First, we discuss the results of the white-beam imaging obtained at the NEUTRA beamline, followed by the ToF-NI performed at the ICON beamline. Each section describes the ex-situ calibrations used to correlate the concentrations of species in the electrolyte with neutron attenuations, and the characterization of concentration profiles in the operando flow cells under various voltage biases and flow configurations. In the NEUTRA section, two sets of experiments are discussed, one with a low attenuating supporting salt (KPF6) and one with a highly attenuating counter-ion (BF4−), to differentiate between the redox active species and supporting ions.White beam neutron imaging (NEUTRA)Attenuation of electrolyte speciesAchieving contrast between the electrolyte constituents (solvent, redox-active species and supporting ions) is critical to identify species and quantify their dynamics within the electrochemical cell. White beam neutron imaging does not technically allow selectivity towards a target component, but it is possible to obtain insights into concentration distributions of individual electrolyte species by careful selection of the redox active species and supporting salt, coupled with the subtraction of reference images. We capitalize on the flexibility in the choice of solvent, supporting electrolytes and redox-active molecule for NAqRFBs, and measure attenuation coefficients for a set of electrolyte types and components using cuvettes (Fig. 2a). The attenuation difference between the pure deuterated solvent (CD3CN) and 0.2 M supporting salt solution (KPF6 in CD3CN) is sufficiently small to be neglected, confirming the negligible attenuation of KPF6 at this neutron energy and concentration. On the other hand, the addition of 0.5 M TEMPO in this electrolyte solution results in an increased attenuation coefficient as it has four methyl groups rich in hydrogen atoms attached to a piperidine ring (molecular formula C9H18NO·). The large number of hydrogen atoms results in a stark contrast between the supporting salt (KPF6) and the active species (TEMPO/TEMPO+). For the concentration range investigated in this study (0−0.5 M), TEMPO and TEMPO+PF6− dissolved in CD3CN show similar neutron attenuations (Fig. 2a, b). The similar cross sections of TEMPO and TEMPO+ are expected given their identical chemical composition (only one electron difference) resulting in almost identical interactions with neutrons, and further confirms the low attenuation of PF6− ions. Finally, when the counter-ion (PF6−) of TEMPO+ is replaced with BF4−, the  solvent-corrected attenuation at the same concentration is nearly doubled (Fig. 2a), which indicates that TEMPO species and BF4− ions have similar microscopic cross-sections. Although the counter-ion contains no hydrogen atoms, BF4− contains boron which features a large neutron absorption cross-section for thermal neutrons53. Figure 2b shows a linear correlation between neutron attenuation vs. concentration for the different species employed in this study, which confirms the validity of the chosen concentration range (0–0.5 M) where the Lambert-Beer law (Eq. (1)) holds. These reference measurements are then used to obtain local concentrations in the electrochemical reactor volume during operation.Fig. 2: Determination of the attenuation coefficient for the chemicals used in this study obtained at the NEUTRA beamline.a The attenuation coefficients of the different species in CD3CN: the solvent only, supporting electrolyte (0.2 M) and species TEMPO, TEMPO+PF6− and TEMPO+BF4− (all 0.5 M). b The linear dependence of the attenuation coefficient of the TEMPO, TEMPO+PF6− and TEMPO+BF4− species on the concentration (0.1, 0.2, 0.3, 0.4 and 0.5 M), where the shaded area represents the attenuation of the solvent.Transport of the active speciesWe performed neutron imaging on an operando redox flow cell to visualize concentration profiles of TEMPO/TEMPO+ (Fig. 3). The cell is connected to tanks with 50% SoC TEMPO/TEMPO+ at 0.5 M concentration on the counter electrode (CE) side and 0.2 M on the working electrode (WE) side, both with 0.2 M KPF6 to provide ionic conductivity and minimize supporting salt impact on neutron attenuation (Fig. 3a). We chose to have an offset in concentrations between both compartments to study the diffusive flux in the absence of reactions. Because the WE and CE compartments are separated by an anion exchange membrane, the transport of cations such as TEMPO+ and K+ is significantly hindered, whereas the anions and neutral molecules such as PF6− and TEMPO can more easily pass through. Furthermore, we elect to use a parallel (flow-by) flow field to limit the convective transport through the porous electrodes. Using this cell architecture and due to the negligible neutron attenuation of KPF6, we can track the movement of TEMPO between the electrodes. The cell is discharged (negative potential applied at the WE) and charged (positive potential applied at the WE) alternately, such that the state-of-charge after each complete cycle does not significantly deviate from the initial condition and two voltage magnitudes were applied to understand their impact on the potential-driven transport processes (e.g., migration). The electrochemical sequence goes through the open circuit voltage (OCV), −0.3, +0.3, −0.6 and +0.6 V steps, each for 20 min at the highest tested inlet flow rate of 15.1 mL min-1 (Fig. 3b). We also studied the impact of flow rate by performing the same electrochemical sequence (without the OCV step) at 5.6 mL min-1 (see Figure S1 in the Supporting Information). The current-time and voltage-time curves of the entire experiment can be found in Figure S2 and a video of the experiment can be found in the Supplementary Materials (see Supplementary Video 1). Operando imaging of the cell during the electrochemical protocol results in transmission images where the attenuation at each location represents the integral of neutron-matter interactions along the neutron path (Fig. 1a). These images are then averaged for the duration of a voltage step (20 min) and result in concentration maps for a given condition at steady-state (Fig. 3c, d). The colour scale represents the cumulative concentration of TEMPO and TEMPO+ and ranges from 0-0.5 M, resulting in a 2D map of the species concentration in the reactor area. The membrane area is omitted as the quantification of concentrations is not reliable in this