Frequency-swept dynamic nuclear polarization

Abstract

Dynamic nuclear polarization (DNP) improves the sensitivity of NMR spectroscopy by the transfer of electron polarization to nuclei via irradiation of electron-nuclear transitions with microwaves at the appropriate frequency. For fields > 5 T and using g ∼ 2 electrons as polarizing agents, this requires the availability of microwave sources operating at >140 GHz. Therefore, microwave sources for DNP have generally been continuous-wave (CW) gyrotrons, and more recently solid state, oscillators operating at a fixed frequency and power. This constraint has limited the DNP mechanisms which can be exploited, and stymied the development of new time domain mechanisms. We report here the incorporation of a microwave source enabling facile modulation of frequency, amplitude, and phase at 9 T (250 GHz microwave frequency), and we have used the source for magic-angle spinning (MAS) NMR experiments. The experiments include investigations of CW DNP mechanisms, the advantage of frequency-chirped irradiation, and a demonstration of an Overhauser enhancement of ∼25 with a recently reported water-soluble BDPA radical, highlighting the potential for affordable and compact microwave sources to achieve significant enhancement in aqueous samples, including biological macromolecules. With the development of suitable microwave amplifiers, it should permit exploration of multiple new avenues involving time domain experiments.

Fig. 1.

The circuit used to generate microwaves at 250 GHz. A 10 GHz signal from an oscillator is fed into a series of passive components, starting with a set of three frequency doublers, and the output at 60 GHz is mixed at an IQ Mixer with a signal from an arbitrary waveform generator operating between 1.5 and 3.5 GHz. The upper sideband of the mixer between 61.5 and 63.5 GHz is filtered before being amplified (active component), and further multiplied using frequency doublers to achieve an output frequency between 246 and 254 GHz at an output power of 220 mW. To protect the multipliers from reflected power, we utilize an isolator with an insertion loss of −1.4 dB, attenuating the power to 160 mW.

Fig. 2.

Block diagram of the initial frequency multiplier design which was rejected. Illustrated are representative frequency measurements made at points (A), (B), and (C) as shown in the circuit diagram in (D), with the LO set to 10.4 GHz and the AWG outputting + 10 MHz, for a nominal final frequency of 249.84 GHz. Plotted spectra are approximate reconstructions of the spectrum analyzer output onto the absolute frequency domain, and powers can be assessed within a plot but not between plots as the downconversion schemes for each frequency differ significantly. (A): output from the IQ mixer shows the desired signal at 10.41 MHz and the suppressed opposite sideband and leakage of the LO. (B): output following a single tripler step. (C): dense comb of sidebands at the final output. (D): schematic of source configuration for these measurements.

Fig. 3.

Frequency stability measurements of the two sources near 250 GHz. Measurements were made using a harmonic downconversion scheme and the FFT function of a digital oscilloscope. The scope’s sampling rate was 5.0 GSa/s and each trace contains 50 kSa (10 µs acquisition time); each plotted point corresponds to the frequency of maximum intensity for a particular trace, and two traces acquired per second.

Fig. 4.

¹H DNP (A-C) and ¹³C DNP (D) frequency profiles obtained with the setup on polarizing agents (A) AMUPol in 6:3:1 glycerol-d₈:D₂O:H₂O exhibiting CE DNP, (B) h₂₁-BDPA and Phe-d₅-BDPA in OTP matrix exhibiting OE and SE DNP, (C) Trityl-OX063 in 6:3:1 DMSO-d₆:D₂O:H₂O exhibiting SE and arguably resonant mixing (RM), and (D) diamond powder containing P1 centers primarily exhibiting CE DNP (to natural abundance ¹³C nuclei). Profiles in (A–C) were collected under MAS (5–7 kHz) at 90–100 K, and (D) under static conditions at room temperature. Specific experimental details are available in the SI.

Fig. 5.

Relative performance of repeated 1 microsecond chirps of varying chirp widths for a sample of 15 mM trityl OX063 and 30 mM 4-amino TEMPO in 6:3:1 glycerol-d₈:D₂O:H₂O. For all experiments, the MAS frequency was 5 kHz and the sample temperature 100 K.

Fig. 6.

Results using the AWG driven solid state source in conjunction with water-soluble NMe₃-BDPA. (A): Bulk ¹H enhancement for 45 mM NMe₃-BDPA in 6:3:1 glycerol-d₈: D₂O:H₂O at 95 K and spinning at 6 kHz, demonstrating the effect of freeze–pump–thaw degassing. (B): ¹³C CP spectra of uniformly labeled GNNQQNY microcrystals wet with a small volume of the same 45 mM NMe₃-BDPA at 95 K and spinning at 7 kHz.