Closed-loop electron-beam-induced spectroscopy and nanofabrication around individual quantum emitters

2.1 Gas-assisted EBIE and CL in diamond

The EBIE and CL microscopy were performed in an FEI Quattro environmental scanning electron microscope (SEM) using a Delmic Sparc CL module to collect luminescence generated by the electron beam into an Andor Kymera 193i spectrograph. A flip mirror was used to redirect the collected CL away from the spectrograph and into a 50/50 fiber splitter for photon correlation function measurements performed with a pair of superconducting nanowire single photon detectors and a Swabian TimeTagger Ultra used to time tag photon detection events.

The emitter was first located and characterized in no water vapor (i.e., pressures below 1E-6 Torr). Then, the EBIE was performed in a 0.75 Torr water vapor background. CL was acquired concurrently during the EBIE process, and reference CL measurements were taken at pressures of less than 1E-6 Torr, where no etching was observed even with extended electron-beam dwell times. A Gatan cryostage was used to acquire CL spectra at 8 K, although, because the water vapor background is incompatible with cryogenic operation, no EBIE was observed at cryogenic temperatures. Notably, cooling the sample to 8 K immediately after switching to high-vacuum operation results in ice formation on the diamond QMC at 8 K, but this water reservoir is quickly removed by electron-beam exposure, and it provides an insufficient environment to facilitate cryogenic EBIE.

In this room temperature (293 K) experiment, the H₂O vapor injected into the chamber adsorbs onto the diamond surface, forming bonds with the diamond surface mediated by the oxygen atoms. The incident and emitted electrons then interact with these adsorbed molecules, dissociating them and generating reactive radicals (i.e., H⁺ and OH⁻) that facilitate local chemical dry etching. The reaction products, such as CO₂ and CO, formed in this process are volatile and spontaneously desorb from the surface, leaving behind etched areas [18], [21]. Several factors, including the dose of the electron beam, the kinetics of the reaction, and the nature and pressure of the injected gas, determine the spatial resolution and etch depth in EBIE. The sub-nm electron beam spot size allows for the local etching of specific structures, enabling the direct writing of a wide range of photonic cavities and structures within a single system that also serves as a photonic interface for quantum technologies. EBIE has been used with various combinations of gases and substrate materials [19], [23], including recent efforts focused on nanoscale patterning of 2D quantum nanophotonic systems [24], [25]. Although this process has traditionally been used for material etching, recent advancements have expanded its applications to material analysis, surface functionalization, and nanostructured material growth [18], [26]–[28].

2.2 Diamond QMC fabrication

A single diamond QMC contains six waveguides, each with a width and thickness of 280 nm and 200 nm, respectively, and connected by diamond trusses. Each waveguide has an array of photonic cavities at the center [11], [29]. To prepare the diamond substrate for color center creation, we etched the top 10 μm of the [001] electronic-grade single-crystal diamond using Ar/Cl₂ plasma etching, followed by O₂ etching to alleviate surface strain. Implantation of ¹²⁰Sn ions was done at Innovion with a dose of 5 × 10¹¹ ions/cm² at 350 keV, targeting a depth of 76 nm as deduced from stopping and range of ions in matter (SRIM) simulations. The diamond underwent annealing at 1200 °C in a vacuum furnace at 10⁻⁷ mbar. The annealing sequence consists of a 120-min linear ramp from room temperature to 400 °C and holding for 400 min, a 120-min linear ramp from 400 °C to 800 °C and holding for 400 min, a 120-min linear ramp from 800 °C to 1200 °C and holding for 300 min, and finally a linear ramp down to room temperature. After high-temperature annealing, we submerged the sample in a boiling tri-acid (1:1:1 nitric:sulfuric:perchloric acid) solution at 345 °C for 2 h to remove graphite formed on the surface. The purpose of high-dose implantation was to ensure, on average, more than one quantum emitter per quantum channel.

Following the color center implantation in diamond, we coated the diamond with a silicon nitride (Si₃N₄) hard mask via plasma-enhanced chemical vapor deposition, patterned by ZEP-520A electron beam resist combined with ESpacer conductive polymer and CF₄ reactive-ion etching (RIE). This was followed by a conformal deposition of an 18 nm alumina atomic layer. Alumina breakthrough was achieved using CF₄ RIE, and an oxygen plasma was employed to undercut the diamond QMC uniformly. Finally, the silicon nitride and alumina masks were removed using hydrofluoric acid. The diamond QMC was transferred to a holey carbon grid for etching and characterization.

