|
|
[next] [prev] [prev-tail] [tail] [table of contents]
A chip containing SOI waveguides was fabricated at the University of Glasgow by Marco Gnan, Marc Sorel and Richard De la Rue. The chip started with a silicon wafer, on top of which lay a 1μm thick layer of silica (which was to become the base of the waveguide), and on top of that a 260nm thick layer of silicon (which was to become the waveguide itself).
An etching mask was applied, marking out the pattern of the waveguide. The chip was coated with a 100nm thick layer of hydrogen silsesquioxane (HSQ). An electron beam was scanned along the desired path of the waveguide, modifying the chemical structure of the HSQ beneath it. The chip was then treated with a developer, which dissolved away the unmodified HSQ, thus leaving a mask in the shape of the pattern drawn by the electron beam.
The chip was then etched using inductively coupled plasma reactive ion etching. A mixture of octafluorocyclobutane (C4F8) and sulphur hexafluoride (SF6) was exposed to a high intensity radio-frequency electric field, superheating and ionising it. This highly reactive plasma was directed towards the chip, where it completely etched away the silicon, except for that beneath the HSQ mask. The silica revealed by the etching was slightly etched itself, causing the waveguide to sit upon a ledge a few tens of nanometres in height.
A schematic of the apparatus used by Wei Ding, William Wadsworth and Jonathan Knight to measure the linear dispersion of the SOI waveguide is given in figure B.1. The group refractive indices over a range of wavelengths were measured, thus providing the data used in section 3.3.1.
|
|
The light was generated by a microchip laser, and passed through a piece of photonic crystal fibre to generate an optical supercontinuum. A long-pass filter was used to remove short-wavelength light beyond the range of interest. The light was then passed through a beam chopper, adding a known periodicity to the signal in order to aid later detection.
The light was directed towards a beam-splitter, to create two separate beams. The first beam was focused into the sample chip with a lens, and then coupled out with a further lens. The second beam passed through a dummy sample consisting of focussing lenses but no chip. It was then delayed by a variable path length, using a computer controlled movable mirror. The two beams were recombined using another beam splitter, thus creating a Mach-Zehnder interferometer. A tunable filter was used to select a particular wavelength, providing an interferogram from which dispersion information could be extracted.
A spatial sample of the interferogram was taken by coupling it into a single mode fibre that lead to a light detector. The detected signal was then passed through a lock-in amplifier, which used the timing of the beam chopper as its reference frequency, allowing the interferogram signal to be separated from any background noise. By scanning the position of the motorised mirror, the relative delay due to the group velocity in the waveguide could be determined by detecting the presence of a packet of interference fringes at a particular point. Subtracting the value taken with the sample absent, gave the absolute group delay. From this, the group index could be calculated.
A schematic of the apparatus used by Wei Ding, William Wadsworth and Jonathan Knight to measure the nonlinear propagation through the waveguide is given in figure B.2. The output spectra and total output powers for a variety of input powers were recorded, thus providing the data used in section 3.3.2. The coupling efficiency into the waveguide was also recorded.
|
|
100fs pulses at a wavelength of 800nm were generated by a Titanium Sapphire mode-locked laser system. These were downconverted to a wavelength of 1500nm using a β-barium-borate optical parametric amplifier (OPA). The pulses were attenuated by varying amounts and focused into the SOI chip with a lens. The outgoing light was focused into a single-mode fibre with another lens, and then directed into an optical spectrum analyser.
To measure the coupling efficiency into the chip, the main laser source could be replaced (via a flip mirror system) with a second low-power continuous wave (CW) laser. This provided a linear propagation regime, and so by comparing the laser’s power to the output power, the total attenuation due to coupling inefficiency and loss within the waveguide could be measured.
[next] [prev] [prev-tail] [front] [table of contents]
