Detection of the Surface of Crystalline Y2O3 Using Direct 89Y Dynamic Nuclear Polarization.

Nuclei with low gyromagnetic ratio (γ) present a serious sensitivity challenge for nulear magnetic resonance (NMR) spectroscopy. Recently, dynamic nuclear polarization (DNP) has shown great promise in overcoming this hurdle by indirect hyperpolarization (via 1H) of these low-γ nuclei. Here we show that at a magnetic field of 9.4 T and cryogenic temperature of about 110 K direct DNP of 89Y in a frozen solution of Y(NO3)3 can offer signal enhancements greater than 80 times using exogeneous trityl OX063 monoradical by satisfying the cross effect magic angle spinning (MAS) DNP mechanism. The large signal enhancement achieved permits 89Y NMR spectra of Y2O3 and Gd2O3-added Y2O3 materials to be obtained quickly (∼30 min), revealing a range of surface yttrium hydroxyl groups in addition to the two octahedral yttrium signals of the core. The results open up promises for the observation of low gyromagnetic ratio nuclei and the detection of corresponding surface and (sub-)surface sites.

Nuclear magnetic resonance (NMR) spectroscopy is one of the most versatile analytical techniques used to understand the structure and dynamics of solid-state materials. The technique can provide detailed atomic-scale information, making it a widely applied approach across biochemistry, polymer science, battery materials, and many other facets of materials science.[1−3] However, solid-state NMR is impeded by its intrinsic low sensitivity as a result of the combination of the low polarization of nuclei in a magnetic field (also present in liquid-state NMR) and the presence of anisotropic interactions, which significantly broaden the NMR spectra.[2] This limitation is especially pronounced for nuclei with a low gyromagnetic ratio γ (arbitrarily defined as lower than that of 15N at −27.116 106 rad s–1 T–1) as the signal-to-noise ratio scales as γ5/2.

Nuclei with low-γ are involved in a wide range of biological and physical processes:[4,5] for example, magnesium (NMR active nucleus 25Mg) is present in many naturally occurring clays, organic compounds, and oxides;[6] potassium (most NMR-sensitive nucleus 39K) is important in biochemistry and solid-state chemistry;[4] calcium (NMR active nucleus 43Ca) is of great interest due to its presence in biomolecules and other minerals;[7] yttrium (89Y) is of interest, for instance, in pyrochlore ceramic materials where it can be used as a probe for cation disorder.[8]

Many approaches have been taken to improve the NMR sensitivity of solids, perhaps most notably magic angle spinning (MAS), used to average out orientation-dependent interactions that can lead to spectral broadening,[9] and cross-polarization (CP), which can enhance the sensitivity of heteronuclei using the higher polarization of receptive nuclei (e.g., 1H, 19F).[10] Combining these techniques with dynamic nuclear polarization (DNP)[11−13] has introduced further, and dramatic, enhancements of the NMR signals from solid samples.[14−18] The DNP approach exploits the transfer of the high polarization of unpaired electrons (∼660 times larger than 1H), which are usually added to a sample in the form of stable free radicals[13] or transition metal ions,[19] to nuclei via suitable microwave (μw) irradiation[11,12] followed by their detection. Thanks to numerous developments, including in hardware,[20−24] sample preparation,[13,14] and polarizing agents,[25−29] DNP can yield a boost of NMR sensitivity of multiple orders of magnitude. The MAS DNP approach is typically combined with CP in order to hyperpolarize abundant 1Hs (or 2H in isotopically enriched samples)[30] and take advantage of their efficient homonuclear spin diffusion to distribute the polarization[31] before its transfer to the heteronucleus X of interest.

