Radiation Pressure and the Scattering Force

Every photon a laser sends toward an atom carries a packet of momentum $\hbar k = h/\lambda$ , and every time the atom absorbs one it is shoved forward by exactly that amount. This article takes two results as given and turns them into a mechanical force. From the quantum optics of the driven two-level atom it borrows the steady-state excited-state population $\rho_{ee}$ — the Lorentzian, power-broadened solution of the optical Bloch equations — and from electromagnetism the fact that a photon of frequency $\omega$ carries momentum $\hbar\omega/c$ . Multiply momentum per photon by the rate photons are scattered, $\Gamma\rho_{ee}$ , and you have the scattering force: the dissipative push light exerts on matter along its travel direction. What makes the force interesting is not that it exists — Kepler guessed at it from comet tails four centuries ago — but that it saturates. An atom must be in its ground state to absorb and wait in its excited state to emit, so it spends at most half its time excited: $\rho_{ee}\le 1/2$ . The scattering rate therefore cannot exceed $\Gamma/2$ , nor the force $\hbar k\Gamma/2$ , however bright the beam. For sodium that ceiling is a deceleration near $10^6\ \mathrm{m/s^2}$ — a hundred thousand times gravity, enough to stop a thousand-metre-per-second thermal beam inside a metre. And because the absorption is sharply resonant, the force is velocity-selective through the Doppler shift — the single fact that seeds laser cooling. Optical molasses, the Zeeman slower, and the magneto-optical trap all grow from the saturating, Doppler-shifted force developed here.

Key takeaways

Each photon carries momentum $\hbar k=h/\lambda$ ; an atom absorbing one recoils by $\hbar k$ along the propagation direction.
Spontaneous re-emission is symmetric over many cycles, so its recoils average to zero — the entire directed force comes from absorption.
The steady-state scattering force is $F=\hbar k\,\Gamma\rho_{ee}$ : photon momentum times the scattering rate $\Gamma\rho_{ee}$ .
Because $\rho_{ee}\le 1/2$ , the force has a hard ceiling $F_{\max}=\hbar k\Gamma/2$ ; for sodium that is a deceleration near $10^6\ \mathrm{m/s^2}$ , about $10^5$ times gravity.
Those same zero-mean recoils have nonzero variance, so scattering also heats through momentum diffusion — this fixes the Doppler cooling limit.
Through the Doppler shift the force is velocity-selective, turning radiation pressure into a coolant in Zeeman slowers, optical molasses, and the magneto-optical trap.
Key idea: the scattering force equals photon momentum times scattering rate, $F=\hbar k\,\Gamma\rho_{ee}$ ; because $\rho_{ee}\le 1/2$ , it saturates at $\hbar k\Gamma/2$ — a fundamental limit on how hard light can push an atom.

The core insight

The scattering force equals photon momentum times scattering rate, $F=\hbar k\,\Gamma\rho_{ee}$ ; because $\rho_{ee}\le 1/2$ the force saturates at $\hbar k\Gamma/2$ , a fundamental limit on how hard resonant light can push an atom.

Where this lives today

The saturating, Doppler-shifted scattering force is the first stage of nearly every cold-atom apparatus. A Zeeman slower uses it to decelerate an atomic beam over about a metre (a tapered field keeping the slowing atoms on resonance); six crossed beams make optical molasses, whose $F=-\alpha v$ drag cools a cloud to the Doppler limit; and a magnetic quadrupole adds the $-\kappa x$ of the magneto-optical trap — the loading stage for atomic clocks, quantum-gas machines, and neutral-atom quantum computers. The same $F=\hbar k R_{\mathrm{sc}}$ physics scales both ways. Upward, radiation pressure supports the envelopes of massive stars, blows cometary tails away from the Sun, and drives proposed light sails. Downward, it appears in optical tweezers — scattering force pushing while the gradient force traps — and in cavity optomechanics, where circulating photons cool micromechanical mirrors toward their quantum ground state. In every case the accounting is the one built here.

