Amplified Overhauser DNP with Selective Deuteration: Attenuation of Double-Quantum Cross-Relaxation

Abstract

We recently used selective ²H labeling of BDPA to investigate the Overhauser Effect (OE) dynamic nuclear polarization (DNP) mechanism in insulating solids doped with 1,3-bis(diphenylene)-2-phenylallyl (BDPA), and established that the α and γ ¹H spins on the fluorene rings are responsible for generating a zero quantum (ZQ) mediated positive bulk polarization. Here, we establish that the phenyl ¹H spins relax via double-quantum (DQ) processes and therefore contribute negative enhancements which attenuate the OE-DNP. With measurements at different magnetic field strengths, we show that phenyl-d₅-BDPA offers >50% improvement in OE-DNP enhancement compared to h₂₁-BDPA attaining a maximum of ~ 90 at 14.1 T and 5 kHz MAS, the highest observed OE-DNP enhancement to date under these conditions. The approach may be utilized to optimize other polarizing agents exhibiting an OE, an important DNP mechanism with a favorable field and spinning frequency dependence.

Graphical Abstract

Nuclear magnetic resonance (NMR) spectroscopy suffers from low sensitivity due to the small size of the nuclear spin energy splittings relative to the thermal energy. Dynamic nuclear polarization (DNP) overcomes this difficulty by using microwave irradiation of electron–nuclear transitions to transfer the polarization of electron spins to the nuclei of interest.^1–3 The increased sensitivity available from DNP has revolutionized the field of NMR, both in liquids and solids, and has enabled the structural studies^4,5 of biomolecules and materials in more complex environments than were previously feasible.^6–11

Particularly effective approaches to DNP rely on coherent manipulation of the electron spin polarization and require that the electron Rabi frequency, ω_1e, be greater than the spin–spin relaxation rate of the electrons, 1/T_2e, where T_2e, the electron transverse relaxation time, is typically on the order of a few microseconds. ω_1e is proportional to the square root of microwave power, and with current technology, this usually ranges from a few milliwatts to tens of watts. Although higher powers have been achieved,¹² microwave power is a limitation frequently encountered at magnetic field strengths exceeding 5 T.¹³ Thus, absent a resonance microwave structure to boost efficiency in coupling ω_1e to the sample, the electron Rabi frequency is low and coherent transfers are presently not practical. However, the achievable Rabi fields are sufficient to support continuous-wave DNP (CW-DNP) mechanisms.^14,15 Accordingly, four known CW-DNP mechanisms are typically utilized in DNP experiments on solids: the solid effect (SE),^16,17 the cross effect (CE),^18–20 thermal mixing (TM),^21–24 and the Overhauser effect (OE).^2,25 Detailed discussions of the first three mechanisms–the SE, CE, and TM–are available in review articles^14,15,26

The OE, which is the subject of this paper, requires coupled electron–nuclear spin pairs. Microwave irradiation at the EPR (electron paramagnetic resonance) frequency of a polarizing agent with a narrow line, such as BDPA, drives allowed single-quantum EPR transitions. Subsequent cross-relaxation in the coupled electron–nuclear spin system via zero quantum (ZQ) and double quantum (DQ) pathways with unequal rates results in enhanced polarization of the nuclear spins. The polarization buildup of the nuclear spins can be positive or negative, depending on whether either the ZQ or the DQ cross-relaxation is dominant, respectively. The advantages of the OE mechanism are that it scales positively with magnetic field and with increasing spinning frequency. Furthermore, because the EPR spectrum of BDPA is narrow, level crossing does not occur during spinning, and therefore, the OE is not subject to depolarization as is the CE.

The OE is usually observed in liquids, conducting solids, or doped semiconductors,^27,28 systems where there is mobility of the spins involved, i.e., where translational fluctuations drive cross-relaxation needed for DNP. More recently, OE DNP was surprisingly observed in insulating solids doped with BDPA.^29–33 Cross-relaxation requires time-dependent modulation of the hyperfine couplings, and the source of the modulation in insulating solids is yet to be definitively established. Recently, it was proposed that intramolecular charge transfer by the mixed-valence nature of BDPA could be responsible for the modulation³⁴ and other mixed-valence polarizing agents that exhibit the OE were recently reported.^32,35 It was also recently suggested that what is generally identified as an OE in BDPA is in fact due to TM;³⁶ however, this topic, while extremely interesting, is not the subject of research results presented here.

