Laser-induced spectral diffusion of T centers in silicon nanophotonic devices

Abstract

Color centers in silicon are emerging as spin-photon interfaces operating at telecommunication wavelengths. The nanophotonic device integration of silicon color centers via ion implantation leads to significant optical linewidth broadening, which makes indistinguishable photon generation challenging. Here, we study the optical spectral diffusion of T centers in a silicon photonic crystal cavity. We investigate the linewidth broadening timescales and origins by measuring the temporal correlations of the resonance frequency under different conditions. Spectral hole burning measurements reveal no spectral broadening at short timescales from 102 ns to 725 ns. We probe broadening at longer timescales using a check pulse to herald the T center frequency and a probe pulse to measure frequency after a wait time. The optical resonance frequency is stable up to 3 ms in the dark. Laser pulses below the silicon band gap applied during the wait time leads to linewidth broadening. Our observations establish laser-induced processes as the dominant spectral diffusion mechanism for T centers in devices, and inform materials and feedback strategies for indistinguishable photon generation.

Figures (9)

• Figure 1: Optical linewidth broadening of T centers. (a) SEM image of the device, a photonic crystal cavity coupled to a waveguide. The zoomed-in schematics illustrate a T center with nearby fluctuating charge traps in the bulk and on the surfaces. (b) Atomic structure of a T center. (c) T center level structure and possible mechanisms leading to linewidth broadening: (i) thermal transition between TX0 and TX1 states, (ii) ionization, (iii-iv) charge transfer between a trap state and (iii) conduction band (CB) or (iv) valence band (VB). (d) The photoluminescence excitation (PLE) spectra and (e) cavity-enhanced lifetimes of two T centers under study.
• Figure 2: Spectral hole burning. (a) Pulse sequence (upper panel) and a typical hole burning spectrum (lower panel) by sweeping the probe laser frequency. (b) Spectral hole width of four lifetime groups with the bar color indicating the total measurement power. The upper dashed line shows the average hole width of the lowest power measurement across the four groups, while the lower dashed line shows the thermal-limited hole width calculated from the cryostat base temperature.
• Figure 3: Check-probe spectroscopy. (a) Pulse sequence (upper panel) and typical unconditional and conditional spectra (lower panel) from the check-probe spectroscopy. (b) Conditional linewidth as a function of the laser pulse power in the waveguide where check and probe pulses are set at the same power. (c) Conditional lineshape at check detunings ($(\omega_{0,j} - \omega_T) / 2\pi$) of (-2.15, -0.96, 0.35, 1.26, 2.28) GHz for $j$ from 1 to 5. The dashed curve shows the unconditional spectrum. (d) Unconditional and conditional linewidths for $\tau$ = 2.5, 12.5, and 52.5 $\mu$s. (e) Unconditional and conditional peak counts per pulse for $\tau$ = 12.5, 52.5, 1000, 3000 $\mu$s.
• Figure 4: Laser-induced spectral diffusion. (a) Pulse sequence with $N$ laser perturb pulses inserted between the check pulse and probe pulse. (b) The linewidth and (c) lineshape of the conditional spectra for $N$ perturb pulses. (d) Top: schematics of the three perturb laser frequency relative to the T center frequency and linewidth with the green curve showing the conditional lineshape and the gray curve showing the unconditional lineshape. Bottom: the peak counts per pulse versus $N$ for the three perturb laser frequency with the color corresponding to the arrow color in the top panel.
• Figure S1: Setup for the spectral hole burning. AOM: acousto-optic modulator, SOA: semiconductor optical amplifier, FPC: fiber polarization controller.