The physical picture

Picture the atom sandblasted by a stream of photons that all move the same way. Each absorbed photon shoves it forward by one recoil momentum $\hbar k$ ; the atom then spits a photon back out into a random direction, so those recoils cancel on average. What survives is a steady one-way push — like standing in rain that only ever comes from one side. The whole net force lives in absorption; emission is what makes the process dissipative rather than a coherent, reversible exchange. The force cannot grow without bound. Absorbing requires the ground state, and after absorbing the atom must wait in the excited state before it can absorb again — so it is excited at most half the time and scatters at most about $\Gamma/2$ photons per second. Multiply by $\hbar k$ and you meet the ceiling $F_{\max}=\hbar k\Gamma/2$ , a hard limit no laser intensity can beat.

Figure

An atom absorbs photons only from the beam direction, gaining +ħk each cycle, then re-emits into random directions whose recoils average to zero — for sodium each kick is v_r ≈ 2.9 cm/s, the rate caps at Γ/2 ≈ 3.1×10⁷ per second, and a_max ≈ 9×10⁵ m/s²

The randomness of spontaneous emission is not merely noise to be averaged away: its mean is zero but its variance is not, so the atom random-walks in momentum space, gaining kinetic energy — a heating that fights cooling and sets the Doppler limit. And because an atom absorbs strongly only on resonance in its own frame, tuning the laser slightly below resonance makes the force strongest for atoms heading into the beam; add a counter-propagating beam and the net force opposes velocity. The second figure plots that molasses force, the object laser cooling is built on.

Figure

Two red-detuned counter-propagating beams with δ = −Γ/2 and s₀ = 2 give a net force nearly linear F ≈ −αv near v = 0 (cooling) and peaking near kv ≈ Γ — for sodium kv/Γ = 1 is v ≈ 5.8 m/s

The core equations

The starting point is at once kinematic and quantum: light comes in indivisible quanta, each carrying momentum inversely proportional to its wavelength — the de Broglie relation read for light.

p_{\gamma}=\hbar k=\frac{h}{\lambda}=\frac{\hbar\omega}{c} \qquad (1)

where: $p_\gamma$ — momentum of one photon; $\hbar$ — reduced Planck constant; $k=2\pi/\lambda$ — wavenumber; $\lambda,\ \omega$ — wavelength and angular frequency; $c$ — speed of light. Physical reading: a visible photon carries only $\sim10^{-27}\ \mathrm{kg\,m/s}$ — the force comes from delivering that tiny kick millions of times a second. The force is that momentum delivered at the rate photons are scattered: absorption supplies the directed momentum, and everything hard is buried in the rate, supplied by the optical Bloch equations.

F_{\mathrm{sc}}=\hbar k\,R_{\mathrm{sc}}=\hbar k\,\Gamma\,\rho_{ee} \qquad (2)

where: $F_{\mathrm{sc}}$ — scattering force; $R_{\mathrm{sc}}=\Gamma\rho_{ee}$ — photon scattering rate; $\Gamma$ — natural linewidth (spontaneous decay rate); $\rho_{ee}$ — steady-state excited-state population. Physical reading: the force is a rate of momentum delivery — whatever caps the rate caps the force. The scattering rate is $\Gamma$ times the fraction of time the atom is excited — a power-broadened Lorentzian in detuning, linear in intensity when weak and clamped at $1/2$ when strong. This is the single most important line here.

\rho_{ee}=\frac{1}{2}\,\frac{s_0}{1+s_0+(2\delta/\Gamma)^2} \qquad (3)

where: $s_0=I/I_{\mathrm{sat}}$ — on-resonance saturation parameter; $\delta=\omega_L-\omega_0$ — laser detuning; $I,\ I_{\mathrm{sat}}$ — laser and saturation intensity. Physical reading: as $s_0\to\infty$ the population saturates at $\rho_{ee}\to1/2$ ; it is Lorentzian in $\delta$ , power-broadened beyond width $\Gamma$ . Substituting the population into the force gives the complete result — intensity through $s_0$ and colour through $\delta$ multiplying a fixed maximum.

