Chinese Chemical Letters  2018, Vol. 29 Issue (6): 747-751   PDF    
Recent progress in investigations of surface structure and properties of solid oxide materials with nuclear magnetic resonance spectroscopy
Jia-Huan Du, Luming Peng    
Key Laboratory of Mesoscopic Chemistry of Ministry of Education, School of Chemistry and Chemical Engineering, Nanjing University, Nanjing 210023, China
Abstract: Solid oxide materials have widespread applications which are often associated with their surface structure and properties. Solid-state nuclear magnetic resonance (NMR) spectroscopy is one of the most powerful methods that give detailed local structural information of solid materials. Recent developments in dynamic nuclear polarization (DNP) NMR spectroscopy and 17O surface-selective isotopic labeling provide more opportunities in investigations of surface structure and properties of oxide materials. We describe in this review some of the latest progress in this field. DNP NMR can enhance the sensitivity of surface sites on the oxides by one to two order of magnitude, making very low concentrated species on the surface of oxides visible in NMR spectroscopy. On the basis of surface-selective 17O isotopic labeling, 17O NMR spectroscopy is now able to distinguish surface oxygen species on the different facets or different surface layers in oxide nanostructures. The nature of these facets can also be probed with help of 31P NMR spectroscopy along with phosphorous-containing probe molecules.
Key words: NMR     Oxide     Surface     17O     DNP    
1. Introduction

A variety of technologically important solid oxide materials are extensively used in catalysis [1], energy storage [2], and environmental management [3]. In many cases, their surface structure and physicochemical properties play an important role in controlling the performances in these applications. Recent advances in nanotechnology show that the detailed surface structure of nanomaterials, such as the exposed facets of oxide nanocrystals, have enormous impacts on the catalytic activity [4]. Therefore, such information is required in order to extract the structure–property relationships, rationally design and finally generate new oxide materials with well-defined surface structure and improved properties.

Diffraction methods, which provide long range order information, represent the most extensively used methods in characterizing the crystal structure of solids.However, the surface of solids often has a disordered nature and lacks the structural regularity required by diffraction approaches. On the contrary, solid state nuclearmagnetic resonance (NMR) spectroscopy is a versatile tool which gives the quantitative local structural information, as well as dynamics of solidsatatomic/molecularlevel.Hence, inprinciple, solid-stateNMR should be an ideal method for characterization of surface of oxides, in which only short range order is present. The low concentration of surface species, as well as the low sensitivity of NMR spectroscopy, however, hindered the applications of solid-state NMR spectroscopy for characterizing the surface of solids. For oxide materials, in particular, applying 17O NMR spectroscopy as a general approach to study the surface meets with the difficulty that 17O, the only NMRactive stable oxygen isotope, has a very low natural abundance of 0.037% and thus isotopic labeling is necessary even for the studies of the bulk structure of oxides. Therefore, in previous studies, NMR spectroscopy was often applied with probe molecules to study surface properties of oxygen containing materials, such as acidic catalysts, as summarized in two recent review papers [5, 6]. Only with the recent developments in dynamic nuclear polarization (DNP) involving transferring polarization from unpaired electrons to nuclei, as well as surface-selective 17O isotopic labeling, there come new opportunities in characterizing the surface structure and properties of solid oxide materials.

This review summarizes important research progress in the studies of surface structure and properties of oxide materials with solid state NMR spectroscopy in the past 5-6 years. Specifically, we focus on the developments of the following approaches for oxides: 17O NMR combined with surface-selective labeling and DNP surface enhanced NMR spectroscopy (DNP-SENS). New results obtained by using 31P NMR along with phosphorus probe molecules will also be discussed.

