1H Hyperpolarization of Solutions by Overhauser Dynamic Nuclear Polarization with 13C-1H Polarization Transfer.

Dynamic nuclear polarization (DNP) is a method that can significantly increase the sensitivity of nuclear magnetic resonance. The only effective DNP mechanism for in situ hyperpolarization in solution is Overhauser DNP, which is inefficient for 1H at high magnetic fields. Here we demonstrate the possibility of generating significant 1H hyperpolarization in solution at room temperature. To counter the poor direct 1H Overhauser DNP, we implement steady-state 13C Overhauser DNP in solutions and then transfer the 13C hyperpolarization to 1H via a reverse insensitive nuclei enhanced by polarization transfer scheme. We demonstrate this approach using a 400 MHz gyrotron-equipped 3.2 mm magic angle spinning DNP system to obtain 1H DNP enhancement factors of 48, 8, and 6 for chloroform, tetrachloroethane, and phenylacetylene, respectively, at room temperature.

Nuclear magnetic resonance (NMR) spectroscopy is the gold standard for characterizing molecules in solution, with its only real weakness in comparison to other spectroscopic methods being its intrinsically low sensitivity. At room temperature, and the magnetic fields typically used for NMR today (e.g., 9.4 T), the nuclear spin states are polarized to only 0.01%.[1] There is thus intense interest in methods for actively increasing NMR sensitivity. One of the most promising approaches to increasing sensitivity is to transiently increase the polarization of the NMR transitions by so-called hyperpolarization methods, with dynamic nuclear polarization (DNP) being at the forefront of these efforts.[2,3] DNP allows the transfer of the much larger polarization of unpaired electron spins to nuclei.[4] DNP has been very successful in solid-state NMR,[5−10] but its implementation in solution-state NMR has proved much more challenging. Transient hyperpolarized NMR signals can be observed at room temperature using the dissolution DNP approach, where a frozen sample is polarized at 1.2 K followed by rapid dissolution and then shuttling of the solution to an NMR spectrometer.[11] This method provides signal enhancements of 4–5 orders of magnitude on target molecules or bulk solvent molecules with promising applications in magnetic resonance imaging.[12,13] However, the hyperpolarization decays rapidly after dissolution and the technique offers a limited sample throughput.[14] A variety of other approaches to producing transient hyperpolarization such as parahydrogen-induced polarization (PHIP),[15] triplet DNP,[16,17] and relay of optically enhanced polarization via the nuclear Overhauser effect (NOE)[18] have also been proposed.

The Overhauser effect (OE) DNP mechanism is the most promising approach for obtaining steady-state continuous hyperpolarization of nuclei in situ and at room temperature.[19−31] Indeed, while the magnetic field dependence of OE is generally unfavorable, it has recently been shown that OE DNP can yield significant 13C or 31P signal enhancements at magnetic fields of ≤14 T.[26,27,29,32−34] For example, Orlando et al. have shown 13C enhancements of ≤600 at 9.4 T for a 35 nL sample in a helix resonator.[27] Dubroca et al. also showed a 13C enhancement[26] of 70 and a 31P enhancement[33] of 160 with sample volumes of 100 μL at 14.1 T, using a custom solution NMR probe and microwave gating to minimize sample heating. In contrast, OE DNP to hyperpolarize 1H in solution at high field is very inefficient.[35] For example, Prisner and co-workers were able to hyperpolarize water with enhancements of −80 at 9.2 T, but only by using microwave superheated temperatures of 160 °C in an ∼1 nL sample volume;[36,37] recent improvements in probe design allowed sample volumes of ≤100 nL.[34] There is thus great interest in hyperpolarizing 1H nuclei in solution.

Here we demonstrate a straightforward approach for obtaining 1H DNP enhancements on chloroform, 1,1,2,2-tetrachloroethane (TCE), and phenylacetylene by transferring 13C hyperpolarization generated by OE DNP to the attached 1H spins using the insensitive nuclei enhanced by polarization transfer (INEPT) pulse sequence (Figure B).[38] Using a Bruker 9.4 T magic angle spinning (MAS) DNP spectrometer and a commercial solid-state MAS DNP probe, we obtain enhancements of 48, 8, and 6 for chloroform, tetrachloroethane, and phenylacetylene, respectively, at room temperature.

