Spying on parahydrogen-induced polarization transfer using a half-tesla benchtop MRI and hyperpolarized imaging enabled by automation

The study addresses the challenge of translating hyperpolarization techniques into routine laboratory and clinical use. While dissolution dynamic nuclear polarization (dDNP) can yield ~50% 13C polarization and enable in vivo metabolic imaging, it requires expensive, complex infrastructure. Parahydrogen-induced polarization (PHIP) is an attractive, lower-cost alternative but lacks robust, portable, and automated polarizers suitable for routine workflows. Prior PHIP/PHIP-SAH implementations often demand intricate manual handling, high-field superconducting magnets, or specialized low-field systems, limiting scalability and accessibility. The authors aim to develop a portable, cost-effective, automated PHIP polarizer operating at an intermediate magnetic field (~0.5–0.55 T) using permanent magnets, enabling larger sample volumes, robust spin order transfer (SOT) in inhomogeneous fields, rapid duty cycles, reaction monitoring, and hyperpolarized imaging. The purpose is to demonstrate practical proton and carbon-13 hyperpolarization, robust SOT despite field inhomogeneity and drift, and automated operation toward quasi-continuous tracer production, thereby advancing PHIP toward in vivo applications.

The paper situates its contribution within the evolution of PHIP technologies. Early PHIP polarizers used magnetic field cycling (MFC) at sub-millitesla fields for 1H→13C transfer but were limited by strong indirect couplings reducing theoretical maxima. RF-induced SOTs at low field improved polarization, and high-field implementations within superconducting magnets achieved some of the highest PHIP-SAH polarizations, particularly with selective deuteration. Systems like SAMBADENA integrated polarization inside MRI scanners but faced constraints in space and materials. Automation in NMR/MRI (MFC, laser control, gas handling, chemical circulation) has improved reproducibility and throughput. Theoretical and simulation work suggests optimal SOT efficiencies at intermediate fields (1 mT–1 T), realizable with portable permanent magnets. Prior devices span low field (2–50 mT), high field (7–9.4 T), and field cycling (µT–T), achieving up to tens of percent 13C polarization, though with variability in tracers and conditions. Recent PHIP-SAH advances include 24% 13C polarization at 22.6 mT and ~9.8% at 50 µT for cinnamyl pyruvate derivatives. SABRE-based methods are also emerging for biomolecules. Despite these advances, a commercial, portable, automated PHIP polarizer remains unavailable, motivating the current development.

Hardware and automation: The polarizer is built around a 0.55 T portable benchtop MRI system (Pure Devices GmbH) with a 10 mm bore permanent magnet and integrated RF electronics. A custom gas–liquid control system operates seven solenoid gas valves, two HPLC valves, a syringe pump, and RF unit via microcontroller-based control boards (ESP32) and a touch display interface. Custom 3D-printed mounts actuate high-pressure PEEK valves by servo motors (ATmega328 control). Stainless steel 1/8" lines, solenoid valves (24 V), gauges, and safety hydrogen sensor (MQ-8) are integrated. The system runs from a domestic power outlet and requires no cryogens. Automatic B0 calibration completes in ~2 minutes after relocation. Reactor and fluidics: A high-pressure 10 mm NMR tube (Wilmad) serves as the hydrogenation reactor, positioned at isocenter using a 3D-printed adapter, typically loaded with 300 µL precursor/catalyst solution. A 1/16" FEP line from a syringe pump injects precursor; a single 1/32" PEEK line into the tube cap (epoxy sealed) doubles as pH2 inlet and sample ejection outlet via a 4-way valve in 3-way configuration. The fluid path enables automated loading, hydrogenation, in situ quantification, and automated disposal, supporting unattended parameter sweeps. Components are pressure-rated to ≥30 bar; pH2 was operated at 5 or 15 bar (up to 30 bar capability), N2 at 5 bar for flushing and shuttling. Chemistry: Precursors (50 mM) included vinyl acetate-d6 (VA-d6), vinyl pyruvate-d6 (VP-d6), 1-13C-vinyl pyruvate-d6, and ethyl phenylpropiolate (EPh), each with 3 mM [1,4-Bis(diphenylphosphino)butan](1,5-cyclooctadien)rhodium(I) tetrafluoroborate ([Rh]) in acetone-d6. Reactions: VA-d6→EA-d6, VP-d6→EP-d6, EPh→EC on bubbling pH2. Parahydrogen enrichment was 52% (LN2 converter) or 92% (two-stage cryo generator). Typical hydrogenation used 5–15 bar pH2 for 20–45 s. Control and timing: The MRI console (Matlab interface) orchestrated the sequence: N2 flush, pressure release, sample injection (300 µL), pressurization and pH2 bubbling for Tb, stop gas and equilibrate (1 s), apply SOT and acquire signal, then eject sample with N2. TTL outputs triggered pumps and valves. A 20 mL syringe reservoir enables ~60 automated runs with a ~1 min duty cycle. RF sequences and SOT: Implemented sequences were φ-FID, 45°-OPE(n,τ), φ-CPMG(n,τ) with composite 180° refocusing (90°–180°–90°), and ESOTHERIC for 1H→13C transfer. For PASADENA-derived PHIP, φ=45° excitations were used; thermal references used 90°. OPE converted anti-phase PASADENA multiplets to in-phase signals, improving SNR in inhomogeneous fields. A modified 45°-CPMG approach “spies” on spin evolution by recording echo amplitudes across many echoes for a single τ, drastically reducing sample consumption. ESOTHERIC parameters (n1,n2,n3,T1,T2,T3) were optimized numerically for 1-13C-EP-d6. Characterization and references: A 9.4 T high-resolution NMR (400 MHz) served for chemical shift and coupling characterization; the 0.55 T system acquired hyperpolarized spectra and images. Field stability and T2 were characterized via acetone-h6/acetone-d6 measurements, with frequency drift SD ~80 Hz due to magnet temperature fluctuations. Relaxation behavior was probed using 90°-CPMG as a function of echo spacing to estimate T2 and apparent diffusion/instability terms. Imaging used 13C-FLASH with 5° excitations, TR 50 ms, matrix 32×32, voxel ~0.312 mm², 10 mm slice, either in situ or after liquid shuttling to a receiver tube containing a 3D-printed phantom.

