Skip to content

High Resolution Temperature‐Resolved Spectroscopy of the Nitrogen‐Vacancy 1 E Singlet State Ionization Energy

FieldValue
Journalphysica status solidi (a)
Published2026-03-11
AuthorsKristine Ung, Connor Roncaioli, Ronald L. Walsworth, Sean M. Blakley
DOI10.1002/pssa.202500538

The 1E singlet state in negatively charged nitrogen-vacancy centers has an ionization energy between 2.29 and 2.33 eV. The measurement covers diamond temperatures from 50 to 150 K. This interval cuts uncertainty by about twofold and resolves the optical energy needed for direct spin-to-charge conversion. The authors obtained it through magnetically mediated photoluminescence quenching in an ensemble of NV− centers.

Magnetic-field control of the ionization pathway

Section titled “Magnetic-field control of the ionization pathway”

Each wavelength and temperature produced a reference spectrum at 0 mT and a signal spectrum at 100 mT. The field was aligned within 1° of the diamond [001] axis and mixed the ms = 0 and ms = -1 states of 3E. This mixing increased population in 1E through the upper intersystem crossing.

Without the field, 1E population was negligible, so the reference reflected direct ground-state ionization or two-photon Auger ionization through 3E. At the same wavelength, the field-on measurement added the competing 1E ionization and shelving pathways. Below threshold, 1E shelving reduced 3E occupancy and lowered NV0 photoluminescence relative to the reference. Above threshold, 1E ionization increased NV0 photoluminescence relative to the reference.

The experiment used a CVD-grown quantum-grade diamond plate with the following reported conditions.

ItemReported condition
Sample3 mm × 3 mm × 0.5 mm diamond
Isotopic content99.99% 12C
Defect concentrations[N] ≈ 16 ppm; [NV] ≈ 4 ppm
Excitation bands450 to 470 nm and 540 to 566 nm, in 2 nm steps
Optical power0.1 mW and 1.0 mW
Temperature50 to 150 K, in 5 K steps
Magnetic field0 mT reference; 100 mT signal, aligned to [001] within 1°
Spectral collection1050 pairs, each using a 100 ms exposure and 10 frame averages

The defects were homogeneous throughout the plate volume, and the polished top face followed the [100] direction. The electromagnet let the plate lie flat while maintaining a field aligned with [001].

The visible source combined an optical parametric oscillator with periodically poled lithium niobate second-harmonic generation. It covered 450 to 520 nm and 540 to 650 nm, with a 520 to 540 nm dead band around the 532 nm pump. The stabilizer adjusted every 25 ms and held output within ±0.25% of the selected setpoint.

The wavemeter showed wavelength accuracy of about 0.5 nm, and sample-stage drift stayed below 1 K during each pair. A 570 nm long-pass filter rejected excitation before a 350 to 1140 nm spectrometer collected the emission. The full matrix produced 1050 signal and reference pairs from 21 temperatures, 25 wavelengths, and two powers.

Measured NV0 photoluminescence amplitude is proportional to its relative population. The analysis divided each field-on spectrum by its field-off reference, without separating the full spectrum into NV− and NV0 components.

The 575 to 605 nm band contains appreciable NV0 emission and negligible NV− emission. A ratio above one indicates added NV0 population from 1E ionization. A ratio below one indicates shelving, while a ratio near one leaves the behavior indeterminate.

The normalized CNV0 contrast used the opposite sign, with positive values denoting shelving and negative values denoting ionization. Values within one standard deviation of zero were classified as indeterminate. The authors also tracked normalized contrast in the 680 to 740 nm NV− band.

They chose the ratio method because low NV0 signal-to-noise can defeat the earlier Bayesian fit, partly because of substitutional nitrogen P1 defects. It also requires less computation than the multiple double integrals and statistical parameter search used by that fit.

At 0.1 mW, a few strong-ionization regions appeared near 470 nm for temperatures above 100 K. The 470 nm photon energy was 2.63 eV, at the edge of the triplet ionization threshold. At 1.0 mW, strong ionization appeared near 470 nm at most temperatures between 50 and 150 K.