Figure 1

(A) Predicted dependence of the coupling factor (ξ) on the applied magnetic field, calculated as described in the Supporting Information, for 1H (red) and 13C (blue) in chloroform. (B) Pulse sequence used here to obtain 1H hyperpolarization by DNP enhanced 13C → 1H INEPT in the presence of continuous microwave irradiation. N is an integer, and π and π′ are delays; further details regarding the pulse sequence are provided in the Supporting Information.

Although 13C NMR is intrinsically less sensitive than 1H NMR, 13C nuclei can in general be much more efficiently hyperpolarized in solution Overhauser DNP experiments. This is primarily due to the large difference in the coupling factors (ξ) between 13C and 1H, as shown in Figure A for factors typical for most protons and for carbons such as the ones in chloroform.[24,25] The coupling factor determines enhancement according to the Solomon equation[39]where γe and γn are the gyromagnetic ratios of the electron and the nuclear spin, respectively, f is the leakage factor (f = 1 – T1n/T1n,dia, where T1n and T1n,dia are the nuclear T1 values in the presence and absence of radicals, respectively), s is the saturation factor (described further below), and the coupling factor is determined by the relative weights of the scalar and dipolar hyperfine contributions (as detailed in the Supporting Information). We also see from Figure A that the efficiency of OE DNP decreases significantly at higher fields due to the magnetic field dependence of the coupling factors.[24,28,40] Although the pure scalar hyperfine mechanism is independent of field, the contribution of the dipolar mechanism results in a decay of the coupling factor at high fields.[28] Consequently, 1H DNP enhancements, which typically rely on direct dipolar contributions between the unpaired electron and 1H spins, are minimal at high fields.[41]

Here, we suggest that a simple way to obtain 1H enhancements is to transfer the hyperpolarization from 13C to 1H using, for example, a reverse INEPT sequence (Figure B)[38,42,43] or cross-polarization[44] in a 13C-enriched substrate. Combining these polarization transfer tools with other approaches to hyperpolarization, including dissolution DNP and PHIP, has been demonstrated.[45−51] Notably, Dey et al. have previously performed 1H→13C transfers to enhance the 13C sensitivity using Overhauser DNP at low fields where 1H hyperpolarization is efficient.[52] INEPT is a well-established method for enhancing the NMR sensitivity of low-gyromagnetic ratio nuclei and/or obtaining through-bond correlation spectra via J couplings.[1] For a fully 13C enriched sample at thermal equilibrium, the 1H magnetization produced by a 13C→1H INEPT experiment will be 25% of that of a directly excited 1H NMR spectrum (in an ideal NMR spectrometer). However, in INEPT experiments under hyperpolarization conditions, the overall sensitivity of the observed nucleus is not only proportional to the ratio of the gyromagnetic ratios of the corresponding nuclei but also determined by the relative hyperpolarization levels. If 13C DNP enhancements of 10–100 can be obtained, neglecting the differences in 1H and 13C T1 relaxation times, the overall 1H sensitivity could be increased by factors of 2.5–25, corresponding to decreases in experimental times of 1–3 orders of magnitude. At natural abundance, the sensitivity gain in comparison to a one-dimensional (1D) 1H NMR spectrum will of course be reduced by 99% due to the low natural abundance of 13C (NA = 1.1%) as only 1.1% of 1H spins will be hyperpolarized. In addition to abundance, the relaxation rates of the involved nuclei, because they are affected by paramagnetic interactions with the radical polarizing agent (vide infra), will affect the INEPT efficiencies, and any differences in efficiency of the two radiofrequency channels in the probe will also impact the sensitivity gains.

