Physicists Turn NMR’s Worst Enemy Into a Precision Tool for Watching Batteries and Fuel Cells Work

For decades, one of the most powerful tools in chemistry has been quietly sabotaged by one of the most common materials in technology. Nuclear magnetic resonance, or NMR, the technique behind hospital MRI scanners and countless laboratory breakthroughs, struggles badly whenever it is pointed at an operating electrochemical cell. The reason is simple and frustrating: batteries, fuel cells and electrolyzers are full of metal, and metal distorts the magnetic fields that NMR depends on. Now a team at Forschungzentrum Jülich in Germany has flipped that problem on its head, showing that the very distortions that ruin NMR experiments can be deliberately harnessed to make the technique more precise than ever before.

The research, published in the journal Magnetic Resonance, describes a workflow that combines finite element method simulations with quantum optimal control, a mathematical technique for sculpting radiofrequency pulses with extraordinary finesse. Instead of trying to cancel out the magnetic havoc wreaked by conductive components, the researchers designed custom pulses that exploit those distortions to excite only specific regions of a sample. In a proof-of-concept experiment, they achieved more than a thirtyfold suppression of an unwanted signal, demonstrating spatial selectivity without any of the magnetic field gradients that conventional imaging requires.

To understand why this matters, it helps to grasp what NMR actually measures. The technique relies on two magnetic fields working in concert. The first, the static field B0, causes atomic nuclei to precess at a characteristic frequency called the Larmor frequency, which acts like a fingerprint revealing the chemical identity of each molecule. The second, an oscillating radiofrequency field B1, tips the nuclear spins so that their signal can be detected. In a clean glass tube of liquid, both fields are well behaved. Inside an electrochemical cell, they are anything but.

Metal electrodes generate eddy currents when exposed to changing magnetic fields, and these currents in turn produce their own magnetic fields that interfere with the measurement. Susceptibility gradients at material interfaces warp the static field, broadening spectral lines and shifting resonance frequencies. The radiofrequency field is attenuated and phase-shifted near conductors through the skin effect, meaning that a pulse calibrated to work perfectly in one part of the cell will deliver the wrong flip angle a fraction of a millimeter away. Conventional selective pulses, such as the widely used BURP family of shaped pulses, simply were never designed for an environment this hostile.

Previous work had shown that these distortions, while troublesome, are at least predictable. Finite element method simulations can accurately reproduce how B0 and B1 fields behave around conductors of various shapes and orientations, and researchers have used this capability to design cells with more uniform fields. The Jülich team, led by Johannes F. Kochs and corresponding author Simone S. Köcher, took a different tack: rather than engineering the distortions away, they built them into the pulse design itself. The key tool was quantum optimal control, specifically the GRAPE algorithm, which uses gradient-based optimization to find pulse shapes that drive a spin system to a desired final state.

The researchers employed a variant known as pattern pulses, in which the optimization runs simultaneously across an entire ensemble of simulated spin systems, each representing a different combination of Larmor frequency and nutation frequency. By assigning different target states to different ensemble elements, the algorithm learns to excite some combinations of fields while suppressing others. The resulting pulses were remarkably efficient: a one-millisecond excitation pulse achieved a mean quality factor of nearly 95 percent across its entire design space, and suppression pulses reached similar performance when extended to two milliseconds.

To test the concept experimentally, the team built an elegant model setup inside a shortened 5-millimeter NMR tube. Two tiny cavities, each just 0.1 millimeters tall, were sandwiched between pairs of coins. One pair was copper, mimicking the conductive environment of a real electrode; the other was polyether ether ketone, a polymer serving as a distortion-free reference. The cavities were filled with two immiscible liquids, n-dodecane in the copper cavity and water in the polymer cavity, chosen so that their signals could be distinguished both by chemical shift and by position.

The results were striking. When the team applied frequency-selective pulses tuned to the water resonance, the water signal was excited efficiently while the n-dodecane signal fell to less than 5 percent of its original value, and vice versa. With longer suppression pulses, unwanted resonances could be driven down to around 1 percent of their reference intensity. Crucially, these pulses remained robust against the nutation frequency variations caused by the nearby copper, something conventional selective pulses could not manage. A direct comparison with a standard E-BURP pulse showed baseline distortions and poor selectivity in the same setup, underscoring how much the optimal control approach gains in conductive environments.

The more visionary result, however, came from inverting the logic. Instead of selecting by frequency, the team designed pulses that select by nutation frequency, exploiting the fact that the radiofrequency field is measurably stronger near metal. Finite element simulations predicted a 25.7 percent enhancement of B1 in the copper cavity, and nutation experiments measured 25.6 percent, an agreement to within a tenth of a percent. Pulses optimized for that specific enhancement selectively excited the liquid near the copper while suppressing the reference liquid, achieving spatial selectivity entirely without magnetic field gradients. Because the selectivity depends on the surrounding material rather than on chemical shift, it works even when different regions of the sample contain the same molecule.

The implications reach well beyond the laboratory bench. Electrochemical reactions happen at interfaces, precisely where standard NMR is blindest, drowned out by bulk solvent signals and hampered by poor surface sensitivity. The new workflow opens a path toward gradient-free, localized in operando NMR, with the prospect of surface selectivity down to the detection limit of the instrument. The authors point to applications such as tracking intermediates during carbon dioxide electrolysis or probing the solid electrolyte interphase that forms inside lithium batteries, questions that individual techniques have struggled to answer alone. Challenges remain, particularly for porous electrodes and inhomogeneous catalyst layers whose field distortions shift during operation. But the core message of the study is a genuinely subversive one for the field: the magnetic field distortions that have plagued electrochemical NMR for fifty years are not merely obstacles to be tolerated. Designed for correctly, they are features waiting to be exploited.

Subject of Research: Quantum optimal control NMR pulse design for spatially selective spectroscopy in electrochemical cells

Article Title: Optimally controlled nuclear magnetic resonance (NMR) in electrochemistry: Larmor versus nutation frequency selective spin excitation for locally selective NMR experiments

Article References: Optimally controlled nuclear magnetic resonance (NMR) in electrochemistry: Larmor versus nutation frequency selective spin excitation for locally selective NMR experiments. (n.d.). https://doi.org/10.5194/mr-7-113-2026

Image Credits: AI Generated

DOI: 10.5194/mr-7-113-2026

Keywords: NMR spectroscopy, electrochemistry, quantum optimal control, GRAPE algorithm, finite element simulation, radiofrequency pulses, magnetic field distortion, in operando, batteries, fuel cells, solvent suppression, spatial selectivity