Charge stabilization of shallow nitrogen-vacancy centers using graphene/diamond junctions Open Access

Negative-charged nitrogen-vacancy (NV⁻) centers in diamond are expected to serve as quantum bits working at room temperature due to their long coherence time.^1–3 Research in quantum sensing has been actively conducted in recent years.^4–6 In particular, NV⁻ centers are expected to be applied to nanoscale nuclear magnetic resonance (NMR) because of their high magnetic field sensitivity with nanoscale spatial resolution.^7–9 For nanoscale NMR, it is important to close the distance between the NV⁻ centers and the sensing targets because the magnetic field sensitivity of NV⁻ centers decreases as the cube of the distance to the targets. Shallow NV⁻ centers with a depth of a few nanometers are necessary for nanoscale NMR. One of the challenges of shallow NV centers is the charge stability. NV centers with an unstable charge state due to switching between a negative-charged (NV⁻) state and a neutral-charged (NV⁰) state are not suitable for quantum sensing. The charge state of shallow NV centers is known to depend on the surface terminations. Charge-state stabilization of shallow NV centers by surface termination has been widely reported. Surface terminations with positive electron affinity, such as oxygen,^10–13 nitrogen,^14–16 and fluorine termination,¹⁷ stabilize the charge state of shallow NV⁻ centers. In contrast, it has been reported that surface terminations with negative electron affinity, such as hydrogen¹³ and silicon termination,¹⁴ destabilize the charge state of shallow NV centers. In addition, the termination methods are also important. For example, a comparison of boiled acid treatment, UV/ozone exposure, and oxygen annealing has been reported for the oxygen termination method.¹⁰ In the present report, charge-state stability comparable to that of deep NV centers was achieved by oxygen annealing at 465 °C for 4 h. Reported studies on the charge-state stabilization of shallow NV centers primarily used (100) diamond; few studies used (111) diamond. One of the advantages of using (111) diamond is the creation of perfectly aligned NV centers,^18,19 which can improve the sensor sensitivity by four times. Perfectly aligned shallow NV centers can be created by nitrogen delta (δ) doping during chemical vapor deposition (CVD) growth²⁰ or overgrowth on nitrogen (N)-terminated diamond.²¹ The charge-state stabilization of shallow NV⁻ centers in (111) diamond is an important issue. In a report on shallow NV centers in (111) diamond, it was shown by simulation that N termination contributes to charge-state stabilization.¹⁶ There is still room to improve the charge stabilization of shallow NV⁻ centers in (111) diamond.

In the present study, we focused on graphene on diamond. We considered that electrons from potassium-doped n-type bilayer graphene²² could be supplied to shallow NV centers. Although it was expected that a metal could supply electrons to the diamond surface, the distance between the shallow NV⁻ centers and the sensing targets would be increased. Therefore, n-type graphene with an atomic thickness was considered to be one of the most appropriate materials to supply electrons from outside the diamond. In addition, graphene can be stably bound to (111) diamond.²³ We evaluated the effect on the charge stability of shallow NV centers brought by the graphene/diamond junctions formed by using potassium-doped n-type graphene.

The sample used in the present study was the (111) single-crystal diamond film grown by plasma-enhanced chemical vapor deposition (PECVD) on a Ib high-pressure high-temperature diamond substrate. The diamond film was lightly doped with phosphorus (P concentration: 10¹⁷ cm⁻³) to stabilize the charge state of the NV⁻ centers.²⁴ Single shallow NV center was created by using nitrogen molecule (¹⁵N₂) ion implantation and post-annealing. We chose ¹⁵N₂ ions to suppress the acceleration energy. The minimum acceleration energy of nitrogen ion implantation in this experiment was 8 keV, which is what we used to perform the ¹⁵N₂ ion implantation. When the ¹⁵N₂ ions hit the diamond sample, each ion separated into two nitrogen atoms. The acceleration energy of each nitrogen atom was equivalent to 4 keV (half the ¹⁵N₂ ion acceleration energy). The ¹⁵N₂ ion fluence was 5 × 10⁸ cm⁻². The implantation depth of the nitrogen ions accelerated with 4 keV was calculated to be 6 nm by the software program Stopping and Range of Ions in Matter (SRIM).²⁵ Post-annealing was performed in an Ar atmosphere, and the annealing temperature and time were 1000 °C and 2 h, respectively. After annealing, the boiled acid treatment was performed to remove any surface contamination and form the oxygen termination. Patterned Ti/Au was deposited on the diamond sample for alignment.