In particular, indirect DNP (i.e., via CP) is becoming an unrivalled method to address the NMR sensitivity issue of low-γ nuclei and has been exploited to probe the local yttrium environment and proton–yttrium connectivity in doped BaZrO3[32] by 89Y NMR, carbonated hydroxyapatite by 43Ca NMR,[33] and silica-supported yttrium–amide complexes by 89Y NMR.[34] These approaches are complementary to those using 1H detection and very fast MAS for the acquisition of the NMR spectra of low-γ nuclei, as very recently illustrated for 89Y, 103Rh, 109Ag, and 183W, offering a significant sensitivity boost compared to standard CP.[35]

However, many solids do not possess protons, therefore ruling out the use of {1H-}X CP and challenging the MAS DNP approach for sensitivity enhancement. One alternative approach can rely on direct transfer of polarization from the radical to the heteronucleus of interest (i.e., direct DNP).[36−41] Unlike in CP, the small homonuclear dipolar coupling of low-γ nuclei limits spin diffusion, rendering the propagation of hyperpolarization from direct DNP throughout the sample extremely inefficient. Nevertheless, it has recently been shown that homonuclear spin diffusion can be used to efficiently propagate DNP into bulk materials containing no protons when exploiting slow spin–lattice relaxation times (e.g., in 29Si, 31P, 113Cd, and 119Sn).[42] However, the spin diffusion coefficient is proportional to γ2 and is expected to be small for low-γ nuclei (estimated around 2 nm2 s–1 for 89Y in Y2O3 based on recent work[42]), and homonuclear spin diffusion might therefore not be effective for low-γ nuclei.

Transition and rare earth metal oxides lacking intrinsic protons find widespread applications in materials science, chemistry, and catalysis. Their NMR investigation has been facilitated by the development of DNP NMR at high magnetic fields, and 17O DNP allows insights into the structure of these materials;[39−41,43,44] however, the DNP NMR detection of the cations themselves has been largely unexplored so far.

In this work, we demonstrate large signal enhancements (ε89Y) for directly polarized 89Y in frozen solutions with both the monoradical trityl OX063[25] (|ε89Y| > 80) and the binitroxide radical AMUPol[28] (|ε89Y| > 30) as sources of polarization. Gains in sensitivity larger than 40 are also demonstrated in Y2O3 nanoparticles that are either pristine or Gd2O3-added. We show that the approach allows the observation of yttrium hydroxyls and yttria surface sites, as well as those from the bulk, demonstrating its versatility.

Materials Preparation. Y(NO3)3·6D2O crystals were prepared by dissolving Y(NO3)3·6H2O (Sigma-Aldrich, 99.8%) in a 100-fold weight excess of D2O (Sigma-Aldrich, 99.9% D) and evaporation of the solvent in air for 24 h.

Y2O3 and 0.1% Gd2O3-added Y2O3 samples were synthesized through a glycine–nitrate combustion route from Y(NO3)3·6H2O and Gd(NO3)3·6H2O (both Alfa Aesar, 99.9%) and glycine (Alfa Aesar, 99.7%) as starting materials. Stoichiometric ratios of the reactants were mixed in deionized water (5–10 mL) with a nitrate-to-glycine ratio of 2:1. Mixtures were then dehydrated on a hot plate, and autoignition followed. Powders were then ground and fired at 1000 °C for 12 h, pressed into pellets and sintered at 1000 °C for 12 h, and finally slowly cooled to room temperature and finely ground. Both materials were shown to be phase-pure and of similar sized nanoparticles by PXRD (Figure S2).

Powder X-ray Diffraction. Powder X-ray diffraction (PXRD) data were collected in Bragg–Brentano mode on a Panalytical X’Pert Pro diffractometer using monochromated Co Kα1 radiation (λ = 1.7890 Å). The crystallite sizes of Y2O3 and 0.1% Gd2O3-added Y2O3 were determined by Debye–Scherrer analysis and found to be 38.2 ± 10.8 and 37.5 ± 12.9 nm, respectively (see Figure S2).

DNP Sample Preparation. A 5 M yttrium nitrate solution was prepared by dissolving Y(NO3)3·6D2O in glycerol-d8/D2O/H2O (6:3:1 volume ratio) containing either 10 mM AMUPol biradical[28] or 40 mM trityl OX063 monoradical (henceforth referred to as trityl).[25] The same amount of the solution (25 μL) was then packed into a 3.2 mm sapphire rotor and closed with a silicone plug and a zirconia drive cap.

Y2O3 and 0.1% Gd2O3-added Y2O3 samples for DNP were prepared by wetness impregnation of finely ground powders (typically 50 mg) with 30 μL of either 10 mM AMUPol biradical or 40 mM trityl in glycerol-d8/D2O/H2O (6:3:1 volume ratio). The same amount of sample was then packed into 3.2 mm sapphire rotors and sealed with zirconia drive caps.

