Part II, Chapter 3

Laser Cooling

From Doppler cooling to sub-Doppler techniques and magneto-optical traps — reaching microkelvin temperatures with light.

3.1 Doppler Cooling

Doppler cooling uses the velocity-dependent scattering force from counter-propagating red-detuned laser beams. An atom moving toward a beam sees it blue-shifted closer to resonance, absorbing more photons from that beam and experiencing a net restoring force.

The cooling process competes with heating from random recoil kicks. Each spontaneous emission imparts a random momentum kick of magnitude ℏk, causing diffusion in momentum space.

Figure. Doppler cooling mechanism: an atom moving toward the laser absorbs a red-detuned photon (Doppler-shifted into resonance), receiving a momentum kick that slows it. Spontaneous re-emission is random, averaging to zero net recoil.

Derivation 1: Doppler Cooling Limit

Step 1. The cooling rate is the energy removed per unit time by the friction force F = -βv:

$$\left(\frac{dE}{dt}\right)_{\text{cool}} = F \cdot v = -\beta v^2 = -\beta \frac{k_B T}{m}$$

Step 2. The heating rate from momentum diffusion (random walks in momentum space):

$$\left(\frac{dE}{dt}\right)_{\text{heat}} = \frac{(\hbar k)^2}{m} R_{\text{sc}} = \frac{(\hbar k)^2}{m}\Gamma\rho_{ee}$$

Step 3. Equilibrium (cooling = heating) gives the steady-state temperature. Optimizing over detuning (δ = -Γ/2):

Doppler Temperature:

$$T_D = \frac{\hbar\Gamma}{2k_B}$$

For sodium (D2 line, Γ/2π = 9.8 MHz): T_D = 235 μK. For rubidium: T_D = 146 μK. For cesium: T_D = 125 μK. This limit depends only on the natural linewidth — a fundamental property of the transition.

3.2 Sub-Doppler (Sisyphus) Cooling

In 1988, experiments at NIST showed that optical molasses could cool atoms well below T_D. The explanation involves polarization gradient cooling (Sisyphus cooling), which exploits the spatially varying polarization in counter-propagating beams with different polarizations.

Derivation 2: Sisyphus Cooling Mechanism

Step 1. In a lin⊥lin configuration, the polarization alternates between σ+, linear, σ-, linear... with period λ/2. The light shifts of magnetic sublevels depend on the local polarization:

$$U_{m_J}(x) = -\frac{\hbar\Omega^2}{4\delta} C_{m_J}^2(\text{pol}(x))$$

where C_{m_J} are Clebsch-Gordan coefficients that depend on the local polarization.

Step 2. An atom in m_J = +1/2 at a σ+ site is optically pumped. As it moves and climbs the potential hill toward a σ- site, it loses kinetic energy. At the hilltop, it is optically pumped back to m_J = +1/2, resetting the cycle.

Step 3. Each Sisyphus cycle removes energy ≈ U₀ (the modulation depth). The equilibrium temperature is:

$$k_B T_{\text{Sisyphus}} \sim \frac{\hbar\Omega^2}{|\delta|} \propto \frac{I}{|\delta|}$$

This can be much lower than T_D. Typical values reach a few μK or even below 1 μK. The ultimate limit is the recoil temperature: T_R = ℏ²k²/(mk_B) ≈ 0.36 μK for sodium.

3.3 Magneto-Optical Trap (MOT)

Optical molasses provides velocity damping but no spatial confinement. The magneto-optical trap (MOT) adds a quadrupole magnetic field that creates a position-dependent force, trapping atoms at the field zero.

Derivation 3: MOT Restoring Force

Step 1. A quadrupole magnetic field B = B'z (gradient B') Zeeman-shifts the m_J sublevels: ΔE = g_Jμ_Bm_JB'z.

Step 2. With σ+ and σ- beams counter-propagating, the effective detuning for each beam becomes position-dependent:

$$\delta_\pm(z) = \delta \mp kv \mp \frac{\mu' B' z}{\hbar}$$

where μ' = (g_em_e - g_gm_g)μ_B.

