Stability of Near-Surface Nitrogen Vacancy Centers Using Dielectric Surface Passivation

We study the photophysical stability of ensemble near-surface nitrogen vacancy (NV) centers in diamond under vacuum and air. The optically detected magnetic resonance contrast of the NV centers was measured following exposure to laser illumination, showing opposing trends in air compared to vacuum (increasing by up to 9% and dropping by up to 25%, respectively). Characterization using X-ray photoelectron spectroscopy (XPS) suggests a surface reconstruction: In air, atmospheric oxygen adsorption on a surface leads to an increase in NV– fraction, whereas in vacuum, net oxygen desorption increases the NV0 fraction. NV charge state switching is confirmed by photoluminescence spectroscopy. Deposition of ∼2 nm alumina (Al2O3) over the diamond surface was shown to stabilize the NV charge state under illumination in either environment, attributed to a more stable surface electronegativity. The use of an alumina coating on diamond is therefore a promising approach to improve the resilience of NV sensors.


Introduction
[5] The negative NV charge state (NV -) is central to such applications in quantum sensing owing to its optically addressable spin states, long-lived quantum coherence and room temperature operation. 68][9] The performance of the NV-diamond sensor can be determined by measurement sensitivity ( .∝ √ 1 + 1  2  . ; Where mag. is AC or DC magnetic sensitivity, C is measurement contrast and navg. is number of NV -photons per measurement) 10 in addition to the spatial resolution (as low as ~ 10 nm) 11 and operational stability under various environments.Achieving the greatest spatial resolution for quantum sensing requires the positioning of NV centers near diamond surface (< 10 nm), however, the brightness and spin properties of NV centers are compromised near the surface. 2 For example, the DC magnetic field sensitivity (mag.)achieved is ~ 17 pT/Hz for NV centers in the bulk, 12 which can be compared to ~ 1 T/Hz for near-surface NV centers. 13[16] The most general diamond surface composition involves non-diamond (sp 2 ) carbon, functional groups (C-Xn,), dangling bonds, metallic traces and adsorbed environmental species. 17,18 mong these, many surface constituents have been identified as a source of local charge traps (e.g.1][22] NVdiamond quantum sensing protocols typically use high power non-resonant laser excitation (~ 532 nm) 12 , which can lead to significant heating, NV spectral diffusion, ionization of nitrogen atoms (P1 centers) and excitation of various surface constituents [23][24][25] .][28][29][30] Ultra-high vacuum (UHV) and cryogenic conditions have been reported to degrade the properties of single NV centers in diamond nanopillar structures, while their stability was partially improved upon surface passivation with ultrapure water. 31The exact origin of NV degradation near surface and under different environmental conditions remains unclear and requires attention in order to develop effective mitigation strategies.
In this article, we investigate the instability of near surface ensemble NV centers under air and vacuum (~1 10 -3 mbar) at room temperature, measuring NV properties such as optically detected magnetic resonance (ODMR) contrast as well as characteristic properties of the material surface, following laser illumination.We found the ODMR contrast to vary over the course of (1-2 mW) laser exposure, due to NV charge state conversion (NV -NV 0 ) caused by the changing surface chemistry.A stable ODMR contrast and NV charge state were achieved by atomic layer deposition (ALD) of aluminium oxide (Al2O3, ~ 2 nm) on the diamond surface.

