WO2018052497A2

WO2018052497A2 - Synchronized-readout for narrowband detection of time-varying electromagnetic fields using solid state spins

Info

Publication number: WO2018052497A2
Application number: PCT/US2017/034256
Authority: WO
Inventors: Ronald L. Walsworth; David Glenn; Dominik BUCHER
Original assignee: President And Fellows Of Harvard College
Priority date: 2016-05-25
Filing date: 2017-05-24
Publication date: 2018-03-22
Also published as: US20200278414A1; WO2018052497A3; US10901062B2

Abstract

A synchronized readout (SR) technique for spectrally selective detection of oscillating magnetic fields with sub-millihertz resolution, using coherent manipulation of solid state spins.

Description

SYNCHRONIZED-READOUT FOR NARROWBAND DETECTION OF TIME- VARYING ELECTROMAGNETIC FIELDS USING SOLID STATE SPINS

CROSS REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of U.S. Provisional Application 62/341,497 filed May 25, 2016, the contents of which are hereby incorporated by reference.

STATEMENT OF GOVERNMENT RIGHTS

[0002] This invention was made with government support under Grant No. HR-0011-11- C-0073 awarded by the Defense Advanced Research Projects Agency (DARPA). The government has certain rights in this invention.

BACKGROUND

[0003] Nuclear magnetic resonance (NMR) spectroscopy is an important analytical tool in modern chemistry, structural biology, and materials research. Conventional NMR relies on inductive detection and requires sample volumes of 0.1 - 1 mL, although alternative detection technologies including microcoils, superconducting quantum interference devices (SQUIDs), atomic magnetometers, and detectors based on solid state spins have been demonstrated to improve sensitivity or allow reduced sample volumes. Current spectral resolution for NMR detection using solid state spins (e.g., nitrogen vacancy centers in diamond) is considerably broader than the resolution needed to observe J-couplings or chemical shifts at typical static magnetic fields used in such NMR experiments, making existing detection techniques unsuitable for many practical molecular NMR applications.

[0004] Two key challenges limited the spectral resolution of nitrogen vacancy (NV)- detected NMR techniques demonstrated to date. First, the interrogation duration was set by the spin state lifetime of the NV (7i ~ 3 ms), or of a proximal solid-state nuclear spin used to create a quantum memory (Γι ~ 5 - 50 ms), both of which are orders of magnitude shorter than typical coherence times of nuclear spins in liquid samples (Ti ~ 1 s). Second, for NV- NMR with nanoscale sample volumes, the thermal spin magnetization at room temperature and Bo ~ 0.1 T is small (~3 x 10^"7 N, for N the number of spins in the volume) compared to statistical fluctuations in the net magnetization (which scale as ]St^/2). All previous NV-NMR experiments therefore measured the spin-noise-induced variance in the local magnetic field, which is large enough detect easily (OB ~ 0.6 μΤ RMS for NV depth dm = 5 nm below the diamond surface and a sample of pure water), but has short correlation time limited by diffusion of sample spins through the sensing volume (e.g., Xc^{" 1} « 10 MHz for water or Xc^{" 1} « 30 kHz for viscous oil, for a sensing volume V∞ [5 nm]³). Therefore, there is a need for systems and methods capable of narrow resolution for detecting time-varying electromagnetic fields.

SUMMARY OF THE INVENTION

[0005] In accordance with one aspect, the present application discloses systems and methods for higher resolution NMR detection using solid state spins, wherein a synchronized readout pulse sequence is utilized to allow coherent interrogation of sample nuclei over many measurements, and implementing this sequence in an instrument with sufficient sensitivity to measure the thermal spin polarization instead of statistical fluctuations.

[0006] In one aspect, a system includes a sensor, wherein the sensor includes a solid state electronic spin system disposed below the surface of the sensor, wherein the solid state electronic spin system has a spin-state dependent fluorescence; a source of light; a source of first external perturbation, wherein the source of first external perturbation generates a magnetic field; a source of second external perturbation; wherein, the source of light and the first and second external perturbations are configured to coherently and independently manipulate the spin states of at least one solid state electronic spin system; a digital timing generator configured to synchronize the light and the first and second perturbations; and a detector to optically measure the solid-state electronic spins spin-state dependent

fluorescence.

[0007] In some embodiments, the sensor includes a high purity diamond crystal lattice. In some cases, the solid state electronic spin system is a nitrogen-vacancy (NV) spin in the high purity diamond lattice.

[0008] In some embodiments, the system further includes a source of third external perturbation such as a radio frequency (RF) electromagnetic field source or an electronic spin resonance (ESR) field source.

[0009] In some embodiments, the source of second external perturbation is a microwave field source.

[0010] In some embodiments, the detector is a CCD camera, or a photomultiplier, or a photodiode. [0011] In some embodiments, the sample to be measured is placed on a surface of the sensor. In some aspects, the sample to be measured is placed in the proximity of the sensor.

[0012] In some embodiments, the digital timing generator is configured to provide a subsequence of pulses of the light and the second perturbation designed to be selective for detecting oscillating magnetic signals near a central frequency,/ In some cases, the subsequence of pulses has a duration that is an integer multiple of l/f.

[0013] In some embodiments, the solid state spin systems spin-dependent fluorescence is changed due to the interaction with the sample.

[0014] In some embodiments, the source of light radiation is an optically pumped laser. In some cases, the optically pumped laser has a wavelength of 532 nm.

[0015] In accordance with another aspect, a method of measuring an oscillating magnetic field includes disposing a sensor in a first external perturbation field, wherein the sensor includes a solid state electronic spin system disposed below the surface of a sensor and the solid state electronic spin system is aligned with the first external perturbation field, wherein the solid state electronic spin system has a spin-state dependent fluorescence; positioning the sensor in proximity to a sample containing a source of oscillating magnetic fields; obtaining a sequence of fluorescence intensity measurements by i) pulsing a light source to generate light to align spins of the solid state electronic spin system; ii) pulsing a source of second external perturbation; and iii) detecting fluorescence of the solid state electronic spin system; wherein steps i) and ii) in each iteration are synchronized using a digital timing generator; and repeating steps i), ii) and iii) for a predetermined number of iterations.