Step 3. Expanding for small z and v:

$$F \approx -\beta v - \kappa z$$

where κ = μ'B'β/(ℏk) is the spring constant. The MOT provides both damping and confinement, trapping up to 10¹&sup0; atoms in a cloud ~1 mm in diameter at temperatures of ~100 μK.

MOT Parameters (Typical for &sup8;&sup7;Rb)

• Laser detuning: δ ≈ -2Γ to -3Γ
• Magnetic field gradient: B' ≈ 10-15 G/cm
• Atom number: 10&sup8; - 10¹&sup0;
• Temperature: 50-200 μK
• Density: 10¹&sup0; - 10¹¹ cm&supmin;³

3.4 Evaporative Cooling

Laser cooling cannot reach quantum degeneracy (phase space density ρ ~ 1). The final step uses evaporative cooling: selectively remove the highest-energy atoms, allowing the remainder to rethermalize at a lower temperature.

Derivation 4: Forced Evaporative Cooling Efficiency

Step 1. In a magnetic trap, RF radiation drives transitions to untrapped states for atoms with energy E > ηk_BT (the truncation parameter η = U_trap/k_BT). Lowering the RF frequency reduces η.

Step 2. Each evaporated atom carries away energy (η + κ)k_BT on average, where κ ≈ 1 accounts for the excess energy. The remaining N atoms rethermalize at lower T.

Step 3. For constant η, the temperature and number scale as:

$$T \propto N^{\gamma}, \qquad \gamma = \frac{\eta + \kappa - 3}{3} - 1$$

For η ≈ 6 (typical): γ ≈ 2/3. The phase space density increases as ρ ∝ N/T³/² ∝ N^{1 - 3γ/2}. With γ < 2/3, the phase space density increases even as atoms are lost. Starting from ρ ~ 10&supmin;&sup6; in a MOT, evaporative cooling reaches ρ > 1 (BEC) with a factor ~100 loss in atom number.

3.5 Sub-Recoil Cooling Techniques

Several techniques cool atoms below the single-photon recoil temperature T_R = ℏ²k²/(mk_B).

Derivation 5: Velocity-Selective Coherent Population Trapping (VSCPT)

Step 1. For a Λ-type three-level atom driven by two counter-propagating beams, there exists a dark state that does not couple to the light:

$$|D\rangle = \frac{1}{\sqrt{2}}(|g_1, \hbar k\rangle - |g_2, -\hbar k\rangle)$$

Step 2. This dark state is an eigenstate only for atoms with velocity v = 0 (momentum p = ±ℏk in the two ground states). Atoms with v ≠ 0 are not in a dark state and continue to scatter photons, performing a random walk in momentum space.

Step 3. Atoms that randomly reach v ≈ 0 fall into the dark state and stop scattering. Over time, atoms accumulate near v = 0:

$$\Delta p \sim \hbar k / \sqrt{\Gamma t}, \qquad T \propto \frac{1}{t}$$

The temperature decreases without bound as 1/t — no fundamental temperature limit! Demonstrated by Aspect et al. (1988) for metastable helium, reaching T < T_R/100.

3.6 Magnetic Trapping

After laser cooling, atoms are transferred to magnetic traps for evaporative cooling. Magnetic trapping exploits the interaction of the atomic magnetic moment with an inhomogeneous field: U = -μ·B = g_Fm_Fμ_B|B|.

3.6.1 Trap Types

Quadrupole Trap

Linear field gradient: B = B'(x&hat;x; + y&hat;y; - 2z&hat;z;)/2. Provides tight confinement but has a zero-field point at the center where atoms undergo Majorana spin flips and are lost.

Ioffe-Pritchard (IP) Trap

Adds a bias field B₀ to eliminate the zero, creating a harmonic potential. The trap frequencies are ω_r = μ'B''/(mB₀)¹/² radially and ω_z from the axial curvature. Most BEC experiments use IP-type traps.

TOP (Time-Orbiting Potential) Trap

A rotating bias field moves the zero point of a quadrupole trap in a circle faster than the atoms can follow. The time-averaged potential is harmonic. Used in the first BEC experiment at JILA (1995).

