Hyperpolarization-Enhanced NMR Spectroscopy with Femtomole Sensitivity Using Quantum Defects in Diamond

Nuclear magnetic resonance (NMR) spectroscopy is a widely used tool for chemical analysis and molecular structure identification. Because it typically relies on the weak magnetic fields produced by a small thermal nuclear spin polarization, NMR suffers from poor molecule-number sensitivity compared to other analytical techniques. Recently, a new class of NMR sensors based on optically probed nitrogen-vacancy (NV) quantum defects in diamond have allowed molecular spectroscopy from sample volumes several orders of magnitude smaller than the most sensitive inductive detectors. To date, however, NV NMR spectrometers have only been able to observe signals from pure, highly concentrated samples. To overcome this limitation, we introduce a technique that combines picoliter-scale NV NMR with fully integrated Overhauser dynamic nuclear polarization to perform high-resolution spectroscopy on a variety of small molecules in dilute solution, with femtomole sensitivity. Our technique advances the state of the art of mass-limited NMR spectroscopy, opening the door to new applications at the picoliter scale in drug and natural-product discovery, catalysis research, and single-cell studies.

Popular Summary

Nuclear magnetic resonance (NMR) spectroscopy is a key analytical tool in chemistry, physics, and the life sciences. However, the intrinsic low sensitivity of conventional NMR limits its applications to macroscopic sample volumes of a few microliters. Here, we overcome this sensitivity limitation with an integrated system that combines detection of NMR signals on a quantum diamond chip with an excess of nuclear spin polarizations.

Nitrogen-vacancy color centers—a type of point defect in diamond—can be used as quantum sensors for optical detection of weak magnetic signals. These quantum diamond sensors can be brought very close to a sample, allowing for high-resolution NMR spectroscopy of picoliter volumes, similar to that of a single biological cell. In our work, we show that the sensitivity of quantum diamond NMR in picoliter volumes can be increased by more than 2 orders of magnitude using nuclear spin hyperpolarization. In this technique, stable electronic spins are added to the sample. The large electronic spin polarization is then transferred to the sample’s nuclear spins by microwave irradiation.

Our approach employs a common infrastructure and microwave pulse sequence for both quantum diamond NMR and hyperpolarization. It allows for high-resolution NMR spectroscopy on small diluted molecules, with an unprecedented femtomole detection limit in microscopic sample volumes. This technique will advance NMR spectroscopy for microscopic chemical analysis in material science, drug discovery, catalysis research, and single-cell studies.

Figure 1

NV NMR spectroscopy with integrated hyperpolarization. (a) Experimental schematic. Microwave loop antenna near the diamond sensor chip drives both NV (purple) and TEMPOL electronic spins (blue). Hyperpolarized NMR signals from the sample nuclear spins (orange) are detected by NV ensemble fluorescence readout from the diamond chip. Inset: The sensor size is defined by the laser spot size on the diamond (scale bar is $30\mu \mathrm{m}$). (b) The NV center in diamond exhibits spin-dependent fluorescence with triplet ground-state spin transitions that can be manipulated coherently by resonant microwaves (left top). With a suitable MW pulse sequence [CASR, see bottom right of panel (c)] NMR signals can be detected from microscopic volumes and read-out optically. (c) Integrated hyperpolarized NV NMR spectroscopy pulse sequence. In the first half of the pulse sequence ($\sim 300\mathrm{ms}$), the electronic drive is used to hyperpolarize proton spins in the sample via interactions with electronic spins in the TEMPOL radical using Overhauser dynamic nuclear polarization (DNP). In the second half of the pulse sequence ($\sim 200\mathrm{ms}$), a $\pi /2$ pulse on the protons induces a free nuclear precession (FNP) signal from the hyperpolarized sample, which is detected by NV sensor spins via a coherently averaged synchronized readout (CASR) pulse sequence. Bottom left: Overhauser DNP. Continuous microwave driving saturates the electronic spin transition of the TEMPOL radical. Relaxation leads to a net polarization of proton spins in an organic molecule (R-H), which diffuses relative to the TEMPOL radical. Bottom right: The proton FNP signal is detected by the CASR readout scheme, based on interspersed blocks of identical dynamic decoupling sequences synchronized to an external clock.

Figure 2

Hyperpolarization-enhanced NV NMR of water. (a) Comparison of NV NMR spectra obtained from pure water with DNP (red circles, 1 spectrum averaged) and without DNP (blue circles, ${10}^4$ spectra averaged) using a coherently averaged synchronized readout (CASR) pulse sequence. Line shape fits (solid lines) indicate a DNP signal enhancement of 230, with a proton number sensitivity of $10\mathrm{pmol}/{\mathrm{Hz}}^{1/2}$ for a signal-to-noise ratio (SNR) of 3. (b) Magnitude of CASR-detected signal from a sample of pure water as a function of DNP drive frequency. The triplet structure arises from hyperfine coupling between the electron and ${}^{14}\mathrm{N}$ nuclear spins in the TEMPOL radical. (c) Magnitude of CASR-detected signal from a sample of pure water as a function of DNP drive power, expressed in units of the electron Rabi frequency (see methods for details). The maximum DNP signal enhancement is reached at a drive Rabi frequency of approximately 10 MHz.

Figure 3

Sensitivity of hyperpolarization-enhanced NV NMR. (a) CASR-detected spectra of DNP-enhanced $t$-BuOD solutions at different millimolar concentrations (650 mM, 195 mM, 58.5 mM, 17.6 mM, 5.3 mM, from left to right) in ${\mathrm{D}}_2\mathrm{O}$. Spectra are fit to represent the $t$-BuOD sample (red solid line) and residual semiheavy water (HDO) in the ${\mathrm{D}}_2\mathrm{O}$ (blue solid line), with red and blue vertical lines indicating the NMR resonance frequencies of $t$-BuOD and HDO, respectively. (b) Plot of CASR signal fit amplitude of $t$-BuOD against number of sample molecules in the sensing volume (bottom axis) and sample molecular concentration (top axis). Gray area at the bottom marks the noise floor after 5000 s of averaging with a signal to noise ratio (SNR) $\approx 3.5$ for a concentration of 5.3 mM, equivalent to $\sim 50\mathrm{fmol}$ of sample molecules in the 10 pL detection volume. Error bars represent standard deviation ($1\sigma$) of CASR signal measured across three independent experiments.

Figure 4

CASR-detected NMR spectra of hyperpolarized small organic molecules in solution. (a) Xylene in $\mathrm{DMSO-}{d}_6$. (b) Dimethyl formamide in ${\mathrm{D}}_2\mathrm{O}$. (c) Trimethyl phosphate in ${\mathrm{D}}_2\mathrm{O}$. (d) Thymine in $\mathrm{DMSO-}{d}_6$. All samples were dissolved at a concentration of 0.8 M. The features of spectra in (a), (b), and (d) are dominated by chemical shifts, whereas in (c), the $J$ coupling between ${}^{31}\mathrm{P}$ and ${}^1\mathrm{H}$ splits the ${\mathrm{CH}}_3$ resonances into a doublet. In the thymine spectrum (d), a line broadening effect is observable due to (N-H) proton exchange with small quantities of water dissolved in the $\mathrm{DMSO-}{d}_6$. In all spectra, the frequency axis has been set to 0 Hz for the methyl resonance line.