Skip to content

Enhancement of Coherence Times of Individual Shallow NV Center due to Ozone Treatment

FieldValue
JournalAdvanced Quantum Technologies
Published2026-04-01
AuthorsJessica fernanda Da Silva Barbosa, A. Kaiser, Alma Beatrice Sergi, Allegra de Gleria Clark, Johannes Lang
DOI10.1002/qute.202500938

Ozone treatment produced a Hahn Echo coherence time of 400 ± 49 µs for shallow NV centers at 30 to 35 nm depth. The study reports up to tenfold higher Ramsey coherence and threefold higher Hahn Echo coherence. For the same centers shallower than 20 nm, Ramsey coherence rose by as much as 3.2 ± 0.7. X-ray photoelectron spectroscopy measured reduced surface sp2 carbon. The authors associated the spin response with that surface change.

Both samples started as 4.0 × 4.0 mm² Element Six single-crystal substrates with {100} surfaces. They were 0.50 mm thick, contained 1.1% 13C, and had substitutional nitrogen below 5 ppb, boron below 1 ppb, and existing NV centers below 0.03 ppb.

Photolithography and oxygen-plasma reactive ion etching created 100 nm deep reference marks. Plasma-enhanced chemical vapor deposition then added a 100 nm layer of isotopically purified 12C diamond.

Sample15N+ implantationLayer after implantationExpected depth and centers measured
A(5 \times 10^8) ions/cm² at 5 keVNone10 to 15 nm; the same six centers before and after treatment
B(1 \times 10^{11}) ions/cm² at 5 keV20 nm of 12C diamond30 to 35 nm; ten centers before and six different centers after treatment

Both samples received a vacuum anneal at 1373.15 K, or 1100 °C. SRIM calculations gave the expected implant depths.

The treatment used a home-built UV ozone apparatus. A constant 40 ml/min flow of 99.999% oxygen passed through a 12 ml chamber. A Xenon XBO 75 arch lamp irradiated the gas for 60 min. Ozone test strips measured at least 150 µg/m³.

Spin characterization used a home-built confocal microscope with a frequency-doubled Nd:YAG laser at 532 nm. A 100X oil objective collected photoluminescence under a 47 mT external magnetic field. The authors measured ODMR, Rabi, Ramsey, and Hahn Echo sequences, while XY8 measurements determined NV depth.

The same six Sample A centers were tracked before and after ozone treatment. For Sample B, increased fluorescence prevented identification of the same centers after treatment. Ramsey decay was fitted with a sine exponential. Hahn Echo used a stretched exponential with the exponent fixed at (p=2). Common-mode noise was removed by alternating the phase of the final pulse.

Tapping-mode AFM scans covered 30 × 30 µm² before ozone treatment. RMS roughness was 710 pm for Sample A and 550 pm for Sample B, compared with 1049 pm for the as-received substrates. The maximum mountain-to-valley differences were 2.5 nm and 1.5 nm, respectively.

Before ozone treatment, XPS measured 91 at% carbon and 8 at% oxygen. C 1s spectra resolved sp3 carbon at 284.8 eV and sp2 carbon at 283.6 eV. Oxygen-related components appeared at 285.6 eV, 286.6 eV, and 287.6 eV.

The bulk-sensitive XPS geometry used a 45° emission angle and represented the top approximately 7 nm. The estimated oxidized sp2 carbon overlayer fell from 0.5 nm before treatment to 0.1 nm afterward. These values corresponded to about five sp3 monolayers and one or two monolayers, respectively. The authors warn that the 0.1 nm estimate exceeded the calculation’s application range.

The 70° surface-sensitive geometry represented approximately the top 3 nm. It also showed reduced sp2 carbon, but shadowing prevented a quantitative comparison. Ozone left the overall 8 at% oxygen abundance nearly unchanged. The relative C-O=C contribution rose slightly.

The authors suggest that ozone forms simple carbon oxides that then desorb from the surface. A separate heteroepitaxial CVD diamond test showed a similar chemical response.

The spin measurements showed a depth-dependent response to ozone treatment. The following values summarize the measured coherence times, contrast, and formation yields for the two samples.

MeasurementSample A, expected depth 10 to 15 nmSample B, expected depth 30 to 35 nm
Ramsey (T_2^*)Best case increased from 2.1 to 6.8 µs. Reported factors ranged from 0.7 ± 0.2 to 3.2 ± 0.7.Before treatment, values were below 2.5 µs. The Results text gives a post-treatment maximum of 14.5 ± 2.7 µs.
Hahn Echo (T_2)The (T_2/T_2^*) ratio increased by 1.4 ± 0.2 to 4.1 ± 0.4.Before treatment, values were below 200 µs. The post-treatment maximum was 400 ± 49 µs.
ODMR and Rabi contrastNo measurable change.Increased by a factor of 2.
NV formation yield2.7% before treatment and 4.1% afterward.0.00093% before treatment and 0.03% afterward.

