Skip to content

Spin Relaxometry with Solid-State Defects - Theory, Platforms, and Applications

FieldValue
JournalOpen MIND
Published2026-02-02
AuthorsRuotian Gong, Alex L. Melendez, Guanghui He, Zhongyuan Liu, Chong Zu
DOI10.48550/arxiv.2602.01521

A single shallow or nanodiamond NV can resolve about 14 external electron spins in about 10 s. Near-surface NV ensembles can image about $10^3$ statistically polarized spins, about 32 net spins, per pixel at resolutions of 500 nm or less. The review finds that quantitative device use depends on calibrated baselines, stable surfaces, and geometry models.

Relaxometry measures field fluctuations rather than static fields. A defect converts transverse magnetic noise into population or coherence decay. In the weak-coupling limit, $\Gamma_1=1/T_1$ rises with the transverse noise PSD at the sensor transition $\omega_s$. The defect therefore acts as a local, frequency-selective magnetic-noise spectrometer.

The measured rate contains background and sample contributions. The review recommends isolating the sample contribution through baseline subtraction and control measurements. Distance, temperature, applied field, and device state provide independent checks.

Extracting a material property requires a source model for spins, currents, or vortices. A second model must propagate those fluctuations through the sample geometry to the defect. A relaxation map therefore does not directly equal a susceptibility, conductivity, or correlation time.

The all-optical $T_1$ sequence initializes the NV in $m_s=0$, waits for $\tau$, and measures photoluminescence. Optical pumping uses a spin-selective intersystem crossing, so the PL contrast follows $C(\tau)=\mathrm{PL}(\tau)/\mathrm{PL}(0)$.

The $T_1$ analysis often uses a stretched exponential when shallow defects, nanodiamonds, or ensembles sample several local environments. Here $\beta<1$ usually indicates a broad relaxation-rate distribution. Values of $\beta>1$ can arise from correlated or non-Markovian dynamics.

NV absorption and emission sidebands peak near 570 nm and 690 nm, respectively. Optical initialization takes about $1\ \mu$s because the metastable singlet lasts about 250 ns.

The excited-state lifetime is only about 10 to 13 ns. Readout therefore yields much less than one spin-projection photon per shot. A single-NV sequence commonly needs about $10^4$ to $10^7$ repetitions, taking seconds to minutes per point. Wide-field ensembles read many defects in parallel and improve per-pixel throughput.

Different control sequences select different parts of the environmental PSD.

ProtocolMain noise window
Ramsey $T_2^*$Near-DC and slowly varying longitudinal noise
Hahn echo and dynamical decouplingkHz to MHz bands set by pulse spacing
Spin lock $T_1\rho$Noise near the Rabi frequency $\Omega_R$
Longitudinal $T_1$Transverse noise near $\omega_s$, typically MHz to GHz depending on magnetic field

Ramsey and echo coherence mainly probe longitudinal noise, whereas $T_1$ responds to transverse noise. For a planar sample, noise power from in-plane wavevector $q$ falls roughly as $e^{-qh}$ or $e^{-2qh}$, depending on geometry.

Standoff, sensor orientation, sample thickness, and shape therefore determine which fluctuations reach the defect. The review proposes combining $T_1$, $T_1\rho$, and coherence probes across fields and depths to separate competing mechanisms.

The review surveys prior experiments rather than reporting a new device. A 2015 silver-film experiment varied a SiO$_2$ spacer and supported the scaling $\Gamma_1\propto T\sigma/d$. This established non-invasive readout of local electron transport without electrical contacts.

A 2019 graphene study mapped asymmetric current noise along the channel. The noise grew exponentially along carrier drift and inverted when current or carrier sign changed. In an Fe$_3$GeTe$_2$/hBN heterostructure, wide-field $V_B^-$ relaxometry produced a peak at $T_c=200\ \mathrm{K}$ in 2022.

For BSCCO, a 2025 study found $\Gamma_1\propto T^2$ at $T\ll T_c$. The rate diverged near $T_c$, and vortex-liquid noise was linear in applied field.

Biological studies included ferritin detection down to about $7.5\ \mu\mathrm{g}/\mathrm{mL}$. Methemoglobin was quantified in volumes below 100 pL. For FNDs attached to $Staphylococcus\ aureus$, $T_1$ recovered to about $170\ \mu$s at the bacterial surface. Recovery began 80 minutes after infection and tracked relative radical dynamics, not absolute concentration.

