Dynamic Nuclear Polarization Efficiency Increased by Very Fast Magic Angle Spinning.

Dynamic nuclear polarization (DNP) has recently emerged as a tool to enhance the sensitivity of solid-state NMR experiments. However, so far high enhancements (>100) are limited to relatively low magnetic fields, and DNP at fields higher than 9.4 T significantly drops in efficiency. Here we report solid-state Overhauser effect DNP enhancements of over 100 at 18.8 T. This is achieved through the unexpected discovery that enhancements increase rapidly with increasing magic angle spinning (MAS) rates. The measurements are made using 1,3-bisdiphenylene-2-phenylallyl dissolved in o-terphenyl at 40 kHz MAS. We introduce a source-sink diffusion model for polarization transfer which is capable of explaining the experimental observations. The advantage of this approach is demonstrated on mesoporous alumina with the acquisition of well-resolved DNP surface-enhanced 27Al cross-polarization spectra.

Dynamic nuclear polarization (DNP) has recently developed into a promising approach to greatly increase the sensitivity of solid-state NMR spectroscopy.[1−3] DNP relies on microwave (μw)-driven polarization transfer from unpaired electrons to nearby nuclei and currently enables 1H signal enhancements of more than 2 orders of magnitude at temperatures from 20 to 100 K and magnetic fields of up to 9.4 T,[4−10] and still relevant enhancements down to 4 K[11] or up to room temperature.[12] This has opened up new avenues for characterization in materials and life sciences, for example, for the detailed characterization of active sites on functional surfaces, for the elucidation of drug formulations or for the structural investigation of complex biological assemblies.[13−20]

Unpaired electrons as hyperpolarization sources are usually introduced into the sample as exogenous organic free radicals in a glass-forming solution.[1,2] For magic angle spinning (MAS) DNP, the most efficient polarization transfer mechanism today is the cross effect (CE), which is effective in particular for binitroxide radicals. Recent progress in the design of binitroxides has led to the introduction of AMUPol[21,22] and TEKPol[7,23,24] like biradicals, which lead to proton enhancements of up to 250 at 9.4 T (400 MHz/263 GHz), 100 K, and 8 kHz MAS frequency.

However, today high magnetic fields (>14.1 T) and fast MAS (>15 kHz) are prerequisites for most applications in life sciences and materials. Unfortunately, the efficiency of the CE is predicted to scale with the inverse of the magnetic field strength.[25,26] The observed drop in the enhancements with magnetic field is more than linear;[27] e.g., enhancements of 10 mM AMUPol in a glycerol/water matrix are reduced to 130 at 14.1 T (600 MHz/395 GHz) and to 35 at 18.8 T (800 MHz/527 GHz).[21,28] Furthermore, binitroxide DNP efficiency also decreases at faster MAS rates, since the MAS modulation of the two electron frequencies results in depolarization processes and significant loss of overall observable nuclear spin polarization.[26,29−32]

Can et al. showed that, in contrast, the Overhauser effect (OE) has a favorable field dependence from 9.4 to 18.8 T.[33] They observed significant OE DNP using 1,3-bisdiphenylene-2-phenylallyl (BDPA) in polystyrene, and sulfonated BDPA in a glycerol/water matrix. At 18.8 T and 8 kHz MAS, we observed that BDPA in o-terphenyl (OTP) yields OE enhancements of up to 80, the largest enhancement observed so far at this field.[12] The favorable field dependence of BDPA has recently been rationalized by fluctuations of the hyperfine interaction.[34]

Here we report solid-state OE DNP enhancements of over 100 at 18.8 T using BDPA in OTP at 40 kHz MAS. This is achieved through the unexpected discovery that enhancements increase rapidly with increasing MAS rates. Similar behavior is observed at 9.4 T and in 1,1,2,2-tetrachloroethane (TCE). We introduce a source–sink diffusion model for polarization transfer that is capable of explaining the experimental observations. The advantage of the approach is demonstrated with well-resolved DNP surface-enhanced 27Al spectra on mesoporous alumina.