Experimental details:
The primary sample investigated here is an electronic grade (100) diamond, ion implanted with nitrogen ( 15 N, 3 keV, 1×10 13 ions/cm 2 ), supplied by Qnami AG. 32The average NV depth was estimated by average range of N + ions and lattice vacancy profile using stopping and range of ions in matter (SRIM) to be ~ 5 nm (Fig. S1; supporting information (SI)).The concentration of NV centers was ~ 5000 NVs/m 2 .The as-procured sample was acid refluxed at 255 o C (H2SO4: HClO4: HNO3 with 1:1:1 v/v ratio) for 2 hours to eliminate non-diamond impurities and increase the oxygen functionalization.We used the acid reflux as a procedure to 'reset' the diamond surface in between experiments in different environments.The acid refluxed sample was termed NVdiamond ('NVD').Sample preparation has been summarized in fig.1(b).After the completion of optical measurements on NVD, a ~ 2 nm layer of aluminum oxide (Al2O3) layer was deposited on the sample (NVD→Acid reflux →Al2O3) using a Savannah S200 atomic layer deposition (ALD) system.The thickness of the Al2O3 layer was found to be ~ 2 nm, measured by ellipsometry on a bare silicon substrate (placed together with NVD during ALD deposition).The sample with the deposited Al2O3 layer is termed alumina coated NV-diamond ('AC-NVD').The experimental setup for optical measurements (see Fig. 1(a)) consists of a home built confocal setup (NA = 0.65) equipped with a Montana instruments s100 cryostation and 522 nm continuous wave laser (LBX-522; Oxxius).The fluorescence signal was filtered through flip mounted 550 nm and 650 nm long pass filters (LPFs) and guided toward a single photon counting module (Excelitas Technologies) and photoluminescence (PL) spectrometer (SpectraPro HRS500, Princeton instruments) for ODMR and PL spectroscopy respectively.For ODMR, the fluorescence signal was collected through a 550 nm LPF in order to observe maximum effect of NV 0 emission in ODMR measurements.A laser power of 40 W was used for ODMR and PL spectroscopy measurements.To study the impact on ODMR contrast from high power laser illumination, ~ 1.2 mW and ~ 2 mW laser powers were used. (+ . )+  where Isat. is the saturated intensity (8.5  0.5) Mcps, Psat. the saturation power (7.1  0.6) mW and  denotes the (non-saturating) background fluorescence.The PL counts shown in Fig. 1(c) as a function of laser power were measured after inserting a neutral density (ND) filter in optical collection paththe actual counts are expected to be ~10 times higher.For vacuum measurements, the sample chamber was evacuated using a rotary pump (110 -3 mbar), while for measurements under ambient conditions the cryostat head was removed.The experimental scheme for ODMR measurements is shown in Fig. 1(e).High power laser illumination was repeatedly applied to the sample with increasing illumination time after each repetition.The ODMR spectrum was measured using lower powers (Pexc.= 40 W) after each high power exposure, following a wait time (t0) of 60 s.After a cumulative exposure time of ~ 1.5  10 4 s at ~ 1.2 mW, laser power was increased to ~ 2 mW up to a total illumination time of ~ 3.0  10 4 s.
Additional material characterization was performed using high purity electronic grade diamond (ELSC20, Thorlabs).As received ELSC20 samples were acid refluxed and termed as electronic grade diamond (ED).A ~ 2 nm layer of Al2O3 was deposited on ED samples and termed as alumina coated electronic grade diamond (AC-ED).In-situ Raman spectroscopy was performed using Renishaw in-Via Raman microscope equipped with 514.5 nm laser.To estimate the effect of laser illumination, spectral features between ~ 1200 -1900 cm -1 were recorded repeatedly under continuous high laser power excitation (2 mW power was applied through an air objective lens of 0.4 NA).For vacuum Raman measurements, the ED and AC-ED samples were placed in a custom made vacuum compatible glass cell and evacuated to ~ 110 -5 mbar.The sealed glass cell was then placed under the microscope for spectroscopy.Other Raman measurements were performed under ambient conditions.Normalized Raman spectra were deconvoluted using Lorentzian line shapes to compare diamond (~ 1332 cm -1 ) and graphitic carbon (~ 1600 cm -1 ; G band) signal intensities.
For XPS sample preparation, the 2D Raman imaging (Area ~ 125 m 2 ) of EDs was performed under different environmental conditions.Two laser exposed samples were prepared under air and vacuum environmental conditions respectively (Figure S5; SI) for XPS.To eliminate adventitious carbon and adsorbed surface species, the laser exposed EDs were annealed at 200 o C (2 hours) under argon atmosphere prior to XPS.The XPS was performed using a Thermo Scientific K-Alpha X-ray photoelectron spectrometer with a base pressure of ~ 2 × 10 −9 mbar, equipped with a monochromatic Al K X-ray source (h = 1486.7 eV).The X-ray spot size was reduced from the standard 400 m to 100 m in order to resolve the laser exposed regions in both samples.The XPS spectra for each sample were recorded at laser exposed position and another unexposed position (situated at ~ 1 mm away from laser exposed position).The maximum XPS probing depth (dXPS) at the maximum kinetic energy of 1486.7 eV, i.e. the photon energy of the Al Kα laboratory X-ray source, was estimated by calculating the relativistic inelastic mean free path (IMFP) (dXPS = 3×(IMFP)) using the TPP-2M model as implemented in the QUASES software package. 33The dXPS for the diamond and diamond with Al2O3 samples were calculated based on C and Al2O3 models available in the QUASES database and were found to be ~ 11.7 nm and 10.2 nm, respectively.XPS analysis was performed using the Thermo Avantage software package.For the estimation of the relative atomic ratios of carbon and oxygen in different samples, the total peak areas of the C 1s and O 1s core levels and in-built atomic sensitivity factors (ASFs) were used.
The change in carbon to oxygen atomic ratio (C/O) due to laser exposure was quantified as × 100; where (sp 2 /sp 3 )exp.and (sp 2 /sp 3 )unexp.represent the laser exposed and unexposed positions on the sample.

