High-Field Solid-State NMR with Dynamic Nuclear Polarization
Microwave-induced dynamic nuclear polarization (DNP) can produce hyperpolarization of nuclear spins, leading to substantial signal enhancement in NMR. This chapter discusses the contemporary application of DNP for solid-state NMR spectroscopy at high magnetic fields. The main mechanisms and polarizing agents that enable this hyperpolarization are presented, along with more practical aspects such as the effect of decreasing sample temperature and analyzing the absolute sensitivity gain from these experiments. Examples of the exploitation of DNP for studies of biomolecules, biominerals, pharmaceuticals, self-assembled organic nanostructures, and mesoporous materials are given as is an outlook as to the future of this powerful technique.
KeywordsDynamic nuclear polarization Solid-state NMR Magnetic field Hyperpolarization Magic angle spinning Signal enhancement Polarizing agents Absolute sensitivity Microwave Biomolecules Materials
Basic Theory and Methodology
There are many ways to achieve a polarization transfer between electron spins and nuclear spins, and the theory of the key contemporary mechanisms has been comprehensively discussed . For the case of MAS-DNP, the theory is being constantly updated. In particular, an Overhauser effect (OE) has been observed for insulating solids , a scenario previously deemed not possible, and also the specific function of the MAS for cross-effect (CE) [7, 8] and solid-effect (SE)  DNP has also been detailed. Simply, the OE originates from the presence of a cross-relaxation process, zero (ZQ) or double quantum (DQ), resulting from a fluctuating hyperfine coupling (between an electron and a nuclear spin). When the difference between ZQ and DQ cross-relaxation rates is sufficient (and allows an efficient redistribution of populations), the microwave (μw)-induced saturation of the electron Zeeman transition results in an enhancement of the nuclear polarization. Therefore, the OE relies on the relative relaxation rates in the system. Contrastingly the SE, which also involves a coupling between an unpaired electron and a nuclear spin, is driven by a single-quantum (SQ) relaxation process and the ability to saturate a resolved “forbidden” ZQ or DQ transition, in order to enhance the nuclear polarization. Therefore, the SE relies on the applied μw power and the spin relaxation rates. The CE, however, requires the presence of three coupled spins (two electrons and one nucleus). The corresponding mechanism is significantly different under static and sample spinning conditions. With MAS, the (fixed-frequency) μw irradiation is used to periodically perturb the polarization of one electron spin (at a time) and generate a large electron polarization difference (compared to the nuclear Boltzmann polarization). The difference in polarization between the two coupled electrons can be transferred to the surrounding nuclei when the CE condition is fulfilled: the difference in electron resonance frequencies matches the nucleus' resonant frequency. Such a MAS-DNP effect does then not require the μw and CE conditions to occur simultaneously. The detailed description and understanding of MAS-DNP processes has been enabled due to efficient quantum calculations, which are continuously being improved and currently allow their implementation on large spin systems [9, 10]. At this point, it is also important to note that in the absence of μw irradiation and under MAS, the periodic repetition of CE matching conditions can induce a perturbation of the nuclear polarization. In some cases, this can lead to a reduction of the detectable nuclear polarization (compared to Boltzmann equilibrium), creating a depolarized nuclear spin state [9, 11].
The OE, SE, and CE all permit the nuclear spin system to evolve to an equilibrium (non-Boltzmann) polarization state. The characteristic time constant for the spin system to reach this equilibrium state via DNP (commonly labeled T B) is then different to the nuclear spin-lattice relaxation time constant, T 1. Subsequent to this hyperpolarization buildup, ssNMR experiments are performed in their usual manner (i.e., no radio-frequency pulse sequence modifications need to be made). Owing to its (relatively) consistently high hyperpolarization efficiency and ease of implementation, the CE is the most commonly used mechanism for high-field MAS-DNP. There are alternative methods for DNP; however these are only currently practicable on static samples at low magnetic field .
Exogenous radicals like those in Fig. 2 are much more frequently used as polarizing agents compared to endogenous or intrinsic radicals because it is easier to chemically control the characteristics of their electron spins, such as concentration, g-tensor anisotropy, couplings, and relaxation properties, leading to more efficient DNP, and also because most NMR samples do not inherently contain unpaired electrons. Nevertheless, high-field ssNMR studies using endogenous DNP have been performed, both for biological systems  and for materials , containing intrinsic paramagnetic sites. The returned signal gains from DNP were far below what could be achieved with exogenous polarizing agents. However, for “directed” (locally selective) DNP, for studies where one does not want to/cannot modify the sample with the addition of solvents or polarizing agents, and/or with improved technology (such as access to lower sample temperatures), there is a growing interest for DNP using endogenous radicals.