2. Surface structure and physicochemical properties of nanostructured oxides 2.1. 17O NMR studies

17O, the only NMR active oxygen stable isotope, is a spin-5/2 quadrupolar nucleus with a moderate quadrupole moment and a large chemical shift range of more than 1000 ppm, making it a sensitive structural probe for oxide materials. Although 17O solidstate NMR spectroscopy has been applied to study the oxygen environments in the bulk part of simple and complex oxides [7], few papers have been published on oxide nanostructures, in which the surface and subsurface are a significant fraction of the whole material. Naturally one would think that 17O NMR spectroscopy is the most direct and general approach to investigate the surface of oxide nanomaterials, without applying additional probe molecules such as trimethylphosphine (TMP) that may alter the surface properties (vide infra). One of the major reason behind is that the commonly used high temperature 17O isotopic enrichment can cause sintering of the nanomaterials.

Wang and co-workers prepared ceria nanoparticles as small as ~3 nm and heated the sample in 17O2 at a relatively low temperature (523 K) for 17O enrichment. The sizes of nanoparticles did not change much and they successfully observed three new resonances centered at 1040, 920 and 825 ppm, in the 17O magic angle spinning (MAS) NMR spectra, in addition to the major peak at 877 ppm, which is known due to oxygen ions in the bulk part of the sample (OCe4) (Fig. 1) [8]. The intensities of these new peaks increase with decreasing particle sizes, indicating that these peaks originate from surface species. With the help of density functional theory (DFT) calculations on the NMR chemical shifts, these new peaks at 1040, 920 and 825 ppm are assigned to the oxygen species in the 1st, 2nd and 3rd layers of ceria {111} facets, respectively. This demonstrates that 17O NMR chemical shift is such a sensitive indicator that it can be used to distinguish the oxygen ions at different layers on the surface of oxides.

Download:
Fig. 1. Solid state NMR spectra of nanosized ceria in comparison with DFT calculations. (A) 17O NMR spectra for ceria nanoparticles enriched at different temperatures and external fields compared with the 17O NMR spectra of micronsized ceria ("bulk" ceria) and the summary of the chemical shifts predicted using a structural model shown in b. The spectra obtained at 14.1 T and 9.4 T were acquired with spinning speeds of 55 kHz and 20 kHz, respectively. Short pulse lengths of 0.1 ~0.4 μs corresponding to π/72 ~π/18 pulses for H217O and optimized recycle delays from 1 ~100 s to ensure quantitative observations of all the resonances were used. (B) The structural model of ceria used in the DFT calculations. Red and white spheres represent oxygen and cerium ions, respectively. The exposed surface is (111) and the calculated chemical shift of 17O in each layer is shown on the right side. Reproduced with permission [8]. Copyright 2015, The Authors; exclusive licensee American Association for the Advancement of Science.

Apart from 17O2 gas, the H217O is also widely used in preparing the 17O labeled samples. Wang et al. further employed H217O to enrich ceria nanostructures at even a lower temperature of 373 K. Only the resonances at 1027 and 920 ppm, corresponding to oxygen ions in the 1st and 2nd layers, as well as the peak centered at 270 ppm arising from surface hydroxyl group, can be observed in the 17O solid-state NMR spectra for ceria nanorods. This data not only confirms the spectral assignment but also exhibits the feasibility of performing highly efficient surface-selective 17O isotopic labeling by using H217O at low temperature, which leads to the desired surface-only observations in 17O NMR spectroscopy. On the basis of these results, the migration process of the surface oxygen species to the bulk part of ceria at elevated temperature can be monitored. Similar approach has also been used to follow the dynamic exchange process between oxygen species in the ZnO nanocrystals and oxygen from water molecules on the surface [9]. Surprisingly, in addition to the surface oxygen, the oxygen ions deeper inside the core of nanocrystals are also involved.

