An ideal medium for transporting photons between two points would have no absorption or scattering losses, no material dispersion, and no nonlinear response. This medium does not exist, but only in the semantic sense. It is all around us in the form of vacuum. Air is not quite vacuum, but still has ultralow loss, dispersion, and nonlinearity as compared to glass. Air core waveguides have existed for some time in the form of antiguiding capillaries, and these can transport light over meters, limited by refractive modal loss. However, when hollow core photonic bandgap fibers (HC-PGBFs) were invented, researchers could begin to think of the implications of transmitting an air-guided mode over kilometer lengths and beyond. One of the great, unrealized, promises of HC-PBGFs was that they might eventually have propagation losses comparable to or lower than standard silica fiber, but with vastly higher nonlinear thresholds, perhaps enabling communications systems with very few optical amplifiers (or none). Of course, many useful applications of HC-PBGFs have been demonstrated for atom optics, gas cells, high power beam delivery, high power pulse stretching and compression, and gyroscopes, to name a few examples. These can all be enhanced by reducing the modal propagation loss, which from a design perspective entails removing surface states and minimizing the field overlap between the air-guided mode and the surrounding glass matrix.
The lowest loss reported in a HC-PBGF to date is 1.2 dB/km, limited by scattering from the ring of glass surrounding the hollow core. This result was in part enabled by using a large, multimode core, so the propagation properties and output beam quality of this fiber are compromised for some applications. Smaller core, single mode HC-PBGFs typically have losses of 10-20 dB/km or more. In 2006, Fini proposed a way we can eat our cake and have it too: in his design, small satellite cores surround a large central air-core, stripping away the higher order modes and leaving a single fundamental mode with low loss. The satellite cores are phase-matched to the higher order modes of the central core, but not to the fundamental mode. The design is like a distributed asymmetric fiber coupler, but the operation is strictly to mode strip - propagation loss in the small satellite cores is high, so power does not couple back to the central core. While the idea is simple, fabricating such a complicated fiber is difficult, and there still is the question of whether the mode stripping would work in practice. Small axial fluctuations in the fiber geometry are all but inevitable under the best conditions, and this can easily detune the phase-matching between the central and satellite cores.
Seven years later, Fini and coauthors have now experimentally realized this design and shown that axial fluctuations (and fiber coiling) do affect the detuning, and that this is entirely acceptable, and desirable. Consider a fiber with no axial variations at all, but for which the central or satellite core dimensions are fabricated slightly off target (which is invariably a result obtained even with state-of-the art HC-PBGF manufacturing methods). In this case, the phase-matching would be destroyed and the satellite cores would not mode strip as effectively. Likewise, if the design works for a straight fiber, it cannot be coiled without destroying the mode stripping, since bending affects the mode index, and would cause the LP1,1-like higher order modes to split into non-degenerate even and odd states (so now there are two or four phase matching conditions to satisfy). However, if we include axial fluctuations in the fiber geometry, the mode stripping becomes more robust. In general, the center and satellite cores are not phase-matched, but the fluctuations ensure that they are phase-matched at some points, and this occurs over a sufficient distributed length of the fiber that the higher order modes are suppressed. If the fiber is bent or twisted, the local phase-matching conditions all change, but globally, the mode stripping remains. The authors coin the term Perturbed Resonance for Improved Single Modedness (PRISM) for this effect. The mode stripping can be empirically optimized by choosing the appropriate bend radius and orientation, and since the PRISM effect is statistical, it can be quantitatively modeled without very high resolution SEMs of the fiber cross-section. The fabricated PRISM fiber had a measured propagation loss of 7.5 dB/km, compared to 5.6 dB/km and 16 dB/km for commercial multimode and single mode HC-PBGFs, and higher order modes were suppressed by -27 dB relative to the fundamental mode, 5 dB less than the nominally single mode, smaller core commercial fiber. There is certainly still room for improvement on these numbers, though how much remains to be seen. Perhaps just as important, the authors have showed that it is feasible to make high-quality multi-core HC-PBGFs, suggesting a variety of novel designs and devices that can now be realized experimentally.
You must log in to add comments.