DEVELOPING A STRONGLY INTERACTING NV-DIAMOND QUANTUM MEASUREMENT EXPERIMENTFOR USE IN UNDERGRADUATE TEACHING
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Stanford Digital Repository |
| Published | 2026-06-03 |
| Authors | Judy Liu, L. Hollberg, Joonhee Choi, Stanford University |
| DOI | 10.25740/rh553th8057 |
The apparatus achieved a magnetic-field sensitivity of 26 nT/√Hz. It resolved eight Zeeman-split nitrogen-vacancy resonances from a 40 ppm ensemble in bulk diamond. A confocal microscope, home-built copper loop-gap resonator, and lock-in detector produced the result at room temperature. The setup measured T1 = 25.00 ± 2.75 µs, while the dense spin bath limited T2* to roughly 28 ns. The thesis also proposes a six-session undergraduate laboratory, but direct Ramsey and Hahn echo measurements remain unfinished.
Sample and apparatus
Section titled “Sample and apparatus”The characterized sample was the darker of two supplied 3D diamond plates with NV ensembles. It was 250 µm thick, contained 40 ppm NV centers, and originated from a 240 ppm substitutional nitrogen concentration. The paper calls these substitutional nitrogen centers P1 centers. The surface normal was along [100]. The second plate was 38 µm thick, with an expected 24 ppm NV density from 130 ppm initial P1 density. It was not characterized in the reported experiments.
Excitation used a 567 nm laser, an acousto-optic modulator, and an NA = 0.22 objective. The modulator rose in 2.6 ns. Red fluorescence passed through a long-pass filter and was collected by a silicon photodiode. A low-noise preamplifier and lock-in amplifier recovered the spin-dependent modulation. The QICK generated microwave tones, a 200 MHz modulation signal for the optical gate, and the lock-in reference. Three coils produced a steerable bias field of up to 30 G.
Microwave power passed through a ZVE-3W-83+ amplifier to about 35 dBm before a coupling loop drove the resonator. The loop-gap resonator was machined from copper tubing and tuned with a 1 pF trimmer. Its S11 measurement showed a dip near 2.84 GHz. Coupling-loop separation provided fine tuning toward the 2.87 GHz NV zero-field splitting.
Resonances and relaxation
Section titled “Resonances and relaxation”Continuous-wave optically detected magnetic resonance applied laser and microwave together. The microwave was chopped at half the lock-in reference rate, and each frequency point used 200 repeats. A bias field split each of four NV orientations into two transitions. This produced eight Lorentzian peaks around 2.87 GHz, with carbon-13 hyperfine satellites roughly 100 MHz from the main peaks.
The fourth peak at about 2.84 GHz had an 8.91 MHz FWHM, corresponding to T2* ≈ 36 ns. Later sections use T2* ≈ 28 ns for the ensemble, including the Rabi analysis. The thesis does not reconcile those estimates. The spin-dependent fluorescence contrast was about 0.5% of total photoluminescence.
Linewidth extrapolation gave a 6.1 MHz floor. The linewidth increased to 9.0 MHz across a 27 dB microwave-power range. The authors attribute the residual floor mainly to optical broadening at the lowest drive, rather than microwave power broadening. A fixed-microwave-power laser sweep would separate optical and inhomogeneous contributions. Pulsed ODMR reproduced the resonances, but its traces were noisier and its average linewidths were broader.
T1 used inversion recovery with 20 µs laser pulses, a 1 ms lock-in reference, and 30 repeats per delay. The total acquisition took 1 minute. A single-exponential fit gave T1 = 25.00 ± 2.75 µs at room temperature without optical pumping during the free delay. This value is well below the millisecond scale reported for isolated NV centers in high-purity diamond. The authors attribute the reduction to the 240 ppm P1 bath.
Rabi response and microwave limit
Section titled “Rabi response and microwave limit”Driving the microwave pulse longer did not produce a sinusoid. The lock-in signal instead rose and saturated within about 300 ns. A fitted saturation time was 324 ns. Combining this with T2* ≈ 28 ns gave ΩR ≈ 1.5 MHz and a π-pulse estimate of 334 ns.
Resolving Rabi oscillations required ΩR to exceed 36 MHz by a large margin and a π pulse well under 14 ns. The ensemble therefore dephased before completing a coherent rotation. The authors describe the observed response as overdamped population transfer. They list resonance-frequency spread, spatial variation in B1, and weak microwave drive as possible causes. A systematic microwave-power study was left for future work.