Results and discussion:
The spin state dependent brightness of the NV -center is a fundamental part of its application as a quantum sensor and can be characterised by the ODMR contrast, or the relative change in PL intensity following a change in the spin state. 6The ground state ( 3 A) spin triplet (ms = 0, 1) of the defect is characterized by an axial zero field splitting (ZFS) of about 2.87 GHz between spin sublevels ms = 0 and ms = 1.5][36] Therefore, in the absence of an external magnetic field, the ODMR spectrum is characterized by a resonance around 2.87 GHz, further split by the non-axial term, as illustrated by the representative ODMR spectrum for NVD in an air environment shown in Fig. 1(d).The maximum ODMR contrast for a single NV is about 30%.8][39][40][41][42] .The evolution of ODMR contrast under laser illumination in different surface and environmental conditions is shown in Fig. 2. For the NVD sample under vacuum, the ODMR contrast exhibits an exponential decay with laser exposure at 1.2 mW, and decays further when the laser power increased to 2 mW (Fig. 2(a)), dropping to a contrast of ~4.5% (or ×0.75 the starting value).The opposite trend is seen when illuminating NVD in air (Fig. 2(b)), where the contrast is seen to rise to ~7.8% (or ×1.09 the starting value).However, for the AC-NVD sample, the ODMR contrast was found to be relatively independent on high power laser illumination under either environments, changing by less than ~ 0.3%.We fit the time evolution of ODMR contrast under the two consecutive periods of laser exposure at different powers, using two exponential functions with a common set of fitting parameters to ensure the evolution of the contrast is continuous across the two periods.Specifically, we use the functions () =  0 + ( 1  −/ 1 +  2 ) for t < 14732 s and () =  0 + ( 1  −14731/ 1 +  2  −(−14731)/ 2 ).Here, 1 and 2 are the decay constants under laser exposure of 1.2 mW and 2 mW respectively.C0 is the ODMR contrast after high laser illumination for infinite time.The C1, and C2 denotes the change in the ODMR contrast after laser illumination of 1.2 mW and 2 mW for infinite time.We attribute the observed changes in ODMR contrast to the conversion of NV -to NV 0 , through a mechanism illustrated in Fig. 3, supported by measurements described in Figs. 4 and 5. Due to near surface NV fabrication (3 keV N + ions), a high N/NV ratio (~ 1%) is present in the NVD sample, 43 and the residual nitrogen atoms (P1 centers) act as a source of electrons to maintain NV - as preferential NV charge state. 44,45 he high electronegativity of the oxygen functionalized surface also helps to maintain their stability. 17During laser exposure, excitation of surface species 46 can result in the detachment of non-diamond carbon and oxygen functionalities and, in the absence of environmental oxygen (e.g. in vacuum), surface electron traps develop which reduce the surface electronegativity, cause upward band bending near the surface and lead to NV charge state conversion.The reduction in NV -PL emission on top of a background fluorescence signal leads to a gradual reduction in the observed ODMR contrast, as a result of this continuously evolving surface composition.Within an oxygen-rich environment (e.g. in air), laser illumination has the same impact on reducing non-diamond carbon at the surface, however, there is an increase in the oxygen adsorption which increases surface electronegativity (downward band bending) and promotes the NV -charge state (NV 0 →NV -).The evolution of the surface described above (and associated NV charge state conversion) appears to be minimised by the presence of the alumina coating in the AC-NVD sample.Any changes at the diamond surface may be compensated by the alumina providing a stable surface electronegativity on surface.
To further investigate the NV center charge state dynamics we monitored the PL features of NV 0 and NV -, which are respectively characterised by zero phonon lines (ZPLs) at 575 nm and 637 nm accompanied by broad sideband emission with maxima around 640 nm and 700 nm. 7nversion from NV -to NV 0 leads to a relative increase (decrease) in PL intensity below (above) ~700 nm, as illustrated in Fig. 4(a).We recorded PL spectra using ~ 40 W laser power, before and after 1.2 mW laser exposure for one hour.The difference PL spectra obtained by normalizing are consistent with the observed changes in ODMR contrast and mechanism described in Figure 3.