These results indicate that while Sample A showed measurable improvements in coherence for most tracked centers, Sample B achieved substantially larger coherence gains and higher formation yields at a greater depth.

The Sample A values came from the same six centers before and after treatment. One center at 8 nm had (T_2^*) below 2 µs in both measurements and became worse after treatment. Its Hahn Echo coherence still increased. The authors linked this exception to an unknown local decoherence source, possibly a nearby crystalline defect.

Sample B values came from different center sets. Its reported tenfold (T_2^*) increase and threefold (T_2) increase are therefore group comparisons rather than paired measurements. The formation yield rose by a factor of 1.5 for Sample A and 32 for Sample B. Photoluminescence background decreased on both samples.

Depth, noise, and comparison with prior work

Section titled “Depth, noise, and comparison with prior work”

Before ozone treatment, (T_2) and (T_2^*) were comparable for Sample A. This indicated that Hahn Echo did not remove the dominant decoherence source, which the authors describe as nonmagnetic. After treatment, both times increased and their ratio widened. The authors associated the response with reduced electric noise, smoother surfaces, and less incoherent photon scattering.

Sample B had a Hahn Echo time two orders longer than its Ramsey time before and after treatment. The authors therefore interpreted the post-treatment magnetic noise as dominant. The 20 nm overgrowth prevents some surface noise from reaching the spins. However, oxygen termination can affect spins as far as 100 nm below the surface. The high nitrogen dose also introduces nitrogen-related defects and a nuclear spin bath.

The untreated Sample B results were in the same order of magnitude as the paper’s cited 5 keV implant with 20 nm of 12C overgrowth. A cited 15 nm hydrogen-terminated NV coated with h-BN had a reported (T_2) of 2.4 µs. Treated Sample B reached 400 ± 49 µs, although the surface and depth conditions differed. Deeply implanted centers have shown coherence times exceeding 2 ms, but they lie much farther from the surface.

The paired evidence covers six Sample A centers. Sample B compares ten centers before treatment with six different centers afterward. Response varied even at equal depth because of local electric and magnetic fluctuations, surface roughness, and nearby lattice defects.

Yield was inferred from bright photoluminescence spots, so P1 centers, vacancy aggregates, and H clusters may confound the count. The authors recommend a 650 nm long-pass filter with excitation below 575 nm to assess NV- centers specifically.

Assignment of the XPS sp2 carbon peak requires caution because charging and complex carbon defects can broaden C 1s. Shadowing blocked quantitative surface-sensitive comparisons. Cu and Pb residues below 1 at% were traced to the soldered wire antenna, leading the authors to recommend less contamination-prone microwave integration.

The conclusion reports (T_2^*) up to 20 ± 4 µs and a 160% contrast increase. The Results text instead gives 14.5 ± 2.7 µs and a contrast increase by a factor of 2. These differences need reconciliation before process acceptance criteria are set.

For device engineering, ozone performance should be validated against surface roughness, implant dose, overgrowth thickness, and NV depth. Tracking the same centers before and after treatment remains the clearest test of local improvement. The study measured coherence but did not report qubit gate fidelity.

Shallow-NV quantum sensing and computing require single-crystal diamond plates with a controlled surface. The diamond materials catalog lists 6CCVD’s single-crystal and polycrystalline MPCVD-grown material as plates and wafers up to 125 mm, with thickness from 0.1 to 500 µm. The paper’s 0.50 mm {100} substrate falls within that thickness range, and the engineering support link provides the contact for custom fabrication needs associated with its surface integration.

View Original Abstract

ABSTRACT Shallow nitrogen vacancy (NV) centers with long coherence times are promising candidates for applications in areas such as quantum computing and quantum sensing. However, as the NVs are only a few nanometers below the diamond’s surface, it is a challenge to remove the different decoherence sources related to the surface, e.g., charge fluctuations and magnetic impurities. Here, the effect of ozone treatment on the chemical composition of diamond surfaces was correlated with its impact on the spin properties of the same individual shallow NV (depth < 20 nm) by comparing its coherence times, Ramsey and Hahn Echo (T 2 ), before and after the treatment. X‐ray photoelectron spectroscopy (XPS) shows a reduction in sp 2 bonds, and it was reported that an enhancement of the coherence times up to a factor of 3. Additionally, we investigate the effects of Ozone treatment on random single NVs engineered to have a depth of ∼30 nm, where an enhancement up to a factor of 10 on and up to a factor of 3 on T 2 have been observed.