Efficient cross-effect dynamic nuclear polarization without depolarization in high-resolution MAS NMR

Abstract

Dynamic nuclear polarization (DNP) has the potential to enhance the sensitivity of magic-angle spinning (MAS) NMR by many orders of magnitude and therefore to revolutionize atomic resolution structural analysis. Currently, the most widely used approach to DNP for studies of chemical, material, and biological systems involves the cross-effect (CE) mechanism, which relies on biradicals as polarizing agents. However, at high magnetic fields (≥5 T), the best biradicals used for CE MAS-DNP are still far from optimal, primarily because of the nuclear depolarization effects they induce. In the presence of bisnitroxide biradicals, magic-angle rotation results in a reverse CE that can deplete the initial proton Boltzmann polarization by more than a factor of 2. In this paper we show that these depolarization losses can be avoided by using a polarizing agent composed of a narrow-line trityl radical tethered to a broad-line TEMPO. Consequently, we show that a biocompatible trityl-nitroxide biradical, TEMTriPol-1, provides the highest MAS NMR sensitivity at ≥10 T, and its relative efficiency increases with the magnetic field strength. We use numerical simulations to explain the absence of depolarization for TEMTriPol-1 and its high efficiency, paving the way for the next generation of polarizing agents for DNP. We demonstrate the superior sensitivity enhancement using TEMTriPol-1 by recording the first solid-state 2D ¹³C-¹³C correlation spectrum at natural isotopic abundance at a magnetic field of 18.8 T.

Fig. 1. The molecular structure of TEMTriPol-1. The unpaired spin density of the trityl radical is highly delocalized and extends well into the linker with the nitroxide. The exchange interaction between the two unpaired electrons can therefore be modulated by changing the chemical structure of the linker.,,

Fig. 2. Experimental (a, b) and simulated (c, d) MAS frequency dependence of ε_on/off (left axis, black), ε_B (left axis, red), and χ_depo (right axis, blue) for trityl-nitroxide (a, c) and nitroxide–nitroxide (b, d) based biradicals. Nuclear polarization was determined via the ¹H signal integral. The μw frequency was 263.67 GHz and the magnetic field set to the position giving maximum ¹H signal intensity. Experimental data from Mentink-Vigier et al. is reproduced here in (b) as a comparison to the data in (a). Definitions for ε_on/off, ε_B, and χ_depo and. the biradical geometry are given in the Materials and method, For trityl-nitroxide J_a,b/2π = 30 MHz, D_a,b/2π = 23 MHz, and for AMUPol, J_a,b/2π = 15 MHz and D_a,b/2π = 35 MHz.

Fig. 3. Black curve: experimental 285.0 GHz CW EPR spectrum of 5 mM TEMTriPol-1 in DNP matrix recorded at 100 K. Blue curve: simulation assuming the presence of two conformations with J-values of 10 and 80 MHz. Red curve: simulation assuming a continuous distribution of J-values. This distribution in J_a,b is shown in the inset.

Fig. 4. (a) Experimental μw power dependence of ε_on/off for TEMTriPol-1 and (b) simulated μw power dependence at ∼9.4 T. Black circles: correspond to the field position of maximum positive enhancement, red squares correspond to the field position of maximum negative enhancement. The experimental output microwave power was adjusted by changing the gyrotron collector current from 30 to 75 mA. Simulations were performed with an exchange interaction of J_a,b/2π = 30 MHz.

Fig. 5. Simulated and experimental DNP Zeeman field profiles for TEMTriPol-1 in frozen glycerol/water at four different magnetic fields (μw frequencies/¹H Larmor frequency): (a, e) 5 T (139.60 GHz/212 MHz), (b, f) 9.4 T (263.67 GHz/400 MHz), (c, g) 14.1 T (395.30/600 MHz), and (d, h) 18.8 T (527.04 GHz/800 MHz). The experiments are displayed on the left-hand side and the corresponding simulations on the right-hand side. The black squares correspond to simulations that use the distribution of exchange interactions obtained from a fit to the 285 GHz EPR spectrum of TEMTriPol-1 (see insert of Fig. 3). The other data correspond to simulations with a single value of the exchange interaction: J_a,b/2π = 10 MHz (orange upside down triangles), 70 MHz (red upright triangles), and 140 MHz (blue diamonds). The trityl linewidth was taken to be 9 MHz T^–1, which corresponds to 45 MHz at 5 T (a), 85 MHz at 9.4 T (b), 127 MHz at 14.1 T (c), and 169 MHz at 18.8 T (d). Note that 0.1 mT corresponds to about 2.8 MHz.

Fig. 6. (a–c) Histograms of the mean individual electron polarizations, P_a and P_b, at steady-state for 144 different crystallite orientations of a bis-nitroxide (red and blue) and a nitroxide-trityl (nitroxide in green, trityl in black) electron spin system with different dipolar couplings: (a) D_a,b/2π = 0 (isolated electron spins), (b) D_a,b/2π = 3 MHz, (c) D_a,b/2π = 23 MHz. (d–f) Histograms of the maximum electron polarization difference |P_a – P_b|_max over one rotor period at the quasi-periodic steady state for 144 different crystallite orientations of a bis-nitroxide (in red) and a nitroxide-trityl (in black) electron spin-system with the same set of dipolar couplings as (a) to (c). |P_a – P_b|_max is normalized by the initial nuclear Boltzmann polarization P_n(0). The vertical dotted line (same color code as for the bars) represents the mean powder-averaged nuclear polarization at steady state scaled by P_n(0), and corresponds to the theoretical mean χ_depo. In all calculations ω_r/2π = 8 kHz, B₀ = 9.399 T, T = 100 K, and ¹H is the nucleus with polarization P_n that is hyperfine coupled (1.5 MHz) to electron a (the nitroxide electron, which also has a hyperfine interaction to a ¹⁴N spin ranging from 15 to 98 MHz).

Fig. 7. DNP-enhanced 2D DQ-SQ ¹³C–¹³C SPC5 (ref. 76) dipolar correlation spectrum recorded on microcrystalline cellulose at natural isotopic abundance at 18.8 T and ∼125 K. The MAS frequency was 6.9 kHz and the number of scans was 32. The spectrum was obtained in 16 hours with a recycling delay of 24 s, and 1.16 ms of SPC5 total mixing time. A corresponding DNP-enhanced {¹H}–¹³C CPMAS spectrum is shown on the top. Solvent peaks, as well as spinning sideband contributions from ¹³C-urea present in the DNP matrix are indicated by a dagger symbol. (Left) Full DQ-SQ cellulose spectrum. Cross-peaks marked with a star represent spinning sideband contributions. (Right) Zoom on the C4 to C3/C5 cross-peaks. Note that the contour levels on the right-hand side were chosen just above the noise level in order to see the correlations from the amorphous C4 peaks.