F_{\mathrm{sc}}=\frac{\hbar k\Gamma}{2}\,\frac{s_0}{1+s_0+(2\delta/\Gamma)^2} \qquad (4)

where: $\hbar k\Gamma/2$ — the maximum force $F_{\max}$ ; the fraction is a saturation- and detuning-dependent factor never exceeding one. Physical reading: no intensity and detuning push past $\hbar k\Gamma/2$ , and being off resonance only makes it smaller. Setting the fraction to its ceiling gives the maximum force and, dividing by mass, the maximum deceleration — set by two atomic constants and enormous.

F_{\max}=\frac{\hbar k\Gamma}{2},\qquad a_{\max}=\frac{\hbar k\Gamma}{2M}=\frac{v_{\mathrm{r}}\Gamma}{2} \qquad (5)

where: $M$ — atomic mass; $a_{\max}$ — maximum deceleration; $v_{\mathrm{r}}=\hbar k/M$ — single-photon recoil velocity. Physical reading: with $v_{\mathrm{r}}$ a few cm/s and $\Gamma\sim10^8\ \mathrm{s^{-1}}$ for alkalis, $a_{\max}\sim10^5$ – $10^6\ \mathrm{m/s^2}$ , up to $10^5\,g$ . The saturation intensity, fixed by the transition wavelength and linewidth, is the natural yardstick for laser power in cooling experiments.

I_{\mathrm{sat}}=\frac{\pi h c\,\Gamma}{3\lambda^3}=\frac{\hbar\omega_0^3\Gamma}{12\pi c^2} \qquad (6)

where: $I_{\mathrm{sat}}$ — intensity at which $s_0=1$ , so $\rho_{ee}=1/4$ ; $\lambda,\ \omega_0$ — transition wavelength and resonance frequency. Physical reading: for sodium D2, $I_{\mathrm{sat}}\approx6.3\ \mathrm{mW/cm^2}$ — cooling needs a resonant laser, not a powerful one. For a moving atom the relevant detuning is the one in its own frame, shifted by $kv$ ; replacing $\delta$ by $\delta-kv$ makes the force velocity-dependent — the mechanism of Doppler cooling.

F_{\mathrm{sc}}(v)=\frac{\hbar k\Gamma}{2}\,\frac{s_0}{1+s_0+\big(2(\delta-kv)/\Gamma\big)^2} \qquad (7)

where: $v$ — atomic velocity along the beam; $kv$ — Doppler shift; $\delta-kv$ — effective detuning in the atom's frame. Physical reading: red detuning ( $\delta<0$ ) tunes an atom moving into the beam toward resonance, so it scatters harder from the beam that opposes it. The zero-mean recoils still diffuse the momentum, $\langle\Delta p^2\rangle=2D_p t$ with $D_p\simeq(\hbar k)^2 R_{\mathrm{sc}}$ , the heating that fixes the Doppler floor (Derivation 2).

Derivation AThe scattering force from the optical Bloch equations

The goal is to obtain the full steady-state force $F=\tfrac{\hbar k\Gamma}{2}\,s_0/[1+s_0+(2\delta/\Gamma)^2]$ from the density-matrix dynamics of a two-level atom (ground $|g\rangle$ , excited $|e\rangle$ split by $\hbar\omega_0$ , decay rate $\Gamma$ ) driven with Rabi frequency $\Omega$ at detuning $\delta=\omega_L-\omega_0$ . The plan: write the optical Bloch equations, impose steady state, solve for the excited population, repackage the drive as the saturation parameter, and multiply by $\hbar k\Gamma$ .

Step 1. In the rotating frame the excited population obeys $\dot\rho_{ee}=-\Gamma\rho_{ee}+\tfrac{i\Omega}{2}(\tilde\rho_{ge}-\tilde\rho_{eg})$ , with the coherences $\tilde\rho_{eg}$ decaying at $\Gamma/2$ . Why: population is fed only from the coherence and leaves only by decay, so the two must be solved together.