To gain further insights into material origin of observed ODMR contrast changes, XPS and in-situ Raman spectroscopy were performed (see Fig. 5).The surface composition dynamics due to high laser power exposure under different environmental conditions were analyzed by XPS (for details of sample preparation, see Fig. S5). 30The ED sample exhibits only carbon (C 1s) and oxygen (O 1s) elements within the XPS probing depth (< 11.7 nm calculated based on the maximum IMFP).
For AC-ED, the Al2O3 layer was also observed (evidenced by characteristic aluminium peaks (Al 2p and Al 2s) and an intense O 1s peak) in addition to carbon and oxygen (Fig. 5(a)). 47The carbon to oxygen relative atomic ratios (C/O) reveal significant surface reconstruction in ED due to laser exposure (Fig. S6).The change in the C/O ratio comparing exposed and unexposed positions indicates that laser exposure in vacuum results in more efficient surface oxygen detachment compared to that in air (Fig. 5(b)).The C 1s core level spectra for different samples were calibrated (peak fitting was used to determine the sp 3 peak position) to the reported binding energy value for diamond (285.0 eV) and are shown in Fig. 5(c). 26,48 he C 1s core level spectra of ED when laser exposed in different environments show a variation in spectral shape because of laser exposure (Fig. 5(c)).Peak fitting of the C 1s core level spectra was performed to disentangle, identify, and quantify the different carbon-related chemical states (Fig. S6).In ED, non-diamond carbon (sp 2 ) is found to be present in addition to diamond (sp 3 ) (SI, Fig. S6(c)). 49,50 he reduction in the ratio of sp 2 to sp 3 under laser exposure is much greater in air than in vacuum (Fig. 5  efficient sp 2 carbon etching and therefore the rates of removal of C-O and C=O remain similar.In summary, these XPS measurements are consistent with the proposed mechanism (see Fig. 3), that surface oxygen is detached as a result of laser illumination.The C 1s core level spectrum for AC-ED (Fig. 5(c)) reveals a more significant fraction of sp 2 carbon compared to sp 3 .This can be explained by a change in signal intensity from the diamond sample itself when the ~ 2 nm Al2O3 layer is added on top.This leads to a relative increase in the signal seen from the Diamond surface (sp 2 ) compared to its bulk (sp 3 ).In addition, the O 1s core level spectrum of AC-ED is dominated by Al2O3, making it difficult to evaluate the diamond surface oxygen functionalization.Due to these factors, we did not perform XPS measurements on laser exposed AC-ED sample.However, it will be interesting to explore the AC-ED surface further to see if Al2O3 alters the diamond functionalization and how the diamond functionality varies under laser exposure.
The material changes due to laser exposure were further studied by Raman spectroscopy (Fig. 5(e,f)).Raman spectra of ED and AC-ED samples acquired under low laser excitation power demonstrated the absence of non-diamond carbon and related defects (Fig. S8) in addition to a sharp peak at ~1331.8 cm -1 characteristic of diamond.To observe the effects of laser exposure, insitu Raman measurements were performed (see Fig. S8 and SI for details).ED showed detectable G band features (centered at 1590 cm -1 ), indicative of graphitic carbon, whose intensity reduced significantly after 4.8  10 3 s of continuous high power laser exposure.For AC-ED sample, a G band feature was observed at ~1550 cm -1 (blue shifted compared to ED) and remained unchanged under prolonged laser exposure.The diamond related Raman features remained unchanged for both samples during high laser power exposure (Fig. S8).These observations are in agreement with XPS and suggest etching of sp 2 carbon in ED during laser exposure in air.We could not find detectable G band for either the ED or AC-ED sample under laser exposure in vacuum, possibly due to significant reduction in laser excitation intensity by vacuum glass cell.

