Skip to content

Mesoscopic Spin Coherence in a Disordered Dark Electron Spin Ensemble

FieldValue
JournalOpen MIND
Published2026-02-19
AuthorsTaewoong Yoon, Sangwon Oh, Junghyun Lee, Hyunyong Choi
DOI10.48550/arxiv.2602.17074

At room temperature under a 446 G bias, an iterative Hartmann-Hahn protocol yielded an estimated 7.4% P1 polarization. This represented about a 740-fold increase over a thermal value of about 0.010%. Collective Rabi oscillations confirmed that the dark spin ensemble carried coherent mesoscopic polarization.

Polarization transfer with a hybrid NV-P1 ensemble

Section titled “Polarization transfer with a hybrid NV-P1 ensemble”

The experiment used a room-temperature CVD diamond sample containing a 40 µm epitaxial layer with 99.9% $^{12}$C enrichment. The sample contained 6.3 ± 0.2 ppm P1 centers and 2.4 ± 0.3 ppm NV centers, representing about 10^5 spins in the probed volume. Each substitutional-nitrogen P1 center hosted an optically dark electron spin with $S=1/2$, strong $^{14}$N hyperfine coupling, and Jahn-Teller distortion.

Each transfer cycle combined 5 µs of double spin locking with 5 µs of 532 nm illumination. The transfer laser power was 50 µW, while initial polarization and fluorescence readout used 10 µW. Under the Hartmann-Hahn condition, the NV centers supplied optical polarization and served as sensors for the P1 state.

Double electron-electron resonance, or DEER, resolved five P1 spectral subgroups with population fractions of {1, 3, 4, 3, 1}/12. The transfer measurements addressed the subgroup containing 3/12 of the P1 population and used 32 cycles for the differential signals. The Hartmann-Hahn condition appeared as a Lorentzian dip in NV spin-lock contrast when the two Rabi frequencies matched.

Readout used the difference between parallel and antiparallel P1 configurations after phase cycling and normalization to bare NV decay. Parallel P1 polarization suppressed energy-conserving flip-flops, while antiparallel polarization accelerated decay. The resulting $\Delta C$ directly tracked net P1 polarization.

At 446 G along the NV symmetry axis, the NV ground-state doublet operated as an effective spin-1/2 system. The NV-P1 frequency mismatch was about 370 MHz, so their lab-frame dipolar interaction reduced to a secular Ising coupling.

Continuous driving rotated the effective quantization axis and converted longitudinal coupling into dressed-state flip-flop exchange. The original longitudinal operators became transverse in the dressed basis. Matching Rabi frequencies, $\Omega_{\mathrm{NV}}=\Omega_{\mathrm{P1}}=\Omega$, made the exchange resonant. The dressed coupling retained the form of lab-frame intragroup flip-flop exchange.

In the strong-driving limit, the effective disorder fell to $W_{\mathrm{eff}}\approx W^2/(2\Omega)$. Strong driving therefore improved transfer across the disordered network by reducing local energy mismatches.

Absolute polarization and exchange dynamics

Section titled “Absolute polarization and exchange dynamics”

Angular momentum conservation in the rotating frame gave the relation $P^{\mathrm{P1}}=P_0^{\mathrm{NV}}(A/2)(1+n_{\mathrm{NV}}/n_{\mathrm{P1}})$. The estimate used $n_{\mathrm{P1}}/n_{\mathrm{NV}}\approx2.6$, an initial NV polarization of 75%, and a saturation amplitude $A_\infty\approx0.143$ in the strong-driving limit. The estimated polarization corresponded to an effective spin temperature of 405 mK at 446 G.

The differential contrast rose and leveled off under an exponential fit, while the full contrast traces used stretched-exponential fits. The exchange occurred over two orders of magnitude faster than intrinsic NV spin-lock relaxation.

QuantityResultConditionInternal comparison
Readout quasiequilibrium time2.2 ± 0.6 µsDifferential double spin-lock readoutNV spin-lock relaxation was about 1.3 ms
P1 spin-lock $T_1^\rho$0.43 ± 0.16 msSpin locking after polarization transferIndependent P1 $T_1$ was about 0.67 ms
P1 Hahn-echo $T_2$4.06 ± 0.78 µsHahn echo on the polarized ensembleNV $T_2$ was 10.7 ± 1.0 µs

A control pulse inserted between polarization and readout produced collective Rabi oscillations after 16 transfer cycles. The fast Fourier transform contained a driven P1 frequency component in the polarized state, but none in the thermal state. This distinguishes direct P1 population dynamics from conventional DEER oscillations caused by an unpolarized bath.