Step 2. Set every time derivative to zero, $\dot\rho_{ee}=\dot{\tilde\rho}_{eg}=0$ , and solve the linear system. Why: the force is a continuous-wave quantity — switch-on transients have died away, so only the fixed point matters.

Step 3. The fixed point gives $\rho_{ee}=\dfrac{\Omega^2/4}{\delta^2+\Gamma^2/4+\Omega^2/2}$ , a Lorentzian in $\delta$ whose width grows with the drive. Why: the $\Omega^2/2$ term is power broadening — driving harder widens as well as raises the resonance.

Step 4. Define the on-resonance saturation parameter $s_0\equiv 2\Omega^2/\Gamma^2 = I/I_{\mathrm{sat}}$ . Why: $\Omega$ is not what a lab measures, but $I/I_{\mathrm{sat}}$ is — one dimensionless number for the drive.

Step 5. In terms of $s_0$ , $\rho_{ee}=\tfrac{1}{2}\,s_0/[1+s_0+(2\delta/\Gamma)^2]$ , tending to $1/2$ as $s_0\to\infty$ . Why: half the population is the most a two-level atom holds, however bright the light.

Step 6. Multiply by $\hbar k\Gamma$ : $F_{\mathrm{sc}}=\hbar k\Gamma\rho_{ee}=\tfrac{\hbar k\Gamma}{2}\,s_0/[1+s_0+(2\delta/\Gamma)^2]$ . Why: each scattered photon delivers one $\hbar k$ , so force is momentum times rate $\Gamma\rho_{ee}$ , and the ceiling on $\rho_{ee}$ becomes the ceiling $\hbar k\Gamma/2$ . *Verified: reproduces Foot, Atomic Physics eq. (9.3) and Metcalf and van der Straten eq. (3.7); $F_{\max}=\hbar k\Gamma/2$ is standard.*

Go deeper · Derivation B · 1 minWhy spontaneous emission gives zero mean force but nonzero heating

Derivation BWhy spontaneous emission gives zero mean force but nonzero heating

The goal is to show why the directed force lives entirely in absorption while emission still matters — contributing no net force but real heating. Count $N$ cycles, treating absorption momentum as deterministic ( $+\hbar k$ along the beam) and emission momentum as random over the symmetric dipole pattern. Separating the mean from the variance turns "spontaneous emission averages away" into two statements: a vanishing first moment and a growing second moment.

Step 1. The absorption momentum after $N$ cycles is $\mathbf p_{\mathrm{abs}}=N\hbar k\,\hat z$ , exactly directed. Why: all absorbed photons come from the same beam, so their kicks add — the coherent, directed force.

Step 2. Each emission recoil is $\mathbf p_i=\hbar k\,\hat n_i$ with $\hat n_i$ random from the dipole pattern and $\langle\hat n_i\rangle=0$ . Why: the pattern is symmetric under $\hat n\to-\hat n$ , so the mean recoil vanishes.

Step 3. Summing, $\big\langle\sum_i\mathbf p_i\big\rangle=\hbar k\sum_i\langle\hat n_i\rangle=0$ . Why: the mean of zero-mean kicks is zero, so the whole directed push is the absorption term from Step 1.

Step 4. The variance does not vanish: $\big\langle(\sum_i\mathbf p_i)^2\big\rangle=\sum_i(\hbar k)^2=N(\hbar k)^2$ . Why: uncorrelated kicks add in quadrature — a random walk in momentum space.

Step 5. Per unit time this is diffusion and heating, $\tfrac{\mathrm d}{\mathrm dt}\langle p^2\rangle\simeq(\hbar k)^2R_{\mathrm{sc}}$ , so $\tfrac{\mathrm dE}{\mathrm dt}=(\hbar k)^2R_{\mathrm{sc}}/2M$ . Why: energy is $p^2/2M$ , so growing variance is literal heating — balanced against cooling, it caps how cold Doppler cooling gets. *Verified: standard in Loudon and Metcalf and van der Straten; absorption adds a diffusion term giving a factor near two in the Doppler temperature.*

Go deeper · Derivation C · 1 minThe velocity-damping coefficient of optical molasses

Derivation CThe velocity-damping coefficient of optical molasses

The goal is to turn the single-beam velocity-dependent force into the viscous drag of optical molasses and show that red detuning is what makes it cool. Place the atom in two equal-intensity counter-propagating beams carrying the force of equation (7) with opposite Doppler signs, add them, and expand to first order in velocity. The linear term defines a damping coefficient $\alpha$ whose sign is set entirely by the detuning.