NMR Methods. All DNP NMR experiments were performed on a 9.4 T Bruker Avance III spectrometer operating at 9.4 T and a gyrotron μw source at 263.66 GHz.[22,24] Experiments were recorded with a 3.2 mm HXY triple resonance MAS probe; for experiments with AMUPol[28], the probe was tuned to ν0(1H) = 400.321 MHz with the X channel tuned to ν0(13C) = 100.403 MHz and the Y channel tuned to ν0(89Y) = 19.700 MHz; for experiments with trityl, the probe was tuned to ν0(13C) = 100.725 MHz and ν0(89Y) = 16.672 MHz on the X and Y channels, respectively; these configurations correspond to the maximum signal enhancement for each nucleus observed in the MAS DNP magnetic field sweep profiles. All experiments were acquired at a MAS rate of νr = 5 kHz and at a sample temperature of T = ∼110 K. 1H pulses and SPINAL-64 decoupling[45] applied during 13C or 89Y detection were performed at a radio frequency (rf) amplitude of 100 kHz. 13C and 89Y directly excited spectra were obtained with a rotor synchronized spin echo sequence with pulses performed at rf amplitudes of 50 and 10 kHz, respectively. {1H-}13C and {1H-}89Y CP experiments were obtained with a 50–100% 1H ramp to 100% 13C/89Y, reaching maxima of 30 and 70 kHz for 1H when matched to 13C at 50 kHz and 89Y at 13 kHz, for a duration of 4 and 10–20 ms, respectively. Recycle delays of 1.3 × τDNP(1H)[46] were used for {1H-}X CP experiments, where τDNP is the measured time constant for the polarization to return to equilibrium after saturation, which was extracted from a saturation recovery experiment with a fit to a stretch exponential function of the form 1 – exp(−t/τDNP)β (where t and β are the variable delays and stretch exponential factor, respectively). The DNP field profiles were recorded by altering the external magnetic field (B0) using the sweep coil of the Bruker Ascend DNP NMR magnet while keeping the gyrotron μw frequency fixed at 263.66 GHz (1H data with trityl were taken from the literature).[47] All 13C and 89Y MAS NMR spectra recorded in the field profile were obtained with a build-up time of 60 s, and all other directly excited 13C and 89Y spectra were recorded with a build-up time of 120 s. All spectra were collected at the optimal microwave power for signal enhancement. 1H, 13C, and 89Y spectra were referenced to H2O at 4.8 ppm, the silicone plug at 0 ppm, and the most intense resonance of Y2O3 at 330 ppm, respectively. All employed pulse programs are depicted in Figure S1.

The signal enhancement for a particular nucleus εn was obtained by scaling the signal recorded in the absence of μw irradiation (μw off) to that with μw (μw on). Where no signal can be obtained without μw irradiation, minimum εn values were given by scaling the noise to the μw on signal; this is the case for the 13C and 89Y enhancement values for the frozen solution of Y(NO3)3 in glycerol-d8/D2O/H2O and for the broad 89Y signals centered at ∼100 ppm for Y2O3 and 0.1% Gd2O3-added Y2O3. All DNP field profile plots were normalized to the maximum signal obtained for the given nucleus using either polarizing agent. The 1H DNP field profile plot was normalized to the maximum enhancement obtained in this work using data reported in the literature for a frozen glycerol/water sample.[47]

To investigate the DNP mechanism responsible for the nuclear hyperpolarization under given experimental conditions, DNP field profiles were recorded. Figure shows the 1H MAS DNP field profiles of 5 M Y(NO3)3 solution with either AMUPol (10 mM) or trityl (40 mM) as a polarizing agent. The typical profile for the solid effect (SE) mechanism giving 1H enhancement is observed with using the trityl radical (Figure b, blue, Table ),[47] which has a relatively narrow EPR signal, showing the characteristic negative and positive enhancement peaks separated by 2 × ω0(1H) (where ω0(1H) is the 1H Larmor frequency). The SE mechanism is expected as the 1H Larmor frequency (400.321 MHz) at 9.4 T exceeds both the homogeneous (δ)[25] and inhomogeneous (Δ ≈ 90 MHz at 9.4 T,[48] extrapolated from a value of 50 MHz at 5 T[49] using the linear relationship between width and field strength for trityl[50]) EPR line widths of trityl, satisfying the selection condition of this mechanism

