I've Got Silica on Silicon on My Quantum Stereo

Linear optics quantum computing, where one combines linear optics with the nonlinear processes of single photon creation and single photon detection, is a relative newcomer onto the scene of possible routes toward quantum computing. Whenever I think about these schemes, what jumps into my head is a crazily filled optical bench, like the one below from the Zeilinger group:

Now, I'm but a mere theorist, but I think even theorists like me understand that trying to build a large scale version of this scheme, which has considerable overhead behind it in terms of the number of modes needed, is a huge challenge. A further bane of these approaches is that one needs to keep the optical paths stable on a length scale less than the wavelength of the light. This requires some pretty sophisticated stable interferometers. Which is why it is nice to see the new paper from Jeremy O'Brien's group in Bristol: A. Politi, M. J. Cryan, J. G. Rarity, Y. Siyuan, and J. L. O'Brien, "Silica-on-Silicon Waveguide Quantum Circuits," Science Express Reports, published online March 27, 2008 (arXiv:0802.0136.)

What this group has done is demonstrated that it is possible to implement the photonic quantum circuits of the linear optics quantum computing using silica waveguides on a silicon chip. The basic setup of integrated optical devices is that light is guided through a material of slightly lower index of refraction (here silicon is the cadding and silica is the material for the core.) By doing this in a regime where optical lithography can be used to construct the waveguides, one can construct pretty good waveguides, which allow a single mode to propogate in the core. Of course, just being able to build waveguides isn't what you need to perform linear optical quantum computing. You also need to be able to enact those linear optical gates. To achieve this, the authors constructed waveguides where the two modes where brought close to each other such that the evanescent modes overlap (directional coupler.) The paper has a schematic of this setup:

Note that because of the reliability of optical lithography, you can guarantee fairly strongly that the mode matching behavior you want between different elements in your linear optics quantum circuit is achieved (disregarding getting the mode into and out of the entire device (40 percent loss going in and 60 percent going out, in this experiment.) As a side note I am amused by one of the words used in describing this coupling: "buttcoupling." Butthead and Beavis would be proud.)

Okay, so that's the setup, what does the group actually achieve? Well they show that their devices can be used to obtain pretty good quantum interference effects (about 95 percent fidelity.) Further they implement a simple controlled-NOT gate in the linear optics scheme (where a CNOT is implemented probabilistically) which achieves a fidelity of about 94 percent. The nice thing about this whole setup, remember, is that you don't need to set up a really complicated interferometer that needs to be extremely stable. Instead you get some amount of this stability directly from fidelity of the lithography. Which makes this a very nice and beautiful step toward building a quantum computer in the linear optics quantum computing scheme.

So, in my head, I always keep a list of different schemes and what I think their ability to scale up will be. Superconducting circuits have always been near the top of this list, but problems of cross-talk have always seemed to me to be a big pain for these setups. But now I'm going to have to move linear optics quantum computing up a few notches on my list. Very cool. Now get back to the lab and make your favorite qubits scalable, damnit!

Oh, and bonus points if you can identify the musical reference in the title of this post.