The graphene/diamond junction was formed by the graphene wet-transfer process, and the formation process is shown in Fig. 1(a). Bilayer graphene grown on copper (Cu) foil by the PECVD method²⁶ was used in this study. Poly(methyl methacrylate) (PMMA) as a support material was spin coated on the as-grown bilayer graphene surface, and the Cu foil was etched by iron chloride (FeCl₃) solution. The bilayer graphene was potassium-doped by dipping of PMMA/bilayer graphene structure in potassium hydroxide (KOH) aqueous solution at room temperature for 30 min.²² The concentration of KOH solution was 5 wt. %. After the rinse in de-ionized water, the PMMA/bilayer graphene structure was transferred onto the surface of half of the diamond surface covered with a glass plate (thickness 0.1 mm) in de-ionized water. By removing the glass plate, PMMA/graphene was formed on half of the diamond surface. Finally, the PMMA on the graphene was removed using acetone at 50 °C for 30 min. Raman microscopy confirmed that the graphene was transferred. The superimposition of the Raman image and optical microscopy image is shown in Fig. 1(b). The Raman peak intensity of the graphene, the G-band (∼1580 cm⁻¹), is drawn in this figure as a Raman image. The graphene was transferred onto the left half of the diamond sample, as shown in this figure.

We observed the NV centers using a home-built confocal fluorescence microscopy (CFM) system. The excitation laser was a continuous-wave green laser (wavelength 532 nm). Laser power at before objective lens was 0.6 mW. Fluorescence from the NV center was detected using an avalanche photodiode via a 633 nm cutoff long-pass filter (LPF) with a pinhole of 30 μm in diameter. Single NV center can be observed by using the CFM. The photon counts of single NV center were 100 × 10³ cps.

The change in work function by graphene/diamond junction was measured by using Kelvin force microscopy (KFM, Park Systems, NX10). Platinum coated cantilever was used, and the tip radius of curvature was 15 nm in this study. Silver paste was used for electrical contact between the sample and the sample holder. The atomic force microscopy (AFM) and KFM images were measured simultaneously in one scan.

Figures 2(a) and 2(b) show the CFM map at the graphene and graphene-free regions, respectively. Isolated fluorescent centers were observed in each region. These isolated fluorescent centers were considered to be single NV centers because reasonable photon counts and the typical fluorescent spectrum were given. Observed single NV centers are highlighted by a white dashed circle in these figures. The brightest fluorescence in Fig. 2(a) is not the NV center because the fluorescence was bleached by the laser irradiation, and the PL spectrum was different from that of a typical NV center. Figure 2(c) shows the PL spectrum from a single NV in the graphene region. As shown in this figure, typical spectrum of NV⁻ center was observed. The first-order (1332 cm⁻¹) and second-order (2422 cm⁻¹) Raman bands of the diamond and the Raman peaks of the graphene, G-band (1580 cm⁻¹), and 2D-band (2700 cm⁻¹) were also observed. In the graphene-free region, only the PL spectra of the NV center and Raman peaks of the diamond were observed. Previous studies have reported that the lifetime of the NV center was shortened, and the fluorescence intensity was reduced due to energy transfer between graphene and the NV center at distances of a few nm to few tens nm.²⁷ However, in these experiments, the CFM map [Figs. 2(a) and 2(b)] showed no major change in fluorescence intensity of NV center between the graphene and graphene-free regions. It suggested that energy transfer between the n-type graphene and the NV center was suppressed, and lifetime was not decreased. Our previous studies have shown that Fermi level for K-doped n-type graphene located above the Dirac point in band diagram.²⁸ The obtained results could be explained by work function modulation due to K-doped graphene formation rather than the energy transfer between graphene and NV⁻ center as discussed later. Next, we evaluated the charge stability of each NV center based on the photon count fluctuation. Figures 2(d) and 2(e) show the time dependence of the photon counts from the stable and unstable NV centers, respectively. Because the fluorescence from the NV centers was observed via the 633 nm cutoff LPF, the fluorescence from the NV⁻ centers passed through the filter, but the fluorescence from the NV⁰ center was blocked. The fluorescence from the NV centers with a stable charge state was constant, as shown in Fig. 2(d), while the fluorescence from the NV centers with an unstable charge state fluctuated, as shown in Fig. 2(e). In the present study, CFM maps of the same area were measured twice to evaluate fluctuations in fluorescence intensity. NV centers with constant fluorescence intensity were defined as charge stable, and NV centers with fluctuating fluorescence intensity were defined as charge unstable during the twice CFM map measurements. The measurement areas were 60 × 60 μm², and the dwell time per pixel was 1 ms. The number of observed NV centers in the graphene and graphene-free regions was 68 and 56, respectively. In the graphene region, the number of stable and unstable NV centers was 63 and 5, respectively. Most of the observed NV centers were in a stable charge state. In the graphene-free region, the number of stable and unstable NV centers was 27 and 29, respectively. About half of the observed NV centers were unstable. Next, we evaluated the creation yield of the stable NV centers. The creation yield was defined as number of stable NV centers/number of implanted N atoms. The number of implanted N atoms in the measured area (3600 μm²) was 36 000, calculated based on the number of implanted N₂ ion fluence. The creation yields of observed NV⁻ centers in the graphene and graphene-free regions were evaluated as 0.18 ± 0.08% and 0.08 ± 0.06%, respectively. The creation yield in the graphene region was nearly two times higher than that in the graphene-free region. These results indicated that the graphene/diamond junction contributed to the charge stabilization of the NV⁻ centers.