Figure 1

1H MAS NMR spectra and MAS DNP field profiles for Y(NO3)3 in solution. (a) 1H MAS DNP NMR spectra of Y(NO3)3 in a solution of glycerol-d8/D2O/H2O in a 6:3:1 ratio (v/v) with AMUPol biradical as the polarizing agent with μw on (orange) and μw off (red). The dagger (†) denotes the isotropic resonance. (b) 1H MAS DNP field profiles for AMUPol[28] (orange) and trityl[25] (blue, adapted from ref (47)). The intensities are normalized to the maximum signal obtained. ω0(1H) is shown centered to the isotropic resonance of the corresponding radicals.

Table 1

Enhancement Values and Build-up Times for 1H, 13C, and 89Y Nuclei in 5 M Y(NO3)3 Solution in a 6:3:1 (v/v) Glycerol-d8/D2O/H2O Solution for AMUPol or Trityl 9.4 T and 110 K

radical	nucleus	DNP mechanism	|εn|	τDNP (s)
AMUPol (10 mM)	1H	CE	60	7.9 ± 0.2
	13C	CE	>58	>2300
	89Y	CE	>30	>2200
trityl (40 mM)	1Ha	SE	29	b
	13C	CE/SE	>35	b
	89Y	CE	>80	>2600

Values extracted from ref (47).

These values were not recorded.

Figure b (orange) shows the 1H field profile with AMUPol as a polarizing agent and is characteristic of the CE MAS DNP mechanism. AMUPol does not fulfill the condition given in eq as the inhomogeneous EPR line width Δ (>600 MHz for the related TOTAPOL)[26] exceeds the 1H nuclear Larmor frequency, and AMUPol instead satisfies the CE conditionThe largest enhancement for 1H with AMUPol was observed at the maximum positive enhancement (B0 = 9.395 T) and is ∼1.2 times larger than the maximum negative signal enhancement, typical for water-soluble binitroxide radicals.[26]

A comparison of the 13C MAS DNP field profiles for AMUPol and trityl at 110 K is given in Figure . The field profile for 13C, a moderate γ nucleus, with AMUPol shows the typical characteristics of the CE mechanism as anticipated as Δ greatly exceeds ω0(13C). At the optimum negative position (B0 = 9.380 T) a maximum signal enhancement of |ε13C| ≥ 58 is obtained (Table ). The 13C MAS DNP field profile for trityl shows a maximum signal (ε13C ≥ 35) at the positive maximum (B0 = 9.403 T) and suggests the presence of the CE mechanism (Table , Figure b), as has previously been reported at 3.4 T[51] and, recently, at 14.1 T.[52] We note that the absence of NMR signal without DNP enhancement for 13C (and 89Y; see below) prohibits exploitation of the more comprehensive analysis of signal enhancements recently discussed in the literature.[21,31,48,53−55]

Figure 2

13C MAS NMR spectra and MAS DNP field profiles for Y(NO3)3 in solution. (a) 13C MAS DNP NMR spectra of Y(NO3)3 in a solution of glycerol-d8/D2O/H2O in a 6:3:1 ratio (v/v) with μw on for trityl (blue) and AMUPol (orange) as the polarizing agent and with μw off (red). The resonance at 0 ppm corresponds to the silicone plug, and the dagger (†) denotes the isotropic resonances (not resolved) of glycerol. (b) 13C MAS DNP field profiles for AMUPol[28] (orange) and trityl[25] (blue) with the inset showing the double minimum observed for trityl. The intensities are normalized to the maximum signal obtained. ω0(13C) is shown centered to the isotropic resonance of the corresponding radicals.