- Portable, automated PHIP polarizer at 0.55 T: Fully automated hyperpolarization and quantification with a footprint of ~1 m²; domestic power; no cryogens; automated B0 calibration in ~2 min; duty cycle ~1 min enabling ~60 experiments/hour from 20 mL stock. - 1H hyperpolarization at 0.55 T: Anti-phase PASADENA spectra (e.g., EP-d6) were converted to in-phase using 45°-OPE, yielding average 1H polarization of 14.4% for EP-d6 and 16.2% for EA-d6, with a 5–6× SNR boost versus PASADENA spectra under inhomogeneous B0. For EP-d6, optimal τ≈10 ms for OPE(3) and comparable results from 45°-CPMG. - Spin order transfer dynamics in intermediate coupling regime: At 0.55 T, 1H chemical shift differences (≤~5 ppm ≈ ≤120 Hz) are similar to J (~10 Hz), leading to intermediate coupling and non-sinusoidal SOT behavior. Simulations and experiments revealed non-linear dependence on τ and number of echoes n. - “Spying” on spins with 45°-CPMG: Recording echo amplitudes across up to 100 echoes for a single τ with one 300 µL injection enabled rapid mapping of SOT landscapes (n vs τ) that would otherwise require thousands of samples with conventional OPE. Entire 2D SOT maps were acquired in <30 min and matched simulations. - Relaxation and field stability: Permanent magnet B0 showed oscillations with SD ~80 Hz (corresponding to 3–4 mK temperature variation). 90°-CPMG measurements yielded T2 = 3.81 ± 0.01 s and D = 13.4 ± 0.2 s−1 from Robs vs echo spacing fits valid up to r ≤ 170 ms; deviations at longer r indicate time-varying B0 contributions. For short echo spacings (≤20 ms), observed relaxation approaches intrinsic T2, supporting efficient SOT under chosen conditions. - Reaction kinetics from hyperpolarized signals: Despite low resolution, hydrogenation rates were extracted via the build-up/decay model [VHH] = ([V0]P0)(e^(−t/T1) − e^(−k1 t)). At room temperature (22 °C): VA-d6→EA-d6 k1(5 bar) = 0.047 ± 0.012 s−1; k1(15 bar) = 0.11 ± 0.025 s−1. VP-d6→EP-d6 k1(5 bar) = 0.0086 ± 0.01 s−1; k1(15 bar) = 0.036 ± 0.027 s−1. VA hydrogenates faster than VP; higher pH2 pressure increased rates. - 13C hyperpolarization and imaging: Using ESOTHERIC (n1=n2=n3=2; optimized delays), 1-13C-EP-d6 achieved ~7% 13C polarization at 0.55 T. The hyperpolarized 50 mM 13C tracer signal corresponds to ~6000 M thermally polarized tracer at 0.55 T. The polarization decayed monoexponentially with Tdecay = 63 ± 3 s under 5°-FID sampling every 6 s. Multiple consecutive 13C-FLASH images (10 acquisitions, 1.5 s each, TR 50 ms, matrix 32×32, voxel ~0.312 mm², 10 mm slice) were obtained both in situ and after liquid shuttling into a receiver tube with a 3D-printed phantom. - Throughput and scalability: The setup supports semi-continuous production due to automation and short duty cycle. Manufacturer-available 20 mm bore magnets could scale reactor volume from 300 µL to >3 mL, approaching preclinical dDNP volumes.