Figure shows the direct 1D 1H and 13C Overhauser DNP enhancements obtained for chloroform, TCE, and phenylacetylene using 10 mM [15N]-d16-TEMPONE (denoted hereafter as 15N-TN). High-power microwaves (50 W, corresponding to an estimated ν1e of ∼1.3 MHz) were applied using a 263 GHz gyrotron microwave source, with 10 μL of the sample placed in 3.2 mm sapphire rotors. While the sapphire rotors that are typical in MAS DNP experiments provide good microwave penetration, the small sample volumes help to minimize temperature gradients. To compensate for the heating caused by microwave irradiation, a low-temperature nitrogen gas flow was applied. With chloroform, we performed experiments using a 100% 13C-labeled solvent. As expected, the direct 1H DNP enhancement was low (ε = 0.6) due to the poor efficiency of the dipolar Overhauser DNP mechanism. However, a large 13C enhancement (ε) of 51 was obtained, which is higher than the enhancement of 17 previously observed at 14.1 T using a similar commercial MAS DNP spectrometer.[27]13C→1H INEPT experiments were performed using the pulse sequence of Figure B to yield a high indirect 1H enhancement (ε) of 48, showing that the enhancement can be transferred from 13C to 1H with >90% efficiency. This corresponds to an overall increase in 1H polarization by a factor 12 (48 × γ/γ).

Figure 2

(A) 1H, (B) 13C, and (C) 13C→1H INEPT NMR spectra obtained at (blue) 9.4 T with and (black) without continuous-wave microwave irradiation using a 263 GHz gyrotron source for (top) [13C1]chloroform, (middle) 1,1,2,2-tetrachloroethane (TCE), and (bottom) phenylacetylene. 10 mM [15N]-d16-TEMPONE (denoted as 15N-TN) radical solutions were used. DNP enhancements (ε) were measured by taking the ratio of the integrated signal intensities of the microwave ON and OFF spectra.

To demonstrate the generality of this method, we performed experiments with (natural abundance) samples of TCE and phenylacetylene. With TCE, a direct 13C DNP enhancement of 9 was observed and a corresponding 13C→1H INEPT enhancement of 8 was obtained. Note that the enhancement decreases from chloroform to TCE likely due to weaker electron–nuclear scalar hyperfine couplings resulting in a coupling factor that is 16% of that of chloroform, assuming the same saturation and leakage factors. Phenylacetylene has a terminal alkyne carbon that interacts with nitroxide radicals and has been shown to display reasonable enhancements.[24,26] With our protocol, we observed a 13C enhancement of 7 and an INEPT enhancement of 5 on the alkyne CH group, which are larger than the enhancements for the other carbons in the molecule, in good agreement with the literature (Figure S1).[24,26]

The frequency of collision between radicals and the target species has been shown to be important to the OE DNP mechanism.[29,41] We studied the enhancements as a function of the degree of 13C labeling and radical concentration to obtain optimal DNP enhancements (Figure ). As the results shown in panels A and B of Figure illustrate, neither the degree of 13C labeling nor the radical concentration affects the direct 13C DNP enhancement of chloroform by >15%. The 13C DNP enhancement is not expected to change with the degree of 13C labeling because the frequency of collision per 13C remains unchanged. Note that the 13C enhancement obtained for chloroform here is larger than that of the terminal carbon of phenylacetylene by a factor of 7, while a factor of 2 was observed in a previous study at 14.1 T using 13C-labeled samples; however, these experiments were performed in dilute solutions.[26] On the contrary, we expected the enhancements to be dependent on radical concentration as demonstrated previously;[27,36] the absence of a significant change in 13C DNP enhancement (Figure B) suggests that changes in the coupling and the saturation factors likely cancel each other here.

Figure 3

Experimental 9.4 T 1H, 13C, and 13C→1H INEPT DNP enhancements for chloroform with different (A) degrees of 13C labeling and (B) concentrations of 15N-TN. All enhancements were measured with the same absolute temperature maintained for microwave ON and OFF spectra (details in the Supporting Information). DNP enhancement is shown in panel C as a function of applied microwave power (with the fitted straight line in blue) and in panel D as a function of sample temperature for 100% [13C1]chloroform with 10 mM 15N-TN.