The double minimum observed for 13C (Figure b, inset) with trityl is indicative of a possible SE contribution to the MAS DNP field profile, which is due to the similar magnitude of the 13C Larmor frequency (100.19 MHz) and the inhomogeneous EPR line width (∼90 MHz at 9.4 T; see above). This dual contribution of CE and SE in the hyperpolarization of 13C with trityl highlights the suboptimal CE polarization transfer when Δ is of the same order of magnitude as the nuclear Larmor frequency (ω0(13C)), resulting in substantially lower overall signal gain for trityl versus AMUPol in this case (Figure and Table ). It has been reported that radical modification by binding a nitroxide moiety such as TEMPO to a trityl radical can increase the efficiency of this CE character; however, this is beyond the scope of the work reported here.[48,56]

The DNP-enhanced 89Y MAS NMR spectra of Y(NO3)3 in glycerol-d8/D2O/H2O (6:3:1 volume ratio) (Figure a) show two poorly resolved resonances that have previously been assigned[32,57,58] to free Y(aq)3+ ions (at −20 ppm) and complex [Y(glycerol)]3+ (at 20 ppm) in frozen solution. The 89Y spectrum of Y(NO3)3 shows a substantially greater maximum enhancement for trityl (|ε89Y| > 80) than AMUPol (|ε89Y| > 30) (see below for further discussion on these values). The 89Y τDNP was of the same order of magnitude for either AMUPol or trityl for CE MAS DNP (Table ), and because all 89Y spectra were recorded with a short recycle delay of 120 s, trityl showed a significant benefit compared to AMUPol for time-efficient signal enhancement, i.e., overall NMR sensitivity.

Figure 3

89Y MAS NMR spectra and MAS DNP field profiles for Y(NO3)3 in solution. (a) 89Y MAS DNP NMR spectra of Y(NO3)3 in a solution of glycerol-d8/D2O/H2O in a 6:3:1 ratio (v/v) with μw on for trityl (blue) and AMUPol (orange) and μw off (red). Only the isotropic resonances are shown. (b) 89Y MAS DNP field profiles for binitroxide radical AMUPol[28] (orange) and trityl[25] (blue). The intensities are normalized to the maximum signal obtained. ω0(89Y) is shown centered to the isotropic resonance of the corresponding radicals.

Both 89Y MAS DNP field profiles of Y(NO3)3 in solution with AMUPol and trityl as polarizing agents (Figure b) showed the typical CE line shapes, consistent with both radicals satisfying eq due to the small 89Y Larmor frequency (19.67 MHz) compared to the inhomogeneous EPR line widths Δ. As has been previously reported for 17O at 5 T,[40] the CE condition is met sufficiently using trityl, and the resulting polarization transfer can produce significantly larger MAS DNP enhancements than those when using a binitroxide polarizing agent with a broader EPR line (such as AMUPol). This is in part a result of the long electron spin relaxation times of trityl compared to those of the employed binitroxide, as well as the more efficient energy level (anti)crossings responsible for MAS DNP with a narrow-line radical.[48] Here, we use AMUPol, which has been shown[28] to have longer spin relaxation times and to result in 3–4 times the DNP enhancement compared to the TOTAPOL binitroxide polarizing agent used in the 17O work at 5 T.[40] Importantly, Figure shows that, even though AMUPol is a preferred binitroxide for CE MAS DNP, trityl is in fact superior for direct MAS DNP of low-γ nuclei.

Notably, the maximum DNP enhancement for 89Y with AMUPol was obtained on the positive enhancement side of the field profile, whereas for trityl the maximum signal gain was obtained on the negative enhancement side (∼1.1 times greater than the positive maximum). All directly polarized 89Y MAS DNP NMR spectra with trityl were subsequently recorded at this field position (B0 = 9.3975 T) and, as such, are plotted with negative phase (Figures –5).

Figure 5

89Y hyperpolarization build-up and direct 89Y DNP-enhanced MAS NMR spectra of 0.1% Gd2O3-added Y2O3 with trityl polarizing agent. (a) Fitted build-up curves of the bulk 89Y (330 ppm, filled triangles; 287 ppm, hollow triangles) and surface 89Y sites (∼200 ppm, hollow squares; ∼100 ppm, filled squares) of 0.1% Gd2O3-added Y2O3. The curves in purple are fit to the data using a stretch exponential function (see the experimental discussion). The 89Y MAS NMR spectra recorded at τDNP = (b) 64 and (c) 4096 s are also given for comparison of relative peak intensities.