It is important to highlight that at 55 K 1Hεon/off = 677, which is a larger value than that given by the ratio of the gyromagnetic ratios of the electron and nucleus, γe/γH-1 = 658, the theoretical maximum nuclear polarization gain in this case. Since nuclei can have an equilibrium polarization that is less than their expected Boltzmann polarization due to depolarization in the presence of the CE but absence of μw irradiation (vide supra), then the ratio of equilibrium polarizations of electrons and protons can be >658. Therefore, the absolute polarization gain obtained with DNP should be determined using a comparison to the Boltzmann polarization of the nuclei, and εon/off does not represent this gain for CE DNP and can thus give misleading values and interpretations.
We introduced the absolute sensitivity ratio (ASR)  as a way to encompass all the positive and negative effects that come with performing high-field DNP-enhanced ssNMR experiments and to thus give a good evaluation of the pertinence of performing DNP when compared to conventional ssNMR. Other methods to rationalize this gain have also been proposed , and these rely on the multiplication of different contributing factors. The ASR encompasses all of the effects because it is an experimentally measured value. The sensitivity (signal-to-noise ratio returned per unit square root of time) is compared between spectra from an analyte recorded under DNP conditions and those commonly employed for conventional ssNMR. Therefore, the ASR takes into account such factors as the DNP enhancement, the temperature change (thermal polarization, thermal noise), spin relaxation and coherence lifetimes (temperature and paramagnetic effects), linewidth changes (vide infra), usable sample mass (paramagnetic bleaching, sample dilution), and the availability of advanced equipment (larger magnetic fields, smaller/larger MAS rotors). Methods to optimize the ASR, including various ways to prepare samples for DNP experiments, have been discussed elsewhere . Figure 3b shows the effect of decreasing the sample temperature on the ASR of nanotubes of cyclo-diphenylalanine prepared for CE DNP. Already by only using the more customary conditions for DNP-enhanced ssNMR (9 T, 108 K), an ASR of 960 could be obtained. However, by reducing the sample temperature to 50 K, the ASR increased by a factor of 5.6. This means that experiments could be recorded an additional (5.62 =) 31 times faster. Compared to conventional ssNMR (9 T, 298 K, no paramagnets), an experiment that would have taken 55 years could now be recorded in 1 min!
With the large achievable sensitivity gains, attention has also been directed at applications of high-field DNP to study systems of biological interest. The biological ssNMR community has always strived for better resolution and sensitivity, particularly pushing the development of higher magnetic fields and faster MAS rates. So the successful establishment of high-field ssNMR with DNP was markedly appealing. Consequently, this technique has been applied to study peptides, proteins, amyloid fibrils, membrane complexes, and even entire intact cells . Unlike solution-state NMR and X-ray crystallography, ssNMR has the potential to study large biomacromolecules that lack long-range order at the atomic scale. Moreover, for those systems that do exhibit long-range order, it is not always straightforward to produce sufficient crystals for X-ray analysis, or sometimes these crystals do not represent the molecule’s structure in its native biological context. In these cases, ssNMR spectroscopy is then the technique of choice, provided adequate resolution and sensitivity is attainable. Indeed, it has been recently shown that the structure of a particular protein at its physiological concentration in a biological environment (of cell lysates) is different to that of the purified system . This study was only possible with the use of high-field DNP. The intrinsic sensitivity of ~1 μM (isotopically-labeled) protein would be too low for the necessary 2D experiments with conventional ssNMR.
