Quantum Spin Sensors Overview

• Quantum spin sensors are devices that utilize two-level quantum spin states to detect external fields through variations in spin transitions, phases, or relaxation rates.
• They encompass diverse platforms like NV centers in diamond, hBN defects, and molecular spins, each offering unique readout architectures and sensitivity profiles.
• Advanced control protocols and quantum correlations, including dynamical decoupling and entanglement, push sensor precision beyond classical limits for applications such as nanoscale magnetometry and inertial sensing.

Quantum spin sensors exploit the quantum state of a spin—typically a two-level system—as the probe of an external field or environment. In present implementations, the sensed quantity is encoded in spin-transition frequencies, accumulated phases, relaxation rates, excitation probabilities, or mechanically transduced magnetization, with representative platforms including nitrogen-vacancy centers in diamond, carbon-related defects in hexagonal boron nitride, molecular electron spins in hybrid circuits and nanomechanical devices, thermally polarized Cr$^{3+}$ defects in sapphire, and electron–nuclear central-spin architectures for rotation sensing (López-García et al., 3 Mar 2026, Gilardoni et al., 2024, Bonizzoni et al., 2023, Wilcox et al., 2021, Ding et al., 2019).

1. Spin Hamiltonians and sensing observables

Across platforms, the sensor degree of freedom is an electronic or nuclear spin, or an effective two-level system derived from a higher-dimensional structure. For NV centers the dominant ground-state terms along the NV axis are $H = D S_z^2 + \mu B S_z$, with $D / 2\pi \approx 2.87\,\text{GHz}$; in the effective qubit subspace $|0\rangle, |-1\rangle$, one writes $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$. For the carbon-related $S=1$ defect in hBN, the ground-state Hamiltonian is

$$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$

which produces three transitions $f_A, f_B, f_C$. For molecular-spin magnetometry the Zeeman interaction is $$\mathcal{H}_Z = g \mu_B \mathbf{B} \cdot \mathbf{S}$$, while the thermally polarized Cr$^{3+}$ sensor in sapphire is described by

$H = D S_z^2 + \mu B S_z$0

The topological-waveguide sensor uses a single two-level quantum emitter with $H = D S_z^2 + \mu B S_z$1, coupled to an SSH bath (López-García et al., 3 Mar 2026, Gilardoni et al., 2024, Bonizzoni et al., 2023, Wilcox et al., 2021, Zhang et al., 2023).

The corresponding observables are platform-specific. In ODMR-based systems the basic signal is a fluorescence contrast,

$H = D S_z^2 + \mu B S_z$2

or, in Ramsey and echo protocols, a phase-dependent population signal. In microwave-only molecular-spin sensors the measured quantities are the amplitude and phase of the spin echo. In microwave-cavity architectures the spin ensemble modifies the cavity reflection coefficient $H = D S_z^2 + \mu B S_z$3 through absorptive and dispersive response. In topological-waveguide sensing the measurand is the quantum-emitter excitation probability

$H = D S_z^2 + \mu B S_z$4

In mechanically detected molecular-spin sensing, the experimentally accessible quantity is a correlation of longitudinal magnetization transduced as a force on a nanowire resonator (Tetienne et al., 2020, Bonizzoni et al., 2023, Wilcox et al., 2021, Zhang et al., 2023, Tabatabaei et al., 5 Mar 2026).

2. Platforms and readout architectures

NV centers remain the canonical solid-state implementation because they combine optical initialization and readout, room-temperature operation, and nanoscale stand-off distances. In high-purity diamond, NV electronic spins exhibit $H = D S_z^2 + \mu B S_z$5 up to milliseconds and $H = D S_z^2 + \mu B S_z$6 up to hundreds of microseconds, and the same defect can function either as a standalone sensor or as part of an imaging system providing spatial resolution down to the atomic scale. A distinct van der Waals implementation is the carbon-related single spin in hBN, where low symmetry and strongly spin-selective direct and reverse intersystem crossing dynamics provide sub-$H = D S_z^2 + \mu B S_z$7T/$H = D S_z^2 + \mu B S_z$8 magnetic-field sensitivity for both on and off-axis bias magnetic field exceeding $H = D S_z^2 + \mu B S_z$9 mT, while the host material enables nanometer-scale proximity to the target (López-García et al., 3 Mar 2026, Tetienne et al., 2020, Gilardoni et al., 2024).

