A team of researchers reckon that they have made a big step towards making ultrafast optical computing a reality with a new manufacturing method.
Scientists have been trying for some time to create light-based photonics computing that makes much faster calculations than conventional electronics, using photons rather than electrons to process information.
While the benefits of such an approach are clear, so far this has proved difficult to manufacture in reality. In order to function properly in a circuit it is necessary to create devices known as microresonators. It has been possible to produce microresonators in the past, but boffins have struggled to create ones which are viable to work for computing.
Current manufacturing involves silicon lithography, etching onto silicon wafers. However, such attempts to create working microresonators have been dogged by an inability to function well due to tiny production imperfections.
Now, a team at the OSF Laboratories claim to have come up with a method that enables precise production, by combining quantum properties of light with previously unknown qualities in optic fibres.
A silica strand of fibre was preffered, rather than the silicon wafer method, using the ability to bend the path of light without causing scattering. This can bounce light back and forth inside its core for many miles without much signal loss.
By making nanoscale changes to the diameter of silica strands the team was able to create functioning microresonators. The microresonators that were produced were able to retain light two orders of magnitude longer than lithographic processes, which could be further improved in more testing.
If enough optical fibre microresonators are coupled together - taking advantages of “quantum leaps” as part of the new design - the scientists think that they can keep light pulses in place long enough for computing purposes. So far this has been managed on a “proof of concept” scale.
The new manufacturing process means that microsresonators could be used in “specialised devices in about two to three years” and could pioneer optical computing.
More information on the method can be seen in the paper in OpticsLetters.