In Wang's work, the DFT calculations also show the surface oxygen species in different ceria surfaces are associated with different chemical shifts [8], which raises the apparent question: can 17O solid-state NMR be used to distinguish different facets? The current approaches to facet characterization are mostly based on electron microscopy methods, while the volume sampled by these techniques is so small that it might not be representative of the whole sample. Thus, this potential application of NMR is very important. Li and co-workers demonstrated that 17O NMR spectroscopy can be used to differentiate faceted oxide nanocrystals, using anatase TiO2 as an example [10]. The nanosheets and nano-octahedra samples were prepared with dominant high energy {001} and low energy {101} facets, respectively. The 17O NMR spectra of the two nanocrystals are very different, after applying low temperature enrichment with H217O (Fig. 2). The peaks at 480 ~ 570 ppm arise from 3-coordinated oxygen (O3c) species on the surface/subsurface, because of similar chemical shift observed for O3c in the bulk anatase TiO2. The resonances at higher frequencies (600 ~ 750 ppm) can be attributed to the O2c species on the surface. Broad peaks owing to surface hydroxyl sites (–150 ~ 300 ppm) and/or water species (centered at -75 ppm) can also be observed. Direct comparison of the spectra from the two samples can reach a conclusion that water molecules tend to undergo dissociative adsorption on the anatase {001} surface, while molecular adsorption of water occurs on the {101} facets.

Download:
Fig. 2. 17O NMR spectra of faceted anatase titania nanocrystals compared to the non-faceted sample. Anatase TiO2 nanosheets with dominant exposed (001) facets (NS001-TiO2), and nano-octahedra preferentially exposing (101) facets (NO101- TiO2) were surface-selectively 17O-labeled and vacuum dried for 2 and 12 h, respectively. The other sample, NF1-TiO2, was nonselectively 17O-labeled. All data were obtained at 9.4 T under a MAS frequency of 14 kHz. A rotor synchronized Hahn-echo sequence (π/6 -τ - π/3 -τ - acquisition) and optimized recycle delays (0.5 s for NS001-TiO2 and NO101-TiO2, and 50 s for NF1-TiO2), with 1H decoupling, were used to obtain the NMR data. 120000, 110000 and 1200 scans were collected for NS001-TiO2, NO101-TiO2 and NF1-TiO2, respectively. The spectra are normalized according to the sample mass and the number of scans. Asterisks denote sidebands. Reproduced with permission [10]. Copyright 2017, The Author(s)..

In order to extract detailed surface structural information, Li et al. established different structure models (i.e., clean surface, water dissociatively adsorbed on the surface, reconstructed surface, etc.), calculated the 17O NMR parameters (chemical shift, quadrupolar coupling constant and asymmetry parameter) of the oxygen species in these models with DFT, simulated the corresponding spectra and finally compared the simulations with the experimental data. Detailed structural information is revealed by this approach: surface reconstruction occurs on the {001} facets, while there are significant concentrations of "step edge" defects on the {101} surface. It is also proposed that the intensity of a specific peak reflects the reactivity of this surface oxygen species with water.

Since any nucleus should be sensitive to its local environments, like 17O NMR, NMR spectroscopy of metal ions should also provide detailed surface/subsurface structural information of nanosized metal oxides. In some cases, it can be more efficient due to the higher receptivity of the nucleus compared to 17O and the fact that expensive isotopic labeling of 17O may be avoided. For example, Chen and co-workers prepared SnO2 nanosheets with a thickness of 1.0 ~1.4 nm and the 119Sn NMR spectra of the sample show 3 resonances at -604, -585 and -618 ppm, which can be assigned to Sn species in bulk, Sn species in the 1st and 2nd surface layers, respectively, according to DFT calculations and 1H →119Sn cross polarization (CP) MAS NMR data [11].

2.2. 31P NMR studies with probe molecules

31P is a spin 1/2 nucleus with a very high gyromagnetic ratio and a natural abundance of 100%, making it readily observable. Along with adsorbed basic probe molecules such as TMP or trialkylphosphine oxides (R3PO; R = CnH2n+1), 31P NMR spectroscopy has been proved to be a powerful method to characterize acidity in aluminosilicates, especially zeolites [5, 6]. Recently, this method is also used to investigate the structure of properties of different facets of oxides nanocrystals, which is often the key in the applications of these materials.