DNP enhancements measured at different microwave powers verified that the optimal condition is at the maximum microwave power attainable with our system because both 13C and INEPT enhancements increase almost linearly with microwave power (Figure C). By using eq and the coupling factor from the literature,[27] we back-calculated the saturation factor at 50 W to be 0.115 (see section S3.1 of the Supporting Information), which supports that the enhancement is saturation-limited. Consistent with our observation, the saturation factor of eq (53) for two coupled hyperfine transitions shows a linear relationship with B12, which is proportional to the microwave power, when γe2B12T1eT2e ≪ 1 as shown in eq .where w1e, w1n, and ωex are the electron spin–lattice relaxation rate [w1e = 1/(2T1e)], the nuclear spin relaxation rate (w1n = 1/T1n), and the Heisenberg spin exchange rate, respectively, and where we findwhen γe2B12T1eT2e ≪ 1. Furthermore, from eq , we estimate that the corresponding B1 field in the sample is approximately 0.05 mT (v1 = 1.3 MHz) by using T1eT2e = 3.5 × 10–15 s2 from the literature[54] (see section S3.1). In comparison, a microwave field of 0.16 mT was obtained previously with 30 W of power on a specialized liquid-state Overhauser DNP probe.[26]

Upon microwave irradiation, significant peak broadening was observed, which can be attributed to a temperature gradient across the sample as we meanwhile observed T2 changes within the peak (see Figure S3). To further examine the impact of sample temperature, we carried out a series of experiments with and without microwave irradiation at different temperatures by actively adjusting the temperature of the cooling gas (Tables S8 and S9). To avoid boiling the chloroform, we tested the behavior below ambient temperature. The measured temperature dependence of the enhancement depicted in Figure D clearly shows decreasing DNP performance with lower temperatures. When the temperature was decreased from 300 to 255 K, an 80% reduction in enhancement was observed. This trend has been observed with CCl4 and can be explained by a less negative coupling factor at low temperature as suggested by theoretical studies.[25,27]

The performance of the INEPT scheme is impacted if there is significant relaxation during the τ and τ′ delays. The INEPT efficiency is related to the measured relaxation times according to

During the τ and τ′ delays of the INEPT sequence, the antiphase coherences leading to the INEPT signal will decay with the relaxation times of both 1H and 13C.[1] (Because the 1JCH coupling constant in chloroform is 210 Hz, both τ and τ′ were therefore set to 1/4JCH, 1.19 ms.) For the samples with a radical concentration of 10 mM, the nuclear relaxation times are longer than τ by at least 1 order of magnitude as Figure A shows, which means that the magnetization of 13C can be successfully transferred to 1H, with INEPT efficiencies of >90% as shown in Figure B. However, when the radical concentration is increased from 10 to >100 mM, the efficiency of INEPT transfer decreases significantly, to ∼25% at 500 mM (Figure B).

Figure 4

(A) Plot showing the variation of measured nuclear spin relaxation times with radical concentration at room temperature with and without microwave irradiation. (B) Plot showing the normalized INEPT signal intensity as a function of the length of the τ delay in the reverse INEPT pulse sequence shown in Figure B. Only radical concentrations of 10 and 500 mM are shown; others are shown in Figure S2.

Because the absolute transfer efficiency is dependent on relaxation, the differences in nuclear relaxation times between microwave ON and OFF experiments due to the temperature gradient will make the DNP enhancement measured by INEPT deviate from the direct 13C enhancement (Figure S2). Among our experiments using chloroform, we consistently observed shorter overall nuclear T2 values under microwave irradiation (Table S5). Consequently, the INEPT efficiency with microwave irradiation is lower than the corresponding efficiency without microwaves, which reduces the INEPT DNP enhancement. However, this effect is negligible when nuclear relaxation times are significantly longer than τ, which is the case at a low radical concentration as we show in Figure B when the radical concentration is 10 mM. When these two effects are combined, the high radical concentration is not favorable for INEPT due to the low absolute INEPT efficiency and the sensitivity of INEPT efficiency to relaxation time fluctuation.

In conclusion, we have demonstrated a simple method for obtaining bulk 1H hyperpolarization in solution by efficiently transferring 13C hyperpolarization generated by the Overhauser effect to 1H with a reverse INEPT pulse sequence. Here, this method yields an overall increase in 1H polarization of a factor 12 for a bulk solution of [13C1]chloroform and is limited by the constraints of our experimental setup. This should be a generally applicable method for hyperpolarizing 1H nuclei that otherwise give intrinsically poor OE DNP enhancement at high magnetic fields. The approach demonstrated here could also be potentially extended to sources containing other abundant heteronuclei such as 31P.