Step 1. Write $F_{\pm}=\pm\tfrac{\hbar k\Gamma}{2}\,s_0/[1+s_0+(2(\delta\mp kv)/\Gamma)^2]$ : the $+z$ beam sees effective detuning $\delta-kv$ , the $-z$ beam $\delta+kv$ . Why: an atom moving along $+z$ runs toward the $-z$ beam and away from the $+z$ beam, so the two are Doppler-shifted oppositely.

Step 2. At $v=0$ the forces cancel; expanding to first order gives $F=F_++F_-\approx \dfrac{8\hbar k^2\,\delta\,s_0}{\Gamma\,[1+s_0+(2\delta/\Gamma)^2]^2}\,v$ . Why: a stationary atom feels no net push, so the leading effect is linear in $v$ — a drag or anti-drag depending on the sign.

Step 3. This is $F=-\alpha v$ with $\alpha=-\dfrac{8\hbar k^2\,\delta\,s_0}{\Gamma\,[1+s_0+(2\delta/\Gamma)^2]^2}$ ; for $\delta<0$ , $\alpha>0$ — genuine viscous damping. Why: only red detuning makes the atom scatter preferentially from the beam it moves toward, so $F$ opposes $v$ ; blue flips the sign and heats. The magnitude peaks near $\delta\approx-\Gamma/2$ , the operating point plotted above. *Verified: reproduces the damping coefficient in Foot Ch. 9 and Metcalf and van der Straten Ch. 3; the optimum near $\delta=-\Gamma/2$ and the $-\alpha v$ form underlie optical molasses.*

Limiting cases

How the steady-state force $F=\tfrac{\hbar k\Gamma}{2}\,s_0/[1+s_0+(2\delta/\Gamma)^2]$ behaves in the regimes that matter ( $s_0=I/I_{\mathrm{sat}}$ , $\delta$ the detuning):

Case	Result	Meaning
Low intensity, on resonance ( $s_0\ll1,\ \delta=0$ )	$F\approx\frac{\hbar k\Gamma}{2}\,s_0=\frac{\hbar k\Gamma}{2}\frac{I}{I_{\mathrm{sat}}}$	linear in intensity — photons scatter independently, rate $\propto I$
High intensity ( $s_0\to\infty$ )	$F\to F_{\max}=\frac{\hbar k\Gamma}{2}$	clamps: $\rho_{ee}\to1/2$ , rate $\to\Gamma/2$ ; more light buys nothing
Large detuning ( $\lvert\delta\rvert\gg\Gamma$ , low $I$ )	$F\approx\frac{\hbar k\Gamma}{2}\,\frac{s_0\Gamma^2}{4\delta^2}\propto\frac{I}{\delta^2}$	dissipative force falls as $1/\delta^2$ , dipole force only as $1/\delta$ — far traps use the dipole force
Moving atom (Doppler)	$\delta\to\delta-\mathbf k\cdot\mathbf v$	effective detuning shifts by $-kv$ , making the force velocity-dependent
Two red beams, small $v$	$F\approx-\alpha v,\ \alpha>0$	viscous drag: optical molasses; diffusion sets the floor temperature

Common trapTurning up the intensity always makes the push stronger

It does not. The scattering force saturates: once $s_0\gg1$ the excited population is pinned at $\rho_{ee}=1/2$ , the rate at $\Gamma/2$ , and the force at $F_{\max}=\hbar k\Gamma/2$ — more light only broadens the line. Two related traps hide here. First, do not confuse the scattering force with the dipole force: the scattering force is dissipative, along $\mathbf k$ , saturating, and involves real spontaneous emission; the dipole force is conservative, along the intensity gradient, grows without bound with detuning, and is purely stimulated. Second, "spontaneous emission averages to zero" is true only of its mean — its variance does not vanish, and the resulting momentum diffusion heats the atom, exactly what sets the Doppler cooling limit.

