NV Centers and Solid-State Defects

Nitrogen-vacancy centers in diamond and related solid-state defects combine localized spin states, optical transitions, phonons, nuclear spins, strain, surfaces, and measurement backaction in one compact platform. They are used as qubits, quantum memories, nanoscale sensors, network nodes, and probes of magnetic, electric, thermal, and mechanical environments.

Optical pumping prepares a spin state, fluorescence reads it out, and the surrounding solid sets relaxation and dephasing. These processes also let the defect probe its environment. See NV-Center Sensing for complete sensing protocols and Defect and Solid-State Spin Qubits for registers and optical interfaces.

Ground-State Spin Model

The negatively charged NV center has an electronic ground-state spin triplet. A common effective Hamiltonian is

$$\frac{H}{\hbar} = D S_z^2 + \gamma_e \mathbf B\cdot\mathbf S + E(S_x^2-S_y^2) + \frac{H_{\mathrm{hf}}+H_{\mathrm{strain}}}{\hbar},$$

Take spin-1 matrices $S_i$ dimensionless, and $D,E,\gamma_e B$ in angular-frequency units; $H_{\mathrm{hf}}$ and the remaining strain terms have energy units. Here $D$ is the zero-field splitting, $\gamma_e$ is the electron gyromagnetic ratio, $E$ represents transverse strain or electric-field splitting, and $H_{\mathrm{hf}}$ contains hyperfine coupling to nearby nuclear spins.

For a field along the NV axis with $|\gamma_e B_z|<D$ and negligible transverse mixing, the positive $m_s=0$ to $m_s=\pm1$ transition frequencies are approximately

$$\omega_\pm = D \pm \gamma_e B_z .$$

This simple expression explains why the center is a magnetometer: a magnetic field shifts spin-resonance frequencies. The actual spectrum can include hyperfine splitting, strain, temperature shifts, and orientation-dependent projections.

Optical Pumping

The essential open-system feature of the NV center is its spin-dependent optical cycle. Green light excites the electronic state. Radiative decay returns photons, but a spin-selective nonradiative intersystem crossing routes population through metastable singlet states. The result is optical pumping into $m_s=0$.

A minimal effective model uses reset-like channels such as

$$L_{\mathrm{pump}} = \sqrt{\Gamma_{\mathrm p}} |0\rangle\langle 1|,$$

where $|1\rangle$ schematically denotes an $m_s=\pm1$ state and $|0\rangle$ denotes $m_s=0$. Real level structures have more states, but the open-system meaning is clear: optical illumination both extracts entropy and prepares a preferred spin state.

This is reservoir engineering in a solid-state defect. The optical bath plus metastable decay path produces an irreversible reset mechanism.

Spin-Dependent Fluorescence Readout

Readout uses the fact that $m_s=0$ tends to fluoresce more brightly than $m_s=\pm1$ during the early part of the optical pulse. A simple photon-counting model is

$$P(n\mid 0) = e^{-\mu_0} \frac{\mu_0^n}{n!}, \qquad P(n\mid 1) = e^{-\mu_1} \frac{\mu_1^n}{n!},$$

with

$$\mu_0\gt\mu_1 .$$

This static Poisson model is an approximation; optical pumping and metastable shelving can produce time-dependent and correlated counts. The same optical pulse that produces fluorescence also pumps the spin toward $m_s=0$. Thus the measurement has finite contrast and time-dependent backaction. Room-temperature single-shot readout of the electronic spin is usually limited by photon collection and optical reset during the measurement; low-temperature resonant readout and repetitive nuclear-assisted schemes can improve fidelity under appropriate conditions.

The open-system description must specify the record: photon number, time-resolved fluorescence, charge-state signal, or resonant optical response. A longer integrated-count window collects more photons, but optical pumping can make the late photons nearly independent of the initial spin. The finite-window calculation below shows both this loss of count contrast and the extra variance from a random pumping time. Quantitative NV predictions require the optical level structure and branching rates, as in the five-level model of Robledo et al. (2011).

Relaxation and Dephasing

A simple spin master equation contains relaxation and dephasing terms:

$$\dot\rho = - \frac{i}{\hbar}[H,\rho] + \Gamma_1\mathcal D[L_1]\rho + \Gamma_z\mathcal D[S_z]\rho + \cdots .$$

For the dimensionless spin-1 operator, this dephasing term gives $\dot\rho_{mn}=-(\Gamma_z/2)(m-n)^2\rho_{mn}$. The $+1,-1$ coherence therefore decays four times faster than either $0,\pm1$ coherence under this channel alone. The coefficient is not the same as a Pauli-$\sigma_z$ qubit dephasing rate. A full spin-1 relaxation model needs the relevant upward and downward transitions, rather than identifying one unspecified $L_1$ with a universal $T_1$.

The relevant channels depend on the operating regime:

• spin-lattice relaxation from phonons and magnetic noise at the transition frequency;
• dephasing from nuclear spins, paramagnetic impurities, surface spins, strain, and electric-field noise;
• optical pumping and ionization during laser illumination;
• microwave drive noise during control;
• charge-state switching between defect configurations.