The primary topic explored here is the role of the ¹H nuclei on the polarization agent in facilitating DNP, and ultimately contributing to the diffusive spread of polarization throughout the sample. In a previous study, we investigated this role via the proton spins on the fluorene moieties (Scheme 1, top left) in determining OE DNP enhancements using selectively deuterated BDPA variants as the agent.³⁷ We established that the strongly coupled (~5.4 MHz) ¹H spins at the α and γ positions of the fluorenes are essential for generating positive enhancements, while the weakly coupled (~1 MHz) ¹H spins on the β and δ positions are likely crucial for efficient spin diffusion of the polarization from the ¹H spins on the radical to the bulk. We also observed a negative DNP enhancement of the bulk ¹H spins for 1,3-[d₁₆]-BDPA, where all the 16 ¹H positions in the fluorenes were deuterated. This suggested that the ¹H spins on the phenyl ring, which exhibit small hyperfine couplings and have heretofore been considered inconsequential, cumulatively contributed to the negative enhancement observed in the bulk. However, their relative contributions to the OE DNP were unknown.

Scheme 1.

Synthetic Route to Phe-d₅-BDPA Radical

In this study, we determine the relative contributions of the ¹H spins on the phenyl ring and the ¹H spins in immediately surrounding oTP molecules to the overall negative polarization observed in the bulk ¹H spins. This was addressed by performing DNP experiments on a fully deuterated sample, and the data suggested that the ¹H spins on the phenyl ring and ¹H spins in the nearby oTP molecules were negatively polarized via a DQ OE DNP mechanism. Based on this, we used minimal chemistry to synthesize a phenyl-deuterated BDPA radical hypothesizing that it would result in improved positive OE DNP enhancement observed in the bulk. We verified this to be the case with DNP experiments performed at magnetic fields ranging from 5 to 14.1 T. Furthermore, we confirm that OE enhancements do scale with B₀ and that there is no depolarization because of the narrow line width of the BDPA EPR spectrum. For these reasons, it is important to develop further examples of OE radicals for use in DNP MAS experiments at high fields.

SYNTHESIS

Scheme 1 details the synthetic route used to synthesize Phe-d₅-BDPA, which is based on a previous report.³⁷ Synthetic routes to BDPA radicals are well established and Phe-d₅-BDPA can be obtained in four steps from readily available fluorene and benzaldehyde-d₆ or benzaldehyde-d₅. First, condensation of fluorene with one equivalent of benzaldehyde-d₅ delivered 9-((phenyl-d₅)methylene)-9H-fluorene (1) in 75% yield. Compound 1 was subsequently subjected to a two-step process consisting of an alkene bromination with bromine followed by dehydrobromination to obtain 9-(bromo(phenyl-d₅)-methylene)-9H-fluorene (2) in 59% yield. Next, 2 was condensed with another equivalent of fluorene via a nucleophilic addition–elimination sequence to form the neutral precursor Phe-d₅-BDPA-H (3) in 62% yield. Finally, radical formation to obtain Phe-d₅-BDPA was performed by deprotonation of 3, followed by one-electron oxidation of the resulting anion using silver nitrate. This last step was performed with an 80% yield. For further experimental details on the synthesis of Phe-d₅-BDPA radical, we refer the reader to the Supporting Information.