Atom Chips

Microfabricated wires on a chip create tight magnetic traps just microns from the surface. Enable compact, portable BEC experiments and have been used in space (ISS Cold Atom Lab).

3.6.2 Optical Dipole Traps for All-Optical BEC

Alternatively, atoms can be evaporatively cooled directly in an optical dipole trap by lowering the trap depth. This approach (all-optical BEC) avoids the need for magnetic traps and works for any atomic species, including those with unfavorable magnetic properties. First demonstrated by Barrett, Sauer, and Chapman (2001) with &sup8;&sup7;Rb.

3.7 Advanced and Specialized Cooling

Beyond standard Doppler and sub-Doppler cooling, several specialized techniques have been developed for specific applications.

3.7.1 Raman Sideband Cooling

Two-photon Raman transitions between hyperfine ground states can drive resolved sideband cooling for neutral atoms in optical lattices. The effective linewidth is determined by the Raman laser parameters, not the natural linewidth, enabling sub-recoil cooling to the 3D motional ground state in each lattice site.

3.7.2 Demagnetization Cooling

Inspired by adiabatic demagnetization refrigeration, atoms in a magnetic trap with multiple Zeeman sublevels are optically pumped to the most weakly trapped state. The released potential energy is carried away by scattered photons. This technique reaches temperatures below standard Doppler limits for atoms with complex level structures.

3.7.3 Narrow-Line Cooling

Atoms with narrow intercombination lines (e.g., &sup8;&sup8;Sr at 689 nm, Γ/2π = 7.4 kHz) have Doppler temperatures in the hundreds of nanokelvin range:

$$T_D = \frac{\hbar\Gamma}{2k_B} \approx 180\,\text{nK (for Sr)}$$

This is comparable to the recoil temperature and enables direct laser cooling to quantum degeneracy without evaporative cooling. Combined with broad-line pre-cooling on the ¹S₀ → ¹P₁ transition, strontium experiments routinely produce BEC starting from a narrow-line MOT.

3.7.4 Sawtooth Wave Adiabatic Passage (SWAP) Cooling

SWAP cooling uses frequency-chirped laser pulses to adiabatically transfer atoms between ground and excited states while always removing kinetic energy. Unlike Doppler cooling, it works even at large detunings and for multilevel atoms. It has been used to cool molecules (CaF, SrF) that lack closed cycling transitions.

3.8 The Path to Quantum Degeneracy

Achieving Bose-Einstein condensation or Fermi degeneracy requires increasing the phase space density ρ = nλ_dB³ from its thermal value (~10&supmin;¹&sup4;) to above 2.612 (for BEC). The journey spans 14 orders of magnitude.

Phase Space Density at Each Stage

Thermal Source (~600 K)

ρ ~ 10&supmin;¹&sup4;. Atoms in an oven with mean velocity ~500 m/s.

After Zeeman Slowing

ρ ~ 10&supmin;¹&sup0;. Atoms decelerated to ~30 m/s in a beam.

MOT (~100 μK)

ρ ~ 10&supmin;&sup6;. About 10&sup9; atoms trapped at 100 μK.

Sub-Doppler Cooling (~5 μK)

ρ ~ 10&supmin;&sup5;. Temperature reduced 20x, density roughly preserved.

Magnetic Trap Compression

ρ ~ 10&supmin;&sup4; to 10&supmin;³. Adiabatic compression increases density.

Evaporative Cooling → BEC

ρ > 2.612. Phase space density increases by 10³-10&sup4; while losing ~99% of the atoms. Final temperature: ~100 nK.

The entire sequence from thermal oven to BEC takes approximately 30-60 seconds and is repeated for each experimental shot. Modern experiments achieve BEC rates of up to 1 Hz, enabling rapid data collection for precision measurements and quantum simulation.

Applications

Atomic Fountains and Clocks

Laser-cooled atoms launched upward in a fountain provide long interrogation times (~1 s) for Ramsey spectroscopy. The NIST-F2 cesium fountain clock has an uncertainty of 10&supmin;¹&sup6;, defining the SI second.

Bose-Einstein Condensation

The combination of laser cooling (MOT) followed by evaporative cooling in magnetic or optical traps is the standard route to BEC. First achieved in 1995 by Cornell, Wieman (JILA) and Ketterle (MIT). Nobel Prize 2001.

Atom Interferometry

Laser-cooled atoms enable high-precision atom interferometers for measuring gravity, gravitational gradients, rotations, and fundamental constants. Proposed for gravitational wave detection (MAGIS, AION) in the 0.1-10 Hz frequency band.

Quantum Computing with Neutral Atoms

Laser-cooled atoms in optical tweezer arrays, combined with Rydberg interactions, form a leading platform for quantum computing. Systems with >1000 qubits have been demonstrated by Harvard/MIT and Atom Computing groups.

3.9 Laser Cooling of Molecules

Laser cooling molecules is far more challenging than atoms because molecules have vibrational and rotational degrees of freedom that open many decay channels, preventing a closed cycling transition.

3.9.1 Diagonal Franck-Condon Factors

Molecules with near-diagonal Franck-Condon matrices (where vibrational quantum number is preserved during electronic transitions with probability >99%) can be laser cooled. Species include SrF, CaF, YO, CaOH, and SrOH. With 3-4 repumping lasers to close the vibrational leaks, these molecules scatter enough photons for Doppler cooling.

3.9.2 Molecular MOTs and Applications

The first molecular MOT was demonstrated for SrF (Barry et al., 2014). Due to the complex level structure, molecular MOTs use RF-modulated lasers to address multiple hyperfine transitions simultaneously. Sub-Doppler cooling has brought molecular temperatures below 5 μK (Cheuk et al., 2018). Applications include electric dipole moment searches, ultracold chemistry, and quantum simulation with dipolar interactions.

3.10 Summary of Fundamental Cooling Limits

Doppler Limit

T_D = ℏΓ/(2k_B). Set by balance of Doppler cooling and photon recoil heating. For alkalis: 100-240 μK. For narrow lines: nanokelvin.

Recoil Limit

T_R = ℏ²k²/(mk_B). The energy of a single photon recoil. For Na: 2.4 μK. For Cs: 0.2 μK. Breached by VSCPT, Raman cooling.

Sisyphus Limit

T_Sis ≈ ℏΩ²/(|δ|k_B). Tunable by intensity and detuning. Can reach T_R or below with careful optimization.

Sideband Cooling Ground State

&bar;n; = (Γ/2ω_trap)². For resolved sidebands (ω_trap >> Γ): &bar;n; << 1. Achieves the motional ground state of the trapping potential.

Key Equations Summary

Doppler Temperature:

$$T_D = \frac{\hbar\Gamma}{2k_B}$$

Recoil Temperature:

$$T_R = \frac{\hbar^2 k^2}{m k_B}$$

MOT Restoring Force:

$$F \approx -\beta v - \kappa z$$

Phase Space Density:

$$\rho = n\lambda_{\text{dB}}^3 = n\left(\frac{2\pi\hbar^2}{mk_BT}\right)^{3/2}$$

Evaporative Cooling Scaling:

$$T \propto N^{\gamma}, \qquad \rho \propto N^{1 - 3\gamma/2}$$

VSCPT Cooling Rate:

$$\Delta p \sim \hbar k / \sqrt{\Gamma t}$$

Historical Context

1975 — Hänsch & Schawlow / Wineland & Dehmelt: Independently proposed laser cooling for neutral atoms and trapped ions respectively.

1985 — Chu et al.: First 3D optical molasses, cooling sodium to 240 μK — confirming the Doppler cooling theory.

1988 — Lett et al.: Measured temperatures far below T_Din optical molasses, sparking the discovery of sub-Doppler (Sisyphus) cooling mechanisms.

1987 — Raab et al.: Demonstrated the first magneto-optical trap, capturing ~10&sup7; sodium atoms.

1995 — Cornell & Wieman / Ketterle: Achieved Bose-Einstein condensation using laser cooling followed by evaporative cooling. Nobel Prize 2001.