[0016] In some embodiments, obtaining a sequence of fluorescent intensity

measurements further comprises iv) pulsing a source of third external perturbation. In some aspects, the source of third external perturbation includes a radio frequency (RF)

electromagnetic field source or an electronic spin resonance (ESR) field source.

[0017] In some embodiments, the method further includes applying a Fourier transform to the sequence of fluorescent intensity measurements that were recorded at equal time intervals during each iteration.

[0018] In some embodiments, the sensor includes a high purity diamond crystal lattice. In some cases, the solid state electronic spin system is a nitrogen-vacancy (NV) spin in the high purity diamond lattice.

[0019] In some embodiments, the source of second external perturbation is a microwave field source. [0020] In some embodiments, the digital timing generator is configured to provide a subsequence of pulses of the light and the second perturbation designed to be selective for detecting oscillating magnetic signals near a central frequency,/ In some aspects, the subsequence of pulses has a duration that is an integer multiple of 1/

[0021] In some embodiments, the sample comprises a liquid sample. In some aspects, the liquid sample volume is between about 10 to 100 pL.

BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The following figures are provided for the purpose of illustration only and are not intended to be limiting.

[0023] FIG. 1 is a schematic block diagram of a solid state magnetometer including a single NV center, in accordance with one embodiment of the present disclosure;

[0024] FIG. 2 is a flow chart of steps for analysis of an oscillating magnetic field in accordance with one embodiment;

[0025] FIG. 3A is a numerical stimulation of synchronized readout (SR) detection of a free induction decay (FID) signal, which oscillates at frequency / and has finite decay lifetime τ;

[0026] FIGS. 3B-3C shows detail of calculated magnetic signal and magnetometry subsequence at the third SR iteration (denoted SR-3) of the simulation in FIGS. 3A;

[0027] FIGS. 4A-4C show the measured SR spectral resolution using signals from a coil antenna. FIG. 4A is a plot of the power spectrum of SR signal obtained with a single-NV magnetic sensor in a confocal microscope;

[0028] FIG. 4B is a plot of the power spectrum of SR signal obtained with an NV ensemble magnetic sensor for a single average for a duration of T = 112.5 s;

[0029] FIG. 4C is a plot of the power spectrum of SR signal obtained with a NV ensemble magnetic sensor for a single average for a duration of T = 3000 s;

[0030] FIG. 5A provides an NV ensemble sensor for SR-detection of proton NMR;

[0031] FIG. 5B depicts the detection of NMR from protons in glycerol; [0032] FIG. 5C is a power spectra of proton resonance frequencies obtained from glycerol FID data for various values of Bo;

[0033] FIG. 5D is a plot of resolved FID signals obtained from protons in glycerol (upper plot) and pure water (lower plot);

[0034] FIG. 6A shows an SR-FID spectrum of xylene (circles), acquired using the SR protocol with XY8-6 magnetometry subsequences, and TSR = 24 and n = 4 x 10⁴. Fitting to a sum of two Lorentzian lineshapes (solid line) indicates a splitting Af= 13 ± 1 Hz due to J- coupling between the central ³¹P nucleus and the methyl protons;

[0035] FIG. 6B shows an SR-FID spectrum of trimethyl phosphate (circles), acquired using SR with XY8-6 subsequences, and TSR = 24 and n = 4 x 10⁴. The relative peak heights obtained from fits to a sum of Gaussian lineshapes (solid lines) are due to the relative abundances of CH and CFb protons in the molecule, and the splitting Af= 20 ± 2 Hz is the result of chemical shifts associated with the two proton positions.

DETAILED DESCRIPTION

[0036] A method and device for magnetic resonance imaging with high resolution and sensitivity under ambient conditions is described.

[0037] FIG. 1 is a schematic showing a solid state electronic spin system having spin- dependent fluorescence 101 in a sensor 100, located below the surface 102 of the sensor 100. The schematic also shows a detector 104 used to measure the changes in the solid-state electronic spin in response to light 105, a first external perturbation 106 and a second external perturbation 107 originating from the source of light 115, a first external perturbation source 116, and a second external perturbation 117, respectively.

[0038] The light 105 generated from the source of light 115, when applied to the solid state electronic spin system 101, causes the electronic spin or spins to align along one or more well-defined directions. In some embodiments, the source of light 115 is a laser. In an exemplary embodiment, the laser may generate light 105 that has a wavelength of about 532 nm and has green color.

[0039] The first external perturbation 106 originating from the first external perturbation source 116 is applied to the solid state electronic spin 101 and causes a detectable splitting in the electronic spin energy level, i.e., causes a detectable Zeeman shift. In some embodiments, the first external perturbation source 116 is a permanent magnet, or a current-carrying coil, and the first external perturbation 106 is a static magnetic field.

[0040] A second external perturbation 107 originating from the second external perturbation source 117 is applied to the solid state electronic spin 101 and causes controllable flipping or rotation of the electronic spin. In some embodiments, the external perturbation 107 is a sequence of pulses of microwave radiation and the second external perturbation source 117 is a microwave field generator. As used herein, microwave radiation is used as the drive field for solid state electronic spins. It is typically 100 MHz - 10 GHz in frequency, depending on the magnitude of the bias magnetic field.

[0041] In accordance with certain aspects of the present invention, additional external perturbations may be applied to the system. For example, a third external perturbation originating from a third external perturbation source may be applied to the sample or source of oscillating magnetic fields to be measured to manipulate the nuclear spins. In some embodiments, the third external perturbation is a sequence of pulses of RF radiation and the third external perturbation source is an RF field generator. In some other embodiments, the third external perturbation is an electronic spin resonance (ESR) field. Radio-frequency radiation is used as the drive field for nuclear spins comprising the sample. It is typically 1 MHz - 100 MHz in frequency, depending on the magnitude of the bias magnetic field.