Hu and co-workers prepared three sulfated TiO2 solid superacid catalysts with dominant {001} facets (ST001), roughly equal percentage of {001} and {101} facets (ST001/101), and dominant {101} facets (ST101) [12]. Four peaks at 66, -4, 32, and 50 ppm can be observed in the 31P solid-state NMR spectrum of ST101 (Fig. 3). The latter two peaks can be assigned to TMP coordinated to Lewis acid sites. The frequencies of these resonances become more positive when the fraction of exposed {001} facets increases in the catalyst. Since the higher 31P chemical shift refers to stronger Lewis acidity, 31P MAS NMR data indicates that ST001, the sulfated TiO2 solid superacid catalysts with mostly {001} facets, is associated with the highest Lewis acid strength. Usually unsaturated metal sites function as the Lewis acid sites and this facet dependent acidity strength can be ascribed to the concentration of unsaturated titanium ions on the {001} and {101} facets: half of the titanium ions on {101} facets of anatase TiO2 are saturated 6- coordinated ions (Ti6c) and only half are unsaturated Ti5c, while all of the titanium ions on {001} facets are Ti5c ions. This trend in Lewis acidity strength was confirmed by catalytic test of Pechmann condensation of phloroglucinol and ethyl acetoacetate. This is also one of the first evidences that NMR spectroscopy can be used to distinguish facets of oxides.

Download:
Fig. 3. 31P MAS NMR spectra of TMP loaded on ST001, ST001/101 and ST101. The dashed curves indicate results of spectral deconvolution. Reproduced with permission [12]. Copyright 2015, The Royal Society of Chemistry.

Peng et al. reported the development of TMP-assisted surface mapping of oxide nanoparticles and crystallites with 31P solidstate NMR spectroscopy, which provides both qualitative and quantitative information on the structure and properties of a variety of exposed facets, by using ZnO as a model material [13]. ZnO nanoplate, nanorod and nanopowder were prepared and the 31P NMR spectra of the three samples adsorbed with TMP are distinct. The majority of the signals come from the TMP bound to Lewis acid sites (–30 ~ –70 ppm), while two resolved peaks in this region are observed for each sample. According to the adsorption energies and chemical shifts calculated with DFT methods, as well as the experimental chemical shift values, these observed peaks can be assigned and the fraction of each sites (i.e., Zn3c, OH, and oxygen vacancy) on each exposed facets (i.e., Zn- terminated (002), O-terminated (002) and (100) facet) can be extracted. Clearly, TMP molecules are bound to different facets differently and this new approach is a surface fingerprinting technique. It can be easily applied to commercial ZnO nanomaterials. According to the ratio of two peaks obtained with 31P NMR, one commercial sample is concluded with a large amount of Zn-/O-terminated polar surfaces, while the dominant 31P NMR signal at -61 ppm indicate high concentration of hydroxyl sites in another sample, and these conclusions have been supported by high resolution transmission electron microscopy (TEM) and electron diffraction data.

More recently, the same group extended this approach to study different TiO2 facets with a variety of different post synthesis treatment and modifications [14]. These procedures, including NaOH wash and calcination which are usually performed in order to remove surface fluorine, as well as surface sulfation, can significantly alter the facet properties and play a vital role in applications such as optical and thermal catalysis. The commonly used methods, such as photoluminescence spectroscopy, X-ray photoelectron spectroscopy (XPS), electron paramagnetic resonance (EPR) and Raman spectroscopy, however, can only provide very limited information on these samples, and have difficulties in illustrating facet-dependent results [14, 15]. For example, in the quantitative measurement of surface oxygen vacancy, photoluminescence spectroscopy does not work in case there is interference from the presence of bulk defects while EPR spectroscopy fails since there are too many factors that affect the EPR signals. In contrast, TMP assisted 31P NMR spectroscopy is very sensitive to the nature of the binding sites on the surface, thus can give both qualitative and quantitative information on the chemical state as well as distribution of surface cations in different facets. These results confirm the properties of facets are adsorbatedependent and the 31P NMR approach will benefit current research in facet control and engineering of oxide nanomaterials.

