Silicon Nanophotonics


Having gained exposure to and experience with MEMS, as an undergrad, I sought to broaden my capabilities and explore other facets of nanoscience. My postgraduate work was with the UW Nanophotonics Lab; I remained at the CNF and focused on the fabrication and characterization of nanoscale optical devices in silicon.

Silicon is a particularly appealing material for moving photons through, due to a high optical index, deep existing CMOS process capabilities and high optical nonlinearities. For example, the Kerr coefficient is factor of 100 greater than that of the silicon oxide used in optical fiber. However, taking advantage of these properties requires small structures, and increased performance and device integration requires ever finer control of fabrication processed.

There is a substantial interest in novel waveguides and structures on the nanoscale. For example, slot waveguides engender tightly confined optical modes in free space. Filling this slot with optical polymer yields tight field confinement in a highly nonlinear material; filling this slot with a liquid enables fine control of biopolymer passage through the waveguide. However, routine fabrication of these devices requires electron beam lithography over large areas with small feature sizes, and presents integration challenges with conventional CMOS approaches. My post-grad work consisted of heading the nanofabrication portion of a collaborative effort to further explore the unique capabilities of this platform. I became a regular user at the University of Washington NanoTechnology User Facility and MicroFabrication Facility, to supplement the work done at the CNF.

While the potential capabilities of the slot waveguide platform were clear, we sought an appropriate demonstration of the results achievable by combing a tightly confined optical mode and a highly non-linear polymer. To this end, we envisioned a Mach-Zehnder interferometer with independent electrical bias through the two legs, and with deposited optical polymer in the slot waveguides.

I developed a fabrication process for this device using electron-sensitive flowable oxide as the electron beam lithography resist, and an inductivly-coupled plasma (ICP) chlorine-based chemistry as the silicon etch step. Of particular concern was pattern transfer fidelity – the waveguides and slots are particularly sensitive to edge roughness, and we sought to minimize optical losses in the silicon layer.

We then demonstrated the switching performance, setting a new lower bound on the voltage required to switch the device with maximal extinction. In this work, as part of the STC I was responsible for device fabrication, the CalTech group was responsible for device test and design, and the University of Washington was responsible for the polymer synthesis. The results were received by APL in late 2007, and have been cited extensively.

In addition to our work with slotted waveguides, I found it interesting to explore other systems that use top-down nanofabrication to address problems in biology. One such case is the on-chip optical interrogation of chemical bonds, using Raman or absorbance spectra. While single-mode fibers, detectors and sources – specifically, quantum cascade lasers – are increasingly available in the mid-infrared, conspicuously absent are a number of components engineers have come to rely on in telecommunications bands. High bandwidth modulators, splitters, and tunable filters rely on guiding light in the first place; more elaborate devices – such as on-chip Raman or FTIR – can be built up from these. With above in mind, we set out to develop CMOS-compatible silicon waveguides operating at a test wavelength of 4.5 micrometers, and then build up a suite of devices already available to designers at working at 1550 nanometers.

Using sapphire as a substrate eliminated issues of mode leakage – and losses! – into the substrate, but required some finesse in the fab. To this end, we used a modified electron beam lithography and tetrafluoromethane RIE processes inspired by the silicon waveguides we had previously made, but appropriately modified to correct for the changed dimensions and roughness sensitivity. Over the course of a number of test runs, I developed and characterized the fabrication procedure to get on-target dimensions. Moreover, I developed the etch chemistry to be compatible with the sapphire substrate.

Moreover, I was instrumental in developing the first round of test equipment for this project – specifically, developing and troubleshooting an optical test setup and understanding the instrument’s failure modes. This was my first experience with optical test; much like my previous experience with electrical test, I quickly understood the limitations of the device and catalyzed the development of the configuration we used to publish this result.

This work was published in Optics Express, stands as the first ever implementation of mid-IR waveguides in silicon, and has been cited over 60 times.

This period in my technical development was particularly interesting. I viewed it as a chance to learn a new field, orthogonal to the work I had done before. More importantly, I viewed it as a chance to gain further perspective on what it takes to get experiments to work, and to work quickly and well.

Moreover, this was a good opportunity to add new skills to my toolkit. While the bulk of my focus had previously been on MEMS fabrication and test, developing optical experiments was a new and worthwhile experience, that later served me well during my tenure at a startup company.

However, the most important lesson during this year were meta-technical: I learned how to manage and own an entire branch of the project – the fab responsibilities were exclusively my own – instead of running to advisors and post-docs at the first sign of trouble. Four time zones away from my collaborators, I learned to deliver independent results and run the collaboration to eventual success.

← Previous
Next →