On-Demand Generation of Neutral and Negatively Charged Silicon-Vacancy Centers in Diamond

Point defects in wide-band-gap semiconductors are emerging as versatile resources for nanoscale sensing and quantum information science, but our understanding of the photoionization dynamics is presently incomplete. Here, we use two-color confocal microscopy to investigate the dynamics of charge in type 1b diamond hosting nitrogen-vacancy (NV) and silicon-vacancy (SiV) centers. By examining the nonlocal fluorescence patterns emerging from local laser excitation, we show that, in the simultaneous presence of photogenerated electrons and holes, SiV (NV) centers selectively transform into the negative (neutral) charge state. Unlike NVs, 532 nm illumination ionizes ${\mathrm{SiV}}^{-}$ via a single-photon process, thus hinting at a comparatively shallower ground state. In particular, slower ionization rates at longer wavelengths suggest the latter lies approximately $\sim 1.9\mathrm{eV}$ below the conduction band minimum. Building on the above observations, we demonstrate on-demand SiV and NV charge initialization over large areas via green laser illumination of variable intensity.

Figure 1

(a) Optical spectrum under 532 nm excitation at an arbitrary but fixed location. The laser power is $100\mu \mathrm{W}$ and the acquisition time is 120 s. Color bands—denoted as BP1, BP2, and BP3—indicate three alternative observation windows. (b) Initialization and optical spectroscopy protocol. The zigzag indicates laser scanning. (c) Confocal images in an area around the point of laser park using bandpass filters BP2 (left) and BP3 (right). The park time is 30 s and the laser power is 1 mW during the park and $100\mu \mathrm{W}$ during imaging at 2 ms integration time per pixel. (d) Averaged optical spectra (632 nm excitation) from an area around the point of laser parking before and after the park (orange and blue traces, respectively). The red laser power during the spectroscopy scan is $500\mu \mathrm{W}$; the acquisition time is 5 ms per pixel.

Figure 2

(a) Relative population of ${\mathrm{SiV}}^{-}$ ($S_{-}/S$) as a function of time for 532 nm laser light of variable power; read-out is carried out via a red laser probe; solid lines are single exponential fits. (b) ${\mathrm{SiV}}^{-}$ ionization rate ($j_{-}$) as a function of laser power for 532 nm laser excitation (green triangles). The dashed green line is a linear fit. Also shown are the extrapolated fits from ${\mathrm{SiV}}^{-}$ ionization at 632 and 692 nm (red and purpled dashed lines, respectively). (c) Electrode and illumination geometry. (d)–(f) ${\mathrm{SiV}}^{-}$-selective images after a 30-s-long red laser park at different depths. From (d) to (f), the applied voltage is 0, 200, and 400 V, respectively. The distance from the top (bottom) image plane to the top (bottom) diamond surface is 10 $(90)\mu \mathrm{m}$. All images share the same color scale and are collected during a red laser scan (1 mW, 2 ms integration time per pixel); all other conditions as in Fig. 1.

Figure 3

(a) Pulse sequence. Red zigzags indicate 632 nm, 1 mW laser scan and the green rectangle denotes a 532 nm, 1 mW laser park. (b) Confocal images (top) of the ${\mathrm{NV}}^{-}$ (left) and ${\mathrm{SiV}}^{-}$ (right) fluorescence after a 10 s green laser park (1 mW) obtained using the BP2 and BP3 bandpass filters. The calculated ${\mathrm{NV}}^{-}$ and ${\mathrm{SiV}}^{-}$ relative populations (respectively, $Q_{-}/Q$ and $S_{-}/S$) are shown in the bottom half images. (c),(d) Same as above, but after initialization using a 532 nm, $50\mu \mathrm{W}$ laser scan. In (b) and (d), the fluorescence within BP3 has been corrected to eliminate contributions from NVs. In all images, the red laser power is $100\mu \mathrm{W}$ and the integration time per pixel is 2 ms; the white arrow indicates the point of laser parking.

Figure 4

(a) (Left) As the green beam scans the diamond crystal, electrons and holes diffuse away from the illuminated spot. (Right) Effect of a strong green scan on ${\mathrm{SiV}}^{-}$; $I$ and $II$ in the inset indicate the areas exposed to the beam and affected by electron capture, respectively. (b) (Right) ${\mathrm{SiV}}^{-}$-selective (top) and ${\mathrm{NV}}^{-}$-selective (bottom) scanning confocal images (632 nm read-out) obtained after a 532 nm, $50\mu \mathrm{W}$ scan. (Left) Optical spectrum averaged over the same area using bandpass BP1. (c) Same as in (b) but for a 1 mW green laser initialization scan. The integration time during the confocal imaging and spectroscopy scans are 2 and 5 ms, respectively; in both cases, the red laser power is $100\mu \mathrm{W}$. All other conditions as in Fig. 1.