[0042] In an aspect, the sensing device can provide information regarding a target material of interest. By way of example, when an individual nuclear (e.g., proton) spin or collection of nuclear spins, such as a can be found in a protein molecule, is placed on the surface of the sensor, the nuclear spin interacts with the solid state electronic spins of the solid-state electronic spin system. The changes detected in the spin-state dependent fluorescence of the solid-state electronic spins can be used for sensing, coherent coupling and imaging the nuclear spin with high resolution. Although the applications apparent to one skilled in the art are innumerable, a non-limiting set of example applications that this approach enables is direct structural imaging of complex molecules that cannot be accessed from bulk studies, a new platform for probing novel materials, monitoring chemical reactions, and manipulation of complex systems on surfaces at a quantum level. [0043] In some embodiments, the sensor is a high purity diamond. In some other embodiments, the solid state electronic spin system is a NV spin. The NV spin is found as a defect in the lattice structure of a single crystal diamond. The NV impurity is based in the lattice of carbon atoms, where two adjacent sites are altered, because one carbon atom is replaced with a nitrogen atom and the other space is left vacant. The vacancies may interact with interstitial atoms, such as nitrogen, and may act as color centers by absorbing visible light. NV spins are visible as red spots when illuminated by laser. In some embodiments, the proximal NV spin is located about 1 nm to 50 nm below the diamond surface. In some other embodiments, the NV spin is located at about 2 to 50 nm, or 2 to 30 nm, below the diamond surface. In accordance with certain embodiments, an ensemble of NV spins located 0 - 10 μπι below the diamond surface is used.

[0044] In some embodiments, the method and device describe magnetic sensing and imaging for narrowband detection of periodically time-varying electromagnetic fields, through coherent manipulation of an electronic spin impurity, including an ensemble of spin impurities. In particular, the technique works by concatenating a series of individually quantum-coherent spin echo or dynamical decoupling sub-sequences applied to the electronic spin, interspersed with quantum measurements of the electronic spin state that give partial information about the sensed field. The full concatenated sequence is synchronized to a digital timing generator (a timing source used to synchronize all pulses during the full loop, e.g., external clock), such that the individual spin echo or dynamical decoupling subsequences are all in phase with each other at the carrier frequency, even though the quantum coherence of the electronic spin is disrupted by the interspersed measurements. One aspect of the present invention is the synchronization of the individual sub-sequences to form a full concatenated sequence, which provides spectral resolution of the detected field proportional to the inverse duration of the full concatenated sequence, while still allowing sensitivity to fields over a bandwidth proportional to the inverse duration of the spin echo or dynamical decoupling sub-sequences. In accordance with certain aspects, the present invention realizes NV spectral resolution <1 Hz to periodically time-varying electromagnetic fields. A particular application of the present invention is the detection of nuclear magnetic resonance (NMR) signals produced by samples placed directly on the diamond surface, for the purposes of spectroscopic molecular identification. [0045] FIG. 2 is a flow chart of steps in accordance with one aspect for analysis of a sample or other source of oscillating magnetic fields in the proximity of a sensor. Step 1 comprises cleaning the surface of the diamond and positioning the diamond in the magnetic bias field such that the solid state electron spin system is aligned with the magnetic bias field. In some embodiments, cleaning is achieved using a 3 acid treatment described elsewhere in this specification. This is followed by Step 2, which comprises placing the diamond in proximity to the source of oscillating magnetic fields that will be measured, more specifically in certain embodiment this step comprises providing a source of nuclear spins on the surface of the diamond. Step 3 comprises programming a digital timing generator with a subsequence of pulses to control light and microwave fields that will be applied to the solid state electron spin system. The pulses are designed to be selective for detecting oscillating magnetic signals near central frequency / The duration of the sub-sequence is an exact integer multiple of l/f. In Step 4, the digital timing generator is utilized to control Steps 5-7, which are repeated in a loop. The maximum number of loop iterations may be determined by the temporal stability of the digital timing generator. Step 5 comprises pulsing the light source to generate light to align the spins of the solid state electron spin system. Step 6 involves the application of pulsed microwave fields to cause the state of the solid state electron spin system to be sensitive to oscillating magnetic signals near central frequency / Step 7 comprises pulsing the light source to generate light to measure fluorescence of the solid state electron spin system and recording the fluorescence intensity at each iteration of the loop. In accordance with certain embodiments, optional Step 8 comprises concurrently pulsing the radio-frequency source to manipulate the nuclear spins. These pulses are applied at defined intervals in the loop, but need not be applied at each iteration. In accordance with Step 9, a Fourier transform is applied to the sequence of fluorescence intensity measurements that were recorded at equal time intervals during the loop over Steps 5-8. Step 9 provides information regarding the target sample in the form of nuclear magnetic resonance spectra with high spectral resolution.

[0046] In accordance with some aspects, steps 4-9 may be repeated, and the outputs from each repetition averaged together, to improve sensitivity to small signals. The repetitions can occur with any desired delay from one repetition to the next (i.e., they need not be

synchronized to each other).

[0047] Experimental [0048] Diamond Samples: The diamond used to construct the NV NMR sensor was a 2 mm x 2 mm x 0.5 mm high-purity chemical vapor deposition (CVD) diamond chip, with 99.999% 12C isotopic purity, and bulk nitrogen concentration [¹⁴N] < 8.5 x 10¹⁴ cm^"3 (Element Six). Modification of the CVD gas mix during the final stage of growth yielded a 13 μπι thick nitrogen-enriched top layer ([¹⁴N] « 4.8 x 10¹⁸ cm^"3, measured by secondary ion mass spectrometry). The diamond was electron-irradiated (1.3 x 10¹⁴ cm^"2 s^"1 flux) for 5 hours, and annealed in vacuum (800° C) for 12 hours, yielding an NV concentration [NV] « 3 x 10¹⁷ cm^"3. The diamond was cut such that the top face was perpendicular to the [100] crystal axis, and the lateral faces were perpendicular to [110]. All four edges of the top face were then polished through at 45° (Delaware Diamond Knives), resulting in a truncated square pyramid, with top face area 1 mm x 1 mm. The ensemble T₂* dephasing time for NV centers in this diamond, measured using Ramsey spectroscopy, was T₂* « 750 ns. The ensemble T₂ decoherence time, measured using a Hahn-echo sequence, was 6.5 μβ.

