Wednesday, November 18, 2009

When was the first BEC observed?

I am getting tired of hearing talks and reading reports which state, "The first Bose-Einstein condensate was observed in 1995." I think a more accurate statement would be "The first BEC in a dilute atomic gas was observed in 1995." Many would argue that superfluid 4He is a BEC. This new phase of matter was first observed around 1930. In 1932 Fritz London proposed that this was a BEC. (BTW, this is the same London as in the Heitler-London wave function, the London penetration depth, and London dispersion forces...).
But it should be noted that the case of a BEC in superfluid He is not as clear cut as in dilute atomic gases. Nevertheless, I dont think these subtleties validate ignoring 80 years of research on superfluid helium. A very useful summary of the history and the associated physical issues is contained in this nice article by Sebastian Balibar.

I think that people who are supervising Ph.D students on BEC's should be familiar with these issues, make sure their students are aware of this history, and present their work in the appropriate context. But then I am just a grumpy old condensed matter physicist....


  1. Not at all Ross, I agree that liquid Helium should not get written out of the story. This seems to be a persistent irritant between condensed matter and AMO trained physicists, and we should know better by now.

    In our defense, when I was studying for my Ph.D the point of view of atomic physicists working on the then new dilute atomic Bose gases was that liquid Helium was very distinct.

    This view hadn't come from nowhere. Many papers and at least some texts, including as I recall Huang, emphasized the strong interactions in Helium, and argued for the inapplicability of a weak coupling description like Bogoliubov's to the system. I certainly had the impression that the consensus was that while liquid helium was a "superfluid" it was not a "BEC".

    Now I have heard it said that this claim that liquid Helium is not a BEC originates with, or was forcefully promulgated by Landau, decaying away only after his death. I'd be interested in your comments on how views on this issue have changed in the condensed matter community over time.

    More broadly, that period when AMO physicists started thinking about essentially condensed matter problems was very interesting since the two communities brought such different language to the same physical problems.

    Obviously there were many ways in which the AMO approach was/is old fashioned or unsophisticated by the lights of the modern understanding of phase transitions, broken symmetry and the renormalisation group.

    On the other hand the optical point of view apparently led to much clearer thinking about the issue of interference of independent condensates, for example.

  2. I think you're being to circumspect in your criticism.

    If 4-He isn't a BEC neither are dilute atomic gases. Both are superfluids.

    An non-interacting Bose gas will form a condensate, but this is not a superfluid. However, this model is essentially unrealisable.

    The superfluidity in both dilute atomic gases and He arrises from the fact that their is a linear dispersion. The linear dispersion is due to the phonons (the Goldstone mode associated with superfluid phase transition). Phonons are only quasiparticles of the interacting system. The non interacting system would have a quadratic dispersion (hence the absence of superfluidity).

    Thus most of the interesting physics of BECs (vortices, superfluidity, phase stiffness,...) result directly from the interactions in both dilute atomic gases and He.

    To my mind the only way it would be interesting to distinguish between BEC and superfluidity in He would be if there were a phase transition (or at least a crossover) as one turned up the interaction strength. I'm not aware of anyone suggesting that.

    It seems to me that the claim that "BEC was first observed in 1995" is mostly a marketing device. But even it that light I think it's stupid. Surely there's enough interesting physics in dilute atomic gases that one doesn't have to ignore 70 years of research on He to make the dilute atomic gases seem interesting? If not why work on it? It seemed to me that a good indicator that people in quantum information were doing interesting things was when they stopped introducing their talks by quoting Schrodingers comments about entanglement and started talking about all the cool things their field had done instead.

  3. Ross,

    I think it is fair to say that BEC was first *directly* observed in dilute vapors. Here the atomic momentum distribution can be directly visualized by time-of-flight after the trap is released. As discussed in the review article, Sokol was able to use high-momentum transfer neutron scattering to determine that the zero temperature BEC fraction in liquid helium is about 10%. However, in these experiments, no distinct condensate peak appears in the data. Richard Silver at Los Alamos remembers this story as part of the "condensate saga":


    You write,

    I certainly had the impression that the consensus was that while liquid helium was a "superfluid" it was not a "BEC".

    The superfluid transition in liquid helium is definitely of a different nature than BEC in the ideal Bose gas. The critical behavior of the two transitions is different: the ideal gas has a kink at in C(T) while liquid helium has the the lambda shape. The formal, generalized definition of BEC to apply to interacting systems (all real ones, including even dilute vapors) is Off-Diagonal Long Range Order (ODLRO) in the one-body density matrix. ρ(r) does not vanish as r --> infinity. This means in the momentum distribution n(p) there is a delta-function spike at p = 0 that comprises the condensate. That is the defining feature of BEC.


    I think it's useful terminologically to distinguish between BEC and superfluidity. Conceptually, BEC means that a macroscopic number of particles occupy a single quantum state and superfluidity refers to transport properties. This makes it a substantive claim, rather than just a definition, that superfluidity is associated with BEC.

    You point out that the linear dispersion gives rise to superfluidity. Well, I'm not sure it's ever been shown that the phonon part of the spectrum is especially connected to BEC as such. Remember in Feynman's theory of the excitation spectrum, he argues that the only low-energy excitations in liquid helium have to be collective density fluctuations. He argues this on the basis of Bose statistics obeyed by helium atoms, but doesn't appeal at all to BEC. Glyde and Griffin have tried to connect the spectrum of helium up with the condensate. But, it is at least conceivable that you could have that kind of spectrum without any BEC existing at all.