28 Sep 2005
Low-cost, polymer-based spectrometer circuits developed in Germany could inspire a whole range of hand-held and miniature products. James Tyrrell finds out more.
Scientists at the Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institute (HHI), Germany, have built a fingernail-sized spectrometer from a polymer chip. Featuring integrated photodetectors and with other functions such as a light source well on the way, the tiny optical unit could lead to a host of new applications.
At the heart of the multichannel spectrometer is an arrayed waveguide grating (AWG) made from a specially developed cross-linked polymer. Containing an arrangement of waveguides of different lengths, the AWG optically filters the incoming signal by using interference effects to couple a specific wavelength to each discrete channel.
The HHI team claims that its AWG device, which features no moving parts, opens the door to spectrometers that are much smaller and far less sensitive to vibration than instruments based on mechanically controlled diffraction gratings.
The Berlin-based group feels that its spectrometer chips could have 8-100 channels with a channel spacing of 0.1-20 nm without compromising the unit's small size. Operating across the infrared range, HHI's initial prototype has eight channels. Monitoring each channel is an indium phosphide (InP) photodetector (PD) located on the optical motherboard's top surface.
A 45 ° mirror connects the outputs from the AWG to the device's integrated PDs. "We can avoid active alignment by using a 45 ° mirror," HHI project manager Norbert Keil told OLE. "For this example we used an industrial pick and placer to set the array of eight photodiodes."
To cut down on the assembly time, Keil and his colleagues are keen to avoid laborious alignment techniques, opting instead for passive methods that suit cheap mass-production.
"60-70% of the total cost [of the device] is in the packaging, which includes fibre-chip coupling," said Keil. "We want to make grooves for the fibre, which you can do very easily in plastic, and then just plug in the [input] fibre."
In recent experiments with a small four-channel AWG, the team has added a taper structure to reduce the optical field mismatch between the input fibre and the waveguide to just 0.2 dB. Without the integrated taper, the value rises to 1.5-2 dB.
Unfortunately for the team, the plug-in concept becomes much harder to achieve as manufacturing is scaled up to produce more than 200 devices per wafer. "At the moment we have 4 inch [polymer] wafers and we are talking about 6 inch, or even larger, to push down the price," revealed Keil. "The technology challenge is to write the same taper structure everywhere on the wafer surface - 8 inches is a large area and the etching process has to be very well controlled."
With the potential to reduce spectrometer costs by a factor of 10, the team's polymer technology is grabbing industry's attention and was recently showcased at LASER 2005. "We had discussions recently with a US medical company that is interested in a minispectrometer for dental applications," commented Keil. "One significant argument for the company was not just price but also size."
Keil says that a compact and robust unit may also suit the cancer-screening market. Here the solution could be a very simple, affordable polymer circuit that can pick out just one or two tell-tale spectral lines.
"To commercialize a device we have to co-operate with companies that are already in the market and [in the case of screening for cancer cells] with large hospitals that know what signals to expect," he added. "We can then design a very small and cheap device that detects exactly at that wavelength."
The researchers are currently developing a transceiver product for the telecommunications sector, taking their hybrid integration process a step further by including laser diodes. With a timescale of two years for the transceiver unit, Keil thinks that a spectrometer product could be as little as 12-18 months away, thanks to the devices' shared manufacturing platform.
Currently the HHI team makes its polymer devices using a reactive ion-etching technique. The process begins by spin-coating a 15 μm thick polymer layer onto a substrate and baking it in an oven at 250-300°C for several hours. This helps to cross-link and stiffen the thick film, which is now ready to receive a thin 3-5 μm layer of waveguide material.
"On that layer you put the etching mask, taking away material on the left and right of the mask to make a channel waveguide," said Keil. "You then remove the etching mask and coat with a top cover to complete the waveguide structure. In principle there are just three or four processing steps."
The researchers considered other fabrication methods, such as moulding or hot embossing, but found that the tiny waveguide structure was often ripped away from the polymer substrate towards the end of the manufacturing process.
"The other thing [with hot embossing] is that you need a thermoplastic polymer that is soft and movable at the processing temperature, which is counterproductive for the high thermal stability of the device," said Keil. "For example, our cross-linked polymer is not thermoplastic."
Keil is confident that his device, designed for polymer processing at temperatures in excess of 250 °C, can tolerate operating conditions of 85 and 90 °C. He points out that, with a glass transition temperature of approximately 100 °C, previous PMMA materials systems have struggled in telecoms-style temperature cycling tests, which are typically in the −40-85 °C range.
Transmission has also proved to be a stumbling block for the use of polymer materials in optical systems. "We started [more than 10 years ago] with PMMA, an acrylate-based polymer with a typical loss of 1 dB/cm at a 1550 nm wavelength," said Keil. "Now, after many industrial driven research projects, for example with Pirelli and Alcatel, we are using a material system with a loss of 0.28 dB/cm."
Additionally, because the material enables a high refractive index contrast between the optical waveguide and the cladding layer, the group can design so-called supercompact AWG's with a radius of curvature of less than 1 mm. For example, its four-channel AWG measures just 10 × 3 mm.
Keil accepts that humidity still presents a challenge to polymer systems, but he has a solution at hand. "Most of the polymers can have water uptake, resulting in [signal] loss and a change in refractive index," he said. "It means that we have to seal the device, but this is not a [major] problem and there are manufacturers here in Berlin that have developed some cheap hermetically sealed packages."