Scientists at Duke University in Durham, N.C., have demonstrated an optical switch capable of switching microwatt light beams employing beams on the order of nanowatts or hundreds of picowatts. Such devices are the optical analogues of electrical transistors and may be called optical transistors, the fundamental building blocks of future optical computers.Because the devices demonstrated at Duke operate at such low light levels, they may eventually lead to single-photon switches in quantum-information networks. In the shorter term, they could dramatically affect the design and operation of optical telecommunications systems.Figure 1. Two counterpropagating laser beams (shown in red) interact through a nonlinear instability in warm rubidium vapor to generate a cone of new light (shown in blue). This cone would appear as a circle (red) on a screen perpendicular to the original beams (A). Two perfectly aligned counterpropagating beams produce a circular pattern (B). If the two beams are slightly misaligned, the cone breaks up into six beams (C). If the power in the counterpropagating, slightly misaligned beams is reduced to near-threshold level for the effect, only two beams are visible (D). Images ©2005 AAAS.In the optical transistors, two counterpropagating laser beams, both linearly polarized in the same orientation, pass through warm rubidium-87 vapor. Normally, the beams would not interact, but strong nonlinear coupling between the beams and the rubidium atoms results in "mirrorless parametric self-oscillation," an instability that results in new, orthogonally polarized light generated in a cone around the two beams (Figure 1A). The effect is enhanced by a resonance with the 780-nm transition in 87Rb. So long as the counterpropagating beams are perfectly symmetric, the cone of new light is symmetric (Figure 1B). However, a slight misalignment causes the cone to break into distinct beams (Figure 1C). If the power in the counterpropagating beams is reduced, only two of the beams remain visible (Figure 1D).Figure 2. A weak switching beam (shown in yellow) rotates the two beams by 60°.Optical switching is effected starting from the condition when only two beams of the offset light are present (Figure 2A). A very weak beam switches the beams to an orientation rotated 60° from their original position (Figure 2B). In the experiment, the power in the two offset beams totaled 1.5 µW, and the switching beam contained only 2.5 nW. The switched power thus was 600 times the switching power. The situation is depicted graphically in Figures 3A and 3B. Figure 3. A switching beam of 2.5 nW rotates the two offset beams (A), containing a total of 1.5 µW, by 60° (B). A much smaller switching power, 230 pW, rotates half the power in the original offset beams to the new locations (C).Partial switching can be achieved with a much lower power switching beam (Figure 3C). In that case, a switching beam of only 230 pW switches half the power from the original two beams into the rotated two beams. The switched power is more than 3000 times the switching power, and the switching is accomplished with only ~2700 photons. Figure 4. The time for complete switching was 4 µs (A). For partial switching, the switching speed was 3 µs (B).The researchers measured the switching speed of their experimental setup by observing the signal from a photodetector placed in the location of one of the switched beams (Figures 4A and 4B). The speeds were 4 µs for complete switching and 3 µs for partial switching.