Skip to content

Local nanoscale probing of electron spins using NV centers in diamond

FieldValue
JournalDiamond and Related Materials
Published2026-03-10
AuthorsSergei Trofimov, Christos Thessalonikios, Victor Deinhart, Alexander Spyrantis, Lucas Tsunaki
DOI10.1016/j.diamond.2026.113530

Double electron-electron resonance after focused helium implantation measured 234 ± 13 ppb substitutional nitrogen in the {111} sector of type IIa HPHT diamond. The method had a calculated detection limit of 5 ppb and a spatial resolution below 1 μm. The concentration of an implantation-related defect reached 15 ppb, depending on the implantation dose. Throughout, ppb denotes atomic parts per billion.

DEER used implanted nitrogen-vacancy (NV) ensembles as sensor spins and substitutional nitrogen P1 centers as target spins. Four noncentral nitrogen resonances supplied the concentration because other electron spin species can contribute to the central line. The central line was analyzed separately for an additional defect. The four-line average was 234 ± 13 ppb in the {111} sector. The result was consistent with an earlier estimate of about 80 ppb for this sector. That estimate came from an EPR average of 22 ppb and an assumption that other sectors contained at least ten times less nitrogen. Photoluminescence saturation estimated the sector concentration ratio as 10.1:1.3:1.0 for {111}, {113}, and {001}. This estimate assumed ensemble photoluminescence was proportional to nitrogen content. The calculated detection limit was 5 ppb for T_B = 0.35T_2 ≈ 100 μs, 5% contrast, and Ω ≫ Γ. Infrared absorption has a detection limit around 1 ppm and spatial resolution of tens of microns. Ultraviolet absorption reaches about 100 ppb with spatial resolution on the order of tens of microns. EPR can detect down to 1 ppb, but quantitative analysis needs a reference sample and faces limited spatial resolution. Long spin relaxation times also complicate EPR measurements. Local DEER combines ppb-scale concentration analysis with submicrometer probe positioning.

The specimen was a (001)-oriented 3 mm × 3 mm × 0.37 mm single-crystal diamond plate. It was cut from type IIa bulk diamond grown at 5 GPa and 1750 K by temperature-gradient HPHT. The plate was mechanically polished to Ra < 5 nm. Growth sectors formed because HPHT diamond grows at different rates on different facets. Ultraviolet luminescence separated the {111}, {001}, and {113} sectors and showed their different defect concentrations. Similar samples had about 500 ppb nitrogen in {111} and below 50 ppb in {001}. The local DEER result was below that typical {111} value. Quantitative EPR gave 22 ppb for the average concentration in the studied sample. Confocal microscopy found no detectable native single NV centers before implantation.

Helium ions were implanted at 30 keV with doses from 5 to 5000 ions per spot. Stopping and Range of Ions in Matter simulations estimated 37 vacancies per ion and a vacancy depth maximum near 120 nm. The vacancy distribution’s full width at half maximum corresponded to a radius of 37.5 nm. The anneal ran in argon at 7 mbar. It included 1 hour at 400°C, 1 hour ramping to 990°C, 2 hours at 990°C, and 2 hours of cooling. The effective probe volume therefore depends on ion straggle and vacancy diffusion during annealing. Each patterned site produced an NV ensemble with bright photoluminescence. Unfocused neutral atoms created a circular background feature about 60 μm across. The authors note that suppressing this artifact requires a specialized ion-column design.

NV and P1 transitions were driven at approximately 1.8 GHz and 1 GHz under a magnetic field near 37 mT. A green continuous-wave laser and a 0.95 NA objective provided excitation and collection. The pulse sequence combined an NV Hahn echo with an added microwave pulse for the target spins. First, a frequency sweep located target resonances. Second, DEER Rabi oscillations at each resonance determined the Rabi frequency and the corresponding DEER pulse length. Third, a second frequency sweep used the calibrated pulse length to fit concentration. The final fit treated resonance positions, linewidths, and concentration as free parameters. P1 analysis used T_A = 20 μs, selected from the measured coherence decay and {13}C modulation. NV concentration measurements used T_A = 80 μs to capture the weaker off-axis signal. QuaCCAToo simulations used B_0 = 37.2 mT and a 0.1° field misalignment. Calculated spin populations were 40%, 30%, and 30% for the three ground-state levels. The measured off-axis transition therefore represented 45% of all NV centers.