Wood’s microwave-free nano-NMR probed about a $(10\ \mathrm{nm})^3$ volume of external proton spins under ambient conditions. Its sensitivity was comparable to microwave-pulse nano-NMR, with the spectrum encoded in $T_1(B)$.

The NV-P1 benchmark occurred near 512 G, and the NV ground-state level anticrossing lies near 1024 G. Cross-relaxation therefore provides field-tunable spectroscopy from electron spins toward nuclear Larmor frequencies.

Diamond NV leads coherence, hBN minimizes standoff, and SiC supports wafer-scale processing.

PlatformDemonstrated characteristicsMain engineering limit
Diamond NVRoom-temperature $T_2$ up to about 1.8 ms in isotopically purified bulk; near-surface $T_2$ of tens of microsecondsSurface noise and integration with a three-dimensional host
hBN $V_B^-$Band gap about 6 eV; $T_1$ about 18 μs; $T_2$ about 50 ns to tens of microseconds with dynamical decouplingODMR contrast below 5% and short lifetimes
SiC $V_V^0$ and $V_{Si}^-$$T_1$ of milliseconds to seconds; $T_2$ about 1 ms and over 5 s with dynamical decoupling in purified materialDefect engineering and surface science still limit sensitivity

Near-surface NVs must usually lie within about 10 nm of the surface to sense external spins. At 5 nm depth, relaxation times fell by as much as about 30 times relative to bulk. An ordered oxygen-terminated surface extended shallow-NV coherence by about one order of magnitude.

Charge conversion between $NV^-$ and $NV^0$ can mimic fast $T_1$ under optical excitation. Nanodiamonds integrate into liquids, but surface chemistry broadens $T_1$ distributions and adds backgrounds.

Scanning probes control standoff but require slower acquisition, drift correction, and stable height. hBN places the defect close to an interface, but dim emission and ODMR contrast below 5% require ensembles or longer averaging.

The dominant systematic uncertainty usually comes from background subtraction or the source model, not the stretched-exponential fit. Reported $T_1$ accuracy is about 5% to 15%, limited by photon shot noise and averaging.

Quantitative inversion from $\Gamma_1(h,B,T)$ to $S_B(\omega)$ is often ill-posed. Small errors in defect depth or scan height can cause large errors because near-field kernels strongly weight spatial wavelengths.

Single-defect imaging remains slow because each point requires repeated averages and careful height control. Surface spins, charge conversion, laser heating, and technical drift can obscure the sample contribution. The review calls for multi-contrast datasets, constrained models, uncertainty estimates, reference samples, and shared analysis pipelines.

At fields above about 1 T, the NV ESR frequency exceeds about 30 GHz, making resonant microwave delivery difficult. All-optical $T_1$ remains available without RF control. Throughput may improve with time-gated single-photon avalanche diode cameras, repetitive readout, spin-to-charge conversion, and adaptive field sampling.

Choose the sensor geometry from the required spatial scale before optimizing a coherence metric. Shallow ensembles suit chip-scale biomolecular assays, scanning probes suit controlled standoff, and wide-field ensembles suit rapid spatial mapping.

Require depth calibration, surface termination records, charge-state monitoring, and on/off controls before converting a rate map into a material parameter. Use $T_1$, $T_1\rho$, and coherence measurements at several fields and heights when mechanisms overlap.

For high-field or RF-restricted work, all-optical relaxometry avoids resonant microwave delivery. Quantitative operation ends with an uncertainty budget that includes fit, baseline, distance, geometry, and source-model terms. Standardize those controls across devices before comparing rate maps between samples or platforms.

The paper’s principal diamond platform needs engineered NV defects, controlled surfaces, and defined sensor-sample geometry. 6CCVD’s diamond materials catalog lists MPCVD single-crystal and polycrystalline plates and wafers up to 125 mm, with thickness from 0.1 to 500 micrometers. NV defect creation and surface preparation are not specified in the catalog, so engineering support should define them before the host geometry is finalized.

View Original Abstract

Spin relaxometry using solid-state spin defects, such as the diamond nitrogen-vacancy (NV) center, probes dynamical processes by measuring how environmental fluctuations enhance the spin relaxation rate. In the weak-coupling limit, relaxation rates sample the transverse magnetic-noise power spectral density through a sensor-specific filter function, turning the defect into a local, frequency-selective noise spectrometer. This review bridges theory and experiment, clarifying how measured relaxation rates map onto noise spectra and how near-field geometry shapes the response. We highlight representative applications across condensed-matter physics, chemical and biological sensing, and relaxometry-based magnetic-resonance spectroscopy. We conclude with emerging opportunities and key challenges.