It is important to note the potential influence of nuclear depolarization[59] on the reported enhancement values. In the absence of μw irradiation and in the presence of CE matching under MAS conditions, polarization can be transferred from a nucleus to the two electron spins in a reverse DNP-type process. This occurs when the polarization of the nucleus is greater than the difference in polarization between the two electron spins.[59,60] Therefore, the enhancement ratio, εn, can be larger than the nuclear polarization gain when compared to thermal Boltzmann equilibrium, and thus, care should be taken when evaluating DNP performance.

Nuclear depolarization has currently been observed for only 1H and 13C, but it is theoretically possible for all nuclei.[59] Here, the 89Y NMR sensitivity in the absence of μw irradiation is too small to perform a detailed analysis of the induced nuclear depolarization (and a comprehensive analysis of signal enhancements).[21,31,48,53−55] Because the CE matching condition is met for 89Y for both AMUPol and trityl, it could be expected that the nuclear depolarization is substantial and accordingly that the quoted enhancement values (ε89Y) may not truly represent the gain in nuclear polarization compared to Boltzmann equilibrium, ε89Y,Boltz. Nevertheless, in the presence of μw irradiation, the ratio of the maximum absolute signal to the minimum (nonzero) absolute signal found from the MAS DNP field profile (Figure ) gives the minimum ε89Y,Boltz; therefore, for AMUPol, |ε89Y,Boltz | > 30, and for trityl, |ε89Y,Boltz | > 80. Furthermore, because these are both obtained in the presence of μw irradiation, it can be safely concluded that trityl is more efficient for direct 89Y CE MAS DNP than AMUPol at 9.4 T using moderate MAS rates (∼5 kHz).

We then turned our attention to crystalline solid materials and targeted Y2O3 due to interest in this phase as semiconductors, as a source of yttrium for chemical doping, or for surface treatment.[61] It is well-known that adding a small amount of paramagnetic metal oxide Gd2O3 circumvents the extremely long spin–lattice relaxation times T1 of 89Y in Y2O3[62] by decreasing them and is verified experimentally (room temperature T1 > 3 h for Y2O3[62] vs ∼3 min in 0.1% Gd2O3-added Y2O3 for the 330 ppm signal[63]). This allows acquisition of 89Y NMR spectra in the 0.1% Gd2O3-added Y2O3 phase with good signal-to-noise in 30 min and shows the two characteristic peaks of Y2O3 at 330 and 287 ppm in a 3 to 1 ratio, respectively (Figure a, red), both sites having a similar DNP build-up time τDNP (vide infra and Figures a and S3). These resonances are typical of the bulk of yttria[63] and correspond to the two octahedral yttrium sites that have 24:8 occupancies in the cubic bixbyte structure (space group Ia3) in Y2O3.[64] It is possible to observe these signals in pristine Y2O3 in the same experimental time (Figure b, red) without DNP, albeit with much worse signal-to-noise than that for Gd2O3-added Y2O3 due to the shortening of the T1 times by the paramagnetic doping in the latter phase (see above).

Figure 4

Direct 89Y DNP-enhanced MAS NMR spectra of (a) 0.1% Gd2O3-added Y2O3 and (b) Y2O3, recorded with μw on (blue) and μw off (red). All NMR data were recorded with recycle delays of 120 s and trityl as a polarizing agent at the negative maximum of the field profile (B0 = 9.3975 T).

The bulk 89Y signals in either 0.1% Gd2O3-added Y2O3 or pristine Y2O3 are barely enhanced by 89Y direct DNP, as revealed by comparing the spectra with and without μw (Figure , blue and red, respectively). Note that the two 89Y NMR signals are inverted in the μw on experiment (versus the spectra without μw irradiation), confirming that these sites are indeed hyperpolarized by DNP. More importantly, in addition to these two narrow peaks, two broad resonances centered at approximately 200 and 100 ppm and spanning ∼300 ppm are now also clearly observed in both 0.1% Gd2O3-added Y2O3 and Y2O3 and are tentatively assigned to surface sites (vide infra).