Microwave-domain architectures broaden the hardware landscape. Molecular spin ensembles embedded in superconducting coplanar resonators implement AC magnetic-field sensing using only echo detection at microwave frequency and no optical readout; the reported dynamical-decoupling protocols synchronized with the AC field enhance the sensitivity up to $D / 2\pi \approx 2.87\,\text{GHz}$0 with a low $D / 2\pi \approx 2.87\,\text{GHz}$1 number of applied pulses. A different non-optical route uses thermally polarized Cr$D / 2\pi \approx 2.87\,\text{GHz}$2 defects in sapphire, where the host crystal also serves as the high quality-factor resonator, yielding a magnetometer with a broadband sensitivity of $D / 2\pi \approx 2.87\,\text{GHz}$3 pT/$D / 2\pi \approx 2.87\,\text{GHz}$4. At the nanoscale, SQUINT combines $D / 2\pi \approx 2.87\,\text{GHz}$5 trityl-OX063 radicals in an attoliter droplet with ultrasensitive mechanical readout and Hamiltonian engineering, extending coherence times to $D / 2\pi \approx 2.87\,\text{GHz}$6s (Bonizzoni et al., 2023, Wilcox et al., 2021, Tabatabaei et al., 5 Mar 2026).

Control infrastructure is also diversifying. “Photonic Links for Spin-Based Quantum Sensors” implements RF-over-fiber control of NV centers using an electro-optically modulated telecom-band laser and a high-speed photodiode, obtaining an RFoF efficiency of $D / 2\pi \approx 2.87\,\text{GHz}$7 at $D / 2\pi \approx 2.87\,\text{GHz}$8 GHz, corresponding to $D / 2\pi \approx 2.87\,\text{GHz}$9 dBm. This architecture directly targets low-noise, thermally isolated, and cryo-compatible ODMR systems (Rahman et al., 29 Jan 2026).

3. Control protocols and spectral selectivity

Ramsey and Hahn echo remain the fundamental control primitives. In the NV setting, Ramsey interferometry prepares a superposition, allows free evolution under $|0\rangle, |-1\rangle$0, and measures a signal of the form

$|0\rangle, |-1\rangle$1

while the Hahn echo sequence refocuses quasi-static and slowly varying noise and gives

$|0\rangle, |-1\rangle$2

These protocols are used both as direct sensing tools and as calibration procedures in simulated NV–impurity systems (López-García et al., 3 Mar 2026).

Dynamical decoupling generalizes this frequency selectivity. In molecular-spin magnetometry, Hahn echo, periodic dynamical decoupling, and CPMG/CP sequences synchronized with the AC magnetic field create a narrowband response around $|0\rangle, |-1\rangle$3, with sensitivity improving as the number of $|0\rangle, |-1\rangle$4-pulses increases until pulse imperfections and RF inhomogeneity dominate. In NV-assisted nanoscale NMR, coherent control of the sensor spin creates a dynamic frequency filter selecting only a few nuclear spins at a time; the filter function

$|0\rangle, |-1\rangle$5

enables a tenfold improvement in spatial resolution, down to atomic scales (Bonizzoni et al., 2023, Ajoy et al., 2014).

A qualitatively different route to wideband sensing uses the AC Zeeman effect. In wide-field NV imaging of spin-wave propagation in YIG, the off-resonant microwave field induces a phase shift

$|0\rangle, |-1\rangle$6

with $|0\rangle, |-1\rangle$7 determined by the microwave amplitude and detuning. This allows imaging of spin waves propagating over a wide frequency range up to a maximum detuning of $|0\rangle, |-1\rangle$8 MHz without changing the external magnetic field (Ogawa et al., 2024).

4. Noise, coherence, and stabilization

The principal limitation of quantum spin sensors is the competition between phase accumulation and decoherence. In NV centers, nearby nuclear spins, P1 centers, other NV centers, and technical noise all contribute. A two-spin quantum simulation of an NV sensor and a single impurity distinguishes a quasi-static nuclear-like regime from a coherent NV–NV regime. In the transmon emulation, the NV–“nuclear” configuration shows a natural $|0\rangle, |-1\rangle$9, reduced to $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$0 when the impurity is prepared in the excited state, whereas the NV–NV configuration has natural $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$1 and $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$2 with oscillations under coupled Hahn echoes. The former corresponds to dephasing without pronounced coherent oscillations; the latter adds coherent phase exchange and short-time entanglement (López-García et al., 3 Mar 2026).

Long-term instability is equally important. In an emulated NV spin gyroscope, the electronic spins of a large ensemble of NV centers monitor local magnetic-field fluctuations and stabilize interleaved Ramsey sequences on the $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$3 nuclear spin. The Allan deviation recovers an approximate $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$4 scaling up to $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$5, and feedback plus mapping reduces the contrast error at an optimal averaging time $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$6 from $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$7 to $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$8, equivalent to a minimum detectable rotation rate $H_S = \frac{\omega_S}{2}\sigma_z^{(S)}$9 (Jaskula et al., 2018).