The measured $T_1$, Ramsey $T_2^*$, echo $T_2$, and dynamically decoupled coherence time are different diagnostics. They should not be collapsed into one “the decoherence time.”

Magnetic Noise and Nuclear Spins

NV centers often couple to nearby $^{13}\mathrm C$ nuclear spins, nitrogen defects, surface spins, and external sample spins. The bath can be slow, structured, and non-Gaussian. A quasi-static longitudinal shift gives Ramsey dephasing, while time-dependent magnetic noise at the spin transition contributes to $T_1$ relaxation.

For sensing, the same coupling is the signal. A target magnetic field $B(t)$ produces a phase

$$\phi = \gamma_e \int_0^T y(t)B(t)\,dt,$$

where $y(t)$ is a control modulation function determined by Ramsey, echo, or a dynamical-decoupling sequence. The measurement is therefore a filtered probe of the magnetic noise or signal spectrum.

This is why Dynamical Decoupling appears both as a coherence-protection tool and as a spectroscopy tool.

T1 Relaxometry

Longitudinal relaxation can be used as a sensor. If environmental magnetic noise has spectral weight near the spin transition frequency, it changes the relaxation rate:

$$\frac{1}{T_1} \propto S_\perp(\omega_\pm),$$

with convention-dependent constants and orientation factors. This is called relaxometry. It is useful for detecting magnetic noise from spins, conductors, materials, and fluctuating fields near the defect.

Relaxometry is not the same as Ramsey magnetometry. Ramsey and echo sense phase accumulation from low-frequency or resonant signals selected by control pulses. $T_1$ relaxometry senses noise near the transition frequency through energy exchange.

Sensing Applications

NV centers and related defects are used to sense:

• DC and AC magnetic fields;
• nuclear magnetic resonance signals from nearby spins;
• current distributions and magnetic textures;
• temperature through shifts of $D$;
• strain and pressure;
• electric fields in suitable regimes;
• surface and material noise.

The sensitivity depends on spin contrast, photon collection, coherence time, initialization fidelity, control errors, defect depth, orientation, and sample backaction. Shallow centers improve spatial resolution but often suffer stronger surface noise. Deep centers may have longer coherence but weaker coupling to a nearby sample.

The relevant figure of merit is rarely just $T_2$. For a sensor, the useful quantity is uncertainty in the target parameter per unit bandwidth, with systematic calibration errors included.

Other Solid-State Defects

Silicon-vacancy centers, divacancies in silicon carbide, rare-earth ions, tin-vacancy centers, germanium-vacancy centers, and other defects share the same open-system themes:

• optical cycling and branching ratios determine readout;
• phonons affect linewidths and spin relaxation;
• strain and electric fields shift optical and spin transitions;
• nearby nuclear spins can act as memories or noise sources;
• charge-state stability is part of the quantum model.

The balance differs by material. Some defects have excellent optical properties but stronger spin–phonon coupling. Others have long spin coherence but less favorable photon collection. Their optical branching ratios and spin–phonon couplings must therefore be modeled for the particular defect and temperature.

Exercises

Transition Frequencies

For the simplified Hamiltonian

$$\frac{H}{\hbar} = D S_z^2 + \gamma_e B_z S_z,$$

with spin projections $m_s=0,\pm1$, find the transition frequencies from $m_s=0$ to $m_s=\pm1$.

Solution

The energy in angular-frequency units is

$$\frac{E_m}{\hbar} = Dm^2+\gamma_e B_z m.$$

For $m=0$,

$$\frac{E_0}{\hbar}=0.$$

For $m=+1$ and $m=-1$,

$$\frac{E_{+1}}{\hbar} = D+\gamma_e B_z, \qquad \frac{E_{-1}}{\hbar} = D-\gamma_e B_z.$$

Thus the transition frequencies are

$$\omega_+ = D+\gamma_e B_z, \qquad \omega_- = D-\gamma_e B_z.$$

Photon-Count Readout

Suppose the photon-count distributions for two spin states are Poisson with means $\mu_0\gt\mu_1$. If the decision rule is “assign state $0$ when $n\ge n_{\mathrm{th}}$,” write the two assignment-error probabilities.

Solution

State $0$ is misassigned as $1$ when the count is below threshold:

$$P(1\mid0) = \sum_{n=0}^{n_{\mathrm{th}}-1} e^{-\mu_0} \frac{\mu_0^n}{n!}.$$

State $1$ is misassigned as $0$ when the count is at or above threshold:

$$P(0\mid1) = \sum_{n=n_{\mathrm{th}}}^{\infty} e^{-\mu_1} \frac{\mu_1^n}{n!}.$$

Optical pumping during the readout pulse modifies this simple static-count model in real NV measurements.

Phase from a DC Magnetic Field

In a Ramsey experiment with $y(t)=1$, show that a constant field $B$ produces phase $\phi=\gamma_eBT$.