1997 — Nobel Prize: Awarded to Chu, Cohen-Tannoudji, and Phillips for development of methods to cool and trap atoms with laser light.

Interactive Simulation

This simulation demonstrates Doppler cooling dynamics, the equilibrium temperature as a function of detuning, MOT trapping, and evaporative cooling trajectories.

Laser Cooling: Doppler Limit, MOT, and Evaporative Cooling

Python

script.py109 lines

import numpy as np
import matplotlib.pyplot as plt

fig, axes = plt.subplots(2, 2, figsize=(12, 10))
fig.patch.set_facecolor("#0a0a0a")
for ax in axes.flat:
    ax.set_facecolor("#111111")

Gamma = 1.0  # natural linewidth
hbar_k = 1.0  # recoil momentum

# Plot 1: Doppler cooling temperature vs detuning
ax1 = axes[0, 0]
delta_vals = np.linspace(-5, -0.01, 500)
s0_vals = [0.1, 1.0, 5.0]

for s0, col, label in zip(s0_vals, ["#a78bfa", "#c084fc", "#f97316"],
                           ["I/Isat=0.1", "I/Isat=1.0", "I/Isat=5.0"]):
    # Temperature: T = (hbar*Gamma/2k_B) * (1+s0)/(4|delta|/Gamma) * (1 + (2delta/Gamma)^2)
    T_doppler = (Gamma / 2) * (1 + s0 + (2*delta_vals/Gamma)**2) / (4 * np.abs(delta_vals) / Gamma)
    ax1.plot(delta_vals, T_doppler, color=col, linewidth=2, label=label)

ax1.axhline(y=Gamma/2, color="gray", linestyle="--", alpha=0.5, label="T_D = hbar*Gamma/(2kB)")
ax1.set_xlabel("Detuning (Gamma units)", color="white")
ax1.set_ylabel("Temperature (hbar*Gamma/kB units)", color="white")
ax1.set_title("Doppler Temperature vs Detuning", color="#c4b5fd", fontsize=14)
ax1.set_ylim(0, 5)
ax1.legend(facecolor="#1a1a2e", edgecolor="#7c3aed", labelcolor="white", fontsize=9)
ax1.tick_params(colors="white")
for spine in ax1.spines.values():
    spine.set_color("#7c3aed")

# Plot 2: 1D MOT simulation (atom trajectory)
ax2 = axes[0, 1]
dt = 0.01
t_max = 30.0
t = np.arange(0, t_max, dt)

# MOT: F = -beta*v - kappa*z
beta = 0.5
kappa = 0.3
z = np.zeros_like(t)
v = np.zeros_like(t)
z[0] = 3.0
v[0] = 2.0

for i in range(len(t) - 1):
    F = -beta * v[i] - kappa * z[i]
    noise = 0.1 * np.random.randn()  # random recoil
    v[i+1] = v[i] + (F + noise) * dt
    z[i+1] = z[i] + v[i+1] * dt

ax2.plot(t, z, color="#a78bfa", linewidth=1, alpha=0.8)
ax2.axhline(y=0, color="gray", linestyle="--", alpha=0.3)
ax2.set_xlabel("Time (arb.)", color="white")
ax2.set_ylabel("Position", color="white")
ax2.set_title("MOT: Atom Trajectory (damped + confined)", color="#c4b5fd", fontsize=14)
ax2.tick_params(colors="white")
for spine in ax2.spines.values():
    spine.set_color("#7c3aed")

# Plot 3: Evaporative cooling trajectory (N vs T)
ax3 = axes[1, 0]
eta_vals = [4, 6, 8, 10]
N0 = 1e9
T0 = 100e-6  # 100 microK

for eta, col in zip(eta_vals, ["#a78bfa", "#c084fc", "#f97316", "#ef4444"]):
    gamma_ev = (eta - 4) / 3.0  # simplified scaling
    if gamma_ev <= 0:
        gamma_ev = 0.1
    N_arr = np.logspace(np.log10(N0), np.log10(N0/1000), 200)
    T_arr = T0 * (N_arr / N0)**gamma_ev
    ax3.loglog(N_arr, T_arr * 1e6, color=col, linewidth=2,
               label="eta = " + str(eta))