[0049] The diamond used for NV ensemble sensing of antenna signals (FIGS 3B -3C) was identical to the NV NMR sensor chip, but without angle-polished edges. The diamond used in single NV experiments was a 4 mm x 4 mm x 0.5 mm high-purity CVD diamond chip, with 99.99% ¹²C isotopic purity near the surface, which contained preferentially oriented NV centers with nitrogen concentration [¹⁴N] « 1 x 10¹⁵ cm^"3 and NV concentration [NV] « 3 x 10¹² cm^"3. The approximate coherence times for the single NV center used in our experiments were Tl « 1ms, T2 « 500 μβ, and T2* « 50 μβ.

[0050] Single-NV Sensor: The single-NV sensor was based on a low NV density diamond chip, as described above. The antenna-generated magnetic signals were measured using a home-built scanning laser microscopy system. Confocal scanning of the diamond chip was done by a three-axis motorized stage (Micos GmbH). A 400 mW, 532 nm diode- pumped solid state laser (Changchun New Industries) was used as an excitation light, and an acousto-optic modulator (Isomet Corporation) operated at 80 MHz was used to time-gate the laser. An oil-immersion objective (lOOx, 1.3 NA, Nikon CFI Plan Fluor) focused the green laser pulses onto a single NV center. Red fluorescence from the NV centers was collected by a silicon avalanche photodetector (Perkin Elmer SPCM-ARQH-12) through a 75 μπι sized pinhole. The NV spin initialization and readout pulses were 3 μβ and 0.5 μβ, respectively. For the single-NV magnetometry, microwave pulses were applied to NV centers through nanofabricated 20 μηι gapped waveguide, and the pulse sequence was generated using a GHz range signal source (Agilent E4428C), which was I/Q modulated by an AWG (Tektronix 5014c) for the microwave phase control.

[0051] NV Ensemble Sensor: The NV ensemble sensor was based on a diamond chip with 13 μπι NV-enriched surface layer, as described above. For detecting antenna-generated magnetic signals, the diamond chip was rectangular. For NMR sensing, an angle-polished chip was employed instead, allowing total internal reflection of the laser beam (FIG. 5A) to prevent direct illumination of the NMR sample. Excitation light was provided by a diode- pumped solid state laser at 532 nm (Coherent Verdi G7), directed through an acousto-optic modulator (AOM) (IntraAction ASM802B47) to produce 5 pulses. The first ~1 of each pulse was used to optically read the spin state of the NV ensemble, while the remainder of the pulse repolarized the NVs. The AOM was driven by a digitally synthesized 80 MHz sinusoid (Tektronix AWG 7122C), amplified to 33 dBm (Mini circuits ZHL-03-5WF), and the total laser power at the sensor volume was 150 mW. The laser was focused to a 20 μπι diameter waist near the position of the NV sensor layer, resulting in an optical intensity 48 kW/cm² (comparable to the NV saturation intensity of 100 kW/cm²).

[0052] For detection of antenna magnetic signals, the diamond was mounted on a glass slide; for NMR detection, it was glued (Epoxy Technology Inc., EPO-TEK 301) to a 3 mm glass prism (Thorlabs PS905) and placed inside a sample cuvette (FireflySci Type4

Microfluorescence Cuvette). In both cases, the diamond was carefully rotated such that a [111] diamond crystal axis was aligned to the static magnetic field Bo. NV centers aligned along this axis were used for sensing, while those along the other three [111] directions were far off-resonance and contributed only to the background fluorescence. The alignment was carried out by overlapping the pulsed electron spin resonance (ESR) frequencies of the 3 non- aligned axes. The static magnetic field strength was Bo = 88 millitesla, such that the resonance frequency of the \m_s = 0) → \m_s = — 1) spin transition for the aligned NV centers was fharmor = 400 MHz. (The \m_s = 0) → \m_s = +1) resonance frequency was 5340 MHz.)

[0053] The NV magnetometry pulse sequences for magnetic resonance detection were carried out on the \m_s = 0) → \m_s = — 1) transition. Microwaves were delivered using a straight length of wire (0.25 mm diameter) above the diamond, approximately 0.4 mm away from the NV sensing volume. Both the 400 MHz carrier frequency and the pulse modulation were synthesized digitally (Tektronix AWG 7122C); pulses were then amplified to 40 dBm (Mini circuits ZHL-100W-52-S+) and coupled into the wire, yielding NV Rabi frequency Ω = 8.3 MHz. An XY8-6 dynamical decoupling sequence was used to selectively detect magnetic resonance signals around 3.755 MHz, the which is the proton Larmor frequency at Bo = 88 millitesla. The phase of the final π/2 pulse of the sequence was optimized to give

fluorescence corresponding to a mixed state of the NV (i.e. equal to the mean fluorescence over one Rabi oscillation), to make the fluorescence signal linearly sensitive to small magnetic field amplitudes. For an ideal two-level quantum system, this condition would correspond to a 90° phase shift between the initial and final π/2 pulses; in practice, small drive detunings associated with ¹⁴NV hyperfine structure required manual optimization of the phase. To reject laser intensity noise and microwave power fluctuations, the phase of the final π/2 pulse of every second SR magnetometry subsequence was shifted by 180° relative to the nominal value, and successive pairs of readouts were amplitude-subtracted. Thus, one SR time-series data point was recorded for every two magnetometry subsequences.

