning
Exact resonance, natural width only	$\sigma_0=\dfrac{3\lambda^2}{2\pi}$	Maximum for an ideal non-degenerate two-level atom broadened only by spontaneous emission; independent of dipole strength since peak and width scale together
Far off resonance	$\sigma(\omega)\approx\dfrac{\sigma_0}{4}\left(\dfrac{\gamma}{\omega-\omega_0}\right)^2$	Lorentzian wings fall as inverse-square detuning; the atom is nearly transparent far from line centre
Doppler-broadened hot vapour	$\sigma_{\mathrm{peak}}\sim\sigma_0\,\dfrac{\gamma}{\Delta\omega_D}$	For $\Delta\omega_D\gg\gamma$ the fixed area spreads over a wider Gaussian, cutting the peak by $\sim\gamma/\Delta\omega_D$
Integrated over all frequencies	$\displaystyle\int_0^\infty\sigma\,\mathrm{d}\omega=\dfrac{\pi e^2}{2\varepsilon_0 m c}\,f$	Fixed by oscillator strength and constants only — the broadening-independent TRK invariant

Common trapA broader absorption line means the atom absorbs more light.

It does not — not integrated over frequency. Broadening never adds absorbing power; the area under $\sigma(\omega)$ is nailed to the oscillator strength by the sum rule, so a wider line is a shorter line by the same factor. Confusing the peak cross section (which falls as the line broadens) with the integrated one (which cannot change) is the single most common error here. Two traps ride along. First, factors of $2\pi$ : a cross section against angular frequency $\omega$ differs from one against ordinary frequency $\nu$ , since $g(\omega)\,\mathrm{d}\omega=g(\nu)\,\mathrm{d}\nu$ — check which you are handed. Second, $\sigma_0=3\lambda^2/2\pi$ is not universal: it holds only for an ideal, purely lifetime-broadened, non-degenerate two-level atom on resonance with correctly polarized light. Real degeneracy, collisions, and Doppler motion all pull the peak below it.

Worked examples

1 · Sodium D line — resonant cross section and saturation intensity (the concrete anchor)

The sodium D $_2$ line has $\lambda=589$ nm and upper-state lifetime $\tau=16.2$ ns. Find $\sigma_0$ , compare it to the geometric atomic area, and compute the saturation intensity $I_{\mathrm{sat}}=\hbar\omega\gamma/2\sigma_0$ at which absorption bleaches.

Step 1. Resonant cross section, equation (6): $\sigma_0=\dfrac{3\lambda^2}{2\pi}=\dfrac{3(589\times10^{-9})^2}{2\pi}=\mathbf{1.66\times10^{-13}\ m^2}=1.66\times10^{-9}$ cm $^2$ . Against a geometric atomic area $\pi a_0^2\approx 8.8\times10^{-21}$ m $^2$ this is larger by $\approx\mathbf{1.9\times10^{7}}$ — twenty million times bigger to a resonant photon.

Step 2. Natural linewidth: $\gamma=1/\tau=6.17\times10^{7}$ s $^{-1}$ (FWHM $\gamma/2\pi\approx 9.8$ MHz). Photon energy: $\hbar\omega=hc/\lambda=3.37\times10^{-19}$ J ( $\approx 2.10$ eV).

Step 3. Saturation intensity: $I_{\mathrm{sat}}=\dfrac{\hbar\omega\,\gamma}{2\sigma_0}=\dfrac{(3.37\times10^{-19})(6.17\times10^{7})}{2(1.66\times10^{-13})}=63\ \mathrm{W\,m^{-2}}=\mathbf{6.3\ mW\,cm^{-2}}$ , matching the accepted sodium value ( $\approx 6.26$ ). It marks where the scattering rate saturates at its ceiling $\gamma/2$ — the number setting a magneto-optical trap's fluorescence.

2 · Integrated strength versus Doppler peak — the sum rule in action

The sodium D $_2$ transition has oscillator strength $f\approx 0.641$ . Find the integrated cross section, then the peak in a $350$ K vapour cell where Doppler broadening dominates, and confirm the peak has fallen by the Doppler-to-natural width ratio.

Step 1. Integrated cross section, equation (3) in frequency form: $\int\sigma\,\mathrm{d}\nu=\pi r_e c\,f=(2.65\times10^{-2})(0.641)=1.70\times10^{-2}$ cm $^2\cdot$ Hz $=1.70\times10^{-6}$ m $^2\cdot$ Hz — fixed, shared exactly by the trap-cold and hot-cell lines.

Step 2. Doppler FWHM at $T=350$ K, $M=23$ u: $\Delta\nu_D=\nu_0\sqrt{8k_BT\ln2/Mc^2}$ . With $\nu_0=c/\lambda=5.09\times10^{14}$ Hz this is $\Delta\nu_D\approx\mathbf{1.42\ GHz}$ , $145\times$ the $9.8$ MHz natural FWHM.