Worked examples

1 · Maximum deceleration of sodium, and how far to stop a beam (confidence builder)

The sodium D2 line has $\lambda=589\ \mathrm{nm}$ , $\Gamma/2\pi=9.79\ \mathrm{MHz}$ , $M=23\,\mathrm u$ . Find the maximum deceleration, and estimate how many photons, how long, and how far it takes to stop an atom leaving an oven at $v_0=1000\ \mathrm{m/s}$ .

Step 1. The photon momentum is $\hbar k=\hbar\,(2\pi/\lambda)=1.055\times10^{-34}\times1.067\times10^{7}=1.13\times10^{-27}\ \mathrm{kg\,m/s}$ , so the recoil velocity is $v_{\mathrm r}=\hbar k/M=1.13\times10^{-27}/(3.82\times10^{-26})=\mathbf{2.9\ cm/s}$ per kick.

Step 2. With $\Gamma=2\pi\times9.79\times10^6=6.15\times10^{7}\ \mathrm{s^{-1}}$ , the maximum rate is $\Gamma/2=3.08\times10^{7}\ \mathrm{s^{-1}}$ and the maximum deceleration is $a_{\max}=v_{\mathrm r}\Gamma/2=0.0295\times3.08\times10^{7}=\mathbf{9.1\times10^{5}\ m/s^2}$ — that is $a_{\max}/g\approx\mathbf{9\times10^{4}}$ .

Step 3. Stopping from $1000\ \mathrm{m/s}$ needs $N=v_0/v_{\mathrm r}=1000/0.0295\approx\mathbf{3.4\times10^{4}\ photons}$ , taking $t=N/(\Gamma/2)\approx\mathbf{1.1\ ms}$ over $d=v_0^2/(2a_{\max})\approx\mathbf{0.55\ m}$ — a resonant beam half a metre long arrests a thermal sodium atom, the design principle of the Zeeman slower.

2 · Saturation intensity and the actual force at the molasses operating point

Compute the sodium D2 saturation intensity, then the scattering force and deceleration at the standard molasses setting $s_0=2$ , $\delta=-\Gamma/2$ .

Step 1. With $I_{\mathrm{sat}}=\pi hc\Gamma/(3\lambda^3)$ : the numerator is $\pi\times6.63\times10^{-34}\times3.00\times10^{8}\times6.15\times10^{7}=3.84\times10^{-17}$ and $3\lambda^3=6.13\times10^{-19}$ , so $I_{\mathrm{sat}}=\mathbf{62.6\ W/m^2=6.3\ mW/cm^2}$ — a milliwatt-scale beam saturates the transition.

Step 2. With $\delta=-\Gamma/2$ , $(2\delta/\Gamma)^2=1$ , so the factor is $s_0/(1+s_0+1)=2/4=0.5$ and the population is $\rho_{ee}=\tfrac12\times0.5=\mathbf{0.25}$ , half its ceiling.

Step 3. The force is therefore exactly half the maximum: $F=F_{\max}\times0.5=\tfrac12(1.13\times10^{-27})(6.15\times10^{7})\times0.5=\mathbf{1.7\times10^{-20}\ N}$ , and $a=a_{\max}\times0.5\approx\mathbf{4.5\times10^{5}\ m/s^2}$ . Cooling runs off the peak and off saturation — detuning makes the force velocity-selective, at a factor-of-two cost in raw push.

3 · The Doppler temperature of sodium

Optical molasses cools until the cooling force balances recoil heating, at $k_BT_D=\hbar\Gamma/2$ . Find the Doppler temperature of sodium and the corresponding one-dimensional rms speed.

Step 1. The Doppler limit is $T_D=\hbar\Gamma/(2k_B)=1.055\times10^{-34}\times6.15\times10^{7}/(2\times1.38\times10^{-23})=\mathbf{235\ \mu K}$ — the textbook floor of simple Doppler cooling for sodium, beaten only by sub-Doppler mechanisms.