# BEC line: T_c ~ n^(2/3) ~ (N/V)^(2/3) ~ N^(2/3) for fixed V
N_bec = np.logspace(5, 9, 100)
T_bec = 0.1 * (N_bec / 1e6)**(2.0/3.0)  # schematic
ax3.loglog(N_bec, T_bec, "w--", linewidth=1, alpha=0.5, label="BEC threshold")
ax3.set_xlabel("Atom number N", color="white")
ax3.set_ylabel("Temperature (microK)", color="white")
ax3.set_title("Evaporative Cooling Trajectories", color="#c4b5fd", fontsize=14)
ax3.legend(facecolor="#1a1a2e", edgecolor="#7c3aed", labelcolor="white", fontsize=9)
ax3.tick_params(colors="white")
for spine in ax3.spines.values():
    spine.set_color("#7c3aed")

# Plot 4: Cooling techniques comparison (temperature ranges)
ax4 = axes[1, 1]
techniques = ["Oven\n(thermal)", "Zeeman\nslower", "Doppler\ncooling", "Sisyphus\ncooling",
              "Evaporative\ncooling", "BEC"]
temps = [600, 0.5, 140e-3, 5e-3, 1e-3, 0.1e-3]  # in Kelvin... using millikelvin scale
temp_mK = [6e5, 500, 0.14e3, 5, 1, 0.1]

colors_bar = ["#ef4444", "#f97316", "#c084fc", "#a78bfa", "#8b5cf6", "#7c3aed"]
bars = ax4.barh(techniques, temp_mK, color=colors_bar, alpha=0.8)
ax4.set_xscale("log")
ax4.set_xlabel("Temperature (mK, log scale)", color="white")
ax4.set_title("Cooling Technique Comparison", color="#c4b5fd", fontsize=14)
ax4.tick_params(colors="white")
for spine in ax4.spines.values():
    spine.set_color("#7c3aed")

plt.tight_layout()
plt.savefig("output.png", dpi=150, bbox_inches="tight", facecolor="#0a0a0a")
plt.show()
print("Plot saved: Laser cooling - Doppler limit, MOT trajectory, evaporative cooling.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Conceptual Questions

Q1: Why must the laser be red-detuned for Doppler cooling?

An atom moving toward a red-detuned beam sees it Doppler-shifted closer to resonance, absorbing more and receiving a decelerating force. An atom moving away sees the beam further from resonance and absorbs fewer photons. With blue detuning, atoms would be accelerated.

Q2: Why can't laser cooling alone reach BEC?

The maximum MOT density is limited to ~10¹¹ cm&supmin;³ by reabsorption. Phase space density peaks at ~10&supmin;&sup5;, still 6 orders of magnitude below BEC threshold. Evaporative cooling bridges this gap by selectively removing hot atoms.

Q3: How does Sisyphus cooling break the Doppler limit?

Atoms climb spatially varying light-shift potential hills (losing kinetic energy), then get optically pumped to the bottom of the next hill. The energy removal per cycle depends on the tunable light shift, not the natural linewidth, reaching far below T_D.

Further Topics

Frontiers of Laser Cooling

• Cooling antihydrogen: The ALPHA collaboration at CERN has laser-cooled trapped antihydrogen atoms (2021) using the 1S-2P Lyman-alpha transition at 121.6 nm, enabling precision spectroscopy to test CPT symmetry and the gravitational behavior of antimatter.
• Sympathetic cooling: Cooling one species by thermal contact with a laser-cooled species. Used to cool molecular ions, exotic isotopes, and antiprotons that cannot be directly laser cooled.
• Cavity cooling: An atom in a high-finesse cavity can be cooled by the cavity-enhanced scattering of photons from a transverse pump beam into the cavity mode. No closed cycling transition is needed, making this applicable to complex molecules.
• Stochastic cooling: Feedback-based cooling where the position or velocity of an atom is measured and a corrective force applied in real time. Demonstrated for single ions and proposed for neutral atoms using cavity-mediated measurements.