[0054] Spin state-dependent fluorescence from the NV centers was collected with a quartz light guide (Edmund Optics 5mm Aperture, 120 mm L, Low NA Hexagonal Light Pipe) and delivered to a balanced photodiode module (Thorlabs PDB210A). To eliminate scatter from the excitation laser, an interference filter (Semrock BLP01-647R) was placed between the light guide and detector. A small fraction of the excitation beam was split off upstream of the diamond chip and directed onto the second channel of the balanced diode module. A glass slide mounted on a motorized stage (Thorlabs PRM1Z8) in the second path allowed automated re-balancing between averages during long SR signal acquisitions. When the NV centers were fully polarized in \m_s = 0), the light-induced fluorescence signal produced a single-channel (unbalanced) photocurrent of 30 μΑ. Immediately after applying a microwave π pulse, the single-channel photocurrent was 28 μΑ, indicating a maximum fluorescence contrast of ~7%. The difference signal of the photodiode module (with onboard transimpedance gain 1.75 x 10⁵ V/A) was further amplified by 3 dB and low-pass filtered at 1 MHz using a low-noise pre-amplifier unit (Stanford Research SR-560), then recorded with a digital to analog converter (DAQ) (National Instruments NI-USB 6281). The DAQ bandwidth was 750 kHz, and the digitization was on-demand, triggered by a TTL pulse from the AWG used to control the experiment. The delay between the rising edges of the AOM gate pulse and the DAQ trigger was 1.9 μβ, optimized for maximum spin state-dependent fluorescence contrast. [0055] SR Protocol Synchronization and Data Analysis: The SR cycle period TSR = 24.06 μβ, the reciprocal central SR detection frequency, )^"1 = 1 / (3.74065 MHz) = 267.3 ns, and the reciprocal NV drive frequency, fharmoi¹ = 1/(400 MHz) = 2.5 ns, were all chosen to be exact integer multiples of the clock period of the timing generator (Tektronix AWG 7122C), rciock = (1/12 GHz) = 0.083 ns. The ultimate frequency resolution of the experiment was therefore determined by the stability of this clock. The NV magnetometry pulse sequence (XY8-6 in all experiments, unless otherwise specified) was saved in the memory of the AWG and its output was gated by a TTL signal from a programmable pulse generator (Spincore PulseBlasterESR-PRO 500 MHz). The PulseBlaster gate duration was used to specify n, the number of SR iterations per experiment. For detecting the NMR signals, the pulse blaster also generated the TTL pulse for gating the proton driving MW pulses. Each readout of the SR protocol was saved in a numerical array, giving a time series of length n. Individual time series were averaged (in the time domain) to improve SNR. The first 20 SR time series data points, which coincided with the proton pulse π/2 pulse plus approximately 50 times the coil ringdown time, were discarded. The averaged time series data were then mean-subtracted before taking the Fourier transformation and fitting using Matlab. Each spectrum was fit to both Lorentzian and Gaussian lineshapes, and the model with smaller residuals (always Lorentzian, except in the case of FIG. 6B) was selected for display. Unless otherwise specified, all spectra shown in the figures are power spectra, calculated as the absolute value of the Fourier-transformed time series data. When uncertainties are quoted for spectral linewidth or splitting parameters, these uncertainties were estimated by repeating the full experiment and fitting procedure several times, then calculating the standard deviation over the ensemble of fitted parameters.

[0056] Electromagnet: The bias magnetic field Bo was produced by an air-cooled electromagnet (Newport Instruments Type A). The pole pieces were cylindrical, 10 cm in diameter, with adjustable gap set to 3 cm. The main coils (each 1900 turns of copper strip, with room-temperature resistance R = 4.5 Ω) were driven (Hewlett Packard HP 6274) with a continuous current of ~650 mA to produce a nominal field Bo « 88 millitesla. A secondary coil pair (diameter 10 cm, gap 7 cm, 15 turns each) were manually wound around the poles to allow precise field stabilization without the need for very small adjustments to main current supply. The secondary coils were driven by a voltage-controlled current supply (Thorlabs

LDC205C), controlled by the analog output channel of a DAQ (National Instruments PCI

6036E). The field strength was monitored using continuous wave electron spin resonance (cwESR) measurements on a secondary diamond chip, spatially separated from the main sensor by ~1 cm. The cwESR microwave frequency modulation was synchronized to the main SR experiment using the same AWG (Tektronix AWG 7122C), to ensure that any cross-talk between the detectors was coherent over averages of the SR protocol and could be removed during data analysis. (This precaution proved unnecessary in the final experiments, as the cwESR drive power was too weak to produce a measureable effect on the SR sensor.) The excitation laser, light collection optics, and microwave drive for the secondary experiment were all independent from those of the main SR magnetic resonance sensor. This enabled feedback control over magnetic field fluctuations (primarily due to current noise in the main coils) with bandwidth ~12.5 Hz, resulting in short-term (~30 minutes) field stability cm ~ 30 nT RMS. To correct slow drifts between the main magnetic resonance sensor and the secondary field-stabilization sensor, we periodically (every 5 minutes) paused the SR protocol between averages and performed pulsed ESR measurements on the primary diamond. Any measured magnetic field drifts were used to correct the setpoint of the fast feedback loop, ensuring long-term (~50 hours) stability cm ~ 50 nT RMS. All cwESR measurements were carried out using both the \m_s = 0) → \m_s = — 1) and the \m_s = 0) → \m_s = +1) transitions of the aligned NV centers, to distinguish resonance shifts due to changes in temperature and magnetic field. For fast feedback measurements on the secondary sensor, only 4 discrete ODMR frequencies were monitored to maximize bandwidth. This system was potentially susceptible to second-order feedback errors associated with simultaneous changes in Bo and temperature. Therefore, the secondary sensor was thermally anchored to a piece of black-anodized aluminum and its temperature actively stabilized using absorption from a separate DPSS laser (Thorlabs DJ532-40). Temperature control was not required for slow feedback on the main magnetic resonance sensor, where a full ODMR spectrum (58 frequency points) was acquired to fully account for all drifts in magnetic field, temperature, and optical contrast.