Microwave-domain sensors expose a different noise hierarchy. In the Cr$S=1$0 sapphire sensor, the measured broadband sensitivity is $S=1$1 pT/$S=1$2, whereas the thermal-noise limit is $S=1$3 pT/$S=1$4; the dominant excess noise source is phase noise of the microwave source. In SQUINT, the bare OX063 ensemble has $S=1$5 ns and Hahn-echo $S=1$6s, so coherence enhancement by XYXYd is not optional but structural to the sensing architecture (Wilcox et al., 2021, Tabatabaei et al., 5 Mar 2026).

5. Quantum correlations, simulation, and enhanced precision

Quantum spin sensors need not be restricted to single-spin interferometry. In the simulated NV–spin system, two-qubit quantum state tomography reconstructs $S=1$7, and entanglement is assessed through the Peres–Horodecki criterion and CHSH inequalities. For the NV–“nuclear” case the minimum eigenvalue of the partial transpose remains non-negative after correcting small systematic offsets, while in the NV–NV case negative eigenvalues appear at early times, confirming entanglement even though all CHSH values remain below the classical bound $S=1$8. The result is a concrete example of a sensing-relevant regime in which entanglement is present but Bell nonlocality is not observed (López-García et al., 3 Mar 2026).

Several recent frameworks push precision beyond the standard quantum limit. In a topological-waveguide sensor, a single-spin qubit coupled to topological-paired bound states achieves sensitivity that can reach the Heisenberg limit across a large field range, with a product initial state and robustness against local perturbations inherited from the SSH bath. In higher-dimensional transverse spin-$S=1$9 Ising chains, dimension serves as a valuable resource, the bound

$$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$0

makes the dimensional dependence explicit, and the time duration appropriate for quantum-enhanced sensing increases with the increase of dimension. In a one-dimensional quantum wire used to estimate Rashba spin-orbit coupling, Heisenberg limited enhanced precision is achieved across a wide range of parameters and does not require fine tuning, for both single particle and interacting many-body probes (Zhang et al., 2023, Singh et al., 2024, Yi et al., 2024).

Collective nuclear-spin memories provide a further route to quantum enhancement. In the semiconductor-dot proposal for rotation sensing, a central electron spin is used to encode, store, and retrieve information in a large ensemble of surrounding nuclear spins; the projected sensitivity is $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$1 per sensor unit and $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$2 for a chip with $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$3 dots. The same architecture is proposed as a route to generating GHZ-like collective nuclear states for further enhancement (Ding et al., 2019).

6. Applications, fundamental tests, and technological directions

The most developed application class is magnetic sensing and imaging. Diamond spin sensors are already used to probe nanomagnetism from the mesoscale down to the nanoscale, including domain walls, skyrmions, superconducting vortices, magnetic nanoparticles, and nanoscale NMR. Wide-field NV imaging now extends this capability to coherent spin-wave transport in YIG over a wide frequency range, while hBN defects add vectorial nanoscale magnetometry at arbitrary $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$4 mT bias-field orientation and down to $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$5 nT/$$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$6 for optimal alignment (Tetienne et al., 2020, Ogawa et al., 2024, Gilardoni et al., 2024).

Spin sensors are also being adapted to inertial and navigation tasks. The dual-spin NV gyroscope uses the $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$7 nuclear spin ensemble as the primary inertial sensor and the NV electronic spin ensemble as a local magnetic-field monitor, with the explicit aim of stable rotation sensing for over several hours and a stated target of stability over several days. This architecture is presented as competitive with MEMS gyroscopes in sensitivity and significantly better in long-term stability (Jaskula et al., 2018).

A separate frontier is fundamental physics. Multi-pulse quantum sensing protocols with NV spin ensembles coupled to moving spin-polarized or unpolarized masses are projected to improve constraints on several hypothetical spin interactions by $$H = D S_z^2 + E (S_x^2 - S_y^2) + \gamma_e \mathbf{B} \cdot \mathbf{S},$$8 orders of magnitude at the micrometer scale. At the systems level, RF-over-fiber control provides a route toward low-noise, thermally isolated, and distributed ODMR networks, while SQUINT establishes a framework that affords molecular-level control over sensor properties and enables direct integration into complex molecular targets (Chu et al., 2021, Rahman et al., 29 Jan 2026, Tabatabaei et al., 5 Mar 2026).

This suggests that the field is converging toward a heterogeneous but increasingly interoperable toolbox: optically addressed defect spins for ambient nanoscale imaging, microwave-only and mechanically detected molecular spins for direct chemical integration, central-spin memories for long-lived inertial sensing, and engineered many-body or topological probes for quantum-enhanced precision.