3. Surface structure of oxides investigated with DNP-SENS

DNP enhances the signal of nuclear spins by transferring polarization of unpaired electrons induced from microwave irradiation. Although first experiment were performed long time ago [16], only very recently, it was demonstrated that DNP can be used to increase the solid-state NMR sensitivity of molecular organic functionalitieson the surface ofhybridsilica materials[17]. Typically, the experiments involve dissolving biradical (e.g., TOTAPOL [18]) in D2O/H2O mixture, with which the solid materials are wetted. The DNP NMR measurement is usually performed at very low temperature (~100 K) and a gyrotron is used to generate very strong microwaves to continuously irradiate electron resonance in order to transfer highly polarized electron spin to nuclear spin.

After thisfirst example, the DNP-SENStechnique was extended to detect 29Si species on the silica surface [19]. Without microwave irradiation (MW = off, i.e., no DNP effects), the signals from the surface Qn sites can hardly be observed while these signals increase 21 times and are clearly visible with microwave irradiation. In terms of acquiring a spectrum with the same signal-to-noise ratio, DNP saves time by approximately 440 (212) times! Clearly, by applying DNP-SENS, some of the surface species with very lowconcentration, which was previously impractical to detect, can now be studied.

DNP experiments can be performed in two ways, direct polarization and indirect polarization. In the former case, the unpaired electron polarization is transferred to the observed nucleus (e.g., 29Si) directly and the electron polarization is transferred to 1H followed by CP (e.g., 1H→29Si) in the latter scenario. It was shown that direct and indirect 29Si DNP NMR spectra provide complementary information of the clay nanoparticles [20]. Direct DNP favors the observation of 29Si species near paramagnetic centers, while 29Si species further away are enhanced in indirect DNP because of the cross polarization and 1H spin diffusion processes. Therefore, in order to obtain the desired information on the materials, DNP-SENS method should be carefully chosen.

DNP-SENS was also applied to observe quadrupolar nuclei, such as 27Al. 27Al has a relatively large gyromagnetic ratio and its natural abundance is 100%, thus is usually a more NMR-friendly nucleus, compared to 17O. Gamma alumina represents the most widely used oxide for catalysis, however, its detailed surface structure still needs investigations. Vitzthum et al. shows that DNP-SENS gives a best 27Al enhancement of about 20 on gamma-alumina [21]. In order to select the surface aluminum species of alumina, indirect DNP with 1H→27Al CP is most efficient, since 1H nuclei are only present on the surface the material.

The first 17O DNP-SENS spectroscopy was demonstrated on 17O labeled MgO, Mg(OH)2 and Ca(OH)2, in both direct and indirect polarization transfer schemes [22]. Although a relatively large enhancement factor were observed, it seemed that the surface (42 ppm) and bulk (47 ppm) oxygen sites in MgO nanoparticles were enhanced by the same factor. Perras et al. improved the sensitivity of 17O DNP-SENS by using phase-shifted recoupling effects a smooth transfer of order (PRESTO) polarization transfer [23], to overcome the problems associated with conventional CP transfer (Fig. 4) [24]. The sensitivity was further gained by using quadrupolar Carr-Purcell-Meiboom-Gill (QCPMG) detection [25]. With PRESTO-QCPMG, high resolution natural abundance DNPSENS 17O MAS NMR spectrum of widely used mesoporous silica SBA-15 can be obtained, with a 14-hour acquisition time. Since it only takes a short time to obtain 1D spectra of simple materials such as Mg(OH)2 and Ca(OH)2 with very good signal-to-noise ratio, 2D heteronuclear correlation (HETCOR) NMR experiments are also feasible.

Download:
Fig. 4. Natural-abundance DNP-SENS 17O MAS NMR spectrum of the mesoporous silica SBA-15, acquired using the PRESTO-QCPMG pulse sequence in 14 h. Reproduced with permission [24]. Copyright 2015, American Chemical Society.

