High-quality nanostructured diamond membranes for nanoscale quantum sensing

TL;DR

This work introduces a low-damage nanostructured diamond membrane platform that hosts shallow NV centers suitable for nanoscale sensing. By employing a controlled quasi-isotropic etch with a protective trellis/tether design, the authors fabricate freestanding nanobeams with high yield and uniform undercut, preserving $T_2$ and $T_1$ properties near the surface while enabling significant photonic enhancement. They demonstrate a pick-and-place transfer method to integrate membranes with diverse targets, achieving up to $7\times$ collection-efficiency enhancements predicted by FDTD simulations and up to $3-7\times$ enhancements depending on the environment. The results establish a practical, photonics-enabled sensing platform for condensed matter applications, with potential extensions to patterned implantation and nanostructured diamond devices like nanosheets and nanowires.

Abstract

Deploying nitrogen vacancy (NV) centers in diamond as nanoscale quantum sensors for condensed matter and materials physics requires placing the NV centers close to the sensing target. One solution is to fabricate diamond nanostructures and integrate them with materials and devices. However, diamond etching and ion milling can introduce subsurface damage and surface defects that degrade the charge stability and spin coherence of NV centers near the surface. Here we report a procedure for fabricating low-damage nanostructured diamond membranes, and we show that this fabrication scheme preserves the optical and spin properties of state-of-the-art shallow NV center quantum sensors, within nanometers of the diamond surface, while providing significant photonic enhancement. Furthermore, we demonstrate a pick-and-place transfer method, which enables integration with diverse sensing targets.

Figures (15)

• Figure 1: (a) Fabrication scheme showing method for fabricating freestanding undercut structures. Steps 1-4 comprise the anisotropic etch of the SiN$_x$ hardmask, the anisotropic etch of diamond, followed by conformal PECVD with SiN$_x$, then the final anisotropic etch of SiN$_x$ in the trench. Step 5 shows isotropic etching of three different width of nanobeams, resulting in different geometry of the underside. (b) Tilted SEM image (left) of nanobeam membrane, with a higher magnification image (right) of an individual nanobeam (720 nm wide and 300 nm thick). (c) SEM image shows the design of the trellis, tethers and gaps between the frame and host sample. Thinning of the diamond on the outer trellis corner section can also been seen. (d) FIB cross-sections of three different gold-coated nanobeam widths fabricated in one device run. The high contrast beads on the top of the beams result from the gold coating. (e) Optical image of the host diamond sample showing many nanobeam membranes across a large area of the 4.5mm x 4.5mm host sample. (f) SEM image overlaid with a confocal photoluminescence scan of a 720 nm wide nanobeam membrane acquired with air objective (NA = 0.9) under 520 nm (400 µW) illumination. Bright spots in the unfabricated host crystal and the nanobeams are NV centers.
• Figure 2: (a) Hahn echo coherence time T$_2$ as a function of NV center depth in nanobeams and the host diamond, plotted together with data from Ref Sangtawesin_SurfNoise for shallow NV centers after the oxygen surface termination procedure including a 1200$^\circ$C and only a 800$^\circ$C anneal (as in the current work). Yellow dashed line shows a $T_2 = kd^n+T_{2,0}$ fit to the 800$^\circ$C annealed data from Ref. Sangtawesin_SurfNoise ($k = 0.047\pm0.04, n = 2.35\pm0.25, T_{2,0} = 3.35\pm3.87$). Grey shaded area marks the error margin for $n$ with fixed $k = 0.047$ and $T_{2,0}=3.35$ showing that most of the data that we measure in the nanobeams falls within this scaling range. (b) T$_2$ measured using an XY-N pulse sequence as a function of the number of pulses, N, (N = 1 is a Hahn echo experiment) for two NV centers in the nanobeams. Dashed line shows $\propto{N^{2/3}}$ dependence. (c) Spectral noise density extracted from a XY-8 decoherence decay using Equation S\ref{['eqS2']} for the same NV centers as in (b). The expected peak position associated with the $^{13}$C Larmor frequency is marked with a vertical dashed line. (d) $T_{1,SQ}$ and $T_{1,DQ}$ as a function of NV center depth in nanobeams (purple) and unfabricated diamond (blue).
• Figure 3: (a) Photon counting histogram of fluorescence from an NV center in a nanobeam under weak orange illumination ($\approx$3 $\mu$W) fitted with a double-Poisson distribution function reflecting the photon number distributions associated with the NV$^-$ (bright) or NV$^0$ (dark) charge states. This distribution corresponds to 77$\%$ NV$^-$ population. (b) Distribution of NV$^-$ population measured based on the histograms as in (a) for NV centers in the nanobeams of various widths and the host diamond. (c) Distribution of NV$^-$ ionization rate coefficients under orange illumination for NV centers in the nanobeams of various widths and for NV centers in the host diamond. (d) Distribution of NV$^-$ spin contrast measured in a Rabi oscillation experiment at 25 G for NV centers in the nanobeams of various widths and for NV centers in the host diamond.
• Figure 4: (a) Widefield image of a membrane mid-transfer, held electrostatically by two glass probes above the host diamond. (b) False-colored SEM of a 15um diamond membrane (red) fully transferred onto a sapphire substrate (grey). (c) Saturated photoluminescence intensity under 520 nm illumination for NV centers in the unfabricated host (left, blue) as compared to the nanobeams of various widths and thicknesses for suspended beams (left, purple) and beams transferred to a sapphire substrate (right, green). The solid red line represents the enhancement in collection efficiency averaged over different NV dipole orientations and lateral displacement. Dashed red lines represent the maximum and minimum collection efficiency enhancement from simulations for a given nanobeam geometry. (d) Saturation power for NV centers in the unfabricated host (left, blue), suspended beams (left, purple), and beams transferred to a sapphire substrate (right, green). Dashed black line in (c) and (d) marks the average values in the unfabricated host, while the center box plot line represents the median of the data.
• Figure S.1: Anisotropic etch depth of diamond in oxygen ICP-RIE as a function of etch time. Blue line is a linear fit to the data that sets the etch rate of the recipe in main text