For 540 to 566 nm at 1.0 mW, shelving dominated at every measured temperature. The exceptions were 548 nm at 95 K and 566 nm at 150 K. The band endpoints correspond to 2.29 eV at 540 nm and 2.19 eV at 566 nm. These observations separate ionization near the triplet threshold from mostly shelving behavior at longer wavelengths.

Earlier direct photoionization measurements placed the energy between 2.25 and 2.33 eV. They used about 13 nm average wavelength resolution and about 50 K average temperature resolution. The new scan used 2 nm wavelength steps and 5 K temperature steps.

That earlier bracket corresponded to 532 nm at 2.33 eV and 550 nm at 2.25 eV. The combined measurements place the onset between 532 nm at 2.33 eV and 540 nm at 2.29 eV for 50 to 150 K. The upper bound remains the previous 2.33 eV limit, rather than a new direct measurement within the laser dead band. The lower bound corresponds to the 540 nm long-wavelength edge of the unavailable 520 to 540 nm region. The 540 nm data showed shelving rather than ionization. The authors describe this interval as a twofold reduction in uncertainty.

The threshold itself lay in the laser’s 520 to 540 nm dead band, so the current scans did not bracket it with two neighboring points. The data came from an ensemble at 50 to 150 K, so the paper reports neither single-center nor room-temperature thresholds.

The 548 nm point at 95 K was attributed to laser power drift during the pair measurement. The stabilizer used a 15 s rolling average, allowing brief excursions of several percent to pass the stability check. Such excursions can shift one spectrum’s amplitude relative to the other and imitate large contrast. The physical cause of the 566 nm point at 150 K remains undetermined.

Planned work includes more laser powers and a combined theoretical and experimental study of Jahn-Teller effects on 1E ionization. Data are available from the corresponding author on request but are not public because of stated privacy or ethical restrictions.

The measured interval gives a tighter photon-energy constraint for direct 1E spin-to-charge conversion than the earlier bracket. The 1E state is the least understood level in the known NV− electronic structure. Its lifetime is about an order of magnitude longer than that of every other NV− state except the 3A2 ground state.

At 1.0 mW, the 540 to 566 nm data show that longer wavelengths mostly shelve 1E without producing direct ionization. Near 470 nm, ionization was strong but sat close to the triplet threshold and depended on power and temperature. Device designs should distinguish singlet ionization from triplet-threshold effects when choosing excitation wavelength and power. Engineers can use the measured interval to bound the optical requirement for spin-to-charge conversion. The measured sample had [N] ≈ 16 ppm and [NV] ≈ 4 ppm, which should accompany comparisons with other diamond material.

The authors cite higher contrast and signal count rate than typical fluorescence sensing as advantages of spin-to-charge conversion and photoelectric detection of magnetic resonance.

The nitrogen-vacancy spectroscopy application requires an isotopically enriched single-crystal diamond plate with controlled nitrogen and NV centers, a polished optical face, and defined crystallographic orientation. The 6CCVD diamond materials catalog includes single-crystal and polycrystalline diamond grown by MPCVD, with plates and wafers up to 125 mm. The 0.5 mm paper plate falls within the catalog’s 0.1 to 500 micrometer range; the linked engineering support can confirm the remaining specifications.

View Original Abstract

The negatively charged diamond nitrogen‐vacancy () center plays a central role in many quantum sensing applications, though much is still unknown about the energy levels in this system. The ionization energy of the singlet state in the center has only recently been measured to be between 2.25 and 2.33 eV. In this work, we reduce this uncertainty by measuring the ionization energy as a function of excitation laser wavelength and diamond temperature via magnetically mediated spin‐selective photoluminescence (PL) quenching that transfers population from the in the into the neutral charge configuration. Measurements are performed for excitation wavelengths between 450 and 470 nm, and between 540 and 566 nm in increments of 2 nm, and for temperatures ranging from about 50 to 150 K in 5 K increments. We determine the ionization energy to be between 2.29 and 2.33 eV, which provides about a twofold reduction in uncertainty of this quantity. Distribution level: A. Approved for public release; distribution unlimited.