Step 2. The 1D rms speed at $T_D$ is $v_{\mathrm{rms}}=\sqrt{k_BT_D/M}\approx\mathbf{0.29\ m/s}$ — roughly ten recoil velocities: cold, but still well above the single-photon recoil limit $v_{\mathrm r}=2.9\ \mathrm{cm/s}$ where the next chapter of cooling begins.

Physics of the portrait

Radiation pressure is one of the oldest ideas in physics and one of the last put to work: guessed from comet tails, predicted by electromagnetism, measured on delicate vanes, and only recently harnessed atom by atom.

Origins · 1 minArthur Ashkin and how this was discovered

Arthur Ashkin

1922–2020 · radiation pressure and the optical manipulation of matter

Arthur Ashkin spent his career at Bell Laboratories turning the faint push of light into a laboratory tool. In 1970 he showed that a focused laser beam could accelerate and even trap transparent micron-sized particles by radiation pressure — the first use of light's momentum to hold matter still, and the direct ancestor of every experiment here. He recognised early that two forces were at play: a scattering force pushing along the beam and a gradient (dipole) force pulling toward high intensity. Balancing them led in 1986 to the single-beam gradient trap now called optical tweezers, and in the same period Ashkin and colleagues used radiation pressure to slow and trap neutral atoms, feeding into the laser-cooling revolution of Chu, Cohen-Tannoudji, and Phillips. His insistence that light momentum was a precise, controllable force reshaped fields far beyond optics — tweezers now stretch DNA molecules, clock molecular motors, and sort living cells. Recognition came late: he shared the 2018 Nobel Prize at the age of 96, then the oldest laureate ever, for the tweezers he had built from the scattering force half a century before.

History

1619

— Kepler argues in De Cometis that sunlight pushes comet tails to point away from the Sun

1873

— Maxwell predicts radiation pressure from electromagnetic theory in his Treatise

1901

— Lebedev measures radiation pressure on solids; Nichols and Hull confirm it independently

1933

— Frisch deflects a sodium atomic beam with resonance light, the first mechanical light force on atoms

1975

— Hänsch and Schawlow propose cooling neutral atoms with laser radiation pressure

1997

— Chu, Cohen-Tannoudji, and Phillips share the Nobel Prize for laser cooling and trapping

Going further

Texts. Foot, Atomic Physics, Ch. 9 (Laser Cooling and Trapping) · Metcalf and van der Straten, Laser Cooling and Trapping, Ch. 3 · Loudon, The Quantum Theory of Light, 3rd ed., chapter on radiation forces · Demtröder, Laser Spectroscopy, Vol. 2, chapter on optical cooling · Saleh and Teich, Fundamentals of Photonics, sections on photon momentum · Hecht, Optics, section on radiation pressure.

Provenance. The steady-state force follows Foot (Atomic Physics, Ch. 9) and Metcalf and van der Straten (Laser Cooling and Trapping, Ch. 3), with the diffusion argument from Loudon. A verifier should confirm: (1) the steady-state population $\rho_{ee}=\tfrac12 s_0/[1+s_0+(2\delta/\Gamma)^2]$ from the optical Bloch equations; (2) $F=\hbar k\Gamma\rho_{ee}$ and its saturation to $\hbar k\Gamma/2$ ; (3) $I_{\mathrm{sat}}=\pi hc\Gamma/3\lambda^3\approx6.3\ \mathrm{mW/cm^2}$ for sodium D2; (4) the sodium numbers $v_{\mathrm r}\approx2.9\ \mathrm{cm/s}$ , $a_{\max}\approx9\times10^5\ \mathrm{m/s^2}$ , stopping distance $\approx0.55\ \mathrm m$ from $1000\ \mathrm{m/s}$ , and $T_D\approx235\ \mu\mathrm K$ ; (5) the linearised damping coefficient $\alpha$ and its optimum near $\delta=-\Gamma/2$ . All atomic constants are standard tabulated values for the sodium D2 line.