By adding windowed-proton detected local-field (wPDLF) technique, which can monitor dipolar coupling between 1H and 17O, Perras and co-workers further developed DNP-enhanced, natural-abundance 17O{1H} wPDLF-QCPMG method to measure the very important O-H distances in a variety of different solid materials [26]. The O-H distances in these materials can be measured with an "unprecedented, sub-pm precision". It was also found that the O-H bond length extracted is inversely correlated with the pH of the zero point of charge, which is a common indicator of the acidity of oxide surfaces.

DNP-SENS can be combined with isotopic enrichment to further increase the sensitivity of detection. For example, the natural abundance of 29Si is 4.7%, a silicated alumina catalyst with a small Si loading of 1.5% means that the 29Si concentration is as low as 700 ppm. By using 29Si isotopic labeling along with DNP, the detailed surface structure of this oxide catalyst can be studied [27]. Similar method has been applied to 17O NMR study. The data of non-selectively 17O labeled ceria nanoparticles obtained by Hope et al. show that the oxygen ions in the first three surface layers can be selectively observed and distinguished by using direct 17O DNP NMR [28]. This high selectivity for the surface sites, similar to the previously mentioned results obtained by combining 17O NMR with surface-selective labeling, can be attributed to the slow spin diffusion of 17O polarization into the bulk. In this case, only the surface/subsurface sites are efficiently hyperpolarized by radicals near of the surface. Indirect 17O DNP NMR spectroscopy is only able to detect surface hydroxyl and adsorbed water, while surface/ subsurface oxygen ions can not be observed. This can be ascribed to the low concentration of protons associated with the surface.

4. Conclusions and perspectives

In conclusion, the approaches summarized in this review can provide complementary information and give a better understanding of the structure and properties of the surface of solid oxide materials. The very high sensitivity enhancement of DNP-SENS has significantly increased the detection limit for the surface of oxide materials, making a lot of low concentrated surface species observable in NMR spectroscopy. With the methodology development that can further increase the sensitivity, DNP-SENS 17O NMR should be able to be applied to a lot more oxide materials with natural abundance, revealing the structure and property relations readily. At the same time, 17O NMR spectroscopy in combination with surface-selective labeling is a much more cost-effective method and is sensitive enough to distinguish different oxygen ions in different layers and facets. The NMR intensities obtained with this technique are related to the reactivity of the surface/subsurface species, as well as dynamic exchange, while the signals observed in DNP-SENS may depend on many factors including 1H concentration, the efficiency of 1H spin diffusion and the proximity to the paramagnetic centers. The probe molecule-assisted 31P NMR spectroscopy is also very convenient to perform and can provide rich information on the nature of facets.

Acknowledgment

This work was supported by the National Natural Science Foundation of China (NSFC, Nos. 21573103 and 91745202) and NSFC – Royal Society Joint Program (No. 21661130149). L. Peng thanks Royal Society and Newton Fund for Royal Society – Newton Advanced Fellowship. This work was also supported by a Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions.

References
[1]
P. Yang, Nature 482 (2012) 41-42. DOI:10.1038/482041a
[2]
X.L. Cheng, J.S. Jiang, D.M. Jiang, Z.J. Zhao, J. Phys. Chem. C 118 (2014) 12588-12598. DOI:10.1021/jp412661e
[3]
Y.J. Yuan, Z.J. Ye, H.W. Lu, et al., ACS Catal. 6 (2016) 532-541. DOI:10.1021/acscatal.5b02036
[4]
X.W. Xie, Y. Li, Z.Q. Liu, M. Haruta, W.J. Shen, Nature 458 (2009) 746-749. DOI:10.1038/nature07877
[5]
A.M. Zheng, S.H. Li, S.B. Liu, F. Deng, Acc. Chem. Res. 49 (2016) 655-663. DOI:10.1021/acs.accounts.6b00007
[6]
A.M. Zheng, S.B. Liu, F. Deng, Chem. Rev. 117 (2017) 12475-12531. DOI:10.1021/acs.chemrev.7b00289
[7]
K. J. D. MacKenzie, M. E. Smith, Multinuclear Solid-state NMR of Inorganic Materials, Elsevier, 2002.
[8]
M. Wang, X.P. Wu, S.J. Zheng, et al., Sci. Adv. 1 (2015) e1400133. DOI:10.1126/sciadv.1400133
[9]
Y. Champouret, Y. Coppel, M.L. Kahn, J. Am. Chem. Soc. 138 (2016) 16322-16328. DOI:10.1021/jacs.6b08769
[10]
Y.H. Li, X.P. Wu, N.X. Jiang, et al., Nat. Commun. 8 (2017) 581. DOI:10.1038/s41467-017-00603-7
[11]
J.C. Chen, X.P. Wu, L. Shen, et al., Chem. Phys. Lett. 643 (2016) 126-130. DOI:10.1016/j.cplett.2015.11.035
[12]
Y.C. Hu, B. Guo, Y.Y. Fu, et al., Chem. Commun. 51 (2015) 14219-14222. DOI:10.1039/C5CC04548G
[13]
Y.K. Peng, L. Ye, J. Qu, et al., J. Am. Chem. Soc. 138 (2016) 2225-2234. DOI:10.1021/jacs.5b12080
[14]
Y.K. Peng, Y.C. Hu, H.L. Chou, et al., Nat. Commun. 8 (2017) 675. DOI:10.1038/s41467-017-00619-z
[15]
Y.K. Peng, Y. Fu, L. Zhang, et al., ChemCatChem 9 (2016) 155-160.
[16]
T.R. Carver, C.P. Slichter, Phys. Rev. 92 (1953) 212.
[17]
A. Lesage, M. Lelli, D. Gajan, J. Am. Chem. Soc. 132 (2010) 15459-15461. DOI:10.1021/ja104771z
[18]
C. Song, K.N. Hu, C.G. Joo, T.M. Swager, R.G. Griffin, J. Am. Chem. Soc. 128 (2006) 11385-11390. DOI:10.1021/ja061284b
[19]
M. Lelli, D. Gajan, A. Lesage, J. Am. Chem. Soc. 133 (2011) 2104-2107. DOI:10.1021/ja110791d
[20]
O. Lafon, A.S.L. Thankamony, M. Rosay, Chem. Commun. 49 (2013) 2864-2866. DOI:10.1039/C2CC36170A
[21]
V. Vitzthum, P. Mieville, D. Carnevale, Chem. Commun. 48 (2012) 1988-1990. DOI:10.1039/c2cc15905h
[22]
F. Blanc, L. Sperrin, D.A. Jefferson, J. Am. Chem. Soc. 135 (2013) 2975-2978. DOI:10.1021/ja4004377
[23]
X. Zhao, W. Hoffbauer, J. Schmedt auf der Günne, M.H. Levitt, Solid State Nucl. Magn. Reson. 26 (2004) 57-64. DOI:10.1016/j.ssnmr.2003.11.001
[24]
F.A. Perras, T. Kobayashi, M. Pruski, J. Am. Chem. Soc. 137 (2015) 8336-8339. DOI:10.1021/jacs.5b03905
[25]
F.H. Larsen, J. Skibsted, H.J. Jakobsen, N. C. Nielsen, J. Am. Chem. Soc. 122 (2000) 7080-7086. DOI:10.1021/ja0003526
[26]
F.A. Perras, Z.R. Wang, P. Naik, I. I. Slowing, M. Pruski, Angew. Chem. Int. Ed. 56 (2017) 9165-9169. DOI:10.1002/anie.201704032
[27]
A.G.M. Rankin, P.B. Webb, D.M. Dawson, et al., J. Phys. Chem. C. 121 (2017) 22977-22984. DOI:10.1021/acs.jpcc.7b08137
[28]
M.A. Hope, D.M. Halat, P. C.M.M. Magusin, et al., Chem. Commun. 53 (2017) 2142-2145. DOI:10.1039/C6CC10145C