Making slow light behave
Slow light has aroused considerable interest in recent years, in part because time delay could benefit a range of applications, including buffering and channel synchronization in telecommunication networks, lightweight and broadband delay lines in phased array antennas, and low power threshold nonlinear switching devices. But advances come at a price. Slow-light systems do not offer ideal response for information to be carried in the form of short pulses, with consequent difference between input and delayed output.
In addition, slowing pulses without distortion theoretically requires a perfectly linear phase response across the entire signal bandwidth. In real systems, this rarely happens. Deviation from linearity causes the short pulse to spread out in time, a phenomenon known as dispersion. Neighboring pulses start to overlap, and the information is scrambled and lost.
Scientists have developed a number of ways to linearize the phase response by using clever structural designs that keep pulse distortion to a minimum.1,2 But a small, inevitable amount of deviation can accumulate over distance to much larger effect, leading to pulse spreading.
Our slow-light technique, by contrast, works with dispersion, which is exactly canceled out in the process. We chose a fiber Bragg grating (FBG) as our medium, as it creates the desired slow propagation without a very long structure. Second, to eliminate dispersion, we specifically launch solitons into our grating. These pulses are sufficiently intense for the glass in the fiber to behave nonlinearly, with the refractive index slightly higher at peaks than in the wings. The nonlinear effect acts as glue, exactly balancing the effect of dispersion. The resulting pulses propagate slowly and continue indefinitely without spreading. The so-called gap solitons3 in the grating differ from conventional solitons, which cancel only the simplest type of dispersion. Gap solitons cancel all dispersion, and this represents another reason to choose grating as our medium.
Our slow-light approach is demonstrated by the experimental setup illustrated in Figure 1.4 The light source is a Q-switched Nd:YAG (neodymium yttrium aluminum garnet) laser emitting pulses of 0.68ns duration. Pulses of the appropriate peak powers are then launched into the FBG to excite the slow-gap solitons. The output is measured by a photodiode, and a sampling oscilloscope times the delay.
Figure 2 shows the measured output pulses (solid lines) at different launch-peak powers. The delay of output pulses is timed from a reference pulse, in this case traveling in the medium at the normal speed of light. At peak power of 1.75kW, the output pulse emerges from the FBG with a delay of 1.6ns, on a par with other slow-light demonstrations. But, more important, the delayed output pulse does not suffer from spreading, indicating that the effect of dispersion has been canceled. In fact, in agreement with simulation, the pulse is slightly narrowed. We also found the delay to be tunable by simply adjusting the launch power, as shown by the descending series in Figure 2. It is also possible to tune the delay for a constant output pulse width by simultaneously applying the appropriate level of strain to the FBG.
The demonstration described here provides a solution to dispersion that introduces loss of information in slow-light systems. However, it is only the first step toward practical applications. We plan next to reduce the required optical power and apply the concept to a more robust and compact environment. To this end, work is under way to implement gap-soliton slow light in waveguides made of chalcogenide glasses, which can potentially reduce the power requirement by a thousandfold.5
