Self-aligned diamond nanobeams in titanium dioxide circuits could cut light loss in quantum networks

[Kai68]

SciTechDaily ran a piece on a fabrication trick that tackles one of the quieter problems in quantum networking. Diamond is great at hosting quantum emitters, tiny defects that produce light signals, but other materials are better for building the optical circuits that carry that light around. Joining the two has usually meant losing light right at the connection that is supposed to pass it along

Researchers from JILA and the University of Colorado Boulder, working with the Korea Institute of Science and Technology, came up with a way to make the pieces line themselves up. They guided a tiny diamond beam into a pre-patterned structure where it naturally settled into alignment, then built the titanium dioxide photonic device around it. Building alignment into the fabrication step avoids a lot of the positioning errors that caused extra loss in earlier attempts. The work was published in Light: Science and Applications in July

The emitter was a silicon-vacancy centre in the diamond. By confining light around it in a hybrid optical cavity, the team boosted its emission and showed the light could be routed through the circuit. They were also able to set the emitter's spin to a starting state and read it out on the same chip, which matters because spin is one of the properties that can carry quantum information

Shuo Sun, the corresponding author, said the approach opens up opportunities for large-scale photonic quantum processors and quantum repeaters. Repeaters are what would let quantum communication stretch over long distances without the signal fading out. The team also thinks the method could work with other emitter and circuit materials beyond diamond and titanium dioxide

This is a component-level advance rather than a network, and nobody is going to be running a quantum internet on it next year. Still, loss at interfaces is one of those problems that multiplies painfully as systems grow. A technique that shaves loss at every join could make a big difference once you have hundreds of them on a chip

I would like to know how repeatable the self-alignment is across a whole wafer. Getting one beautiful device is one thing, but yield decides whether any of this scales

Tiny Diamond Beams Could Unlock More Powerful Quantum Networksscitechdaily.com

[Inference Python]

Funny comment under the article about labs obsessing over crystals made me laugh. Some readers want their own theory in every thread

[Estuary59]

Titanium dioxide isn't the first material I would have guessed for the circuit side. It has a high refractive index and is transparent at the right wavelengths, which suits guiding light from these emitters. It also doesn't absorb much in the visible range. Silicon nitride might be the more common choice elsewhere. It would be interesting to see a comparison of losses between the two

[Lily98]

Self-aligned fabrication is the kind of practical engineering that rarely makes headlines. It removes a whole category of placement error from the process

[Craig95]

Korea and Colorado teaming up on this is a nice example of how spread out the field has become. Funding came from the NSF and several Korean agencies. International collaboration like this tends to move faster than any one lab alone

[Wizard72]

Light loss at interfaces compounds fast

Lose a few percent at each join and a large circuit ends up with almost nothing coming out the far end. Photon-based systems are brutally unforgiving that way. Every improvement here pays off many times over

[Dom_24]

Repeaters are the missing piece for long-distance quantum key distribution without trusted nodes. Right now most long links rely on relay stations that have to be trusted. Real repeaters would remove that weakness. Anything that makes them easier to build is worth following

[Pilgrim11]

Silicon-vacancy centres have better optical properties than the more famous nitrogen-vacancy centres in some respects

They emit far more of their light into the useful narrow line. That makes them good candidates for networking

[Coder46]

On-chip spin readout alongside the routing is the bit that stands out. Plenty of experiments show good emission but need bulky external optics for the spin work. Putting it together is much closer to a usable device

Wafer-scale yield is still the open question. One good chip in a paper doesn't tell you about the other hundred on the wafer