To increase resolution for biological studies, which are usually complicated by many overlapping resonances, the highest magnetic fields possible are generally desired. To date, the highest static, continuous field for ssNMR is 24 T, and DNP combined with ssNMR is commercially available up to 19 T. However, the efficiency of the SE and CE DNP mechanisms generally decreases with increasing field strength, with relationships expected to be proportional to B 0 −2 and B 0 −1, respectively. To circumnavigate this Catch-22 for high-resolution and high-sensitivity biological ssNMR with DNP, further technological and methodological advancements still need to be made; the use of the OE  or trityl-nitroxide biradicals  seem promising. Additionally, DNP enhancements, 1Hεon/off, on real systems are usually much inferior to the maximum currently obtained for small molecule model systems – signal gains depend heavily on the nature of the sample and its preparation – and this is obviously then especially important at 19 T. As such, values at 19 T for 1Hεon/off of 8 (100 at 9 T) and 15 (60 at 9 T) have been obtained for a membrane-embedded potassium channel protein  and a cell-embedded megadalton protein complex , respectively; model systems (small amino acids) prepared in a similar manner would usually return 1Hεon/off ~ 35. The latter study involved probing the large protein complex in its native environment. This study showed that the 3D fold seen in in vitro crystals is also present in the native cellular preparation. For the former experiment on the membrane-embedded potassium channel, the low-temperature conditions and addition of paramagnetic polarizing agent were useful not just for the DNP, which facilitated an ASR of 2.6 (meaning an experiment that would usually take 1 week could be recorded in only 1 day), but they also provided further information. The exogenous polarizing agent was used to highlight solvent exposed regions of the protein through paramagnet-induced signal modulations of those regions. Additionally, highly dynamic areas of the channel could be detected through increased linewidths of 13C resonances from certain residues at low temperature compared to ambient temperature measurements.
One way to lessen resolution issues is specific isotopic labeling, as this will decrease the number of overlapping resonances, and, as such, this has been comprehensively employed for conventional NMR of biomolecules . Combining labeling schemes with DNP can therefore be pivotal and has been used, for instance, to investigate the interface, and thus interactions, between proteins in oligomeric complexes within a lipid bilayer . This interface is relatively small and thus would normally be undetectable without the sensitivity gains afforded by DNP. Specific isotopic replacement of 1H with 2H has also been used. Although reducing the 1H concentration may reduce the efficiency of 1H–1H spin-diffusion to distribute the hyperpolarization to all the 1H in the system, it has been shown that it can improve CE DNP, leading to larger 1Hεon/off for biomolecules [32, 34]. However, the reduced 1H content can also impact the efficacy of CP, meaning that there may actually be no overall net gain in sensitivity (ASR) .
It is interesting to note that studies of biomolecules can be made using high-field ssNMR with OE DNP at ambient temperature, thereby maintaining molecular mobility . This premise shows much potential. However, it has only so far been demonstrated with singly tunable radio-frequency NMR circuits, with restricted amounts of (static) sample, and at magnetic fields strengths of 9 T. Large technical advances are required to make this interesting approach widely applicable.
Applications of high-field ssNMR with DNP to materials ranging from organic to inorganic, with hybrids in between, have been plentiful in recent years , with particular interest paid to surfaces . Unlike biomolecules, isotopic labeling of (non-bio)materials can be challenging or expensive (or both). Therefore, working at NA, if possible, can be highly advantageous. This was shown (vide supra) to be now possible for 43Ca (NA = 0.14%) with the sensitivity gain provided by CE DNP. Furthermore, high-field CE DNP has already made NA ssNMR of 15N (NA = 0.37%) , 17O (NA = 0.04%)  as well as 2H (NA = 0.01%)  readily accessible.
Along with the examples of 43Ca and 17O, high-field DNP has provided access to fast ssNMR studies of other quadrupolar nuclei, such as 35Cl , 27Al [47, 48], and 14N . It has thus opened the flood gates to a whole range of studies that were previously inaccessible. This is especially important because quadrupolar nuclei account for >¾ of all NMR-active isotopes and are present in most materials. Taking the technique further still, as for the case of biomolecules, will require efficient DNP at even higher fields. Here, studies of quadrupolar nuclei could benefit greatly since their observed ssNMR linewidth, when dominated by their quadrupolar interaction (as is usually the case), is inversely proportional to the field strength. However, as stated above, there is still much work to be done to realize efficient DNP at ≥19 T.
Conclusions and Outlook
There is a current revolution in solid-state NMR, and it owes a lot to the commercialization of an experimental setup that provides low-temperature MAS NMR along with suitable continuous microwave irradiation of the sample, which followed on from the groundbreaking work of Griffin and co-workers. The experimental setup is under constant development, with access to lower sample temperatures using cold helium gas in combination with MAS and opportunities arising from alternative microwave sources such as those that are pulse, phase, and frequency adjustable, providing large interest. Furthermore, the introduction of NMR probes capable of rapid sample spinning using smaller diameter rotors (currently down to 1.3 mm) at low temperatures advances the use of higher-field DNP (currently up to 19 T)  and also permits experiments in a new regime.