2.3 Optical measurements

We pre-characterized the diamond QMC quantum emitters using cryogenic free-space confocal photoluminescence (PL) microscopy at 4 K equipped with an objective lens (ZEISS, NA = 0.9, 100×). The collected signal is sent to either an avalanche photodiode (APD) for initial intensity mapping (Figure 1b) and for high-resolution photoluminescence excitation (PLE) spectroscopy measurements or to the spectrometer (SP-2500i, Princeton Instruments) for initial spectral analysis. Figure 1d shows a representative PLE spectrum of the SnV⁻ quantum emitters from the chiplets. The corresponding g⁽²⁾ characterization shows an average (g)⁽²⁾(0) < 0.1, confirming the presence of single-photon emission. A representative autocorrelation measurement of the SnV⁻ emitters is presented in Figure 1d. The extended optical system design is illustrated by Linsen Li et al. [30].

3.1 Simultaneous EBIE and CL around SnV⁻ in QMC

Figure 1c presents an SEM micrograph alongside a CL hyperspectral map of a diamond QMC obtained at an accelerating voltage of 10 keV. A spatially isolated color center was selected for the analysis to enable the independent examination of emission characteristics of individual SnV⁻ color centers. The CL hyperspectral map revealed a distinct emission peak at 620 nm, indicative of localized emission at room temperature (293 K) from the SnV⁻ color centers. This emission manifests as a broad spectral feature composed of several overlapping peaks. Upon cooling the system to 8 K, a pronounced enhancement of the SnV⁻ emission peak was observed, along with a significant narrowing of the spectral full-width at half-maximum (FWHM), suggesting enhanced emission purity at reduced temperatures. Because the high-energy electron-beam excitation can excite multiple excited states that are inaccessible in resonant or near-resonant PL microscopy, no antibunching was observed in CL photon correlation function measurements, and broadband background fluorescence contaminated many of the measurements [31], [32]. However, the close correlation between the PL map shown in Figure 1b and CL maps acquired on each QMC, together with the low-temperature CL spectrum shown in Figure 1e, provided strong confidence that the localized CL observed in Figure 1c and e was from a single SnV⁻ center.

To investigate the coupling between fabricated photonic cavities and SnV⁻ color centers, time-dependent CL measurements were performed in an H₂O environment (0.75 Torr) with a beam energy of 30 keV and a current of 2.8 nA for an electron beam incident on the SnV⁻ color center shown in Figure 2a. The resulting CL time series spectra in Figure 2c illustrate the change in SnV⁻ luminescence as a function of electron beam etching time; the time-dependent CL and secondary electron signals are consistent with etching rates of approximately 4 nm/s for these beam conditions. Interestingly, as we etch through the diamond chiplet at the location of the SnV⁻ color center, we observed a gradual increase in the 620 nm emission. Figure 2d–i shows the spectral fitting results within the 4.5 s–12 s timeframes to investigate the CL enhancement further. Three emission peaks were fitted and deconvoluted, where red, green, and blue correspond to diamond emission from lattice disorder (e.g., A-band emission from dislocations and grain boundaries) [33], [34], SnV⁻ color center, and potentially other defects complexes, including nitrogen impurities with a CL peak around 1.7 eV [35]–[37], and vacancies that are mobile above 900 °C. The key effect observed in Figure 2d–i is the marked emission enhancement from the SnV⁻ color center at 620 nm (green curve). This is in contrast to the red peak from the host matrix that showed almost constant emission throughout the process. Thus, EBIE can be potentially used to dynamically control and locally optimize the coupling from a single diamond color center in a diamond photonic cavity.

Figure 2:

Cavity fabrication with in situ cathodoluminescence. Hyperspectral cathodoluminescence mapping (top panel) with concurrent scanning electron microscopy (bottom panel) before etching (a) and after etching (b). (c) Time-resolved cathodoluminescence with 1.5 s time interval. Start time and stop time are indicated by blue and red, respectively. (d–i) Individual spectra acquired from 4.5 s to 12 s highlighting the enhancement at 620 nm. The spectra were fitted and deconvoluted with three peaks corresponding to extended defects in diamond, SnV⁻, and potential transition-radiation induced by the electron-beam.