Summary and conclusion:
The instability of near surface NV centers under non-ambient conditions is a long-standing challenge for diamond-based quantum sensing, particularly for work at cryogenic temperatures.
To understand the origin of this instability, we have combined a study of the optical properties of the NV centers with an analysis of the material composition of the diamond surface.We observe a change of the ODMR contrast for near-surface (~ 5 nm deep) NV centers under laser illumination in oxygen functionalized diamond which we attribute to surface reconstruction under different environmental conditions.In vacuum, the ODMR contrast was reduced from around 6% to below 4.5%, whereas it increased under air up to over 7.5%.A ~ 2 nm layer of Al2O3 was deposited on diamond surface which successfully led to a stable ODMR contrast, even under laser exposure.
The origin behind these changes in ODMR contrast was revealed by PL, XPS and Raman spectroscopies to arise from NV charge state switching caused by surface dynamics.In vacuum, owing to lack of atmospheric gases, electron traps develop on surface and the NV -charge state is converted into NV 0 .Atmospheric oxygen inhibits the development of such traps and increases the NV -charge state fraction.The Al2O3 layer prohibits both the degradation of surface as well as adsorption of environmental oxygen, achieving a more stable NV charge state.The Al2O3-oxygendiamond surface is shown to be resilient against optical excitation in vacuum, but requires further investigation under low temperature conditions, while the NV spin coherence properties should be analyzed in such material to assess its potential for quantum sensing.The use of alumina coating could be extended to help stabilise single near surface NV centers in planar and nanopillar diamond structures.A single NV-diamond AFM probe with stability under different environmental conditions might be achievable using such optimized passivation of the surface.

