The authors were targeting sensing applications at 3.4 μm and note that the sapphire limits the devices to a maximum wavelength of about 5.5 μm. However, the use of a solid substrate means that recent demonstrations of silicon film transfer to Calcium Fluoride substrates can be exploited, providing a pathway to utilising the full transparency bandwidth of the silicon out to just over 8μm, with the added benefit of a lower substrate refractive index enabling broader slow light bandwidth. This provides a wider range of applications than perhaps has been previously anticipated.
The authors also observed a low minimum propagation loss of ~12dB/cm and good sample-to-sample repeatability, which confirms that the fabrication tolerances are indeed relaxed at longer wavelengths. Based on experiences with MIR loss spectra in Chalcogenide and Silicon on Sapphire waveguides, a considerable part of this loss may come from absorption from surface chemisorbed hydrocarbons and/or Si-F surface bonds from the HF smoothing process. This would imply that at longer wavelengths the losses may be reduced to a few dB/cm or possibly less with further optimisation. This would allow longer devices and coupled with the slow light characteristics could enable very sensitive devices to be built.
The careful nature of the work performed also showed that tuned arrays of devices can be reliably built in the MIR, which is also important in a sensing context to target multiple bands/species. The use of silicon means that the huge efforts directed at stringent dimensional control under mass scale manufacturing conditions in the VLSI sector can be directly leveraged to ultimately gain exquisite control/reproducibility and large volume production with little effort, something currently unavailable in any other waveguide materials system. The solid underclad also provides a level of robustness that overcomes many of the issues that have previously limited PC waveguide devices, and will also play an important role in obtaining high yield (particularly important if many different PC devices are integrated onto a single chip for broad spectral coverage or multispecies operation). There are very important attributes in a practical device platform.
Lastly, there is a considerable body of work in the NIR on enhanced nonlinear interactions in slow light PC waveguides, and clearly this can be exploited in the MIR with the system demonstrated in this paper. This itself could lead to a range of interesting functionalities that could not previously be easily integrated, for example, low cost high sensitivity MIR upconversion based detection. Again, this makes the platform quite attractive for a range of applications.
You must log in to add comments.