While the ability to localize the emission from a single color center in a prepatterned QMC using operando EBIE and CL microscopy, as shown in Figure 2 is certainly compelling, the approach described here also allows for the direct writing of nanophotonic structures in diamond QMCs around single color centers. As an initial proof of concept of this approach, the morphology of photonic cavities patterned by RIE and EBIE on this diamond QMC is compared in Figure 3. A large area image of the QMC is shown in Figure 3a, and images of the same cavity design fabricated by RIE and EBIE are shown in Figure 3b. The cavities fabricated via EBIE were created by rastering the electron beam in a circular pattern to intentionally increase the cavity diameter, matching that of the RIE process. Figure 3c illustrates a similar cavity design with a reduced hole diameter of 56 nm achieved by dwelling the electron beam at a single point for a user-defined time. The ability to fabricate cavities above and around the quantum emitters on these length scales provides a critical testbed for complex photon cavity designs in diamond with in situ sub-diffraction-limited optical feedback inaccessible with modern nanofabrication tools.

Figure 3:

Patterning of nanophotonic structures in diamond waveguides. (a) Diamond QMC suspended on holey carbon support films for E-beam in situ fabrication. (b) Photonic cavities fabricated by reactive ion etching (RIE) (top panel) and electron beam etching (bottom panel). (c) Ultimate resolution limit for periodic photonic cavities achieved with electron beam etching. Scale bars are 200 nm.

3.2 Origin of SnV⁻ CL enhancement

To identify the origin of CL enhancement at 620 nm during the etching shown in Figure 2 and to better understand how the EBIE modifies the photonic local density of states (LDOS) around the emitter, we performed finite-difference time-domain (FDTD) simulations of the diamond QMC. The FDTD simulations were designed to replicate the experimental conditions as closely as possible. We modeled a SnV⁻ embedded in a diamond waveguide, incorporating the geometric dimensions as determined by SEM and optical measurement: a width (w) of 280 nm, a height (h) of 200 nm, and a depth for the SnV⁻ centers of 100 nm (Figure 4a). The emitter was positioned near the simulated etched dimple, offset from the center by δx and δy in the x and y directions, respectively, to account for potential misalignments (Figure 4d–f). To examine the spectral characteristics, we monitored the normalized intensity of the emission peak at 620 nm.

Figure 4:

Single photon emission enhancement. (a) Illustration of an etched dimple in a diamond waveguide. Characterization with scanning electron microscopy (SEM) and optical measurement revealed a width (w) of 280 nm, a height (h) of 200 nm, and a depth for the SnV⁻ centers of approximately 100 nm. The location of the emitter is offset from the center of the etch by δx and δy in the x and y directions, respectively. (b) Normalized intensity versus electron beam etching time for emission peak at 620 nm showing over 2.5× increase in emission intensity and (c) reference peak at 405 nm. (d–f) Finite-difference time-domain (FDTD) simulation results indicate that the collection efficiency of the emitter, which depends on the etch depth, δx and δy, is not significantly affected by the local electromagnetic environment. The emission is assumed to be circularly polarized, with the numerical aperture (NA) of the CL optics being 0.97.

The FDTD simulations accounted for the emitter’s assumed circular polarization and the collection CL optics’ high numerical aperture (NA), set to 0.97, to ensure accurate collection efficiencies. At low temperatures, the upper branch of the excited state rapidly decays to the lower branch, leading to only two transitions near 620 nm visible under excitation. Each of these transitions is linearly polarized and orthogonal to each other [38]. Because the excitation and the thermal relaxation are incoherent, the combination of two transitions can be modeled as circular polarization, which we assumed throughout the simulations. By varying the etch depth, δx and δy, we evaluated the impact of these parameters on the collection efficiency. The results indicated that the collection efficiency of the emission does not significantly change as the LDOS is modified by the etching process – less than 20 % change in collection efficiency is expected for a wide range of structural parameters.

Therefore, we conjecture that the CL enhancement could be from other effects, including the removal of the surface diamond and changes in electron–emitter interaction as a function of etching depth rather than photonic modification. Further studies are needed to identify the exact dynamics. For example, the implantation energy of Sn defects can be swept to vary the depth of the emitters and to identify the correlation with the emission intensity. Alternatively, photonic crystal cavity structures that can significantly modify LDOS can be used to see the photonic enhancement by increased LDOS [39].