The work demonstrates that an intermediate-field (0.55 T) permanent-magnet platform with robust automation can deliver practical PHIP hyperpolarization, address inhomogeneous B0 constraints via nonselective OPE-based SOT, and enable efficient reaction monitoring and imaging. Converting anti-phase PASADENA multiplets to in-phase via OPE mitigates line cancellation in inhomogeneous fields, yielding 5–6× SNR gains and enabling straightforward quantification. The modified CPMG strategy capitalizes on echo amplitude readouts to track spin order evolution over many refocusing events with minimal sample consumption, providing a rapid, experimentally tractable method to optimize SOT parameters, crucial under intermediate coupling where polarization transfer deviates from simple sinusoidal kinetics. Although the achieved 13C polarization (~7%) is lower than high-field ESOTHERIC reports (~60%), the results highlight the feasibility of benchtop PHIP with quasi-continuous operation and minute-scale duty cycles sufficient for repeated imaging. The system’s mobility, lack of cryogens, and self-calibration reduce barriers for widespread adoption and potential integration near preclinical MRI. Reaction kinetics could be quantified despite limited chemical shift resolution by leveraging hyperpolarization labeling, and pH2 pressure was shown to proportionally increase hydrogenation rates. The relaxation and B0 stability characterization confirm that, for short echo spacings relevant to SOT, the system is stable enough for efficient polarization transfer. Overall, the approach addresses key translational bottlenecks—cost, complexity, and throughput—by providing a portable, automated polarizer capable of producing hyperpolarized 1H and 13C tracers, monitoring hydrogenation, and performing imaging, thereby advancing PHIP toward in vivo applications.

This study introduces a portable, automated PHIP polarizer operating at 0.55 T that enables semi-continuous hyperpolarization with a ~1 min duty cycle. The device achieves substantial 1H polarizations (14.4% for EP-d6 and 16.2% for EA-d6) and 7% 13C polarization in 1-13C-EP-d6, sufficient for repeated 13C-FLASH imaging. OPE-based SOT overcomes challenges from inhomogeneous fields by converting anti-phase to in-phase signals, and a 45°-CPMG “spy” method efficiently maps SOT with minimal sample use. Automated operation supports extensive parameter sweeps and reaction monitoring, and the system’s mobility eases deployment near imaging sites. These contributions lower practical barriers for PHIP and point toward in vivo translation. Future work should target increased polarization and throughput via improved hydrogenation (higher temperature/flow, optimized mixing and catalysts), enhanced magnet temperature stability and field homogeneity, stronger gradients for higher-resolution imaging, elevated pressure-rated reactors, and scaling to larger bore magnets or continuous-flow/membrane reactors to reach milliliter-to-tens-of-milliliter doses required for clinical translation.

- Magnetic field inhomogeneity and drift in the permanent magnet (≈80 Hz SD in 1H frequency; temperature fluctuations of a few mK) limit spectral resolution and can reduce SOT efficiency; relaxation/echo decay deviates from simple models at longer echo spacings. - Achieved polarizations (1H ~14–16%, 13C ~7%) are lower than high-field implementations (up to ~60% 13C), attributable largely to field inhomogeneity and non-ideal hydrogenation conditions. - Hydrogenation time (~30–45 s for 50 mM precursor) is longer than best-in-class reports (≈5 s), likely due to temperature, gas flow, and mixing; current system lacks elevated temperature control. - Pressure constraints from standard NMR tubes (≤~13.7 bar used in some experiments despite hardware rated higher) limit further acceleration via higher pH2 pressure. - Gradient performance limited imaging resolution (matrix size and voxel dimensions), restricting spatial resolution. - Reactor volume (≈300 µL) limits single-shot dose; scaling to milliliter-scale or continuous production requires larger bore magnets/reactors and re-engineering. - Relaxation and sample diffusion/convection contribute to signal decay; sequence imperfections and B1 inhomogeneity can reduce SOT efficiency.