from OE Reports April 1995 issue
Cells filled with cesium vapor may provide a cheap alternative to the expensive crystals and other difficult-to-fabricate materials often used in optical processing. Using a new figure of merit, physicists and electrical engineers at the University of Southern California, Los Angeles, have determined that cesium cells are almost 100 times more useful than competing nonlinear optics. To demonstrate this high performance in practice, researchers have successfully used the cells to achieve efficient degenerate four- wave mixing in simple optical correlator experiments.
Nonlinear optical (NLO) materials have one basic thing in common: they undergo a localized change in refractive index that varies with the exposure to a particular wavelength of light. The problem is that this can be all they have in common. Such materials can be gas, liquid, or solid; need anything from huge to tiny amounts of energy to produce an effect; retain their altered state indefinitely or for picoseconds; and operate at many different wavelengths. For engineering purposes, this can make NLO materials difficult to assess. Whether or not they are practical depends on the application.
At USC, researchers were experimenting with cesium vapor cells as NLOs for optical processing.1-3 Sodium vapor, another alkali- metal, had been considered for this kind of application in the past. It was eventually neglected by most researchers, though, because it had to be maintained at high temperatures and needed an argon-pumped dye laser to operate. Instead, the preferred choices became bismuth silicon oxide (known as BSO) and iron- doped lithium niobate (Fe:LiNbO3) crystals, joined more recently by multiple-quantum-well (MQW) structures. With the advent of commercial laser diodes operating in the near infrared, however, the USC scientists determined that atomic vapor was worth a second look. Because the melting and boiling points are much lower for cesium than for sodium, cesium vapor was now a practical option: it could be contained in simple, cheap cells and operated by small, cheap lasers.
To demonstrate their effectiveness in optical processing, researchers determined a figure of merit that could be used to compare all competing materials: one designed to allow the fastest throughput of data with the smallest energy expenditure. They decided to relate this quantity to what is probably the most widely used optical processor to date, the joint-transform correlator. It works by interfering the fourier transforms of two images (one reference, one test) to form a hologram, which is then read out using a probe beam. A large correlation peak occurs at the point where the images are most similar, thus identifying the input object. This kind of processor has been proposed for all sorts of target recognition tasks, and is now used in industry for quality control.
To determine the third-order NLO material that will give the best results in any given correlator, researchers came up with was PPP: the number of photons per pixel required in a correlator to generate an output of one photon per pixel within the response time. This figure of merit
2/3 1/3 I tau dm^2 z PPP = ( ---------) ( -) , h n Rtakes into account the total input intensity required (I), the amount of the probe beam that will be reflected from the correlator (R), the minimum pixel size (dm), and the proportion of the overall input allotted to the probe z. As is shown in table 1, even though the pixel size has to be large with cesium vapor (because of the movement of the atoms within the gas) the material's overall sensitivity (I) and fast response time to allow it to perform almost two orders of magnitude better than the best of the rest. Also, according to researchers, inserting a buffer gas with the cesium could improve dm and so make it even more competitive.
Researchers say that, as well as having the lowest PPP, the cesium vapor has the advantage of flexibility: it is cheap and easy to make the glass containers in different sizes and shapes. Disadvantages include optical pumping and the need for single- frequency lasers. To demonstrate their use in practice, USC scientists used cesium vapor cells in joint-transform correlator set-ups. Figure 1 shows the theoretical and experimental results of comparing a letter A (the test object) with a reference set of the letters A, B and C.
Sunny Bains, based in Boston, writes about optoelectronics, holography, robotics, and electronic imaging.