Magnetic sensitivity
Section titled “Magnetic sensitivity”The apparatus parked at 2847.5 MHz on the steepest slope of the largest CW ODMR peak. The fitted slope was 1.108 V/MHz. Lock-in input noise was 4.0 mV rms, and a 10 ms filter time constant gave a 25 Hz noise-equivalent bandwidth. These conditions produced 26 nT/√Hz using 28 GHz/T for the NV gyromagnetic ratio.
Electronic noise set the floor because it did not change when the laser was on or off. The authors therefore treat 26 nT/√Hz as a conservative upper bound and expect shot-noise-limited performance to be better. Increasing photocurrent or collection efficiency would raise photon shot noise above the electronic floor. Pulsed ODMR and Ramsey protocols would separate initialization, manipulation, and readout, but lower-impurity diamond remains necessary for a narrower intrinsic resonance.
Two prior undergraduate setups reported sensitivities of 1 µT/√Hz and 28.32 nT/√Hz. Optimized research-grade systems using similar lock-in detection reported 2 to 5 nT/√Hz. The teaching apparatus therefore reached sensitivity close to research-grade measurements while retaining a simple photodiode readout.
Undergraduate laboratory module
Section titled “Undergraduate laboratory module”The proposed module spans six classes within a two to three week Physics 106 laboratory. The apparatus arrives assembled, but the resonator is detuned and its coupling loop is displaced. Students must restore the resonance with a vector network analyzer and modify provided Python templates for CW and pulsed ODMR.
- Class 1 aligns excitation and collection optics, records photoluminescence, and measures the detector noise floor.
- Class 2 finds the resonator resonance, adjusts coupling and capacitance, and extracts Q from S11.
- Class 3 acquires CW ODMR, fits the eight peaks, and applies four-orientation vector magnetometry.
- Class 4 measures pulse-dependent fluorescence and fits T1 from inversion recovery.
- Class 5 acquires pulsed ODMR and attempts Rabi pulse calibration.
- Class 6 compares Ramsey T2* with Hahn echo T2.
The last two classes depend on pulse calibration that the present apparatus could not provide. Clean Rabi oscillations, direct Ramsey T2*, and Hahn echo T2 therefore remain curriculum goals rather than completed thesis measurements.
Material selection and open issues
Section titled “Material selection and open issues”The measurements show that sample selection sets the operating regime. High NV density gave enough fluorescence for modest green-laser power and a simple silicon photodiode. It also increased spin-spin broadening and shortened coherence. The authors treat 40 ppm as strongly interacting because their interaction estimate places dipolar coupling in the tens-to-hundreds of kilohertz range.
For magnetometry, engineers must balance ensemble fluorescence against impurity concentration, strain, and coherent-control requirements. The authors suggest lower-impurity diamond for narrower resonances. They also recommend stronger B1 through a redesigned resonator, a printed circuit board loop-gap resonator, or a broadband microstrip or coplanar waveguide. Their preferred resonant option concentrates limited microwave power at the sample.
Fast π and π/2 pulses would unlock Ramsey interferometry, Hahn echo, dynamical decoupling, and interaction-aware control. A direct time-domain measurement of T2 and T2* remains open. The practical selection rule is to use dense ensembles when photoluminescence dominates the design, and lower-impurity material when narrow resonances or coherent rotations are required.
Materials and support
Section titled “Materials and support”The sensing element is an NV-active diamond plate for ensemble magnetometry, with the studied plate at 250 µm and a [100] surface orientation. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal diamond as plates and wafers up to 125 mm, with thickness from 0.1 to 500 µm. The catalog facts do not establish the study’s 40 ppm NV and 240 ppm P1 concentrations, so defect density, impurity content, and orientation should be confirmed through engineering support.
View Original Abstract
Nitrogen-Vacancy (NV) centers in diamond are a compelling solid-state platform for quantum sensing due to their long coherence times, room-temperature operation, and optical spin readout. Several groups have recently demonstrated that NV centers, capable of running at room temperature, are practical for undergraduate laboratories. However, Stanford currently has no dedicated NV-center teaching lab for physics undergraduates. Here, I (1) design of the experimental apparatus and initially can build from existing resources at Stanford; (2) experimental validation of basic features of optical pumping, optical-microwave double resonance, magnetic sensitivity; (3) time variable quantum state preparation and detection using various optical and microwave pulse sequences to measure coherence times; (4) experimental design and validation of pulse sequences; (5) practical curriculum and syllabus for a 2-3 week laboratory in Physics 106. On the second front, I characterize a strongly interacting NV ensemble in diamond.
High Concentration Quantum Grade Single Crystal Diamond
View Specs →
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →