Laser-induced spectral diffusion and excited-state mixing of silicon T centres

TL;DR

This work probes laser-driven spectral diffusion and laser-induced excited-state spin mixing in silicon T centres integrated with nanophotonic cavities. By employing two-colour spectral correlation measurements and resonance-check spectroscopy, the authors quantify diffusion dynamics, achieve a 35× narrowing of the emitter linewidth to $110\pm10$ MHz, and demonstrate stability for up to $1.5$ ms in the dark, enabling scalable multi-emitter synchronization. They show that spectral diffusion is driven predominantly by the excitation laser and model it with an Ornstein–Uhlenbeck process, while also revealing laser-induced spin mixing in the TX0 excited state that grows with power and can be enhanced by off-resonant light, pointing to a broadband charge-environment mechanism. These findings have immediate implications for improving entanglement rates and resource-state generation in silicon-based spin-photon interfaces and guiding engineering strategies to mitigate spectral diffusion in practical devices.

Abstract

To find practical application as photon sources for entangled optical resource states or as spin-photon interfaces in entangled networks, semiconductor emitters must produce indistinguishable photons with high efficiency and spectral stability. Nanophotonic cavity integration increases efficiency and bandwidth, but it also introduces environmental charge instability and spectral diffusion. Among various candidates, silicon colour centres have emerged as compelling platforms for integrated-emitter quantum technologies. Here we investigate the dynamics of spectral wandering in nanophotonics-coupled, individual silicon T centres using spectral correlation measurements. We observe that spectral fluctuations are driven predominantly by the near-infrared excitation laser, consistent with a power-dependent Ornstein-Uhlenbeck process, and show that the spectrum is stable for up to 1.5 ms in the dark. We demonstrate a 35x narrowing of the emitter linewidth to 110 MHz using a resonance-check scheme and discuss the advantage for pairwise entanglement rates and optical resource state generators. Finally, we report laser-induced spin-mixing in the excited state and discuss potential mechanisms common to both phenomena. These effects must be considered in calibrating T centre devices for high-performance entanglement generation.

Figures (14)

• Figure 1: Spectral diffusion in integrated colour centre devices. (a) A 1D silicon photonic crystal cavity is addressed by a resonant laser through an above-chip optical fibre and a grating coupler. (b) Cross-sectional view showing an integrated T centre at the device centre. The laser resonates in the single-mode cavity (red). Charge traps at surfaces, interfaces, impurities, and vacancies within the silicon device layer and oxide insulator generate electric fields that perturb the colour centre emission frequency. We expect the closest interface to a T centre in a cavity to be less than 100nm away. (c) An illustration of spectral diffusion. The T centre emission frequency fluctuates as the charge environment is reconfigured.
• Figure 2: A representative integrated T centre device (T Centre II). (a) The photoluminescence excitation (PLE) spectrum at zero magnetic field (blue), and the cavity spectrum measured in reflection (orange). (b) Correlation measurements of the centre luminescence under pulsed resonant excitation exhibit a high degree of antibunching with $g^{(2)}(0) = 0.186\pm0.004 \ll1$, confirming the emission is dominated by a single centre. $g^{(2)}(N)$ shows a positive correlation at low pulse separation number $N$, consistent with spectral diffusion Sallen_2010_subnanosecondSDquantumdots.
• Figure 3: Two-colour correlation measurement. (a) Laser detunings $\delta_{1,2}$ from the centre of the inhomogeneous line (solid) and modelled homogeneous lines (dashed). The vertical blue line indicates $\delta_1$. (Inset) Pulse sequence showing consecutive excitation pulses at detunings $\delta_{1,2}$. Correlations are calculated between photons emitted in detection windows indicated by the grey luminescence transients. (b) Correlations between frequencies $\delta_{1,2}$ as a function of the number of intermediate pulses. Solid lines display a joint single-parameter ($A$) fit to an O-U model of spectral diffusion. (c) Power dependence of the two colour correlation for two different powers. The correlation amplitude reduces with higher power. (d) Dark-time dependence of the two-colour correlation. The curves show closer agreement when plotted as a function of pulse number compared to when they are plotted as a function of time (inset).
• Figure 4: Resonance check (RC) PLE scheme. (a) The experimental pulse sequence. We excite at check frequency $\delta_1$ until a photon is detected, which triggers a probe pulse at frequency $\delta_2$ after delay $\tau$. (b) Background subtracted RC PLE (solid blue: high resolution, dashed blue: low resolution) compared to regular PLE (black) showing the linewidth narrowing to $\Gamma = 110\pm10MHz$. (c) RC PLE with dark wait time, $\tau$. We measure broadening in the absence of laser excitation as a function of wait time and observe no significant broadening out to $1.5ms$. (d) Simulated $N$-qubit entangled state preparation using parallel RC-PLE, with parameters corresponding to T centre II. The speedup, defined as $\tau^*/\tau_{\textrm{RC}}$, where $\tau_{\textrm{RC}}$ is the time required to obtain $N$ resonant photons from $N$ independent T centres prepared with RC PLE and $\tau^*$ is the time it takes to obtain $N$ resonant photons from $N$ independent T centres prepared conventionally. White indicates the region where no speedup is expected.
• Figure 5: TX$_0$ spin populations as mixed by laser power. (a) The level diagram of T centre II under an applied field and experiment schematic. (b) Time-resolved decay measurements showing the populations converging at higher excitation powers, indicating increased spin mixing. Powers are quoted in the at-field saturation power, $P_\text{sat,~B}$. (c) Population ratio $n_1 / n_2$ as a function of excitation power. The insets show the at-field PLE spectrum for this T centre (top) and the extracted mixing rate $\gamma$ versus power (bottom).