Ramsey correlation spectroscopy with phase cycling using a single quantum sensor
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Open MIND |
| Published | 2026-03-05 |
| Authors | Inbar Zohar, Santiago Oviedo-Casado, Andrej Denisenko, Rainer Stöhr, Amit Finkler |
| DOI | 10.48550/arxiv.2603.05650 |
RESOLUTE extended the effective coherence time of a single NV center from (T_2^*=0.38 \pm 0.2) µs to (T_2^p=5.1 \pm 0.2) µs. The authors described the resulting coherence as 15 times longer than Ramsey. It also detected (^{13}\mathrm{C}) nuclear spin precession near 53 kHz at 49.5 G, below the 110 G limit reported for Hahn Echo correlation sensing.
Coherence gain and low-frequency result
Section titled “Coherence gain and low-frequency result”The coherence comparison used the same NV center at (64.3 \pm 0.1) G. Ramsey gave (T_2^*=0.38 \pm 0.2) µs, while Hahn Echo gave (T_2=4.36 \pm 0.01) µs. RESOLUTE gave (T_2^p=5.1 \pm 0.2) µs with a 10 µs correlation time. RESOLUTE also exceeded Hahn Echo under the same bias field.
RESOLUTE correlated two sensing periods so that static phases cancel in their difference. Fast fluctuations remain in the difference and determine the residual decay. When an oscillating field period matches an integer multiple of (T_{\mathrm{corr}}), its phase shift becomes (2\pi) and the field cancels. Scanning (T_{\mathrm{corr}}) therefore produced coherent revivals.
Pulse sequence and phase cycling
Section titled “Pulse sequence and phase cycling”After green-laser initialization, a (\pi/2) microwave pulse starts the first Ramsey period. The second (\pi/2) pulse converts accumulated phase into a population imbalance. This imbalance is stored for (T_{\mathrm{corr}}), with (T_1) as its limiting relaxation process.
A third pulse starts the second Ramsey period, and a final (\pi/2) pulse maps the result to populations. Green laser fluorescence then reads those populations. Four phase-cycled repetitions use x and y middle-pulse phases with opposite final projection phases.
The added x and y outputs isolate fields that vary during (T_{\mathrm{corr}}). The subtracted output cancels DC contributions that remain unchanged. Fields changed during (T_{\mathrm{corr}}) remain in the signal, including manipulated dipolar coupling. Flipping the final projection phase suppresses systematic errors.
Pairwise correlation during acquisition preserves target phase after averaging. Earlier phase-sensitive low-frequency proposals required continuous accumulation and post-processing correlation. RESOLUTE uses a fixed pair separation and least-squares fitting.
Carbon-13 nuclear spin detection
Section titled “Carbon-13 nuclear spin detection”With the sensing time fixed at (\tau=6) µs, the authors scanned (T_{\mathrm{corr}}) under three bias fields. The lower fitted frequency at each field was assigned to (^{13}\mathrm{C}) Larmor precession.
| Bias field | Fitted revival frequency | Expected (^{13}\mathrm{C}) Larmor frequency |
|---|---|---|
| (49.5 \pm 0.2) G | (53.2 \pm 1.6) kHz | (52.9 \pm 0.2) kHz |
| (64.3 \pm 0.1) G | (71.2 \pm 1.3) kHz | (68.8 \pm 0.1) kHz |
| (107.2 \pm 0.2) G | (117 \pm 1) kHz | (114.5 \pm 0.2) kHz |
The fitted and expected values were closest at 49.5 G, while the other two pairs differed by a few kilohertz. A second fitted line near twice each Larmor frequency was attributed to a second harmonic of the correlation sequence.
Prior Hahn Echo correlation sensing reported no detectable signal below 110 G. RESOLUTE therefore reached 49.5 G with (\tau=6) µs. These measurements characterize a (^{13}\mathrm{C}) bath around one NV, not individual carbon nuclei.
Frequency estimation and operating limits
Section titled “Frequency estimation and operating limits”Fisher information links a signal model to the smallest mean-squared frequency error allowed by that model. The target signal was modeled as a pure tone, (A\sin(\omega t+\phi)), with random phase (\phi). RESOLUTE’s accessible band lies between (1/T_1) and (1/T_2^p) because (T_{\mathrm{corr}}<T_1).
The protocols were compared for equal experiment duration. In the 500-sequence comparison, RESOLUTE and Hahn Echo used a 5 µs coherence time, while Ramsey used 0.5 µs. All protocols assumed (T_1=1000) µs and a 3 µs overhead between measurements.
Within the applicable band, RESOLUTE gave the strongest frequency-estimation information. Low-frequency calculations favor (\tau) near (T_2^p), although limited (T_{\mathrm{corr}}) may favor shorter (\tau) for resonance. The longest feasible (T_{\mathrm{corr}}) is best because the target completes more oscillations during correlation.
High frequencies require shorter (\tau) and accurate matching, reducing contrast and Fisher information. The model predicts linear information accumulation with experiment duration. The approximate formula matches exact calculations over most relevant frequencies. It underestimates information at very low frequencies and overestimates it at high frequencies when phase accumulation is large.
The calculations set signal amplitude and contrast to one, so real values reduce the predicted information. Ordinary Ramsey remains suited to DC detection, while Hahn Echo covers higher frequencies. RESOLUTE fills the low-frequency gap but does not replace high-frequency dynamical decoupling.