Figure a plots the direct 89Y MAS DNP build-up curves for surface and bulk sites after saturation in 0.1% Gd2O3-added Y2O3, from which the DNP build-up times, τDNP, can be extracted. While the full build-up is well captured for the ∼200 and ∼100 ppm sites, yielding polarization times, τDNP, on the order of 410 ± 90 and 260 ± 30 s, respectively, a hyperpolarization plateau could not be reached (in a time-efficient manner) for the Y2O3 bulk sites at 330 and 287 ppm. This is due to the long build-up time, in excess of 15 min for these sites, which is consistent with the T1 times for crystalline Y2O3.[62] The shorter τDNP of the sites at ∼200 and ∼100 ppm compared to that of the yttrium bulk site likely indicates a lower degree of order and/or the presence of additional relaxation mechanisms such as a closer proximity to paramagnetic centers (e.g., trityl radical) or dipolar couplings to protons (e.g., yttrium hydroxyls;[5,65] see below), all demonstrating that these resonances arise from surface sites. Consequently, at short τDNP, the broad surface sites are the dominant feature of the spectrum (Figure b), with small contributions from the narrow bulk resonances.

DNP-enhanced {1H-}X CP MAS NMR of materials without intrinsic protons prepared by wetness impregnation with nonsolvents containing protons has been shown to provide surface-selective signal enhancement.[66]Figure shows the DNP-enhanced {1H-}89Y CP MAS NMR spectra of both 0.1% Gd2O3-added Y2O3 and Y2O3 materials recorded under the optimum 1H CE MAS DNP matching condition with AMUPol. In both cases, the dominant resonances are those centered at ∼200 and ∼100 ppm, leading to the confirmation of their assignments as surface sites, specifically to yttrium hydroxyl Y(OH) sites.[5,65] The 89Y resonance of Y(OH)3 appears at 66 ppm, upfield from Y2O3 (at 330 and 287 ppm), and as such, a range of inhomogeneously broadened Y(OH) surface sites resonating between 0 and 300 ppm are consistent with yttrium centers being bound to one (at ∼200 ppm) or two hydroxyl groups (at ∼100 ppm) on the surface of Y2O3. Note that the broadening does not stem from the proximity to Gd3+ because the broad Y(OH) surface peaks are also observed in the undoped Y2O3 (Figure b). A third narrower resonance is observed in the {1H-}89Y CP spectrum of pristine Y2O3 (Figure b) at around 330 ppm, which is likely signal from bulk 89Y. This site is largely enhanced by the long CP contact time (20 ms) that allows some hyperpolarization to penetrate into the bulk of the nanoparticles. This peak is less pronounced, but still observed, in the DNP-enhanced {1H-}89Y CP MAS NMR spectrum of 0.1% Gd2O3-added Y2O3 (Figure a) and results from a combination of shorter CP contact time (8 ms) used to record this CP spectrum and shorter rotating frame longitudinal relaxation times 1H T1ρ (due to the presence of Gd2O3).

Figure 6

DNP-enhanced {1H-}89Y CP MAS NMR spectra of yttrium oxides. (a) 0.1% Gd2O3-added Y2O3 with a CP contact time of 8 ms and (b) Y2O3 with a contact time of 20 ms, recorded with μw on (orange) and μw off (red). Note that fewer scans were collected in (b). All {1H-}89Y CP MAS NMR spectra were recorded with AMUPol at the positive maximum of the 1H CE (B0 = 9.395 T).

We have demonstrated that it is possible to hyperpolarize low-γ 89Y efficiently in the solid state by directly transferring polarization from electron spins in exogenous radicals, without requiring nearby proton spins. CE MAS DNP with trityl OX063 monoradical yields significant signal enhancement (|ε89Y| > 80), larger than that observed with AMUPol biradical (|ε89Y| > 30) for a frozen solution of Y(NO3)3. This approach was then employed to interrogate Y2O3, and comparison of DNP-enhanced 89Y spectra obtained with and without 1H CP reveals the presence of a range of yttrium hydroxyl sites at the surface of Y2O3. This was possible for both the Gd2O3-added and, notably, the unmodified Y2O3. This work demonstrates that DNP is a highly viable approach to facilitate the analysis of the atomic-scale environments of low-γ nuclei in the absence of protons in a time-efficient manner without additional modification of materials.