In previous reports, the charge state of the NV⁻ centers was stabilized by surface termination. In the case of graphene, it was bonded by van der Waals force without changing the surface termination. We focused on the graphene-induced change in the work function as the reason for the charge-state stabilization of the NV⁻ centers. Reports on semiconductor work functions changed by graphene are well known.^23,28 Since the work function corresponds to the gap between the Fermi level and the vacuum level, it has been suggested that the work function changes as the Fermi level changes. We investigated whether the change in work function due to the graphene/diamond junctions stabilizes the charge state of the NV⁻ centers. The change in work function was measured by using Kelvin force microscopy (KFM). Figure 3(a) shows an atomic force microscopy (AFM) image of the graphene (left) and graphene-free (right) regions. Based on this result, the steps of the graphene cannot be confirmed even by AFM with a nanometer-scale resolution. Figure 3(b) shows a work function image measured by using KFM at the same region as in Fig. 3(a). As seen in this figure, the work function differs between the graphene region and graphene-free region. Figure 3(c) shows the line profile of the work function at the solid blue line in Fig. 3(b). As seen in this figure, the work function decreases at the graphene region compared with that at the graphene-free region. Similar results were obtained by photoelectron microscopy for the work function measurement on hexagonal boron nitride (h-BN) covered with graphene.^29,30 Therefore, it is considered that n-type graphene tunes the work function at the Fermi level at the diamond surface. It was difficult to quantitatively estimate the work functions of the graphene-covered diamond by using KFM, since it detected electric forces from both the graphene and the diamond. However, the observed results suggested that the graphene raised the Fermi level at the diamond surface. From these results, it can be considered that graphene contributes to the stabilization of the charge state of the NV⁻ centers.

In conclusion, we attempted to stabilize the charge state of shallow NV⁻ centers in (111) diamond using K-doped n-type graphene. The K-doped graphene was transferred onto half of the diamond sample, and the charge states of the shallow NV centers at the graphene-transferred region and graphene-free region were evaluated. First, fluorescent intensity of the NV center was not different compared with the K-doped graphene region and the graphene-free region. It was suggested that K-doped graphene contributes to suppressing energy transfer due to the Fermi level for K-doped graphene located above the Dirac point in the band diagram. The comparison of energy transfer between the NV center and graphene with K-doped graphene and undoped graphene is an area for future work to discuss the model in more detail. Second, the charge states of most of the NV centers at the K-doped graphene region were stable, whereas about half the NV centers at the graphene-free region were unstable. The creation yield of the stable NV⁻ centers was evaluated. The creation yield at the graphene region was about twice as high as that at the graphene-free region. We achieved stabilization in the charge state of the shallow NV⁻ centers by using graphene/diamond junctions. Finally, the work function of the graphene and graphene-free regions was evaluated by using KFM. The results confirmed that the work function decreases in the graphene region compared with the graphene-free region, which suggested that the Fermi level raises. Therefore, we consider that K-doped graphene can stabilize the charge state of the NV⁻ centers. For practical applications, it is necessary to improve the coherence time of the shallow NV⁻ centers. Degradation of the coherence time due to surface defects is known to be a key issue for shallow NV⁻ centers. In this study, the charge state of random axis orientation NV⁻ centers created by ion implantation was stabilized. We expected that perfectly aligned NV centers can be stabilized by K-doped graphene/diamond junctions. The coherence time improvement and charge-state stabilization of perfectly aligned shallow NV⁻ centers in (111) diamond by graphene/diamond junctions will greatly advance nanoscale NMR using diamond.

This work was supported by the MEXT Quantum Leap Flagship Program (Q-LEAP) under Grant No. JPMXS0118067395 and JSPS KAKENHI under Grant No. 20K21136.

The authors have no conflicts to disclose.

Moriyoshi Haruyama: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Validation (equal); Visualization (equal); Writing – original draft (lead); Writing – review & editing (lead). Yuki Okigawa: Data curation (equal); Formal analysis (equal); Investigation (equal); Resources (equal); Writing – review & editing (equal). Mitsuhiro Okada: Investigation (equal); Writing – review & editing (equal). Hideaki Nakajima: Investigation (equal); Writing – review & editing (equal). Toshiya Okazaki: Investigation (equal); Writing – review & editing (equal). Hiromitsu Kato: Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Toshiharu Makino: Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal). Takatoshi Yamada: Conceptualization (equal); Funding acquisition (lead); Investigation (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.