The DNP experiments were performed using four different NMR/DNP spectrometers operating at the following fields, ¹H frequencies, and microwave frequencies for g = 2 electrons: 5.0 T/211 MHz/140 GHz, 8.9 T/380 MHz/250 GHz, 9.4 T/400 MHz/263 GHz, and 14.1 T/600 MHz/395 GHz. The DNP experiments on the 8.9 T/380 MHz/250 GHz spectrometer were performed with microwave irradiation from a home-built 250 GHz gyrotron delivering ~5 W of output power, and a low-power 160 mW solid-state source from Virginia Diodes, Inc. (VDI) (Figure 2a). The low-power microwave system uses an arbitrary waveform generator (Keysight AWG M8190A) and an amplifier multiplier chain (VDI AMC 691) and is capable of producing microwaves over a frequency range of 8 GHz from 246–254 GHz. The details of this instrument will be discussed in detail in a future publication. The gyrotron source of this DNP spectrometer is capable of producing microwaves at 250.02 GHz, which is required to observe the OE DNP in the BDPA samples studied here. The DNP experiments on the 9.4 T/400 MHz/263 GHz spectrometer were performed with microwaves from a 280 mW amplifier multiplier chain (VDI AMC 749). The DNP experiments on the 14.1 T/600 MHz/395 GHz spectrometer were performed with a 30 W gyrotron at a fixed frequency of 395.285 GHz. The NMR magnetic field is swept to obtain DNP Zeeman profiles with a gyrotron, while the microwave frequency is swept at a fixed field with the solid-state sources. All the experiments were performed at a MAS frequency of ω_r/2π = 5 kHz, except for the experiments on the 5 T spectrometer which was performed at a MAS frequency of ω_r/2π = 3.5 kHz. The temperature of the sample was 90 K (5 and 8.9 T) or 100 K (9.4 and 14.1 T).

Figure 2.

(a) DNP field profile obtained at 5 T (b) DNP frequency profile obtained at 8.9 T with a 160 mW low-power microwave source, (c) DNP frequency profile at 9.4 T with a 280 mW low-power microwave source, and (d) DNP field profile at 14.1 T with a gyrotron for h₂₁-BDPA (blue) and Phe-d₅-BDPA (red). The data at 8.9 T was obtained at 90 K, while the data at 9.4 and 14.1 T were obtained at 100 K. The measurements were made on a sample spinning at a MAS frequency of 3.5 kHz (a) or 5 kHz (b–d).

The DNP samples were prepared by doping 2.5 wt % of the radical, h₂₁-BDPA, Phe-d₅-BDPA, or d₂₁-BDPA, all in 1:1 complex with benzene, into a mixture of 95/5 mol % d₁₄- and h₁₄-ortho-terphenyl (oTP).³⁸ In molar concentrations, this corresponds to ~60.0 mM of h₂₁-BDPA, ~ 59.4 mM of Phe-d₅-BDPA, and ~ 57.5 mM of d₂₁-BDPA. h₂₁-BDPA was purchased from Sigma-Aldrich, while the Phe-d₅-BDPA was synthesized as discussed above. The perdeuterated d₂₁-BDPA was prepared using procedures described previously.³⁷ The doped mixture was dissolved in CDCl₃ which was removed by evaporation under a vacuum. The resulting thin film was finely ground and packed in a 4 mm sapphire rotor for experiments on the 5 and 8.9 T DNP spectrometers or a 3.2 mm sapphire rotor for experiments on the 9.4 and 14.1 T DNP spectrometers. The 4 mm sapphire rotors were subsequently warmed in a sand bath maintained at 60 °C for 5 min to melt the oTP and then quenched in liquid nitrogen to generate a glassy matrix. Subsequently, the sample was inserted into the precooled probe. A similar melt–quench procedure was used for the 3.2 mm sapphire rotors; however, a heat gun was used in place of a sand bath.

The NMR experiments involved a train of saturation pulses on the ¹H channel with a Rabi frequency of ω₁/2π = 83 kHz and 3 μs (5 and 8.9 T), ω₁/2π = 152 kHz and 1.65 μs (9.4 T), and ω₁/2π = 100 kHz and 2.5 μs (14.1 T). In all experiments, the interpulse delay of the saturation pulses was 220 μs. The signal was acquired following a recovery period using a solid–echo sequence. The spin–lattice relaxation times T₁ and the DNP buildup times T_B were estimated by fitting mono-exponential buildup curves to the signal intensities with varying recovery periods. The DNP enhancement was determined using ε = (I/I₀) – 1, where I denotes the NMR signal intensity with the microwaves on and I₀ denotes the NMR signal intensity with no microwave irradiation, both obtained at a recovery period of 1.3T₁ or 1.3T_B, whichever is longer.

