Quantum Nanophotonic Interface for Tin-Vacancy Centers in Thin-Film Diamond

TL;DR

This work addresses scalable quantum networks using SnV- centers in thin-film diamond by embedding them in 1D photonic crystal cavities to enhance photon emission and enable optical spin readout. A two-transition spontaneous emission model for the C and D ZPL transitions is developed to extract the C/D branching ratio and the individual Purcell factors from lifetime measurements, yielding η_BR = 0.7815 and Purcell factors up to $F_C \approx 26.21$ and $F_D \approx 5.12$ (angled device). Key results include measured quality factors up to $Q \approx 6\times 10^3$, lifetime reductions up to ~12×, and orientation-dependent Purcell enhancements that differ between parallel and angled cavities. The findings provide a pathway toward on-chip, high-fidelity spin readout for SnV- qubits and inform optimal cavity orientation and fabrication strategies for scalable quantum-networks.

Abstract

The negatively charged tin-vacancy center in diamond (SnV-) is an excellent solid state qubit with optically-addressable transitions and a long electron spin coherence time at elevated temperatures. However, implementing scalable quantum nodes with high-fidelity optical readout of the electron spin state requires efficient photon emission and collection from the system. In this manuscript, we report a quantum photonic interface for SnV- centers based on one-dimensional photonic crystal cavities fabricated in diamond thin films. Furthermore, we develop a model describing the spontaneous emission dynamics of our system, allowing for rigorous determination of Purcell factors and the C/D branching ratio from cavity enhancement of the C and D transitions of the SnV- zero phonon line. We observe quality factors up to ~6000 across this sample, and measure up to a 12-fold lifetime reduction. By considering the lifetime reduction of both the C and D transitions independently, we determine the C/D branching ratio to be ηBR=0.7815, in line with previous theoretical and experimental findings. Finally from our analysis, we extract a Purcell factor of up to Fc=26.21(0.01) for a single SnV- transition.

Figures (14)