It has been demonstrated herein that there has been much progress in high-field ssNMR with DNP, which has resulted in great sensitivity gains. These have not just facilitated, but in numerous cases enabled, intricate experiments on a variety of systems ranging from biomolecules to inorganic-organic hybrid nanomaterials. For biomolecular studies, DNP has, for example, permitted the comparison of a protein in its natural biological context and as a purified system, showing that the structure differed depending on the environment. For inorganic-organic hybrid nanomaterials, DNP has allowed, for example, studies of the surface grafting of functionalizing moieties, showing that lateral and vertical functionalization could be differentiated.
Although high-field ssNMR with DNP is fast becoming a routine atomic-level characterization tool, substantial advancements can still be made. Spectral resolution, sample preparation, ultra-fast MAS, and sensitivity could all still benefit further from methodological and technological improvements. Therefore, this technique, which is already showing broad and potent employment, can only become more useful, ubiquitous, and influential.
- 3.Wind RA. Dynamic nuclear polarization and high-resolution NMR of solids. In: Harris RK, Wasylishen RL, editors. Encyclopedia of magnetic resonance. Chichester: Wiley; 2007. p. 1–10.Google Scholar
- 10.Mentink-Vigier F, Vega S, De Paëpe G. Large spin ensembles for MAS-DNP. Phys Chem Chem Phys. 2017;19:3506–22.Google Scholar
- 13.Sauvée C, Casano G, Abel S, Rockenbauer A, Akhmetzyanov D, Karoui H, Siri D, Aussenac F, Maas W, Weber RT, Prisner T, Rosay M, Tordo P, Ouari O. Tailoring of polarizing agents in the bTurea series for cross-effect dynamic nuclear polarization in aqueous media. Chem A Eur J. 2016;22:5598–606.CrossRefGoogle Scholar
- 17.Riikonen J, Rigolet S, Marichal C, Aussenac F, Lalevée J, Morlet-Savary F, Fioux P, Dietlin C, Bonne M, Lebeau B, Lehto V-P. Endogenous stable radicals for characterization of thermally carbonized porous silicon by solid-state dynamic nuclear polarization 13C NMR. J Phys Chem C. 2015;119:19272–8.CrossRefGoogle Scholar
- 24.Bouleau E, Saint-Bonnet P, Mentink-Vigier F, Takahashi H, Jacquot J-F, Bardet M, Aussenac F, Purea A, Engelke F, Hediger S, Lee D, De Paëpe G. Pushing NMR sensitivity limits using dynamic nuclear polarization with closed-loop cryogenic helium sample spinning. Chem Sci. 2015;6:6806–12.CrossRefGoogle Scholar
- 33.Maciejko J, Mehler M, Kaur J, Lieblein T, Morgner N, Ouari O, Tordo P, Becker-Baldus J, Glaubitz C. Visualizing specific cross-protomer interactions in the homo-oligomeric membrane protein proteorhodopsin by dynamic-nuclear-polarization-enhanced solid-state NMR. J Am Chem Soc. 2015;137:9032–43.CrossRefGoogle Scholar
- 36.Lee D, Leroy C, Crevant C, Bonhomme-Coury L, Babonneau F, Laurencin D, Bonhomme C, De Paëpe G. Interfacial Ca2+ environments in nanocrystalline apatites revealed by dynamic nuclear polarization enhanced 43Ca NMR spectroscopy. Nat Commun. 2017;8:14104.Google Scholar
- 38.Rossini AJ, Zagdoun A, Lelli M, Canivet J, Aguado S, Ouari O, Tordo P, Rosay M, Maas WE, Copéret C, Farrusseng D, Emsley L, Lesage A. Dynamic nuclear polarization enhanced solid-state NMR spectroscopy of functionalized metal-organic frameworks. Angew Chem Int Ed. 2012;51:123–7.CrossRefGoogle Scholar
- 47.Vitzthum V, Miéville P, Carnevale D, Caporini M a, Gajan D, Copéret C, Lelli M, Zagdoun A, Rossini AJ, Lesage A, Emsley L, Bodenhausen G. Dynamic nuclear polarization of quadrupolar nuclei using cross polarization from protons: surface-enhanced aluminium-27 NMR. Chem Commun (Camb). 2012;48:1988–90.CrossRefGoogle Scholar