Figure 1 shows a superposition of the ¹H DNP Zeeman frequency profiles recorded as a function of microwave irradiation frequency for 1,3-[α,β,γ,δ-d₁₆]-BDPA (reproduced from ref 37) and d₂₁-BDPA, the latter newly recorded on the 8.9 T DNP spectrometer with the low-power (160 mW) solid-state source. The d₂₁-BDPA sample was degassed using the procedure employed for the 1,3-[α,β,γ,δ-d₁₆]-BDPA sample with the freeze–pump–thaw protocol reported in our previous study.³⁷ Deuteration of the fluorene moieties in 1,3-[α,β,γ,δ-d₁₆]-BDPA resulted in a negative OE enhancement of ε = −13. However, perdeuteration to yield d₂₁-BDPA resulted in a smaller negative OE enhancement ε = −6.4. These results indicate that the ¹H spins on the phenyl ring are polarized negatively due to dominance of DQ cross-relaxation mechanisms in the e-−¹H systems. The conclusion led us to propose a version of BDPA with only the phenyl ring deuterated so as to quench its negative contributions to the DNP enhancement and obtain a larger positive DNP enhancement than with h₂₁-BDPA.

Figure 1.

DNP frequency profile obtained 8.9 T spectrometer on 1,3, [α,β,γ,δ-d₁₆]-BDPA (blue) and d₂₁-BDPA (red), both spinning at a MAS frequency of 5 kHz and at 90 K. The incident microwave power was 160 mW.

Figure 2 shows a series of DNP frequency/field profiles observed for nondegassed h₂₁-BDPA and Phe-d₅-BDPA on the 5 T DNP spectrometer (Figure 2a), the 8.9 T spectrometer (Figure 2b), the 9.4 T spectrometer (Figure 2c), and the 14.1 T spectrometer (Figure 2d). These data were recorded on nondegassed samples as the 5 and 8.9 T DNP spectrometers utilized a 4 mm rotor, while the 9.4 and 14.1 T spectrometers utilized a 3.2 mm rotor, for which the samples were transported to a remote lab for experiments, where degassing equipment was not available. The OE enhancements, defined by the intensity of the central peak in each profile, for h₂₁-BDPA and Phe-d₅-BDPA samples were respectively 29 and 64 on the 5 T spectrometer, 36 and 67 on the 8.9 T spectrometer (160 mW), 40 and 70 on the 9.4 T spectrometer, and 49 and 88 on the 14.1 T spectrometer. The results demonstrate that Phe-d₅-BDPA consistently exhibits a higher OE DNP enhancement than the h₂₁-BDPA, at least up to a field strength of 14.1 T.

Table 1 summarizes the ¹H spin–lattice relaxation times for the sample doped with 2.5 wt % h₂₁-BDPA and Phe-d₅-BDPA as measured at 8.9 and 14.1 T DNP spectrometers. Table 2 lists the enhancement values measured for the two radicals with an incident microwave power of 2 W at 5 T, with an incident microwave power of 160 mW at 8.9 T, and 4.4 W at 250.020 GHz, with an incident microwave power of 280 mW at 9.4 T and 264.250 GHz, and with an incident microwave power of 30 W at 14.1 T and 395.285 GHz. Using UV–vis absorption measurements, we determined the relative concentrations of active free radicals between the two samples, both prepared at 2.5 wt %, to be 1.0:1.2 for h₂₁-BDPA and Phe-d₅-BDPA, respectively (see Figure S8). The enhancement values adjusted for the relative concentrations are also listed in Table 2. This is done since the commercially available BDPA samples were not 100% active. We also note there is no evidence as yet for linear dependence of the enhancement on concentration.

Table 1.

Experimentally Measured Spin–Lattice Relaxation Times (T₁) and DNP Buildup Times (T_B) for h₂₁-BDPA and Phe-d₅-BDPA at 8.9 T and 14.1 (DNP Spectrometers Described in the Text)

	h21-BDPA		Phe-d5-BDPA
	T1 (s)	TB (s)	T1 (s)	TB (s)
8.9 T	32.1 ± 2.2	37.5 ± 1.5	38.0 ± 0.4	29.6 ± 0.3
14.1 T	31.9 ± 1.5	30.1 ± 1.5	37.1 ± 3.9	28.9 ± 2.0

Table 2.

