Voltage brings silicon-carbide quantum emitters into tune

Turning the crystal sideways

Jannik H. Schwarberg and colleagues studied V2 silicon vacancies: defects where a silicon atom is missing from a particular lattice site in 4H silicon carbide. Their optical transitions respond to electric fields, offering a way to adjust their resonance frequencies and sense their surroundings.

The device starts with a change in crystal orientation. In conventional c-plane material, the vacancies’ optical dipoles point approximately perpendicular to the wafer surface, complicating resonant illumination from above. The team instead used a-plane wafers, placing the dipoles parallel to the surface and enabling direct resonant optical access through it.

They fabricated lateral p-i-n diodes, with heavily doped p-type and n-type regions separated by a lightly doped active region. An optical window preserved a thin protective oxide above that region. After semiconductor processing, electron irradiation and annealing created the vacancies.

That sequence matters: the experiment demonstrates useful emitters in fully processed devices, rather than showing that pre-existing vacancies survived every fabrication step. The authors used a scalable CMOS fabrication process, but did not demonstrate a complete quantum photonic circuit.

Narrow resonances, adjustable frequencies

At 4 kelvin, the researchers scanned lasers across the defects’ optical transitions and detected fluorescence at longer wavelengths. For one emitter, the two measured optical resonances had linewidths of 69.4 and 75.6 megahertz at zero applied bias.

Different emitters nevertheless started well apart in frequency. Across six investigated vacancies, one transition’s zero-bias resonance frequency had a standard deviation of 23.9 gigahertz. Narrow individual lines therefore did not automatically mean matching light sources.

Applying reverse bias changed the local electric field, shifting the optical resonances through the Stark effect. An emitter near the p-type junction shifted by 4.79 gigahertz per volt, compared with approximately 0.32 gigahertz per volt for another farther into the active region.

The researchers overlapped both optical transitions of two vacancies in spectra recorded at different applied voltages: one at −5 volts and the other at −20 volts. This shows that electrical tuning can compensate for their frequency differences. It is not a demonstration of simultaneous matching at one shared bias, or of interference between photons from the two emitters.

Defects that check the device model

The same frequency shifts offered a microscopic probe of the diode’s internal electric field. Using a Stark-response model and previously reported coupling parameters, the team inferred the field component along the crystal’s c-axis.

At −10 volts, the inferred field increase was 25.0 kilovolts per centimetre near the junction and 1.13 kilovolts per centimetre at the second emitter.

Those measurements exposed limitations in the simulations. A two-dimensional semiconductor model, calibrated against room-temperature forward-current measurements, predicted concentrated fields near the implanted junction. But the cryogenic optical measurements did not agree quantitatively: the inferred field at one vacancy was 224% higher than simulated, while at the other it was 41% lower.

Uncertain defect depths, charge trapping and reduced dopant ionization at low temperature could contribute, the authors say. The comparison is therefore not a fully validated field map. It shows how embedded defects can reveal local behavior that a room-temperature-calibrated model misses.

Stability comes after settling

Electrical control also carried a waiting period. Extended exposure to the off-resonant initialization laser temporarily shifted the resonances, as did changes in the previously applied voltage. The lines then relaxed toward their earlier positions over minutes; one fitted relaxation time was 187 seconds.

The authors propose that nearby defects and interfaces trap and release charge, temporarily changing the field around a vacancy. That explanation remains an interpretation, not a direct measurement of the individual traps.

Once these transients settled, a measurement at a constant −40 volts showed little spectral drift over four hours.

Reported as an arXiv preprint, the work establishes electrical tunability, narrow resonances and sustained spectral stability under the tested cryogenic conditions. Demonstrating indistinguishable photons and two-photon interference remains a separate experimental step. Matching the resonances addresses an important prerequisite—not the complete quantum-network connection.