[0057] NMR Drive Coils: Radio frequency (RF) pulses for driving sample protons (e.g., with a π/2-pulse at the start of an SR-FID experiment) were produced by a pair of solenoid coils wound around the sample cuvette. This geometry, with 1.1 cm coil diameter and 1.2 cm center-to-center spacing, provided a combination of strong drive fields and convenient optical access to the NV ensemble sensor. The coils were 24 turns each, connected in series and coupled to the current source (Rigol DG 1032) with a standard network of variable matching and coupling capacitors. After tuning, the resonance frequency was 3.75 MHz, and the coil Q was 140. Driving the coils on resonance, a maximum nuclear Rabi frequency Ω = 2.8 kHz was obtained.

[0058] In accordance with one aspect, a synchronized readout (SR) technique is used to coherently measure oscillating magnetic signals (e.g., the free induction decay (FID) in liquid-state MR) for a duration greatly exceeding the T2 coherence time of the NV. The SR protocol (FIGS. 3A-3C) includes a series of concatenated NV magnetometry sub-sequences, interspersed with projective NV spin state readouts, all synchronized to an external clock. The protocol is defined with respect to a particular central frequency fo. The magnetometry sub-sequences are all identical, each comprising an initial π/2 pulse, followed by a train of π pulses applied at a rate of 2 fo, and ending with another π/2 pulse. The initial and final π/2 pulses are chosen to be 90° out of phase, such that the final NV spin population is linearly dependent on the amplitude of the oscillating magnetic field signal. This is different from previous NV-detected NMR protocols, where magnetometry pulse sequences yielded a quadratic dependence of the final NV state on the magnetic field signal in order to sense statistical fluctuations in the sample magnetization. At every SR readout step, the

accumulated NV spin population is measured via spin-state-dependent fluorescence, and the NV spin is repolarized.

[0059] The delay between the start of successive magnetometry sub-sequences in the SR protocol is an integer number of periods at the central frequency, TSR = k I fo. Thus, a magnetic signal at fo produces exactly the same NV phase accumulation during each repetition of the sub-sequence, and the mean NV fluorescence intensity is the same at each readout step. By contrast, a magnetic signal that is slightly detuned from the central frequency, f=fo + 5f is advanced in phase at the start of each SR iteration by δφ = TSR 5f In this case, the mean NV fluorescence changes at every readout, and the time series of readouts will oscillate at frequency 5f. In effect, the SR protocol mixes down the magnetic signal by fo. The frequency resolution of the fluorescence time series data is set by the total duration of the SR protocol and is therefore limited, in principle, only by the stability of the external clock. The range of signal frequencies / that can be detected without aliasing is given by 1 / (2 TSR). One condition for the successful application of the SR technique, which is not necessarily satisfied in other NV-detected NMR protocols, is that the NV should be weakly coupled to the magnetic signal source to avoid line broadening due to back-action at each SR

measurement iteration. [0060] In accordance with an exemplary embodiment, the SR technique was applied to a magnetic sensor including a single NV center, imaged in a confocal microscope (FIG. 4A). A nearby coil antenna was excited continuously (without gating or triggering of the sources) to produce a magnetic signal consisting of three closely-spaced frequencies around 3.7325 MHz; and the SR protocol was performed at a central frequency fo = 3.7313 MHz (period 268 ns). Magnetometry was carried out using an CPMG-32 dynamical decoupling sequence; the SR cycle period was TSR = 75.04 μβ; and the total experiment duration was T= n TSR = 112.5 s, for n = 1.5 x 10⁶ the number of readouts. The signal strength was tuned such that magnetic field amplitude at the NV center was ~3 μΤ, corresponding to the maximum fluorescence contrast for a single CPMG-32 sequence. Due to finite optical collection efficiency, each SR fluorescence readout detected a mean of only ~0.03 photons. The SR protocol was repeated 100 times, and a periodogram was contructed from the full data set. In the resulting power spectrum, the three signal peaks were clearly distinguishable, and the spectral resolution was 5.2 mHz (FWHM).

[0061] To improve sensitivity and eliminate the need for signal averaging, the same measurement was repeated using an NV ensemble magnetic sensor that integrates fluorescence from a total of ~5 x 10⁸ NV centers. A spectral resolution of 5.2 mHz (FWHM) was observed again, this time in a single SR experiment without averaging (FIG. 4B). The SR protocol duration was extended to T = 3 x 10³ s, using TSR = 1.2 ms and n = 2.5 x 10⁶, and linewidths of 0.4 mHz (FIG. 4C) were recorded, again without averaging. This spectral resolution was approximately 5 orders of magnitude narrower than previously demonstrated for non-SR magnetic signal detection using NV centers, and was likely limited by pulse timing jitter and/or oscillator phase noise in the waveform generator used to synthesize the SR magnetometry sub-sequences.

[0062] Detection of magnetic resonance signals with high spectral resolution using the SR technique requires long sample coherence times. Because molecular diffusion necessarily limits the correlation times of magnetic noise in nanoscale samples, larger volumes were used such that the mean thermal magnetization, M_z, is greater than the distribution width of statistical fluctuations in the magnetization, M. The sample temperature and the magnitude of the static magnetic field Bo determine the thermal polarization, and hence the minimum detection volume. For experimental convenience in driving magnetometry pulse sequences at the NV Larmor frequency, testing was typically conducted at Bo = 88 millitesla. Then, taking protons in water as a characteristic sample, the condition M_z > <JM sets a lower bound on the

(2 /c 1

— ~ — ^¾ (9 μ^ηι)³. Here y„ = 1.41 x 10^"26 J/T is the proton Yp ^Bo Pp

magnetic moment, ke = 1.38 x 10^"23 J/K is Boltzmann's constant, T = 300 K is the temperature, and p_p = 6.7 x 10²⁸ / m³ is the density of protons in water.

[0063] An NV ensemble magnetic sensor was constructed, which was designed to detect NMR signals due to the thermal sample magnetization (FIG. 5A). The sensing volume included the overlap region between a 13 μιη thick NV-doped layer (NV density 2 x 10¹⁷ cm^" ³) at the diamond surface, and the ~10 μιη diameter waist of the optical excitation beam, which was totally internally reflected off the diamond surface at an angle of ~45°. Applying a sinusoidal magnetic test signal from a nearby coil antenna at /= 3.742 MHz, an SR sensitivity was measured (using/o = 3.74066 MHz) of 50 pT/Hz^½ . By comparison, the expected signal size due to a large sample [V » (9 μηι)³] of protons in water is ~81 pT.