Figure a and Table S4 give the 1H OE DNP enhancements (εH) of 60 mM BDPA in partially deuterated OTP (d14-OTP/OTP, 95/5 w/w) as a function of the sample spinning frequency. The experiments were recorded using a 1.3 mm Bruker MAS DNP probe at 18.8 T and ∼128 K (see SI for details). At 0.5 kHz MAS, an enhancement of 23 was observed, which increases with spinning rate. At 40 kHz, the maximum spinning frequency attainable with the probe, an enhancement of 105 was obtained. We note that at 10 kHz MAS εH = 74, in agreement with previously reported values (80) using a 3.2 mm rotor under otherwise similar conditions.[12]Figure a also reports 1H DNP build-up times TB,ON,[35] which significantly increase with the spinning frequency. As previously reported,[12] the TB,ON are relatively long for BDPA in OTP. Experiments recorded at 9.4 T show a similar trend with spinning frequency, with maximum εH ≈ 90 (Table S5, Figures S8 and S9). Figures S10 and S11 report εH and TB,ON as a function of the MAS frequency for 60 mM BDPA in TCE. Again, we observed that εH and TB,ON increase with increasing MAS rate, with εH = 19.4 at 9.4 T and εH = 23 at 18.8 T, both at 40 kHz spinning frequency. In all these experiments, εH and TB,ON show an almost linear correlation (Figure S12). Similar trends were observed in a completely degassed TCE solution (Figure S13), and for the solid effect (SE) of BDPA in OTP (Figure S14).

Figure 1

a) MAS frequency dependence of the 1H OE DNP enhancements (red) and of 1H TB,ON (blue) at 18.8 T for the solvent resonance of 60 mM BDPA in 95% d14-OTP and 5% OTP (w/w). b) Contribution factor as a function of MAS rate.

Figure b shows the contribution factor θ, which is defined as the ratio of the signal intensities per unit of mass recorded in the absence of μw irradiation for solutions with and without the polarizing agent (eq 1 in the SI).[36] It accounts for signal reduction in the presence of paramagnetic species without μw irradiation, due to paramagnetic effects (quenching, depolarization, anisotropy of the hyperfine interactions, etc.) induced by the radical. A high contribution factor close to unity is favorable for a high overall sensitivity gain (eq 2 in the SI).[29,30,36,37] Here, we find that the contribution factor remains almost constant at around 1.0 over the whole spinning frequency range. Similar data were obtained at 9.4 T (Figure S15a). This behavior is in contrast to that of the CE in binitroxides, where θ decreases significantly with increasing MAS frequency due to depolarization, e.g., to 0.46 at 40 kHz MAS with 10 mM AMUPol in glycerol/water.[28] A stable high contribution factor is in line with expectations, since BDPA is not an efficient paramagnetic relaxation agent (confirmed by the long TB,ON) and since depolarization processes are absent in isotropic narrow-line radicals.

Even though the TB,ON values of the BDPA/OTP system are longer than in most binitroxide radicals, in the bulk radical solution the overall sensitivity at 18.8 T of BDPA/OTP is higher than for 10 mM AMUPol in a glycerol/water matrix under otherwise similar conditions (see Table S3).

If we assume that the intrinsic efficiency of Overhauser DNP does not depend significantly on MAS rate in this spinning regime, then the spinning speed dependence of εH and TB,ON observed here can be explained in terms of 1H spin diffusion dynamics. Spin diffusion models have previously shown that the enhancement depends on the polarization time if the spin diffusion rate from the radical source to the bulk of the sample is the rate-determining step.[13,35,38] A detailed description for different cases is given in ref (35). We observe such behavior in the BDPA/OTP system (Figure S16), suggesting that the dependence of εH and TB,ON on the spinning frequency could be modeled in a similar fashion. The spinning speed behavior would then be due to progressive isolation of the paramagnetic relaxation sinks as the spin diffusion constant is reduced with increasing spinning rates.[39,40] If the BDPA is the relaxation sink (as is often assumed in models for polarization of frozen solutions of binitroxides[35]), then we cannot reproduce an increase of the enhancement with faster spinning (see Figure S17) at the radical concentrations relevant here (tens of mM). However, as mentioned above, BDPA is not an efficient relaxing agent, and so we extend the previous models to include separate polarization sources and relaxation sinks (Figure a). The sinks could be traces of paramagnetic impurities or of dissolved oxygen,[39,41,42] surrounded by a sphere in which the longitudinal relaxation times are short. A detailed explanation of the model and the simulations is given in the SI. Within this hypothesis, we were able to successfully reproduce the experimentally observed 1H OE enhancements and build up times as a function of the MAS rate (Figure b). If the model is correct, it shows that even a very low concentration of impurities (here 14 μM) can generate a MAS dependence in the DNP enhancements. Fast sample spinning decouples the relaxation sinks from the bulk solution, which increases the overall apparent build-up time constant, and in turn leads to more efficient propagation of the hyperpolarization from the BDPA sources. The model also predicts that εH increases with increasing polarization time, which is expected for a system having two domains, one being a polarized domain and the other being a relaxation sink with longer and shorter relaxation times, respectively,[13,35] and this agrees with experiment (Figure S16). We note that the source–sink model is in principle general, and a possible further increase in the enhancement is expected at higher MAS frequencies. Recently, we observed a MAS dependence of the TB,ON for AMUPol in a water/glycerol matrix, though the observed effect is much smaller due to shorter TB,RS.[28]