The measured P1 spin-lock lifetime was close to the independently measured longitudinal relaxation. This agreement led the authors to infer that intrinsic relaxation, rather than diffusion into the unpolarized bulk, set the rotating-frame lifetime.

Hahn-echo dephasing was shorter for P1 than for NV, and their ratio matched the defect-density ratio. The authors associated both echo decays with intragroup dipolar flip-flops, while echo refocusing suppressed intergroup Ising couplings.

Polarization buildup followed $A(N)=A_{\mathrm{sat}}(1-e^{-N/N_{\mathrm{sat}}})$, with $N_{\mathrm{sat}}\approx3$ cycles. At 10 µs per cycle, this corresponded to a saturation timescale of about 30 µs. Saturation occurred when polarization injection balanced dissipation.

A semiclassical network model required an effective laser-on P1 relaxation time of about 32 µs. The authors attributed this fast depolarization to 532 nm photo-ionization of P1 centers and charge-state dynamics near NV centers. The model reproduced qualitative saturation but overestimated absolute polarization by a factor of 2 to 3. This result indicates additional dissipation beyond the pairwise description.

Rabi-dependent saturation followed $A_{\mathrm{sat}}(\Omega)=A_\infty\Omega^2/(\Omega^2+W^2)$, with a fitted disorder scale of $W=1.36\pm0.19$ MHz. For $\Omega<W$, local detuning suppressed flip-flops, while $\Omega>W$ allowed coherent driving to overcome the detuning.

The P1 subgroup used in the network model had an estimated concentration of about 1.6 ppm and a mean spacing of 15.3 nm. Simulations at $\Omega=6.40$ MHz gave $D\approx0.22$ nm²/µs and a diffusion length $L_D\approx6.2$ nm over the saturation time. The diffusion estimate used 1,000 disorder realizations and a finite-size extrapolation.

The mean nearest-neighbor distance of the addressed NV subgroup was about 11.7 nm, larger than the diffusion length. Poisson statistics placed a neighbor within $2L_D$ for about 57% of those NV centers. The authors link overlapping diffusion fronts to reduced polarization gradients and longer steady-state polarization lifetimes.

Device engineering requirements and open questions

Section titled “Device engineering requirements and open questions”

For device engineering, the sample specification cannot stop at defect density. Isotopic purification suppresses $^{13}$C nuclear-spin bath contributions, while crystal alignment sets the NV symmetry axis relative to the bias field. Controlled P1 and NV populations matter because transfer efficiency and total resource volume depend on their ratio. The authors identify the NV-to-P1 ratio as a synthesis target requiring further optimization. Charge stability during optical initialization requires attention because laser-on relaxation dominated cycle saturation.

Quantum correlations, leakage into unaddressed P1 subgroups, and position-dependent photo-ionization could explain why the model overestimated polarization. The paper lists these as possible sources of error, not confirmed causes. The polarization estimate applies to the addressed P1 spectral subgroup, not the unaddressed groups.

Earlier approaches often relied on isolated NV centers, whose polarization diffused into the bulk thermal reservoir. The experiment addresses the stated gap that mesoscopic polarization in a disordered P1 ensemble had not been realized. The dense NV network supplied this mesoscopic polarization without isolating individual P1 centers.

Before using the method for sensing, the authors propose Hamiltonian engineering and algorithmic cooling to reduce the effective spin temperature further. The proposed uses are environment-assisted sensing, nuclear-spin polarization, and quantum many-body simulation. Those applications were proposed rather than demonstrated in the reported experiment.

For this NV-P1 quantum-sensing application, the material requirement is a 40 µm, 99.9% $^{12}$C-enriched CVD diamond epitaxial layer with dense P1 and NV centers. The diamond materials catalog includes MPCVD-grown single-crystal and polycrystalline diamond plates and wafers with thicknesses from 0.1 to 500 µm. The paper’s isotope and defect-density requirements should be discussed with 6CCVD through the linked engineering support.

View Original Abstract

Harnessing dipolar spin environments as controllable quantum resources is a central challenge in solid-state quantum technologies. Here, we report the observation of a coherent mesoscopic spin state in a disordered ensemble of substitutional nitrogen (P1) centers in diamond. An iterative Hartmann-Hahn protocol transfers polarization from dense nitrogen-vacancy (NV) centers to a P1 ensemble, yielding a 740-fold enhancement over room-temperature thermal equilibrium as revealed by differential readout. The resulting mesoscopic P1 spin ensemble exhibits collective Rabi oscillations and long-lived spin-lock and Hahn-echo coherences. We identify a crossover in the saturation polarization arising from the competition between coherent driving and local disorder, providing a quantitative measure of the system’s intrinsic disorder. These results establish a foundation for utilizing dark electron spin ensembles as robust resources for quantum sensing and quantum many-body simulation.