At IBM Research – Zurich, the advanced materials and photonics group collaborate on photonics technology for future computing systems and the teams are currently researching the use of silicon photonics for intra-chip data communication to overcome bandwidth limitations imposed by copper wires. Optical link technology -- or "CMOS integrated silicon photonics" -- could ultimately enable data transfer rates of several 100 gigabits per second.
CMOS integrated silicon photonics combines electrical and optical functions in the same technology platform: silicon. The high integration density of optical and logical components in a 3D design will increase the performance, and significantly improve the energy efficiency of these new chips. And because it can be seamlessly integrated into standard CMOS fabrication lines, the cost of the new technology will be kept low.
|Stefan Abel, a Pre-Doc at IBM in Zurich working in his lab.|
Today, high speed electro-optic modulators and switches are realised in silicon photonics by exploiting the plasma dispersion effect: free charge carriers, in which electrons or holes in the semiconductor can move around, are injected or depleted from the optical waveguide to induce a local change of the refractive index.
By applying an electrical current, this effect can be exploited to build switches or electro-optic modulators. In telecommunications, lithium niobate modulators are well established as high-speed electro-optic devices. This material system only requires an electrical field and does not require a steady current flow.
This has tremendous power efficiency advantages in switches or in device trimming functions. In the increasingly miniaturized field of silicon photonics, however, lithium niobate could not be exploited yet due to integration challenges.
IBM scientists recently succeeded in embedding an oxide called barium titanate with very large optical nonlinearities into the silicon platform. Their results were recently published in Nature Communications.
Stefan Abel, an IBM PhD. student and co-author of the paper explains.
What presents the biggest challenge in silicon photonics?
|Polarization changes of a laser operating|
at the telecommunication wavelength 1550nm
were measured in transmission to the sample
in order to determine the
Credit: Nature Communications
Stefan Abel: Currently a tremendous effort is invested in the field of silicon photonics, and considerable progress has been made bringing this technology closer and closer to commercialisation.
The fabrication is rather challenging due to the very tight tolerances required for photonics. Also temperature variations on the chip may have to be compensated for. While some devices such as waveguides and modulators are very mature, some others still need improvement, such as power efficient tuning elements or integrated light sources.
Q. How did the team come up with the idea to test barium titanate?
SA: To realise electro-optic devices, non-linear materials that convert an electrical signal into a change of the optical material properties, for example a shift in the refractive index. One drawback of silicon is that it does not exhibit strong optical non-linearities due to its crystal symmetry.
You may not know it, but we benefit from non-linear optical materials such as lithium niobate everyday, as it plays a major role in the backbone of the Internet. Telecommunications companies wouldn’t be where they are today without it because it features properties for encoding electrical signals into optical signals, a key requirement for making phone calls. Barium titanate has one of the strongest optical nonlinear properties among all materials, but so far it cannot be produced on large wafers with high quality.
For us, barium titanate is a well-known material. IBM has decades of experience in working with it, particularly on silicon -- but for completely different kinds of applications. The link between the superior optical properties and the integration into silicon photonics is the key of our new findings.
Q. The "Nature Communication" paper mentions that you achieved a 5x improvement of the optical non-linearity. Were you surprised?
SA. I'll put it this way: initially we didn’t know if any effect would occur. The challenge is that the nano-scale properties of the material, and the properties of thin films, could be very different from those on a larger scale. So, we were very happy to see such a dramatic improvement and it’s only the beginning. With some adjustments that we are preparing now, we could see greater improvement.
Q. Can you talk about any initial applications?
SA. We anticipate the integration of thin films made of barium titanate on silicon to first appear in ultra-compact integrated devices, including modulators, tuning elements, and bistable switches. Devices of that type are novel in the field of silicon photonics and could help to advance the functionality and power efficiency of optical networks.
The paper, "A strong electro-optically active lead-free ferroelectric integrated on silicon”, 10.1038/ncomms2695, was published online in Nature Communications’ on 9 April.
Labels: CMOS, IBM Research - Zurich, photonics