Step 3. Normalized-Gaussian peak $g(\nu_0)=(2/\Delta\nu_D)\sqrt{\ln2/\pi}=0.939/\Delta\nu_D$ , so $\sigma_{\mathrm{peak}}=\int\sigma\,\mathrm{d}\nu\cdot g(\nu_0)=(1.70\times10^{-6})(0.939)/(1.42\times10^{9})=\mathbf{1.1\times10^{-15}\ m^2}$ . Against the natural peak $1.66\times10^{-13}$ m $^2$ this is smaller by $\approx 150$ — the same factor as the width ratio, exactly as the sum rule demands: the area held, so the height fell.

3 · Optical depth of a vapour cell

A $5$ cm sodium cell holds $n=10^{16}$ m $^{-3}$ atoms at $350$ K. Using the Doppler-limited peak from example 2, find the on-resonance optical depth and transmission.

Step 1. Optical depth, equation (8): $\tau_{\mathrm{opt}}=n\sigma_{\mathrm{peak}}L=(10^{16})(1.1\times10^{-15})(0.05)=\mathbf{0.56}$ , so transmission $e^{-0.56}=\mathbf{0.57}$ — the cell removes about $43\%$ of resonant light in one pass at this modest density.

Step 2. With the natural peak $\sigma_0$ instead, $\tau_{\mathrm{opt}}\approx 83$ — utterly opaque. This is why alkali spectroscopy runs off line centre, at low density, or Doppler-free.

Physics of the portrait

The absorption cross section is where nineteenth-century classical dispersion theory made peace with the quantum atom — and the man who brokered it gave us the number still at the centre of the formula: the oscillator strength.

Origins · 1 minRudolf Ladenburg and how this was discovered

Rudolf Ladenburg

1882–1952 · the f-value linking classical dispersion to quantum absorption

Rudolf Ladenburg, a German-American physicist trained under Röntgen, solved a puzzle that had nagged spectroscopy since Drude and Lorentz: the classical dispersion formula predicted absorption as if each atom carried some integer number of electron oscillators, yet measured strengths were fractional and varied wildly from line to line. In 1921 Ladenburg reinterpreted that count quantum-mechanically — the effective "number of dispersion electrons" became the oscillator strength $f$ , the ratio of a real transition's absorbing power to one classical electron's. This is precisely the $f$ in $\sigma_{\mathrm{int}}=(\pi e^2/2\varepsilon_0 m c)\,f$ , letting the Einstein coefficients, the dipole matrix element, and the old dispersion constant all sit on the same page. In 1926, with Hans Kopfermann, he measured negative dispersion in excited neon — the first laboratory fingerprint of stimulated emission, two years before its full quantum treatment and decades before the laser. The Thomas–Reiche–Kuhn sum rule (1925) capped his programme, proving the oscillator strengths he measured must add to the number of electrons. Emigrating to Princeton in 1932, Ladenburg left the cross section its enduring shape: a universal constant, an atom-specific $f$ , and a line shape.

History

1729

— Bouguer describes the exponential attenuation of light through an absorbing medium

1852

— Beer ties the absorption coefficient to the density of absorbers

1916

— Einstein introduces the A and B coefficients governing absorption and emission

1921

— Ladenburg recasts absorption strength as an oscillator strength, the f-value

1925

— Thomas, Reiche and Kuhn establish the oscillator-strength sum rule

Going further

Texts. Foot, Atomic Physics (2005), §7.3–7.5 and Eq. (7.24) for $\sigma_0=3\lambda^2/2\pi$ · Loudon, The Quantum Theory of Light (3rd ed., 2000), Ch. 1–2 · Demtröder, Laser Spectroscopy (Vol. 1, 4th ed.), §2.7–2.8 · Saleh & Teich, Fundamentals of Photonics (3rd ed.), Ch. 13 · Hecht, Optics (5th ed.), Ch. 3–4 · Milonni & Eberly, Laser Physics (2010), Ch. 3.

Provenance. Equations cross-checked against Foot (§7.3–7.5, $\sigma_0=3\lambda^2/2\pi$ ), Loudon (dipole and $B$ -coefficient forms), Demtröder (§2.7–2.8, Beer–Lambert and integrated cross section), and Saleh & Teich (Ch. 13). Anchor numbers recomputed here: $\sigma_0(\mathrm{Na})=1.66\times10^{-13}$ m $^2$ ; $r_e=2.818\times10^{-15}$ m; the integrated-cross-section constant $\pi e^2/2\varepsilon_0 m c=2\pi^2 r_e c=1.67\times10^{-5}$ m $^2\cdot$ s $^{-1}$ ( $\pi r_e c=2.65\times10^{-2}$ cm $^2\cdot$ Hz per unit $f$ in the $\nu$ -integral); $I_{\mathrm{sat}}(\mathrm{Na\ D_2})=6.3$ mW·cm $^{-2}$ ; Doppler FWHM $1.42$ GHz at $350$ K. A verifier should confirm the $\gamma$ -cancellation in derivation 1 and the sum-rule constant $2\pi^2 r_e c$ .