Experimentally Measured OE DNP Enhancement for 2.5 wt % h₂₁-BDPA and Phe-d₅-BDPA at 5 T with a Gyrotron, at 8.9 T with both a Low-Power Microwave Source and a Gyrotron, at 9.4 T with a Low-Power Source, and 14.1 T with a Gyrotron^a

	h21-BDPA		Phe-d5-BDPA
	measured	adjusted	measured
5 T/2 W	29.2	35.0	63.6
8.9 T/160 mW	36.5	43.8	66.6
8.9 T/4.4 W	40.7	48.8	70.0
9.4 T/280 mW	39.7	47.6	70.3
14.1 T/30 W	49.2	59.0	88.2

^aThe enhancement values adjusted for the concentration difference (1.0:1.2) as measured by UV–vis are listed for h₂₁-BDPA, normalized for Phe-d₅-BDPA

Figure 3a shows the dependence of the measured DNP enhancement as a function of MAS frequency at the 9.4 T DNP spectrometer. The 3.2 mm sapphire rotor at about 100 K could only spin at a maximum of 12.5 kHz. Figure 3b shows the dependence of the measured DNP enhancement as a function of microwave irradiation power at the 9.4 T DNP spectrometer equipped with a solid-state microwave source with a maximum output power of 280 mW (filled circles). There are additional points (empty circles) in the plot showing the DNP enhancements with 4.4 W of irradiation power for the two polarizing agents obtained on the 8.9 T DNP spectrometer.

Figure 3.

Experimental ¹H DNP enhancements of h₂₁-BDPA and Phe-d₅-BDPA as a function of (A) MAS frequency and (B) incident microwave power obtained at the 9.4 T DNP spectrometer with a maximum power of 280 mW. The empty circle and square are data points obtained on the 8.9 T DNP spectrometer with an incident microwave power of 4.4 W from a gyrotron. The data in part a were obtained with an incident microwave power of 280 mW and the data in part b were obtained with a MAS frequency of 5 kHz.

The OE ideally involves a two-spin system of one electron and one nucleus, hyperfine coupled to each other. The nucleus in our discussion is an ¹H spin–either on the BDPA molecule itself or on the surrounding oTP molecules. The excitation of the electron spin transition is followed by unequal cross-relaxation rates in the electron–nuclear two-spin system that can polarize the nuclear spin either positively or negatively. The sign depends on whether the dominant cross-relaxation mechanism is ZQ or DQ, respectively. Cross-relaxation requires time-modulation of the hyperfine coupling and the isotropic component leads to ZQ cross-relaxation while the anisotropic component leads to DQ cross-relaxation. The observed NMR signal is that of the bulk ¹H spins in the oTP matrix after they equilibrate with the directly polarized ¹H spins in and immediately around the radical. Thus, the observed enhancement in the bulk nuclear spins results from the cumulative contribution of both positive and negative contributions originating in the ¹H spins coupled to the electron. In an earlier study, we used the DNP frequency profiles of selectively deuterated BDPA radicals to determine that the strongly coupled ¹H spins on the α,γ sites of the fluorene moieties generate a significant positive OE enhancement. In contrast, the weakly hyperfine coupled ¹H spins on the β,δ sites on the fluorene moieties primarily aid efficient spin diffusion of the enhanced polarization from the α,γ sites to the bulk ¹H spins in the matrix.

