Condensed-matter physics

Really cool molecules

Article metrics

Ultracold molecules have been made by applying a changing magnetic field to a quantum gas of 'fermionic' atoms. This raises the prospect of creating novel superfluids and molecular Bose–Einstein condensates.

At the quantum level, any particle can also be considered to be a wave, its momentum corresponding to a wavelength. In ultracold atomic gases, at temperatures much lower than a millionth of a degree above absolute zero, the wavelength of the atoms becomes larger than the mean distance between them, giving rise to some remarkable quantum mechanical properties that are at the forefront of contemporary physics research1. The properties of such cold quantum gases depend on whether the atoms are bosons or fermions. Bosonic atoms have integer values of the quantum number known as spin, and can form a Bose–Einstein condensate in which the trapped atoms occupy the single ground state of the system. Fermionic atoms, on the other hand, have half-integer spin values and each identical fermion must occupy a different quantum level. But if two fermions paired up to make a bosonic molecule, an exotic kind of superfluidity — flow without resistance — could be the result2,3,4. On page 47 of this issue, Regal et al.5 describe an important step in this direction.

Their experiment began with a quantum gas of fermionic potassium atoms, a mixture of equal numbers of atoms with spin quantum numbers −9/2 and −5/2. The different spin states are necessary for the fermions to undergo collisions with each other with low collision energy. To create molecules from these atoms, Regal et al. took advantage of a special molecular state known as a Feshbach resonance, which may be thought of as a weakly bound pair of atoms. The energy of such a state can be tuned, using a magnetic field, to lie close to that of two separated atoms. By ramping the magnetic field so that the energy of the resonance state moved from being above to being below the energy of two separated atoms, colliding pairs of atoms were induced to form a lower-energy molecule that is bound with respect to the separated atoms (Fig. 1). Regal et al. were able to convert up to half of the atoms to diatomic molecules.

Figure 1: Making cold molecules.

Regal et al.5 have succeeded in creating molecules in a mixture of cold gas atoms. The atoms exist in two different spin states (represented by red and blue dots) at a certain energy, but can be made to pair up through a Feshbach resonance. Varying the magnetic field applied to the system changes the energy of the resonance. When the resonance energy becomes the same as the energy level of the atomic mixture, colliding atoms can be converted to resonant-state molecules. As the resonance energy decreases further, the molecules finally arrive in a lower-energy state, lower than the atomic state by the amount of molecular binding energy.

Producing a molecular gas in this ultracold regime has proved to be a challenge. There have been several proposals to make molecules, and possibly a molecular Bose–Einstein condensate, by combining bosonic atoms in a Bose–Einstein condensate using a molecular resonance state6,7,8,9,10. Experiments have come tantalizingly close. Wynar et al.11 were the first to infer the formation of molecules nearly at rest, by observing the loss of condensed rubidium-87 atoms as they were photoassociated into molecules using a two-colour pulse of light. Donley et al.12 have also given evidence for coherent atom–molecule interconversion13,14 in experiments with a rubidium-85 condensate that involved manipulating the energy of a Feshbach-resonance state using a sequence of magnetic-field pulses. However, neither experiment could detect any molecules directly: the imaging methods used for atoms do not work for molecules, because they have very different light-scattering properties.

Part of the beauty of the new work by Regal et al.5 is that they have conclusively demonstrated the presence of cold molecules. They exposed the molecules to a radio-frequency electromagnetic field of the right frequency to make them fall apart into single atoms with spin quantum numbers −9/2 and −7/2. It is quite easy to detect atoms in different spin states separately, and as there were no −7/2 atoms present in the original mixture (only −9/2 and −5/2), counting the number of such atoms determines the number of molecules that had been present. By measuring the response of the molecules to different field frequencies, Regal et al. were also able to measure the extremely small binding energy of the molecules — of the order of only a nanoelectronvolt, and in excellent agreement with calculations based on the known properties of low-energy collisions of potassium atoms.

This experiment opens up several promising avenues for development. The molecules should have a temperature comparable to that of the atoms from which they were made — about 150 nanokelvins. A study of the momentum distribution and the coherence properties of such a molecular gas is clearly in order, to see whether it might actually be a molecular Bose–Einstein condensate. There are also challenges to theory, such as understanding why only half of the atoms were converted to molecules. Another major experimental goal is to demonstrate the existence of superfluidity in quantum degenerate fermionic gases, which might arise through the pairing of fermions in different spin states, in the same way that Cooper pairing of electrons is the basis of superconductivity. Theory even suggests that the very strong coupling associated with a Feshbach-resonance state should make such an effect much easier to observe4.

A new era of research into ultracold molecular gases may be about to begin. There is much to do. For example, little is known about the collisions or chemical reactivity of ultracold molecules, which will determine the lifetime of such gases. Ultracold molecules also have potential uses in precise time and frequency measurements, in the search for an electron dipole moment15 and in quantum computing16.


  1. 1

    Nature Insight on Ultracold Matter Nature 416, 205–246 (2002).

  2. 2

    Houbiers, M. & Stoof, H. T. C. Phys. Rev. A 59, 1556–1561 (1999).

  3. 3

    Bruun, G., Castin, Y., Dum, R. & Burnett, K. Eur. Phys. D 7, 433–439 (1999).

  4. 4

    Holland, M., Kokkelmans, S. J. J. M. F., Chiofalo, M. L. & Walser, R. Phys. Rev. Lett. 87, 120406 (2001).

  5. 5

    Regal, C. A., Ticknor, C., Bohn, J. L. & Jin, D. S. Nature 424, 47–50 (2003).

  6. 6

    Julienne, P. S., Burnett, K., Band, Y. B. & Stwalley, W. C. Phys. Rev. A 58, R797–R800 (1998).

  7. 7

    Timmermans, E., Tommasini, P., Côté, R., Hussein, M. & Kerman, A. Phys. Rev. Lett. 83, 2691–2694 (1999).

  8. 8

    Heinzen, D. J., Wynar, R., Drummond, P. D. & Kheruntsyan, K. V. Phys. Rev. Lett. 84, 5029–5032 (2000).

  9. 9

    Mies, F. H., Tiesinga, E. & Julienne, P. S. Phys. Rev. A 61, 022721 (2000).

  10. 10

    Kokkelmans, S. J. J. M. F., Vissers, H. M. J. & Verhaar, B. J. Phys. Rev. A 63, 031601 (2001).

  11. 11

    Wynar, R. H., Freeland, R. S., Han, D. J., Ryu, C. & Heinzen, D. J. Science 287, 1016–1019 (2000).

  12. 12

    Donley, E. A., Claussen, N. R., Thompson, S. T. & Wieman, C. E. Nature 417, 529–533 (2002).

  13. 13

    Kokkelmans, S. J. J. M. F. & Holland, M. J. Phys. Rev. Lett. 89, 180401 (2002).

  14. 14

    Köhler, T., Gasenzer, T. & Burnett, K. Phys. Rev. A 67, 013601 (2003).

  15. 15

    Hudson, J. J., Sauer, B. E., Tarbutt, M. R. & Hinds, E. A. Phys. Rev. Lett. 89, 023003 (2002).

  16. 16

    DeMille, D. Phys. Rev. Lett. 88, 067901 (2002).

Download references

Author information

Correspondence to Paul S. Julienne.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Julienne, P. Really cool molecules. Nature 424, 24–25 (2003) doi:10.1038/424024a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.