• Figure 1: Cavity design, fabrication, and characterization. (a) A white light image of the fabricated devices on the thin film membrane. Representations of the two classes of devices, parallel and angled, are indicated by the white dashed boxes and zoomed to be shown in greater detail. The scale bars indicate 15 $\mu m$ and 5 $\mu m$ for the zoomed out and in images, respectively. (b) SEM of fabricated devices. Scale bars indicate $1 \mu$m. (c) The cross polarized reflectivity spectrum for the parallel device. The broadband spectrum is background corrected, then fit to a Fano model, yielding a quality factor of $6032$. The inset shows the broadband reflectivity spectra for both the resonance and background. The fitted region is indicated by the dashed box.
• Figure 2: SnV$^{-}$ level structure, PL confocal scan and cavity enhancement. (a) Schematic of the orbital energy states of the SnV$^{-}$. Characteristic of a group-IV color center, the ground ($\ket{g}$) and exited state ($\ket{e}$) are split via the combined effects of spin-orbit coupling and the Jahn-Teller effect. For the SnV$^-$ the ground state splitting is $\sim$850 GHz, and the excited state splitting $\sim$3000 GHz; these state splittings yield four separate ZPL transitions. In cryogenic conditions, PL signal is dominated by the two longer wavelength, lower energy transitions, labeled as C and D Görlitz_2020. (b) PL confocal scan of the parallel device. The emitter cluster addressed is indicated by the white circle. The scale bar indicates 3$\mu$m. (c) PL enhancement via gas tuning. The cavity resonance is first red-shifted via Ar gas condensation past all SnV$^-$ transitions of interest. The sample is then naturally 'back-tuned' over the course of repeated PL scans. The two spectra of interest are indicated by the white dashed lines. The gray dashed line is an approximate guide for the eye of the cavity resonance. (d) PL spectra on and off resonance with the cavity. We see that the most strongly enhanced SnV$^-$ transition demonstrates a $\sim$10-fold PL enhancement.
• Figure 3: Lifetime reduction and branching ratio analysis. (a) Spontaneous emission rate vs cavity resonance wavelength for the parallel device (black). The data are fit to a double Lorentzian model. The PL spectrum (red) is overlayed. We extract from this fit model the amplitude ratios $\zeta_C=4.672$ and $\zeta_D=1.985$. The three vertical dashed lines indicate the lifetime slices that are plotted separately, representing the traces with the C transition on resonance (pink), D transition on resonance (purple), and both transitions off resonance (black). (b) Spontaneous emission rate vs cavity resonance wavelength for the angled device (black). As in (a), the PL spectrum (red) is overlayed. For this device, we extract from this fit model the amplitude ratios $\zeta_C=12.230$ and $\zeta_D=1.514$. We note that due to filtering limitations, there are contributions from two separate emitters, and the data are thus fit to a quadruple Lorentzian model. The dashed lines indicate the lifetime slices of interest, as in panel (a). (c) Normalized lifetimes of the SnV$^-$ with C/D transitions on and off resonance for the parallel device. We measure an off-resonance lifetime of $9.412 \pm 0.09$ ns. When the C(D) transition is on-resonance with the cavity, the lifetime is reduced to $1.847 \pm 0.002$ ns ($4.570 \pm 0.02$ ns). The lifetimes are fit to amplitude normalized and background corrected single exponential models. (d) Normalized lifetimes of the SnV$^-$ with C/D transitions on and off resonance for the angled device. We measure an off-resonance lifetime of $10.507 \pm0.2$ ns. When the C(D) transition is on-resonance, the lifetime is reduced to $1.079 \pm 0.002$ ns ($8.109 \pm 0.2$ ns). The lifetimes are fit to amplitude normalized and background corrected single exponential models. We note that due to the contributions of a second emitter transition, at longer timescales, the data begins to deviate slightly from a single exponential mode. However, within the fitting range, the observed fluorescence lifetime is primarily dominated by a single decay timescale. (e) Schematic description of the cavity orientation with respect to the lattice sites and dipole moments of the C and D transitions. $\theta$ represents the angle between one of the dipoles and the cavity mode, and $\phi$ represents the collective fabrication angular offset from the diamond $\braket{100}$ lattice axis. (f) Solution of the analysis in Section \ref{['subsec:lifetime_reduction']}, resulting in a branching ratio $\eta_\text{BR}$ of 0.7815 and fabrication offset $\phi$ of $1.1 \pm 0.1$ degrees.
• Figure 4: Schematic of SnV$^-$ implanted thin-film diamond preparation. The process starts with a bulk electronic grade diamond that is polished to a precise miscut angle (a). The sample is then heavily implanted with He$^{2+}$ ions to form a vertically localized graphitized layer, around 400 nm below the sample surface (b). A pristine layer of single crystalline diamond is overgrown over the implantation damaged layer (c). Sn$^{2+}$ is implanted $\sim$90 nm below the sample surface, and SnV$^-$ centers are formed after high temperature, vacuum annealing (d). Any surface graphitization is removed via a tri-acid clean. A SiN hard mask is then deposited via CVD, and spun with photoresist (e). 200 $\mu$m by 200 $\mu$m squares are patterned by photolithography; these squares define the final size of each membrane (f). The membranes are defined in the bulk diamond by an anisotropic RIE etch (g). Individual membranes are then released via electrochemical etch (h), and bonded to a Si carrier wafer with HSQ (i). Finally, the film thickness is tuned to 180 nm via RIE etching (j).
• Figure 5: Schematic of cavity fabrication procedure. The starting material consists of thin film diamond bonded to a Si handling wafer via HSQ (a). A thin Al$_{2}$O$_{3}$ had mask is deposited by ALD (b). The sample is spun with ZEP (c) and patterned via ebeam lithography (d, e). The resist pattern is transferred into the hardmask by ICP-RIE etching (f). Any remaining ZEP is then stripped (g), and the pattern transferred into the diamond membrane (g). Lastly, the devices are suspended by HF vapor and XeF$_{2}$ dry etching (i).