Solution

Use

$$\phi = \gamma_e \int_0^T y(t)B(t)\,dt.$$

For $y(t)=1$ and $B(t)=B$,

$$\phi = \gamma_e \int_0^T B\,dt = \gamma_e BT.$$

Optical Pumping and the Readout Window

Consider a coarse-grained two-state model: $|1\rangle$ pumps irreversibly into the brighter state $|0\rangle$ at rate $r>0$. Conditional on the state history, detected photons form a Poisson process with rates $R_0>R_1\ge0$. The unobserved pumping event is independent of these counted emissions and does not itself add a count. This model neglects optical-cycle antibunching, explicit shelving levels, reverse pumping, and detector memory; its time resolution must exceed the eliminated optical-cycle times.

For an integration time $T$, find the mean and variance of the total count $N$ for each initial state. Does increasing $T$ indefinitely improve the count-separation score

$$\mathcal S(T)=\frac{\mu_0-\mu_1}{\sqrt{v_0+v_1}}, \qquad \mu_j=\mathbb E[N\mid j],\quad v_j=\operatorname{Var}(N\mid j)?$$

This score compares two count distributions; it is not a readout fidelity or the optimum inference from photon arrival times.

Solution

For initial $|0\rangle$, $\mu_0=v_0=R_0T$. For initial $|1\rangle$, the survival probability is $e^{-rt}$, so the pumping time $\tau_p$ is exponentially distributed. Let $A=\min(\tau_p,T)$ be the time spent in $|1\rangle$ and $\Delta R=R_0-R_1$. Conditional on $A$, the count is Poisson with mean

$$\Lambda=R_1A+R_0(T-A)=R_0T-\Delta R\,A.$$

Using the survival probability to integrate the moments gives

$$\mathbb E[A]=\frac{1-e^{-rT}}r,\qquad \mathbb E[A^2]=\frac{2[1-e^{-rT}(1+rT)]}{r^2}.$$

The laws of total expectation and variance then yield

$$\mu_1=R_0T-\frac{\Delta R}{r}(1-e^{-rT}),$$ $$v_1=\mu_1+\frac{(\Delta R)^2}{r^2} \left(1-2rT e^{-rT}-e^{-2rT}\right).$$

The extra term is $\operatorname{Var}\Lambda$. Thus replacing the mixture by a Poisson law with its mean still misses pumping-induced fluctuations. For example, $R_0=2r$, $R_1=r/2$, and $T=1/r$ give $\mu_1\simeq1.052$ but $v_1\simeq1.342$.

At short times, $\mu_0-\mu_1\simeq\Delta R\,T$ and $v_0+v_1\simeq(R_0+R_1)T$, so $\mathcal S$ grows as $\sqrt T$. At long times the mean difference saturates at $\Delta R/r$, while $v_0+v_1\sim2R_0T$. Therefore $\mathcal S\to0$ as $T^{-1/2}$: there is a finite best window for this integrated-count score.

This does not mean a longer time-resolved record contains less available information. An estimator retaining arrival times can ignore late counts and reproduce any shorter-window decision. The limitation here comes from adding all counts into one number after the spin has largely reset. See Photon Counting for the distinction between time tags and integrated counts.

References

• G. Balasubramanian et al., “Nanoscale imaging magnetometry with diamond spins under ambient conditions,” Nature 455, 648-651 (2008).
• C. L. Degen, F. Reinhard, and P. Cappellaro, “Quantum sensing,” Reviews of Modern Physics 89, 035002 (2017).
• M. W. Doherty, N. B. Manson, P. Delaney, F. Jelezko, J. Wrachtrup, and L. C. L. Hollenberg, “The nitrogen-vacancy colour centre in diamond,” Physics Reports 528, 1–45 (2013). arXiv:1302.3288.
• A. Gruber, A. Dräbenstedt, C. Tietz, L. Fleury, J. Wrachtrup, and C. von Borczyskowski, “Scanning confocal optical microscopy and magnetic resonance on single defect centers,” Science 276, 2012-2014 (1997).
• J. R. Maze et al., “Nanoscale magnetic sensing with an individual electronic spin in diamond,” Nature 455, 644-647 (2008).
• L. Robledo, H. Bernien, T. van der Sar, and R. Hanson, “Spin dynamics in the optical cycle of single nitrogen-vacancy centres in diamond,” New Journal of Physics 13, 025013 (2011). arXiv:1010.1192.
• L. Robledo, L. Childress, H. Bernien, B. Hensen, P. F. A. Alkemade, and R. Hanson, “High-fidelity projective read-out of a solid-state spin quantum register,” Nature 477, 574-578 (2011).
• L. Rondin, J.-P. Tetienne, T. Hingant, J.-F. Roch, P. Maletinsky, and V. Jacques, “Magnetometry with nitrogen-vacancy defects in diamond,” Reports on Progress in Physics 77, 056503 (2014).