Hyperpolarized long-lived nuclear spin states in monodeuterated methyl groups† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c8cp00253c

Dissolution-dynamic nuclear polarization is implemented to hyperpolarize long-lived nuclear spin states in monodeuterated methyl groups.


Dissolution-dynamic nuclear polarization methods
Solutions of 0.375 M (N-CH 2 D)-2-methylpiperidine in the glass-forming mixture D 2 O:glycerol-d 8 (50:50 v/v) were doped with 25 mM TEMPOL (purchased from Sigma-Aldrich). The synthesis of (N-CH 2 D)-2-methylpiperidine is reported elsewhere [1]. The solution was sonicated for 2 minutes. Ten frozen pellets of the "DNP mixture" (10 µL volume per pellet) were inserted in the polarizer. The sample polarized at ∼1.3 K and 6.7 T for ∼48 minutes in a home-built polarizer by applying frequency-modulated microwave irradiation at 188.3 GHz frequency and 100 mW power [2,3]. The microwave modulation frequency and amplitude were 10 kHz and 50 MHz, respectively. The polarized pellets were dissolved with 5 mL CD 3 CN solvent (degassed via bubbling with nitrogen gas for 5 minutes) preheated to 410 K at a pressure of 10 bar. The liquid sample was transferred in 10.7 s to a 11.7 T (500 MHz) NMR magnet by pushing with helium gas at 6 bar through a PTFE tube (1.5 mm inner diameter) running inside a magnetic tunnel (0.91 T, 5 m length) [4]. 1 s was taken for sample injection and bubble dissipation. Spectra were recorded by using a 11.7 T Bruker Avance II console and processed with home-written Python software.

Solid-state polarization
Zeeman polarization p solid Z was accumulated in the solid-state for 0.375 M (N-CH 2 D)-2methylpiperidine in the presence of 25 mM TEMPOL radical and glassy D 2 O:glycerol-d 8 matrix in a field of 6.7 T and at a temperature of ∼1.3 K under the action of negative dynamic nuclear polarization (DNP), see the dissolution-dynamic nuclear polarization methods section for more details. A Zeeman polarization of p solid Z = -59±5% was achieved in ∼48 minutes, see Fig. S1a. The solid state enhancement solid Z was approximately -360±20 compared to a spectrum recorded with the microwaves off, see Fig. S1b. The thermal spectrum was acquired after a 1 hour equilibration period at ∼4.2 K.

T 00 filter and singlet to magnetization pulse sequence
In the current study, we use the combination of a T 00 filter and the S2M (singlet-tomagnetization) pulse sequence to retrieve singlet order generated directly from DNP. The T 00 filter and S2M pulse sequence are shown in Fig. S2.
The singlet state is a magnetically silent arrangement of nuclear spin configurations and is unperturned by the T 00 filter, which employs the optimized parameters shown in Table S1 to remove signals deriving from residual magnetization. Details of the T 00 filter are also given in Refs [1,5].
The S2M pulse sequence converts hyperpolarized singlet order into hyperpolarized transverse magnetization. The S2M pulse sequence consists of two spin-echo trains generated by a recurring sequence of a composite 180 • pulse (90 • 0 -180 • 90 -90 • 0 ) sandwiched between two evolution periods τ J of duration 1 4J , where J is the in pair scalar coupling (11.7 Hz). The first and second spin-echo trains are repeated n 2 and n 1 times, respectively, with n 1 ∼ n 2 /2. An additional τ J -90 • 0 module is inserted between the two echo trains [6,7]. The parameters of the S2M pulse sequence were as follows: τ J = 21.4 ms, n 1 = 3 and n 2 = 1.
In order to determine the efficiency of the S2M sequence η S2M we converted the magnetization of a thermally polarized sample to singlet order by using the M2S (magnetization "to" singlet) pulse sequence. The M2S pulse sequence is the time-reversal of the S2M pulse sequence including an initial 90 • 90 -τ J segment. The M2S pulse sequence employs the same parameters as the S2M pulse sequence. Any remaining magnetization was quenched by using a T 00 filter and the singlet order was back-converted to magnetization using the S2M pulse sequence. The signal was recorded and compared to a separate signal which was acquired following an excitation with a 90 • 0 pulse. The ratio of the two signals was found to be 0.4. The efficiency of the S2M pulse sequence is therefore: η S2M = 0.4 1/2 = 0.63 ± 0.02, which is close to the theoretical maximum of 2/3 [8]. The experiment was carried out on a sample of 5 µL "DNP mixture" in 0.5 mL CD 3 CN solvent.  The pulse sequence for measuring the 1 H T 1 of the CH 2 D peak obscured by the suspected water impurity is shown in Fig. S3. The scheme commences with a "saturation comb" (90 • 0 -delay) 100 which crushes all observable magnetization. The delay between 90 • 0 pulses was 5 ms. After an evolution period τ EV , ordinary magnetization is accrued and converted into singlet order by the M2S (magnetization to singlet) pulse sequence [6,7]. The T 00 filter destroys all signals not orgininating from the proton singlet order [1,5], which is subsequently back-converted into observable magnetization by the S2M pulse sequence. The S2M applies the same transformations as the M2S but in reverse chronological order. The triplet-singlet-triplet conversion has an efficiency of 40%, see the main text for more details. NMR spectra were acquired as a function of τ EV and the CH 2 D T 1 of 5.9 ± 0.7 s was determined from the integrals of the resulting singlet resonances.

Impurity
The resonance position of the impurity, thought to be residual water [9], was found to be dependent on the volume of DNP mixture ((N-CH 2 D)-2-methylpiperidine, glass-forming D 2 O:glycerol-d 8 (50:50 v/v) and TEMPOL) dissolved in CD 3 CN solvent, see table S2. At higher concentrations of "DNP mixture", the water impurity was shifted sufficiently far downfield such that the CH 2 D resonance was unobscured in the proton NMR spectrum. At a 10 µL volume of "DNP mixture" dissolved in CD 3 CN solvent, the CH 2 D peak was observed at 2.19 ppm. For "DNP mixture" concentrations <2 µL, such as those achieved after dissolution, the CH 2 D peak was obscured by a more intense water resonance. The resonance shift of the water impurity as a function of "DNP mixture" volume is approximately linear, but is not currently understood. A plausible mechanism would be an exchange interaction between the -OH protons of the TEMPOL radical with those of the residual protonated water belonging to the glassy matrix. Such an exchange interaction could simultanesouly lead to a downfield peak shift for water and broader NMR lines. We are not aware of previous reports of similar phenomena. Table S2: Resonance position of the suspected water impurity for different volumes of (N-CH 2 D)-2-methylpiperidine, glassforming D 2 O:glycerol-d 8 (50:50 v/v) and TEMPOL mixture dissolved in 0.5 mL degassed CD 3 CN solvent at 11.7 T (500 MHz) and 25 • C. Chemical shifts were referenced with respect to the CD 3 CN solvent peak.