Figure 1 .
Figure 1.(a) Schematic of optical measurement setup used for PL and ODMR.(b) Summary of the sample preparation and optical measurement steps.(c) Fluorescence saturation measurements (squares) for NVD sample with a fit (curve) to the saturation curve (see text).Red shaded regions show the PL signal counts at powers used in subsequent laser exposure experiments.(d) Continuous wave (CW) ODMR spectrum for NVD sample (black dots) and double Lorentzian fitting (solid red line).(e) Pulse sequence to study ODMR contrast under the influence of periodic high power laser pulses of increasing duration.

Figure 1 (
c) shows the PL intensity as a function of laser power P, fitted to the function: . .

Figure 2 .
Figure 2. Evolution of ODMR contrast as a function of the total duration of laser power exposure for NVD and AC-NVD samples under (a) vacuum and (b) air environments.A series of laser exposures of increasing duration (from 1 s to 3.6 ks) is first applied using 1.2 mW laser power.After a cumulative exposure time of about 15 ks, the laser power is increased to 2 mW, and the exposure time per datapoint reset to 1 s time and subsequently increased.The data are fit to exponential decay functions, with separate time constants for the periods of 1.2 mW and 2 mW laser exposure (see text).

Figure 3 .
Figure 3. Schematic for proposed mechanism (a) The acid refluxed diamond sample (NVD) has a shallow NV-doped layer up to 5 nm from the surface, which consists of non-diamond carbon (grey region) and oxygen functionalities (orange dots) on the surface.(b) Laser exposure under different environments can cause desorption of non-diamond carbon and oxygen-containing functional groups (in vacuum), or an increase in oxygen termination on the surface (in air).These changes lead to charge state conversion between NV -and NV 0 .

Figure 4 .
Figure 4. (a) Cartoon representation showing how the PL spectrum evolves as the NV -/NV 0 fraction changes, with the difference shown in the lower panel.Measured difference PL spectra for NVD and AC- (d)).Peak fitting of the O 1s core level spectra (Fig. S7) reveals that the rate of elimination of specific oxygen functionalities (ether/alcohol (C-O-C/C-O-H) or ketone (C=O)) depends on the environment: Due to efficient etching of sp 2 carbon during laser exposure under air, the concentration of C-O bonds increases whereas that of C=O bonds decreases.Laser exposure under vacuum induces less

Figure 5 .
Figure 5. Surface spectroscopy (a) XPS survey spectra of ED and AC-ED samples before laser exposure.(b) Changes in the total carbon to oxygen (C/O) ratio of the ED sample as a result of laser exposure, in different environments.(c) C 1s core level spectra showing results from regions in ED exposed to laser (pink, grey), as well as unexposed regions (black, red).The spectra are normalized to the maximum peak height.(d) Changes in the sp 2 /sp 3 carbon ratio of the ED sample as a result of laser exposure, in different environments.(e) Normalised Raman spectra for ED and AC-ED samples under high power laser illumination in air.Red dotted lines are the baselines used for estimating G band intensity for ED and AC-ED.(f) The change in graphitic to diamond carbon ratio (IG/IDia) as a result of high-power laser exposure under air environment.

Figure S2 .
Figure S2.ODMR spectra before and after laser exposure for (a) NVD under vacuum (b) AC-NVD under vacuum (c) NVD under air and (d) AC-NVD under air respectively.

Figure S3 .
Figure S3.Normalized PL spectra under different conditions before and after high power laser exposure.(a)and (b) show normalized spectra for BD sample under air and vacuum respectively.(c) and (d) show the PL spectra of AC-ED sample under air and vacuum respectively.

Figure S4 .
Figure S4.Normalized PL spectra of NVD and AC-NVD samples before high power laser exposure under (a) air and (b) vacuum respectively.

Figure S5 .
Figure S5.Sample preparation for XPS measurements.(a) shows the Raman mapping pattern over diamond.(b) and (c) show the laser exposed region of different samples under air and vacuum respectively.

Figure S6 .Figure S7 .
Figure S6.(a) Al 2p core level spectrum for AC-ED sample.(b) C/O atomic ratio for EDs after high power laser exposure under different environments.(c) Representative C 1s fitted core level spectrum for sample ED after laser exposure under air environment.(d) The sp 2 /sp 3 carbon ratio for different ED samples estimated from peak fit analysis of the corresponding C 1s core level spectra.

Figure S8 .
Figure S8.a) Experimental scheme for Raman spectroscopy.(b) Raman spectra for ED and AC-ED samples at low excitation laser power and corresponding zoomed D and G band regions.There is no observable feature related with defects (D band) and graphitic carbon (G band) at low laser excitation power.(d) and (e) show the comparison of characteristic diamond Raman features (spectra were acquired at low laser power) before and after high laser power exposure under air environments for ED and AC-ED samples respectively.