Principal Investigator Craig Savage Project r62,u57

Department of Physics and Theoretical Physics Machine VP

Modelling Atom optics

One of the goals of atom optics is the creation for atoms of efficient analogues of optical elements. Considerable progress has been made demonstrating mirrors, lenses, and beamsplitters. These elements might be combined into devices such as atomic interferometers.

Evanescent wave atomic mirrors have been demonstrated in a number of experiments. In one particular experiment of a reflective diffraction for atoms, diffraction of a slowed metastable neon beam from an evanescent optical grating formed by counterpropagating laser beams was observed. Up to 3% of incident atoms were diffracted by up to 50 mrad from the main reflected beam. This constitutes a large angle reflective beamsplitter for atoms.

Two-level atom models have dominated the theoretical work on evanescent wave devices. Although two-level models predict diffraction we have previously reported that they cannot explain the particular diffraction observed. In the current project we have shown that multi-level atoms can explain the observed diffraction

The multi-level model not only explains the observed atomic diffraction but also suggests further experiments. In particular we find a sensitive dependence on the polarization of the laser beams, and on the incoming atomic Zeeman level.

What are the basic questions addressed?

Can we quantitatively model atom optics experiments from first principles? Can we understand existing experimental results? Can we gain new insight into atom optics and hence suggest new experiments?

Our work suggests a preliminary answer of yes to all these questions.

What are the results to date and future of the work?

Our results show that control of the polarization of both laser beams is crucial in reflection grating atomic diffraction experiments. Furthermore we found that under the conditions of the experiment of Christ et al. the Zeeman m=-2 state produced most of the diffraction. Hence optical pumping into this state could potentially increase the diffraction by up to a factor of four.

We also predict reduced reflection due to the formation of a dark state at pure p-polarization. (Dark states are only possible in multi-level models.) Associated with this dark state we predict strong high order diffraction for laser polarizations close to p-polarization.

Our model is easily adapted to other atoms and we plan to use it to model a Caesium evanescent wave diffraction experiment currently underway in our department.

What computational techniques are used and why is a supercomputer required?

We solve the single atom time-dependent Schrodinger equation from first principles. The internal atomic structure and its interaction with light is included. We make a minimum of assumptions which ensures a versatile and realistic model.

The solution is found by a split-operator method which reduces the problem to matrix multiplication and Fourier transforms, which vectorise well. A supercomputer is required to maximise the realism of our model. We must also examine a large number of cases to cover the uncertainties in experimental parameters. During 1996 we hope to work in "real time" with an experiment in our department to provide diagnosis, interpretation, and to suggest new directions. The VPP300 upgrade will be crucial to the success of this enterprise.


Numerical modeling of evanescent wave atom optics, C. Savage, D. Gordon, and T. Ralph, Phys. Rev. A 52, 4741 (1995).

Evanescent wave diffraction of multi-level atoms, D. Gordon and C. Savage,, preprint