Prerequisites. Einstein A and B Coefficients · Electric Dipole Approximation · Transition Dipole Moment · Lorentzian Line Shape. Leads to. Resonant Absorption · Oscillator Strength · Thomas–Reiche–Kuhn Sum Rule · Radiative Lifetime and Natural Linewidth.

Ask the tutor: Want me to work out $\sigma_0$ , $I_{\mathrm{sat}}$ , or the optical depth of a cell for a specific transition — say rubidium 780 nm or a molecular line from HITRAN — step by step?

Why does a cross section have units of area if nothing physically that big is blocking the light?

Because "cross section" is defined by the rate of an interaction, not by any solid object. Formally $\sigma$ is the constant making absorbed power equal $\sigma$ times incident intensity, so it carries units of area — the size of a perfectly absorbing disk that removes photons at the same rate the real atom does. On resonance that disk is of order $\lambda^2$ , far larger than the electron cloud, because a resonantly driven atom radiates and re-absorbs like an antenna reaching out a wavelength in every direction. Nothing literally blocks that area — it is a probability per unit flux, dressed as geometry.

How can the resonant cross section not depend on the dipole moment — surely a stronger transition absorbs more?

A stronger transition does absorb more integrated over frequency: $\int\sigma\,\mathrm{d}\omega\propto f\propto|\mathbf d_{ba}|^2$ . But for a purely lifetime-broadened line the peak is a different story: a larger dipole moment makes the transition stronger, yet it also makes the excited state decay faster, widening the line by the same factor — taller area, wider line, same peak. When you evaluate the peak the matrix element cancels between strength and width, leaving $\sigma_0=3\lambda^2/2\pi$ . The cancellation is special to natural broadening; if collisions or Doppler motion set the width, the peak does depend on the dipole moment through the fixed area.

What is the oscillator strength, really, and why is it never bigger than one per transition?

The oscillator strength $f$ compares a real quantum transition to a single classical electron on a spring — the fraction of that classical absorbing power the transition carries. It is capped by a bookkeeping identity: the Thomas–Reiche–Kuhn sum rule proves the oscillator strengths of all transitions from a given state sum to the number of electrons available (one for a single-electron problem). So no line can borrow more than the whole electron's worth, and strong resonance lines like the alkali D lines ( $f\approx0.3$ – $0.7$ ) already claim most of it. This is what makes the integrated cross section a hard invariant rather than a fitted parameter.

If broadening conserves the area, why do spectroscopists care so much about linewidth?

Because most measurements read the peak or the shape, not the area. A Doppler-broadened line has the same integrated strength as a cold one but a peak lower by the width ratio, so a hot cell can look nearly transparent on resonance while a cold one is opaque — worked example 3 gives optical depth 0.6 hot versus 80 cold for the same atoms. Linewidth also sets resolution: two nearby transitions blur together once broadening exceeds their separation, which is why Doppler-free spectroscopy exists to recover the narrow natural line under the Gaussian. The area is invariant, but almost everything you observe depends on how it is distributed.

How does the Beer–Lambert coefficient α relate to the extinction coefficient κ from the refractive index?

They are the same absorption counted two ways. The complex index $\tilde n=n+i\kappa$ describes a bulk medium, its imaginary part giving $\alpha=4\pi\kappa/\lambda_0$ ; the cross-section picture describes the same medium as discrete absorbers, giving $\alpha=n_{\mathrm{atoms}}\sigma$ . Equate them: $\kappa=\lambda_0 n_{\mathrm{atoms}}\sigma/4\pi$ — the extinction coefficient is the cross section repackaged with the number density. The microscopic view is natural for dilute gases and single atoms, the macroscopic view for dense solids and liquids. Kramers–Kronig ties $\kappa$ to the dispersive part $n$ — the statement that the line shape $g(\omega)$ has a dispersive partner.

Does the same cross section govern stimulated emission and laser gain?

Yes — one of the deep economies of the two-level atom. The cross section for a lower-state atom absorbing a photon also describes, sign flipped, an upper-state atom emitting one by stimulation, since $B_{12}$ and $B_{21}$ are equal (up to degeneracy factors). In a medium with populations $N_1,N_2$ the net gain coefficient is $\gamma_{\mathrm{gain}}=(N_2-N_1)\sigma$ : absorption when the lower state wins, optical gain when an inversion makes $N_2>N_1$ . And the saturation intensity $I_{\mathrm{sat}}=\hbar\omega\gamma/2\sigma_0$ where absorption bleaches is the very intensity where a laser amplifier saturates — one cross section, absorption and amplification both.