Chu’s optical molasses: slowing motion while atoms still escape

Load a region, then watch it empty

In 1985, Chu, Hollberg, Bjorkholm, Cable and Ashkin sent precooled sodium atoms into six intersecting laser beams at Bell Laboratories. The experiment joined three-dimensional velocity damping to a directly recorded residence signal: fluorescence persisted after the incoming pulse had passed.

Their report compared the remaining population with diffusion through an escape boundary set by the beams. Its approximately one-tenth-second half-loss time described survival in a finite illuminated region. It did not describe permanent storage, and the initial fast-atom fluorescence spike was a separate contribution to the trace.

Read the evidence: S. Chu, L. Hollberg, J. E. Bjorkholm, A. Cable and A. Ashkin (1985), Three-dimensional viscous confinement and cooling of atoms by resonance radiation pressure, PRL 55, 48–51

Drag remembers velocity, not an address

Consider an invented particle whose acceleration is proportional to its velocity with the opposite sign. Integrating this motion makes the velocity decay, while its position approaches a value determined by where it started and how fast it entered. There is no selected spatial center.

Translate every initial position by the same distance and the whole subsequent trajectory translates with it. A harmonic restoring force breaks that symmetry by referring to a fixed center. Chu’s lecture explicitly distinguishes the early molasses from such a trap; a long residence time alone cannot make that distinction.

Read the evidence: S. Chu (1998), Nobel Lecture: The manipulation of neutral particles, Reviews of Modern Physics 70, 685–706 — optical molasses and the retrospective thermometry discussion

Steady velocity spread, growing displacement

Now add independent random kicks to that one-dimensional drag model. Assume constant coefficients, stationary Gaussian velocity fluctuations, and no walls or gravity. The spatial diffusion coefficient, measured in length squared per time, is the stationary velocity variance multiplied by the relaxation time. The formula uses these quantities; brackets denote an ensemble mean.

Integrating the velocity covariance gives the displacement variance below. At short times the spread is ballistic; after many relaxation times it grows approximately linearly in time even though the velocity variance stays fixed. Our independent moment integration checks both regimes. A stationary velocity distribution therefore does not imply a stationary spatial distribution.

\begin{gathered}\langle v(t)\rangle=\langle v(0)\rangle e^{-t/\tau}\\[3pt]\operatorname{Var}[x(t)-x(0)]\\=2D_x\!\left[t-\tau\!\left(1-e^{-t/\tau}\right)\right]\end{gathered}

Read the evidence: S. Chu, L. Hollberg, J. E. Bjorkholm, A. Cable and A. Ashkin (1985), Three-dimensional viscous confinement and cooling of atoms by resonance radiation pressure, PRL 55, 48–51 S. Chu (1998), Nobel Lecture: The manipulation of neutral particles, Reviews of Modern Physics 70, 685–706 — optical molasses and the retrospective thermometry discussion

Keep the lights on, or interrupt them

The original experiment also extinguished the six beams for a variable interval and measured the population remaining when illumination resumed. Atoms then escaped ballistically during the dark interval. This release record tested a different evolution from the gradual population loss with the cooling light present.

Reading either curve requires the observation region and prepared spatial distribution. Faster departures can reflect directed drift as well as random motion. Comparing the two protocols asks which dynamics are operating during each interval; assigning both curves one universal lifetime would erase the intervention that makes them informative.

Read the evidence: S. Chu, L. Hollberg, J. E. Bjorkholm, A. Cable and A. Ashkin (1985), Three-dimensional viscous confinement and cooling of atoms by resonance radiation pressure, PRL 55, 48–51

The retrospective changes how to read agreement

The 1985 paper reported a temperature near 240 microkelvin and described agreement with its cooling expectation. In his later lecture, Chu revisited the analysis: an initially lower inferred temperature was moved toward the expected value through a correction for the spatial distribution. He criticized the influence of theoretical expectations.

That recollection is a warning about inference, rather than a new analysis of the surviving records. A shape correction can be physically motivated while still needing an independent constraint. The companion thermometry article examines that measurement problem; the later below-Doppler discoveries should not be silently projected back onto the original claim.

Finite beams make capture a transport question

Even without noise, velocity damping needs distance: a sufficiently fast incoming particle can leave a finite beam region before slowing. With noise, particles that have slowed can eventually wander out. Loading probability, residence distribution and velocity spread consequently answer different questions and require different denominators.

Use the existing cooling Lab to compare its declared damping, diffusion and trapping models. It supplies neither the historical fluorescence records nor an independent sodium thermometer. Our constant-coefficient calculation omits internal sublevels, polarization gradients, beam profiles and loss processes; it explains a logical distinction, not the full performance of Chu’s apparatus.