Prerequisites. Photon Momentum and the de Broglie Relation · Two-Level Atom and Rabi Oscillations · Optical Bloch Equations and Saturation · Natural Linewidth and Spontaneous Emission. Leads to. Doppler Cooling and the Doppler Limit · Optical Molasses · Zeeman Slower · Magneto-Optical Trap · Dipole Force and the Optical Dipole Potential · The Recoil Limit and Sub-Recoil Cooling.

Ask the tutor: Give me any atom's wavelength, linewidth, and mass, and we will work out its maximum deceleration, saturation intensity, and Doppler temperature together.

Is the scattering force the same thing as radiation pressure?

Yes — "radiation pressure," "scattering force," and "dissipative force" all name the same push light exerts along its travel direction on an absorbing or scattering object. For a two-level atom it is $F=\hbar k\,\Gamma\rho_{ee}$ : photon momentum times the rate photons are scattered. "Pressure" is a holdover from the continuum picture (force per area); "scattering force" is the atomic-physics name contrasting it with the conservative dipole force. Same physics at different scales — the push on a star's envelope and the kick that slows a sodium atom are one accounting.

Why does the force saturate instead of growing with laser power?

Because an atom can only be in one state at a time. To scatter it must absorb (from the ground state) and then emit (from the excited state, waiting for spontaneous decay). Even with an infinitely bright laser, stimulated absorption and emission drive the population toward a fifty-fifty mix, so the atom is excited at most half the time: $\rho_{ee}\le1/2$ . The rate cannot exceed $\Gamma/2$ , nor the force $\hbar k\Gamma/2$ . Intensity beyond $I_{\mathrm{sat}}$ only power-broadens the line; it cannot make the atom cycle faster than its excited-state lifetime allows.

If emission recoils average to zero, how can they heat the atom?

The mean and the variance are different quantities. Emitted photons leave in random directions from a symmetric pattern, so their average recoil is zero — no net force. But each recoil is full-sized, $\hbar k$ , and successive recoils are uncorrelated, so they add in quadrature: the momentum spread grows as $\langle\Delta p^2\rangle=2D_p t$ , which is literally rising kinetic energy. Spontaneous emission is a coin fair on average but paying out on every flip — the accumulated variance is a random walk that warms the atom and sets the Doppler floor.

Why does laser cooling detune off resonance and off saturation?

Because a force largest on resonance is useless for cooling — it pushes a stationary atom as hard as a moving one. Cooling needs the force to depend on velocity, and that dependence comes entirely from the Doppler shift moving the atom on or off resonance. Red detuning ( $\delta<0$ ) makes an atom moving into a beam scatter more, so the net force from two beams opposes the velocity. The price (worked example 2): at $\delta=-\Gamma/2$ , $s_0=2$ the force is only half its saturated maximum — you trade raw push for velocity selectivity, the trade that makes molasses cool.

How is the scattering force different from the dipole force?

They are the two forces light exerts, scaling oppositely with detuning. The scattering force is dissipative: along the beam, from real absorption plus spontaneous emission, saturating, falling as $1/\delta^2$ far from resonance. The dipole force is conservative: along the intensity gradient, from stimulated redistribution with no net scattering, falling only as $1/\delta$ . That difference is decisive — far-detuned light gives a strong trapping potential ( $\propto I/\delta$ ) with negligible heating ( $\propto I/\delta^2$ ), which is why dipole traps and tweezers run far off resonance.

Does this explain radiation pressure on surfaces and in stars?

Yes, with the microscopic detail averaged out. On a macroscopic absorber the pressure is intensity over $c$ (doubled for a perfect mirror, which reverses each photon's momentum) — the coarse-grained version of "momentum $\hbar k$ per photon at the rate photons arrive." In a hot star, radiation pressure on partially ionised gas, through bound-bound and bound-free absorption (the many-line cousin of a single resonance), helps support the outer layers and drives stellar winds. Comet tails, light sails, and optomechanical mirrors are the same bookkeeping — photon momentum times scattering rate, whatever the scatterer.