Figure 1 illustrates an OE enhancement of ε = −13 for 1,3-[α,β,γ,δ-d₁₆]-BDPA, which we reported in our earlier study.³⁷ This suggests that the ¹H spins on the phenyl ring and the oTP molecules in the surrounding matrix cumulatively make a negative contribution to the overall OE enhancement. However, the relative contributions between the ¹H spins on the phenyl ring and the surrounding oTP molecules to the overall negative DNP enhancement were unknown. To determine this, in this work, we measured the OE enhancement on d₂₁-BDPA and obtained an enhancement of ε = −6.4. The relative concentrations between the two samples–1,3-[α,β,γ,δ-d₁₆]-BDPA and d₂₁-BDPA–were determined using UV–vis spectroscopy to be 1.4:1.0, respectively, and this is attributed to the partial decay (reduction) of the radicals to different levels. Adjusting for the concentrations, the negative DNP enhancement of 1,3-[α,β,γ,δ-d₁₆]-BDPA was about 45% larger than the negative DNP enhancement obtained for d₂₁-BDPA. The result suggests that the ¹H spins on the phenyl ring of the BDPA radical and the ¹H spins on the surrounding oTP molecules were both individually polarized via a dominant DQ cross-relaxation mechanism in their respective electron–nuclear spin systems. That is, both groups of ¹H spins directly contribute to a negative enhancement. This finding leads us to propose that 2H labeling of the phenyl ring in the BDPA radical should lead to higher OE DNP enhancement than observed with the standard h₂₁-BDPA.

Our hypothesis is proven correct, as evidenced by DNP measurements performed in various DNP spectrometers and presented in Figure 2. For example, on the 8.9 T DNP spectrometer with the 160 mW microwave source, Phe-d₅- BDPA results in an OE enhancement of 67, in contrast to an enhancement of 36 observed for the standard h₂₁-BDPA. However, UV–vis measurements determined the relative radical concentration between h₂₁-BDPA and Phe-d₅-BDPA samples to be 1.0:1.2 (Figure S8). Adjusting for this, Phe-d₅-BDPA exhibits an OE enhancement ~ 52% higher than that of h₂₁-BDPA on the 8.9 T spectrometer. Similarly, the increase in enhancement is ~ 82%, ~ 48%, and ~ 50% on the 5, 9.4, and 14.1 T spectrometers, respectively.

Finally, we verified the dependence of the OE DNP enhancements on MAS frequency and microwave irradiation power for both radicals, as illustrated in Figure 3. The experiments were performed on the 9.4 T DNP spectrometer with the low-power microwave source and there is an additional data point in Figure 3B performed with 4.4 W microwave irradiation at the 8.9 T DNP spectrometer under otherwise similar experimental conditions. Both the radicals exhibit increasing enhancement with increasing MAS frequency, while the power dependence appears to saturate with well under 1 W of microwave irradiation, consistent with prior investigations of OE DNP in insulating solids.^30,32,39 This is also encouraging for use of compact, lower-cost, low-power solid-state microwave sources.

Based on prior DNP data on selectively deuterated BDPA radicals and newly obtained data on a perdeuterated BDPA radical, we concluded that the ¹H spins on the phenyl ring of the BDPA radical participate in OE DNP primarily through DQ cross-relaxation, contributing a negative enhancement. We demonstrated that deuteration of the phenyl ring in the BDPA radical improves the overall positive OE enhancement compared to the standard h₂₁-BDPA by removing contributions to DQ cross-relaxation. We verified the hypothesis by synthesizing the newly proposed Phe-d₅-BDPA radical and examining its DNP properties in several spectrometers. In the magnetic field range of 5–14.1 T, the Phe-d₅-BDPA polarizing agent improves the OE DNP enhancement by about 50% when adjusted for the concentrations of active radicals in the sample, compared to the standard h₂₁-BDPA in the oTP matrix, and has achieved a maximum enhancement of ~ 90 at 14.1 T at a MAS frequency of 5 kHz at 100 K–the highest OE at these conditions to date for a nondegassed sample–and is expected to improve further by ~ 70% with freeze–pump–thaw degassing as demonstrated in our earlier study.³⁷ These results offer insight into designing new or modifying existing polarizing agents to exhibit amplified OE DNP. Considering the improvement of OE with stronger magnetic field strengths as shown here, in stark contrast to other CW-DNP mechanisms, the results discussed are of prime importance to achieve efficient sensitivity enhancements in high-resolution solid-state NMR of biological macromolecules. The work presented here also underscores the rationale for pursuing low-power solid-state microwave sources for use at high magnetic field strengths.