P1 spectra were measured from 35 to 5000 ions per spot in {111}. The central P1 resonance developed a shoulder at the 5000-ion dose. One simulated spectrum used 200 ppb P1 and 13 ppb of the additional defect, designated X. The additional peak was not visible at 50 and 250 ions per spot. It became detectable from 1000 ions per spot and increased at 5000. The fitted concentration of X reached 15 ppb, depending on the implantation dose. The authors suggest a vacancy-related origin for this implantation-induced defect. One candidate was the WAR10 defect, comprising interstitial nitrogen and interstitial carbon with spin 1/2. The measurements support the spin-1/2 model but do not establish the defect identity. The NV concentration also rose with implantation dose from 50 to 2500 ions per spot. DEER and photoluminescence saturation agreed qualitatively on this increase. Lower P1 values at some doses may reflect conversion into NV and other defects, or local nitrogen variation. Both explanations remain unresolved.

Implanted NV ensembles in {111} had an average Hahn echo coherence time of 284 ± 14 μs. A 5000-ion site in the supplement gave T_2 = 313 ± 5 μs. The coherence approached the limit imposed by the natural 1.1% {13}C abundance. Combining DEER concentration with the photoluminescence-derived NV count gave V_NV = (4.8 ± 0.3) × 10^-2 μm³. Assuming a spherical volume gave r_NV = 226 ± 13 nm. The difference in squared radii gave a vacancy diffusion length of 223 ± 13 nm. The derived diffusion coefficient was 1.2 ± 0.1 nm²/s over 7200 s. Published estimates were 1.8 nm²/s at 850°C and 1.1 nm²/s at 750°C. Those comparisons support the inferred vacancy diffusion coefficient. The implantation-defined geometry produced spatial resolution below 1 μm.

The quantitative P1 result applies to the {111} sector only. The authors expected about 20 ppb in {113}, but they detected no central P1 resonance. Saturation curves may have overestimated NV numbers in {113} because other luminescent defects, including NV0, were present. Fewer NV centers also increased the averaging time required for DEER. The {001} sector was not measured because its estimated NV count was similar to {113}. The 5 ppb limit is a calculated threshold that assumes 5% visible contrast and a favorable drive condition. The central line cannot identify defect X alone because other electron spin species share its g-factor. The X defect remains unidentified. For material selection, bulk absorption can hide sector-scale nitrogen variation in HPHT diamond. Local DEER can test sector-scale nitrogen variation after controlled implantation and annealing. The implantation dose must therefore balance NV production against additional paramagnetic defect creation. For NV device engineering, nitrogen content must be low enough to limit magnetic noise while retaining enough substitutional nitrogen for conversion into sensors. That balance should be checked locally under the intended implantation and annealing conditions rather than inferred from a bulk value.

The DEER sensing experiment required a low-nitrogen, (001)-oriented type IIa single-crystal diamond plate measuring 3 mm × 3 mm × 0.37 mm. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal plates with thickness from 0.1 to 500 micrometers, but nitrogen equivalence to the HPHT sample was not established. The reported process also requires focused helium implantation and a 990°C argon anneal, so process compatibility should be checked through the engineering support page.

View Original Abstract

Substitutional nitrogen atoms in a diamond crystal (P1 centers) are, on one hand, a resource for creation of nitrogen-vacancy (NV) centers, that have been widely employed as nanoscale quantum sensors. On the other hand, P1’s electron spin is a source of paramagnetic noise that degrades the NV’s performance by shortening its coherence time. Accurate quantification of nitrogen concentration is therefore essential for optimizing diamond-based quantum devices. However, bulk characterization methods based on optical absorption or electron paramagnetic resonance often overlook local variations in nitrogen content. In this work, we use a helium ion microscope to fabricate nanoscale NV center ensembles at predefined sites in a diamond crystal containing low concentrations of nitrogen. We then utilize these NV-based probes to measure the local nitrogen concentration on the level of 230 ppb (atomic parts per billion) using the double electron-electron resonance (DEER) technique. Moreover, by comparing the DEER spectra with numerical simulations, we managed to determine the concentration of other unknown paramagnetic defects created during the ion implantation, reaching 15 ppb depending on the implantation dose.

  1. 2020 - Nitrogen in diamond [Crossref]
  2. 2008 - High-sensitivity diamond magnetometer with nanoscale resolution [Crossref]
  3. 2011 - Electric-field sensing using single diamond spins [Crossref]
  4. 2013 - High-precision nanoscale temperature sensing using single defects in diamond [Crossref]
  5. 2006 - Room-temperature manipulation and decoherence of a single spin in diamond [Crossref]
  6. 2020 - Decoherence of ensembles of nitrogen-vacancy centers in diamond [Crossref]
  7. 2022 - Decoherence of nitrogen-vacancy spin ensembles in a nitrogen electron-nuclear spin bath in diamond [Crossref]
  8. 1990 - Growth-sector dependence of optical features in large synthetic diamonds [Crossref]
  9. 2013 - The origin of color in natural C center bearing diamonds [Crossref]