Electron spin sensing with chirped pulses
Section titled “Electron spin sensing with chirped pulses”To recover a DC dipolar interaction, the authors drove the target electron during (T_{\mathrm{corr}}) so its state flipped between sensing periods. The correlation time was fixed to the carbon bath period, allowing that noise to cancel while the electron pulse flipped the target.
A 2 µs chirped pulse with adiabaticity (Q=5) produced about 6% normalized contrast in RESOLUTE. DEER-(\tau) with the same pulse produced 1.4% contrast and a Fourier spectrum dominated by spurious noisy peaks. The dipolar oscillation occurred at (2\pi\omega_{dd}=0.6) MHz. RESOLUTE produced a strong Lorentzian feature around the fitted dipolar coupling.
With pulse duration fixed at 2 µs, (Q=5) gave (5.9 \pm 0.1%) contrast, while (Q=1) gave (3.2 \pm 0.1%). The authors linked the decline above (Q=5) to reduced effective pulse span and spin-flipping efficiency.
In a separate appendix comparison using a 1.6 µs chirp, four NVs at average depth 10 nm all had (T_2) above 10 µs. One NV gave 62% DEER contrast with the chirp and 31% with the rectangular (\pi) pulse.
Diamond sample and measurement setup
Section titled “Diamond sample and measurement setup”The three samples were cut from the same original stock and thinned to approximately 50 µm. Each had a polished surface roughness of (R_a=1) nm. The samples used the following implantation settings.
| Sample | Energy | Dose | Typical depth | Final NV concentration |
|---|---|---|---|---|
| JWR2 | 5 keV | (6 \times 10^9) cm(^{-2}) | (8 \pm 3) nm | 1 µm(^{-2}) |
| 205-4 | 9.8 keV | (8 \times 10^9) cm(^{-2}) | (14 \pm 5) nm | 1 µm(^{-2}) |
| AM-E2 | 9.8 keV | (8 \times 10^9) cm(^{-2}) | (14 \pm 5) nm | 1 µm(^{-2}) |
Ar/SF6 strain relief and an O2 soft ICP finish preceded nanopillar etching for photon collection. Measurements used a custom confocal fluorescence microscope, QM-OPX pulse control, and QuDi software.
Engineering consequences for NV devices
Section titled “Engineering consequences for NV devices”RESOLUTE makes (T_{\mathrm{corr}}) a tunable frequency-matching parameter for shallow NV sensors. For diamond material selection, the relevant operating limits are the measured NV (T_1) and (T_2) under the intended field and temperature. Long (T_1) permits longer (T_{\mathrm{corr}}), while long (T_2^p) permits longer (\tau).
Operation also requires (T_{\mathrm{corr}}>\tau), small (\omega\tau), and (\tau<T_2^p). Target decay and experimental constraints can prevent using the maximum (T_{\mathrm{corr}}). The authors project detection near 0.5 G after compensating Earth’s field and other low-frequency noise. That value remains a theoretical projection.
Lower temperatures may reduce phonon-driven relaxation and extend the limit. An external memory qubit with (T_M>T_1) was proposed for lower frequencies. The operating point should therefore be chosen from measured sensor times, target lifetime, and field noise.
Materials and support
Section titled “Materials and support”For shallow-NV magnetometry, 6CCVD’s diamond materials catalog includes MPCVD-grown single-crystal diamond plates and wafers. The paper’s sensor uses a roughly 50 µm diamond plate with shallow NV centers, an (R_a=1) nm polish, and nanopillar patterning for photon collection. The engineering support link can be used to discuss custom preparation for the paper’s thickness, polish, and patterned-surface requirements.
View Original Abstract
Magnetic spectroscopy at the nanoscale provides unique insights into material properties and dynamics, with quantum sensors like nitrogen-vacancy (NV) centers being ideally suited for these scales. However, detecting low-frequency signals remains a challenge due to finite coherence times ($T_2^$), as signals oscillating slower than $1/T_2^$ decay before sufficient phase accumulation occurs. We present RESOLUTE (Ramsey corrElation SpectroscOpy puLse seqUence wiTh phasE cycling), a protocol that overcomes these limitations by combining Ramsey measurements with correlation spectroscopy. By storing accumulated phase as a population imbalance during a correlation period ($T_\mathrm{corr} < T_1$) between two sensing periods, RESOLUTE generates an effective coherence time $T_2^p > T_2^$. This shifts the frequency-matching condition to the correlation time, enabling detection in the previously inaccessible spectral region between $1/T_1$ and $1/T_2^p$. We experimentally demonstrate an extension of the effective coherence time from $T_2^ = 0.38,μs$ to $T_2^p = 5.1,μs$, surpassing Hahn Echo measurements. The technique successfully detects $^{13}$C nuclear spin Larmor precession at fields as low as 49$,$G ($\sim$50$,$kHz). We further provide theoretical insight using Fisher information to characterize RESOLUTE’s frequency estimation capabilities compared to existing protocols. Finally, by integrating adiabatic pulses and phase cycling, we demonstrate robust spin control and effective DC signal extraction. These advancements provide enhanced sensitivity to weak dipolar interactions, essential for single-molecule imaging and quantum sensing applications.
High Concentration Quantum Grade Single Crystal Diamond
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →