Quantum Symmetry Trick Makes Ultralow-Field NMR Simulations 50 Times Faster

Nuclear magnetic resonance is one of the most information-rich tools in science, but it has always fought a battle against sensitivity. The magnetic signals from atomic nuclei are extraordinarily faint, and for many nuclei of interest — carbon-13, nitrogen-15, phosphorus-31 — the intrinsic weakness of the signal makes even basic experiments slow and difficult. Hyperpolarization techniques promise to change that, and among them, signal amplification by reversible exchange, or SABRE, has become one of the most versatile. Now a team of researchers from the International Tomography Center in Novosibirsk and the University of Miami has tackled a quieter but equally important problem: the sheer computational cost of predicting how SABRE works in realistic molecules. Their solution, published in the journal Magnetic Resonance, is a symmetry-based framework that makes simulations of large multi-spin systems not just faster but, for the first time, genuinely practical.

To understand why the achievement matters, it helps to grasp what SABRE actually does. The technique exploits parahydrogen, a special quantum spin state of hydrogen gas in which the two proton spins are locked into an antisymmetric singlet configuration. When parahydrogen binds transiently to a metal complex — typically an iridium catalyst — alongside a substrate molecule, the nuclear spins of the hydrides and the substrate become strongly coupled. During that fleeting encounter, the singlet spin order of the parahydrogen is coherently transferred to the nuclei of the substrate. When the substrate dissociates, it carries away a dramatically enhanced nuclear polarization, boosting NMR signals by orders of magnitude without any permanent chemical modification of the molecule being studied.

The twist in this new work is the magnetic field regime in which the transfer happens. In zero- and ultralow-field NMR, often abbreviated ZULF, experiments are performed in fields so weak — microtesla and below — that the Zeeman interactions of the spins with the external field are comparable to or weaker than the scalar J-couplings that connect the spins through chemical bonds. In this regime, spin states that would remain separate at high field mix coherently, revealing interactions and dynamics that are effectively invisible in conventional spectrometers. ZULF NMR also offers practical advantages: it can be carried out in compact magnetically shielded setups with optically pumped magnetometers instead of superconducting magnets, and the absence of a strong static field eliminates line broadening from field inhomogeneity, yielding exceptionally sharp spectra.

Simulating this physics rigorously, however, is brutal. The standard approach treats the quantum state of the spin system in Liouville space, where the dimension of the problem grows as four to the power of N, with N the number of spins. A realistic SABRE complex containing a substrate plus two hydride ligands can easily reach fourteen spins, which corresponds to a matrix of roughly 4.3 billion elements before any dynamics is even considered. Worse, a faithful SABRE model must simultaneously track coherent spin evolution, relaxation, and reversible chemical exchange between the free substrate and the metal complex. For anything beyond small molecules, full simulations become computationally intractable on ordinary hardware, and the field has lacked a general, scalable tool for predicting how polarization builds up and what the resulting ZULF spectra look like.

The insight of Danil Markelov, Alexander Snadin, Alexey Kiryutin, Danila Barskiy, and Alexandra Yurkovskaya is that the equations governing SABRE at ultralow fields hide a powerful symmetry. The Hamiltonian, the relaxation superoperator, and even the chemical exchange superoperators all commute with the z-projection of the total spin, where the z-axis is set by the residual ultralow magnetic field. In the language of NMR, the dynamics conserves the coherence order. Because the initial state of the system — an unpolarized substrate and a parahydrogen singlet — also has zero coherence order, the entire evolution is rigorously confined to the so-called zero-quantum coherence subspace. Everything outside that subspace remains identically zero for all time and can be discarded without any approximation whatsoever.

The team layered a second reduction on top of this. Molecules frequently contain groups of magnetically equivalent nuclei, such as the three protons of a methyl group or the two protons of a methylene unit. Quantum mechanically, such a group can be treated as a single effective pseudo-spin whose total spin quantum number takes a small set of discrete values, each weighted by well-defined statistical factors. Combining the effective-spin treatment with the zero-quantum coherence restriction shrinks the problem dramatically. For a system of N non-equivalent substrate spins, the matrix dimension falls by a factor of roughly πN, and the computation time drops by a factor proportional to N. Crucially, the reduction is exact: no physics is thrown away.

The validation was unambiguous. For isotopically labeled acetonitrile containing nitrogen-15 and two carbon-13 nuclei — eight spins once the hydrides are included — the symmetry-reduced simulations reproduced the full, unreduced calculations with a relative numerical residual of about one part in one hundred thousand, while running roughly thirty to fifty times faster. The researchers computed the magnetic field dependence of the hyperpolarization for each nucleus in the molecule, revealing broad, structured profiles extending up to several microtesla, with multiple maxima, minima, and even sign changes arising from the coherent interplay of the nitrogen, carbon, and proton spins. They also mapped how the polarization depends on the substrate dissociation rate, finding that the optimal polarization field sits near half a microtesla across a wide range of exchange kinetics.

The real payoff came with butyronitrile, a twelve-spin substrate whose SABRE complex contains fourteen spins in total. A full Liouville space simulation of a single magnetic field point would take roughly three hundred hours on a desktop workstation; with only the effective-spin reduction, it would still take about a day. The zero-quantum coherence reduction cut the computation to about three and a half hours per field point — an eighty-six-fold speedup — making it feasible to compute complete ZULF NMR spectra in around one hundred hours of computing time. The simulated spectra showed the expected low-frequency features near ten hertz, broadened by the dense network of couplings, along with characteristic high-frequency fingerprints of the methyl and methylene groups at multiples of the one-bond carbon-hydrogen coupling of 136 hertz.

Beyond the immediate results, the framework establishes something the field has lacked: a predictive, quantitative bridge between the chemistry of the catalyst and the spectra observed by atomic magnetometers. With it, researchers can systematically search for optimal polarization transfer fields, estimate how exchange rates and relaxation times shape the outcome, and design new hyperpolarization protocols before committing to the bench. The authors note that the approach applies across the full range of coupling regimes, from the J-coupling-dominated limit to the Zeeman-dominated limit, and covers the common relaxation mechanisms relevant to SABRE, including dipolar relaxation and chemically shifted anisotropy with axial symmetry. The main limitation is that transverse radiofrequency pulses, which break the coherence-order symmetry, fall outside its scope — but at ultralow fields, where the most interesting dynamics happens spontaneously, that is rarely a constraint.

The broader implications stretch toward applications that have energized the hyperpolarization community in recent years: metabolic imaging with hyperpolarized pyruvate, sensitive detection of disease-marker enzymes, and spectroscopy of biomolecules at natural isotopic abundance, all of which benefit from heteronuclear detection that avoids the overwhelming water background of proton NMR. By making rigorous simulation of chemically diverse, multi-spin systems routine, the symmetry-based framework turns a computational bottleneck into a design tool. As ZULF NMR instruments shrink from laboratory curiosities into compact, magnetometer-based devices, having theory that keeps pace with experiment may prove just as decisive as the hardware itself.

Subject of Research: Symmetry-based computational modeling of SABRE parahydrogen hyperpolarization and spin dynamics in zero- and ultralow-field NMR

Article Title: Scalable modeling of multi-spin ensembles in SABRE hyperpolarization: a symmetry-based framework for zero and ultralow fields

Article References: Markelov, D., Snadin, A., Kiryutin, A., Barskiy, D., & Yurkovskaya, A. (2026). Scalable modeling of multi-spin ensembles in SABRE hyperpolarization: a symmetry-based framework for zero and ultralow fields. Magnetic Resonance, 7(1), 53-79. https://doi.org/10.5194/mr-7-53-2026

Image Credits: AI Generated

DOI: 10.5194/mr-7-53-2026

Keywords: SABRE, hyperpolarization, parahydrogen, ZULF NMR, zero-field NMR, spin dynamics, Liouville space, zero-quantum coherence, magnetic resonance, computational simulation, J-coupling, optically pumped magnetometers