100 cascaded microrings form on-chip optical buffer

All-optical information processing on a chip will require controllable buffering of optical signals.

All-optical information processing on a chip will require controllable buffering of optical signals. While “slow-light” approaches are natural buffers of light, they are bulky and operate at bandwidths only up to the megahertz range. Researchers at IBM (Yorktown Heights, NY) have developed a non-slowed-light approach that shunts the optical signal into a series of up to 100 cascaded low-Q microring resonators; the whole device fits in a 0.09 mm2 area. Delay times of up to 500 ps were demonstrated; a delay of 10 bits was achieved at a 20 Gbit/s data rate.

A silicon-on-insulator wafer with a 226 nm thick upper silicon layer was etched to form photonic-wire waveguides that confined light within a 0.1 µm2 cross section. Two designs were examined: coupled resonators with many rings cascaded in a series and all-pass filters, in which rings in parallel coupled light from one waveguide to the next with their individual delays summing to the total delay. The experimental control was a bent-waveguide delay line. The two designs were five times smaller than the control, but had three times the insertion loss. Contact Yurii Vlasov at yvlasov@us.ibm.com.

More in Optics