Can resonant driving produce coherent quantum light from a single impurity-bound exciton?

Resonant driving replaces above-band pumping

Earlier optical studies of impurity-bound excitons in ZnSe pumped the emitter with light above the bandgap. That incoherent route introduces random time jitter from carrier relaxation and fluctuating local electric fields, both of which broaden the emission line and destroy coherence. It also cannot address the internal energy levels of the bound exciton directly.

Resonant excitation instead uses a narrow-bandwidth tunable laser tuned to the bound-exciton transition, producing resonance fluorescence. The difficulty is practical: the resonant laser's own scattering and reflection mask the weak emission, and resonant driving can change the emitter's charge state and quench it. The paper addresses both problems with a cross-polarized detection scheme whose extinction ratio exceeds 10^6 and with charge stabilization described below.

A chlorine donor in a ZnSe nanopillar at 3.2 K

The emitter is a single chlorine atom incorporated as a substitutional donor in a 4.7-nm ZnSe quantum well, sandwiched between ZnMgSe barriers on a GaAs substrate. The wafer is fabricated into a nanopillar 250 nm in diameter and 75 nm high, topped with a hydrogen silsesquioxane hemispherical lens to improve out-coupling. The device is cooled to 3.2 K and excited through a confocal microscope.

Above-band photoluminescence at 3.06 eV (405 nm) shows free-exciton emission between 2.827 and 2.832 eV and a discrete donor-bound exciton line at 2.8233 eV. That line is the transition the resonant laser later addresses. The above-band saturation power is 2.9 µW, a reference value used to express the much weaker stabilizing powers as fractions of saturation.

A weak above-band laser keeps the emitter charged

With only the resonant laser, the bound exciton emits little light. Adding an above-band laser at 3.06 eV while resonantly pumping the bound exciton restores the signal. Powers from 3 nW to 32 nW, or 0.1% to 1.1% of the above-band saturation power, stabilize the impurity while adding negligible photoluminescence of their own.

This is the condition that qualifies the headline answer: coherent resonant emission from this emitter is observed together with a weak charge-stabilizing pump, not from resonant excitation alone. A detuning scan with no above-band laser, where the impurity is discharged and optically inactive, shows only background from imperfect cross-polarization.

The line is 28 times broader than its lifetime limit

The resonance fluorescence intensity peaks at the bound-exciton energy and falls off with detuning; a Voigt fit gives a linewidth of 133 ± 11.6 µeV. The measured radiative lifetime of 141 ± 1 ps implies a lifetime-limited linewidth of 4.7 µeV, so the observed line is 28 times broader.

The resonant pump power of 10 nW is about 0.1 of the resonant saturation power, so power broadening is unlikely to explain the excess. The authors attribute the broadening to spectral wandering, possibly from charge fluctuations at defects and trap states near the chlorine impurity. This matters for any comparison with emitters quoted at their lifetime limit: the 0.94 Debye-Waller factor and the coherence demonstration should not be read as a lifetime-limited line.

Emission keeps the drive phase and favors the zero-phonon line

Polarization interferometry shows that the emitted field interferes with the resonant drive, meaning the emission preserves the phase of the pump. Second-order correlation measurements confirm the light is quantum emission from the bound exciton rather than scattered laser light. The paper also reports intensity-dependent interference of the reflected field at low incident powers, which it frames as a route to low-photon-number nonlinear phase gates.

Resonant excitation makes the Debye-Waller factor directly measurable. The value of 0.94 means most emission goes into the zero-phonon line rather than into phonon sidebands, and the paper describes it as one of the highest reported for defect emitters. For comparison, the paper cites 0.04 for nitrogen-vacancy centers in diamond and 0.6 to 0.8 for SiV and SnV centers. The Debye-Waller factor and the linewidth describe different things: a high zero-phonon fraction does not by itself make the line narrow.

Ionization quenches emission, and a pump pulse restores it

Time-resolved resonance fluorescence shows the emission turning on and then decaying to zero with a rate of 1.013 µs^-1. That rate is far below the radiative decay rate and grows linearly with resonant power, which the authors argue rules out a two-photon process and points to optically induced discharging of the emitter while it is resonantly pumped. They attribute this fast process to Auger recombination.

A second, slower decay appears when the delay between pulses is varied from 0.5 µs to 200 µs; a stretched exponential fit gives a 21 µs time constant, attributed to spontaneous discharging by charge tunneling to nearby trap or surface states. Turning on an above-band pulse after the resonant laser recharges the emitter: the recovery time constant falls from 200 ns at 20 nW to 9.3 ns at 90 nW. That 9.3 ns figure is the fastest demonstrated recovery in the paper and is the basis for the proposed fast optical gating.

The device operates in a few-photon regime

To judge whether the nonlinear response reaches the single-photon level, the paper compares the emitter's extinction cross-section with the input mode area. Assuming the homogeneous linewidth approaches the lifetime limit gives an ideal cross-section of 1.14 × 10^-2 µm^2; using the measured 1/28 ratio of lifetime-limited to inhomogeneous linewidth gives a time-averaged value of 4.0 × 10^-4 µm^2. Finite-difference time-domain simulations give a mode area of 4.2 × 10^-2 µm^2.

The ratio of mode area to ideal extinction cross-section is 3.7, so the device works in the few-photon regime but not at the single-photon level. Spatial misalignment and pure dephasing would make the ratio worse, while integrating the emitter with a cavity could reduce it through the Purcell effect. This is the paper's own threshold for calling the nonlinearity single-photon.

A bright zero-phonon emitter with a spin-1/2 ground state

The paper's comparison across solid-state emitters rests on two measured features: a radiative lifetime about an order of magnitude shorter than typical diamond color centers and single molecules, comparable to strongly confined quantum dots, and a Debye-Waller factor above that of NV, SiV, and SnV centers. The inhomogeneous broadening is treated as a fabrication problem that could be reduced by limiting surface trap states, and prior work cited in the paper found lifetime-limited homogeneous linewidths for bound excitons in ZnSe.

The forward-looking claim is that the chlorine shallow donor naturally has a spin-1/2 ground state, so the resonant methods demonstrated here are a route to optical control and readout of that spin, and cavity integration could enable spin-photon hardware. The priority the evidence supports is narrower than a general promise: this emitter combines near-unity zero-phonon emission with resonant, phase-preserving driving, and the practical condition for using it is charge stabilization on nanosecond timescales with a weak above-band pulse.

Sources used in this answer