[0064] To motivate this choice of sensor geometry, a single NV center, located a depth d below the diamond surface, and a sample consisting of a half-space of Larmor-precessing spins (density p_p = 6.7 x 10²⁸ / m³) above the surface are considered. Comparing the standard deviation of magnetic field fluctuations at the position of the NV, σ_Β, to the mean magnetic field due to the sample magnetization, B, one can determine that the condition for the latter to dominate is d_NV > 3 μηι. (Note that both a_sand B are obtained by projecting the magnetic field onto the dipole axis of the NV, which has been aligned parallel to the direction of Bo.) On the other hand, the effective detection volume for the mean sample magnetization, defined here as the radius r of a hemisphere above the sensor such that spins within the hemisphere contribute exactly half of B, is given by V = ^ (κ d_NV)³ , for κ « 2.4 a geometric constant estimated by numerical integration (FIG. 5A inset). The present design in accordance with this aspect, with mean NV depth d= 6.5 μπι, thus represents a tradeoff between (i) suppression of magnetic noise due to near-surface spin fluctuations, and (ii) minimization of the effective sensing volume for the thermal sample magnetization.

[0065] Using the NV ensemble sensor, SR spectroscopy was applied to detect NMR from glycerol (C3H8O3) molecules. The diamond was placed in a cuvette filled with glycerol (volume 0.64 mL) and aligned in the bias field of an electromagnet (Bo = 88 millitesla). At the start of the experiment, a π/2 pulse was applied to tip the sample protons into the transverse plane of the Bloch sphere; their precession frequency was then measured by probing the NV centers with an SR sequence (parameters TSR = 24.06 and n = 4 x 10⁴). The SR sequence duration was chosen to allow full population relaxation of the sample spins (Γι « 3 x 10^"2 s). After 7 x 10⁴ averages, the nuclear FID was readily observable (FIG. 5B). Near the end of the SR sequence, after the sample spins were fully dephased, a coil antenna was used to drive a calibrated oscillating magnetic field pulse (zero-to-peak amplitude 90 pT, offset frequency 5f= 1.4 kHz from fi, duration 12 ms). Comparison of integrated peak intensities in the SR power spectrum (FIG. 5B, inset) yielded a signal amplitude of 105 pT (zero-to-peak) for the glycerol FID, approximately consistent with the calculated value of 79 pT for a glycerol proton spin density of p_p = 6.7 x 10²⁸ / m³. To exclude the possibility of spurious detection associated with room noise or sensor imperfections, Bo was swept over 0.02 millitesla and the SR protocol was repeated at each value. A linear fit to the measured FID line centers gave the correct value for the proton gyromagnetic ratio, γ_ρ = (42.574 ± 0.002) MHz / T (FIG. 5C).

[0066] To assess the spectral resolution limits of NMR detection using this technique, SR spectroscopy was carried out on a sample of pure water. The experimental conditions were identical those of the glycerol measurements, except the full SR sequence duration was extended (T= 2 s) to account for water' s longer decoherence and population decay lifetimes (J2, Ti > 2s). The water FID linewidth in the SR power spectrum was 9 ± 1 Hz FWHM (FIG. 5D), approximately a factor of ~25 narrower than the best spectral resolution obtained using NV correlation spectroscopy. Nevertheless, the observed lineshape was notably broader than the limiting value (TFWHM ~ 0.2 Hz) associated with the intrinsic bulk decoherence lifetime of the sample.

[0067] A number of effects that could contribute to sample dephasing in the SR spectroscopy of water were inveestigared. Temporal inhomogeneity of the bias field Bo was excluded by active stabilization using the electromagnet current supply, with residual fluctuations < 0.05 μΤ RMS (equivalent to a proton linewidth Γ< 2.5 Hz FWHM) over the course of the experiment. Gross spatial gradients in Bo were ruled out by continuous wave electron spin resonance (cw-ESR) measurements at defined positions across the diamond surface, which showed Bo variability < 0.3 μΤ over an area of ~1 mm². To check for broadening due to far-detuned proton driving by the NV magnetometry sequence (which should scale approximately as Γ ~ (i¾-¾) /γκν )² I A ~ 1 Hz, for i¾ « 15 MHz the NV Rabi frequency, A ~ 400 MHz the detuning, and ¾_> = 42.58 MHz/T and γΝγ = 28.02 GHz/T the proton and NV gyromagnetic ratios, respectively), the NV Rabi frequency was reduced by up to a factor of three, but no narrowing of the proton resonance was observed. Finally, to test whether the proton line was broadened by magnetic field gradients associated with repolarization of the NV electronic spins at each SR readout iteration, (an effect estimated to contribute to broadening at the Γ ~ 1 Hz level, the duty cycle of the magnetometry subsequence was varied (relative to the SR cycle period TSR) from 0.18 (using XY8-2) to 0.53 (using XY8-6). This produced no significant increase in the proton linewidth, indicating that NV back-action was not the dominant dephasing mechanism.

[0068] Having thereby ruled out both Bo inhomogeneity and interactions with the NV spins as primary determinants of the observed water FID signal width, the limited spectral resolution of the measurements was attributed to a combination of (i) micron-scale magnetic gradients due to susceptibility differences between sensor components, and (ii) dephasing due to diffusion of sample molecules close to magnetic defects at the diamond surface. This hypothesis was tested by applying π-pulses to the protons at times t = 40 ms and t = 120 ms after the start of the SR protocol, to refocus proton spin dephasing due to local gradients. This resulted in a proton linewidth of 2.8 Hz FWHM (FIG. 5D), approximately consistent with the measured distribution of Gaussian temporal fluctuations in Bo (~2.5 Hz FWHM) recorded by the electromagnet feedback controller over the full duration of the experiment. Gradient- induced broadening is commonly observed in sub^L volume NMR spectroscopy with microcoils, and can be mitigated by improved susceptibility matching in the sensor design. Dephasing due to shallow paramagnetic impurities may be reduced by careful diamond surface preparation using a combination of wet-etching and annealing in oxygen.

