Spotlight on “Si-CMOS Compatible Materials and Devices for Mid-IR Microphotonics”

Having in mind the huge success of optical fibers, everyone would be tempted to say that silica is a material the photonic world cannot do without. This is almost true in the near-IR wavelength range, where silicon dioxide (together with its doped compounds) is one of the most used materials in free-space, fiber and planar integrated optic devices. Unfortunately, moving to different wavelengths, your “best-friend” material can become your biggest problem. And this is the unhappy end of silica in the mid-IR range. The wavelength range from 2 um to 8 um is becoming of large interest for applications in biochemical detection and toxic gas/liquid monitoring. Photonic integrated circuits would allow finger-printing spectrometry and multiple chemical sensing on a single photonic chip, thereby reducing the cost and size of the detection system. Yet, at wavelengths above 3.6 um the optical loss of silicon dioxide dramatically increases, making its use impractical, and, if we want to play with mid-IR photonics, we are bound to get rid of it. The good part of this story is that we do not have to reinvent mid-IR photonics from scratch. For instance, silicon has a good transparency up to a wavelength of 8 μm and luckily can be saved from the mid-IR-material trash bin. Silicon photonics has evolved from a niche research field to one of the most prominent technologies for near-IR integrated photonics, enabling a dramatic size and cost reduction of photonic components, and providing also the possibility to leverage CMOS fabrication technology and to densely integrate CMOS electronics with photonics. The problem to be solved with current silicon photonic devices is that they are conventionally fabricated on a silicon-on-insulator (SOI) platform, where silicon dioxide is used as under-cladding material. To make silicon integrated devices work above a wavelength of 3.6 μm alternative under cladding materials have been investigated, such as porous silicon, sapphire and silicon nitride. However silicon waveguides based on these technologies suffer from either high propagation loss or a limited transparency wavelength range. The best solution would be a silicon waveguide completely surrounded by air, but the realization of suspended structures poses severe technological challenges. The idea proposed in the work by A. Agarwal and co-workers is that the optical loss of an SOI waveguide can be decreased considerably without the need for removing completely the silicon dioxide beneath the waveguide. The novel waveguide concept demonstrated in this work is realized by symmetrically undercutting a conventional SOI waveguide, leaving a narrow pedestal of less than 20% of the waveguide width (about 3 m) to sustain the silicon waveguide core. Engineered Silicon on Oxide Undercladding Pedestal (SOUP) waveguides exhibit a 10 dB/cm improvement in optical loss for λ > 5 μm. The SOUP waveguide can be advantageously fabricated through fully CMOS-compatible multistep dry/wet etching processes and exhibit good optical propagation over a broad mid-IR regime. To realize fully integrated mid-IR on-chip systems, the authors address also the problem of on-chip light detection. To this aim, they propose the use of thermally evaporated polycrystalline PbTe thin film mid-IR detectors. Compared to alternative approaches, PbTe detectors exhibit good chemical stability, can be realized through an easy, low-cost and CMOS compatible deposition process, and offer good room temperature responsivity. An efficient solution to integrate PbTe detectors with chalcogenide glass (ChGs) waveguides is also proposed, which is based on a suitable spacer between the waveguide core and the PbTe detector. ChGs form a viable materials solution for mid-IR photonics applications due to their wide IR transparency window which extends beyond that of SOI and the ability to use low-cost fabrication processes. After reading this work, we are happy to know that a variety of building blocks, including low loss waveguides and integrated thin film detectors, are ready for broad-wavelength-range mid-IR applications. The possibility of “recycling” Si-CMOS compatible materials and devices used in the near-IR range, such as SOI waveguides, makes these results even more attractive. The main challenges to cope with are now on the on-chip integration of CMOS compatible light sources. This issue is still open in the near-IR range and will become soon a hot topic for mid-IR photonics. Or better, the race has already started...