The samples we use are fabricated by PECVD in the cleanroom located next to COM or by NKT Integration A/S. This company produces
the highest quality samples we know of, so get your wallet out and give them a call.
The PECVD process yields very high quality samples in which we can achieve excellent waveguide performance. However, in a production scenario, the potential benefits
of UV writing could be exploited much more efficiently if samples could be fabricated without the need for cleanroom based deposition techniques.
Others have demonstrated the feasibility of this (see for instance Gawith et al. (2002), or Mairaj et al. (2002) in our
reference list).
Prior to UV writing the samples are loaded with molecular deuterium at a pressure of up to 2000 bar. This greatly enhances the photosensitivity
permitting index changes up to 0.022 to be achieved. Our loading setup was custom made by Granzow A/S and
features a wide range of safety measures to ensure that nobody gets killed.
Shown below is the central part of our direct UV writing setup. We use a Sabre FreD UV laser from Coherent Inc. to produce a beam of 257 nm radiation.
This beam is stabilized in direction and position, corrected for astigmatism, spatially filtered and finally focussed onto the sample with a MicroSpot
UV objective from OFR Inc.
Here is a closer view of the sample mounting apparatus. A sample is lying on the copper plate, which can be thermoelectrically cooled to -40 C during
UV processing. Thereby the rate of deuterium outdiffusion is dramatically lowered, thus stabilising the photosensitivity. Without cooling we can have as little
as 15 minutes (depends on sample type) of UV processing time to achieve good waveguide performance, with cooling we have more than 24 hours.
The UV writing has begun. The cooled sample is housed in a small vaccum chamber (actually, the chamber in the picture is flushed with nitrogen,
but we stopped doing this due to air-fluctuations at the sample surface which lead to waveguide roughness). The sample is scanned using computer
controlled DC stages (PM500) from Newport Inc.
A close-up view of direct UV writing in progress. The waveguides are clearly visible, as is the small spot of UV induced luminescence that is
associated with the index-change process.
Besides direct writing of waveguides we also can use the setup for various other UV processing techniques. Below is a colorful view of Bragg
grating fabrication in UV written waveguides. In this case, the waveguides are flood illuminated through a phase mask. Other techniques for combining
waveguide and grating fabrication are discussed in the talk from OFC'04.
Below is shown a configuration that combines UV writing of waveguides with UV trimming. Fiber in/output leads can probe the component performance
during or immediately after fabrication and changes can subsequently be made very easily.
We only use UV trimming in special cases, since it adds to the number of process steps and thus also the overall process complexity.
We prefer to make the waveguides and then anneal the sample to remove residual deuterium. Annealing also removes part of the induced index change,
so that what remains exhibits a much larger degree of long-term stability at the specified operating conditions. Below is a picture of our measurement
setup, located on the same optical table as the writing setup. Here we can image waveguides with sub-micron resolution, measure losses and do
spectroscopy.
So, what do the UV written samples look like? Here are a few examples:
A chip with waveguides and components:
A multichannel VOA chip with gold electrodes:
A 10-40 GBit/s OTDM:
An intersection of a UV written waveguide and a microfluidic channel:
Luminescence microscopy of UV written waveguide core:
The result of a exposing a sample for 10 seconds with a stationary UV spot:
All images on this page are copyright Research Center COM and may not be used without permission.