Figure 2

a) Schematic model for spin diffusion in the sample: BDPA molecules are assumed to be randomly distributed in an OTP matrix (pink). The diffusion constant D depends on the MAS rate. Relaxation sinks (green) are included with a short T1 at low concentration (here 14 μM), leading to a mean distance between the sinks of 60 nm. b) Comparison of experimental and simulated results at 18.8 T. TB,RS is the modeled build-up time of the BDPA/OTP solution in the absence of sinks. In the fit TB,RS is taken to be 73 s, T1,sink = 1 ms, ε0 = 260 and D varies from 1.54 × 10–4 μm2/s at 0.5 kHz MAS to 0.080 × 10–4 μm2/s at 40 kHz. The same parameters were used for fitting both the enhancement and TB,ON. Details in the SI.

The practicability of this new approach and the importance of high enhancements at high fields is illustrated here in the context of DNP surface-enhanced NMR spectroscopy (DNP SENS),[2,3] where quadrupolar nuclei at surfaces are of particularly high interest, playing key roles in many functional inorganic or hybrid materials. Several 27Al spectra have been reported in this context on a variety of mesoporous and nanoparticulate systems.[43−46] In this case, spectral resolution is limited by second-order quadrupolar shifts and broadenings, and resolution and sensitivity increase with B02. In parallel fast MAS separates spinning sidebands from the isotropic shift range and provides spectra from which isotropic chemical shifts, quadrupolar couplings and asymmetry parameters can be reliably extracted. DNP at high magnetic fields and fast MAS is thus particularly attractive for surface quadrupolar NMR.

Figure shows 27Al DNP-SENS spectra of mesoporous alumina at 18.8 T with MAS at 10 and 40 kHz. The sample was prepared by impregnation with 40 mM BDPA in 95% d14-OTP and 5% OTP (see SI for details). OE enhancements of εH(solvent) = 60 and ε = 16 were measured at 40 kHz MAS on the solvent and on the 27Al resonances, respectively (Figure S18). At 10 kHz MAS, significantly lower values were obtained (εH(solvent) = 37 and ε = 8), as expected from above. We note that, as expected, DNP at fast MAS yields high sensitivity spectra free of spinning sidebands and spectral distortions. The quadrupolar parameters extracted from the spectrum are given in Table S6.

Figure 3

OE DNP-enhanced 27Al cross-polarization (CP) MAS spectra of mesoporous alumina at 18.8 T at (a) 40 kHz and (b) 10 kHz MAS.

In conclusion, we have obtained solid-state Overhauser effect DNP enhancements of over 100 at 18.8 T using BDPA in OTP at 40 kHz MAS. This is, to the best of our knowledge, the highest εH achieved so far at this magnetic field. We found that enhancements increase rapidly with increasing MAS rates. This is consistent with a theoretical description using a source–sink diffusion model. The study shows that BDPA in OTP forms an efficient polarizing matrix for DNP at fast MAS, as exemplified on a model sample of mesoporous alumina. The BDPA/OTP pair at fast MAS will open new perspectives, notably for surfaces which may react with TEMPO-based radicals, and should stimulate the development of other narrow line radicals for OE DNP.