A wave of integration

14 March 2013

Researchers in Japan have designed and fabricated a compact silicon wire array waveguide, which can minimise phase error associated with the wavelength division multiplexing necessary for all silicon photonics applications.  Two features allowed the team to effect this: the efficient and simple design of the device, and the use of ArF immersion lithography to optimise the manufacturing process.  This is a project undertaken by a number of institutions and corporations, with researchers contributing from Oki Electric Industry, AIST and NTT. The target of the project is advancing the technology for the LSI and photonics integrated systems to realise on-chip data centres.  It began in 2010 and ends in March 2014.

Combine. Split.

Silicon photonics, the use of silicon as an optical medium in integrated circuits, has potential to speed up and simplify a range of applications.  Optical interconnects can transfer data considerably faster than electronic interconnects and, as modern integrated circuits are already silicon based, silicon waveguides can easily be incorporated.

Another example is that of optical routers and signal processing.  Fibre optic communication currently requires a complex set-up at the transmitter and receiver to transfer the optical data to the electronic components.  However, such communication usually uses infrared light, which silicon is transparent to.  This means that silicon waveguides, integrated into the transmitter or receiver circuit, can simplify these devices.

These two applications, along with others in long range telecommunication, require the optical signals to be (de)multiplexed, usually by wavelength.  If the number of channels for this multiplexing is greater than eight, the standard approach is to use an array waveguide grating.

Wide lenses

The array waveguide grating the team used, as Dr Hideaki Okayama explained, “consists of a very small Si waveguide core and a surrounding lower refractive index cladding material. Due to the large refractive index difference between the Si core and the clad, the light can be confined in a very small region - the core size can be smaller than 500 nm.”

To achieve multiplexing, first the input signal is diffracted and shared among the arrayed waveguides, with channel as a function of wavelength. Each signal then propagates along its own path, passing through a series of narrow channel waveguides and wide optical waveguides.  Okayami explained the effect of this process: “The wide and the narrow arrayed waveguides function like lenses and a grating respectively.”

Okayami continued to explain the final stage of the multiplexing, saying that “at the output wide waveguide, the light propagated in the arrayed waveguides are focused into output waveguides. Due to the length difference of the waveguides in the array, the propagation direction of the light becomes a function of wavelengths in the output wide waveguide. Different wavelength lights are focused into different output waveguides.”

Immersing the substrate

The design Okayami described very effectively helps to mitigate phase noise in the system – the problems caused by any unwanted phase discrepancies in the multiplexed light.  However, manufacturing inaccuracies will also have a drastic effect on the phase error, and the team have used a technique known as ArF immersion lithography to overcome this.  Electron beam lithography is the standard process, but Okayami explained that “this uses area-scanning and is thus a time consuming process.”

ArF immersion lithography is photolithographic technique that uses laser light generated by an ArF excimer, focussed by a liquid onto the substrate to produce a device pattern. Analogously to photography, the pattern can then be chemically etched into the substrate, producing the object.  This is a one shot process (simple, compared to the electron beam method) and the high frequency of the ArF generated light, combined with the highly refractive liquid, means a very focussed beam, and thus an accurate image can be generated.

However, a number of problems still need to be overcome before the design can be used in real world applications: the crosstalk between adjacent channels is still too high and needs to be reduced below to -30 dB.  Additionally, polarisation sensitivity must be reduced, temperature stability improved, and a flat-top response of the wavelength transmission peak introduced.

Dr Okayami concluded that several origins of these problems have been identified and that the team are working to improve the device.  He said that “our main target is the optical interconnection for multiple electronic processor units. The ever increasing signal speed and power consumption in the interconnects will require optical technology in the near future.”

Further reading

This article is based on the Letter: Si wire array waveguide grating with parallel star coupler configuration fabricated by ArF excimer immersion lithography (new window).

PEcst: http://www.pecst.org/index_en.html (new window)

A PDF version (new window) of this feature article is also available.

Journal content

Cover of Electronics Letters, volume 50, issue 08

View all content

Browse or search all papers in the latest or past issues of Electronics Letters on the IET Digital Library.