[0069] To illustrate the applicability of SR to molecular NMR, liquid-state FID spectra of trimethyl phosphate [PO(OCH₃)₃] and xylene [(CH₃)2C₆H4] were acquired. Trimethyl phosphate (TMP) is a standard reagent known to have large scalar coupling (J[P,H] « 11 Hz) between the methyl protons and the central ³¹P nuclear spin. Xylene is an aromatic solvent with substantial chemical shifts (~5 ppm of the proton Larmor frequency) due to different electron densities associated with the carbon ring structure and satellite methyl groups. Data were acquired using the same procedure as for glycerol and water FID spectra; the SR protocol parameters were TSR = 24 and n = 4 x 10⁴. The SR NMR spectrum for TMP

(FIG. 6A) shows two clearly resolved peaks due to the J-coupled nuclei, with splitting Afi « 13 ± 1 Hz. The SR NMR spectrum for xylene (FIG. 6B) also shows two peaks, split by Afcsj « 20 ± 2 Hz, consistent with the previously reported value for the chemical shift. The observed peak intensity ratio of ~2.2: 1 in the xylene SR NMR power spectrum is as expected for the relative nuclear abundance of 6:4, with the protons in high electron-density methyl groups shifted to lower frequency. These measurements constitute the first demonstration of NV-detected NMR with spectral resolution sufficient to resolve frequency shifts due to molecular structure.

[0070] Those skilled in the art would readily appreciate that all parameters and configurations described herein are meant to be exemplary and that actual parameters and configurations will depend upon the specific application for which the systems and methods of the present invention are used. Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments of the invention described herein. It is, therefore, to be understood that the foregoing embodiments are presented by way of example only and that the invention may be practiced otherwise than as specifically described. The present invention is directed to each individual feature, system, or method described herein. In addition, any combination of two or more such features, systems or methods, if such features, systems or methods are not mutually inconsistent, is included within the scope of the present invention.

Claims

CLAIMS What is claimed is:

1. A system comprising: a sensor, wherein the sensor comprises, a solid state electronic spin system disposed below the surface of the sensor, wherein the solid state electronic spin system has a spin-state dependent fluorescence; a source of light; a source of first external perturbation, wherein the source of first external perturbation generates a magnetic field; a source of second external perturbation; wherein, the source of light and the first and second external perturbations are configured to coherently and independently manipulate the spin states of at least one solid state electronic spin system; a digital timing generator configured to synchronize the light and the first and second perturbations; and a detector to optically measure the solid-state electronic spins spin-state dependent fluorescence.

2. The system of claim 1, wherein the sensor comprises a high purity diamond crystal lattice.

3. The system of claim 2, wherein, the solid state electronic spin system is a nitrogen- vacancy (NV) spin in the high purity diamond lattice.

4. The system of any one of claims 1-3, further comprising a source of third external perturbation comprising a radio frequency (RF) electromagnetic field source or an electronic spin resonance (ESR) field source.

5. The system of any one of claims 1-4, wherein, the source of second external perturbation is a microwave field source.

6. The system of any one of claims 1-5, wherein the detector is a CCD camera, or a photomultiplier, or a photodiode.

7. The system of any one of claims 1-6, wherein the sample to be measured is placed on a surface of the sensor.

8. The system of any one of claims 1-6, wherein the sample to be measured is placed in the proximity of the sensor.

9. The system of any one of claims 1-8, wherein the digital timing generator is

configured to provide a sub-sequence of pulses of the light and the second perturbation designed to be selective for detecting oscillating magnetic signals near a central frequency,/

10. The system of claim 9, wherein the sub-sequence of pulses has a duration that is an integer multiple of l/f.

11. The system of any one of claims 1-10, wherein the solid state spin systems spin- dependent fluorescence is changed due to the interaction with the sample.

12. The system of any one of claims 1-11, wherein the source of light radiation is an optically pumped laser.

13. The system of claim 12, wherein the optically pumped laser has a wavelength of 532 nm.

14. A method of measuring an oscillating magnetic field, comprising: disposing a sensor in a first external perturbation field, wherein the sensor comprises a solid state electronic spin system disposed below the surface of a sensor and the solid state electronic spin system is aligned with the first external perturbation field, wherein the solid state electronic spin system has a spin-state dependent fluorescence; positioning the sensor in proximity to a sample comprising a source of oscillating magnetic fields; obtaining a sequence of fluorescence intensity measurements by i) pulsing a light source to generate light to align spins of the solid state electronic spin system; ii) pulsing a source of second external perturbation; and iii) detecting fluorescence of the solid state electronic spin system; wherein steps i) and ii) in each iteration are synchronized using a digital timing generator; and repeating steps i), ii) and iii) for a predetermined number of iterations.

15. The method of claim 14, obtaining a sequence of fluorescent intensity measurements further comprises iv) pulsing a source of third external perturbation.

16. The method of claim 14 or 15, further comprising applying a Fourier transform to the sequence of fluorescent intensity measurements that were recorded at equal time intervals during each iteration.

17. The method of any one of claims 14-16, wherein the sensor comprises a high purity diamond crystal lattice.

18. The method of claim 17, wherein the solid state electronic spin system is a nitrogen- vacancy (NV) spin in the high purity diamond lattice.

19. The method of claim 15, wherein the third external perturbation comprises a radio frequency (RF) electromagnetic field source or an electronic spin resonance (ESR) field source.

20. The method of any one of claims 14-19, wherein the source of second external

perturbation is a microwave field source.

21. The method of any one of claims 14-20, wherein the digital timing generator is

22. The method of claim 21, wherein the sub-sequence of pulses has a duration that is an integer multiple of l/f.

23. The method of any one of claims 14-21, wherein the sample comprises a liquid sample.

24. The method of claim 23, wherein the liquid sample volume is between about 10 to 100 pL.