region due to the high hydrogen content of the polymer membrane (perfluorinated with a polyketone reinforcement) and the reduced membrane thickness (130 μm). Finally, we calculate the concentration profiles across the thickness of the electrodes and compute these between the flow field-electrode interfaces of both half cells. Using this approach, one-dimensional concentration profiles, parallel with the membrane plane, are obtained.Fig. 3: Operando imaging of the active species transport in the NEUTRA beamline with the low attenuating KPF6 supporting salt.a Schematic representation of the non-aqueous cell designs during charge and discharge mode, where the counter electrode (CE) corresponds to 0.5 M TEMPO/TEMPO+PF6− at 50% state-of-charge in 0.2 M KPF6 and the working electrode (WE) to 0.2 M TEMPO/TEMPO+PF6− at 50% state-of-charge in 0.2 M KPF6. b Electrochemical sequence over time showing the applied potential steps and measured averaged current output at an inlet flow rate of 15.1 mL min-1. c, d Cumulative active species (TEMPO/TEMPO+) concentration profiles over the electrode thickness at an inlet flow rate of 15.1 mL min-1. The averaged snapshots of the cell after image processing and the concentration profiles are shown for various applied potential steps: c OCV, −0.3 V and +0.3 V and d −0.6 V and +0.6 V.The experiment begins with an OCV step where no current is drawn from the cell. The brighter colour of the CE side in the OCV radiograph represents a higher total active species concentration compared to the WE side, as expected by the concentrations of the electrolyte fed (0.5 M and 0.2 M TEMPO/TEMPO+). An advantage of neutron radiography is that electrolyte wetting of the porous electrodes can be visualized because of the low attenuation of gasses, which will appear as dark spots (i.e., lower concentration) in the radiographs54. During the OCV period, the concentration on both sides does not show dark regions (Fig. 3c, d), suggesting full wetting, at least to the spatial resolution of the measurement. Moreover, over the course of the OCV period, the concentration profile remained fairly constant which can be attributed to the low diffusion rate of TEMPO/TEMPO+ through the dense anion exchange membrane. Overall, the concentration profiles under cell polarization do not strongly deviate from the initial OCV state, except at positive potentials near the membrane area on the WE. The OCV profile can be explained by the reactor configuration (i.e., anion exchange membranes, flow-by flows fields) and the low ionic conductivity of the nonaqueous electrolyte, resulting in low current densities and charge consumption (< 15% of the tank capacity, see Figure S3). In this experiment, the concentration gradient is from left to right due to the higher cumulative TEMPO/TEMPO+ concentration on the CE side. Under negative potentials (−0.3 V and −0.6 V), TEMPO is converted to TEMPO+ in the CE side, resulting in a build-up of TEMPO+, while the opposite reaction is taking place in the WE, resulting in TEMPO+ depletion. To compensate for the charge, PF6− crosses through the membrane towards the CE side, which is not visible in the images due to its low attenuation. Although favourable to sustain the electrochemical reaction, we do not expect TEMPO to cross to the CE side on the timescale of this experiment as this would be against its concentration gradient, thus the images and profiles for negative potentials are nearly identical to the OCV conditions (Fig. 3c).On the contrary, when positive potentials are applied (+ 0.3 V) a concentration gradient develops in the WE and does not return fully to the OCV concentration profile when a positive potential step is applied afterwards. This result is supported by the bright concentration front in the corresponding radiographs near the membrane and could be attributed to the accumulation of TEMPO+ at the membrane due do Donan exclusion. Increasing the potential to +0.6 V amplifies this trend, exacerbating the concentration gradient and extending the concentration front deeper within the WE (Fig. 3d). Decreasing the flow rate to 5.6 mL min-1 further intensifies the concentration front in the WE (Figure S1) for both the positive and negative potentials. Interestingly, this concentration front does not appear in the CE at negative potentials, motivating the use of energy-selective neutron radiography to deconvolute the concentrations of the different species. The concentration front reveals mass transfer limitations, determined by the membrane properties, applied potential, electrolyte velocity, species concentration and electrolyte and electrode properties8. To visualize such limiting phenomena, we utilized a flow-by flow field design that induces limited convection within the porous electrode. However, the intensification of the concentration fronts at lower flow rate suggests that this flow field does have convective transport contributions in the electrode. Nevertheless, we anticipate that a convection-enhanced flow field (such as interdigitated or flow-through) would further increase species replenishment and reduce concentration gradients55. In this first set of experiments, the low attenuation coefficient of the KPF6 salt was utilized to maximize the contrast of TEMPO and TEMPO+ compared to other electrolyte components. To visualize the motion of anions, we then employ a strongly attenuating counter-ion (BF4− instead of PF6−) without any additional supporting salt to amplify the contrast between all species in the electrolyte.Transport of the counter-ionSupporting ions are essential in RFBs to provide ionic conductivity56,57. Here we leverage BF4− as counterion due to its high neutron attenuation (see Attenuation of electrolyte species)58. To assess the influence of migration on the charged species transport (i.e., stoichiometric operation), we do not add a supporting salt in the electrolyte (Fig. 4a), which negatively impacts the obtained current density (Figure S4) but enables visualization of the counterion. To amplify the counterion concentration, we utilize 0.5 M TEMPO on the CE side and 0.5 M TEMPO+BF4− on the WE side. Furthermore, the use of an anion-exchange membrane significantly restricts TEMPO+ movement between compartments, which leaves BF4− as the main charge carrier. As we perform subtractive neutron imaging, the isolation of [TEMPO], [TEMPO+] and [BF4−] is not possible with white beam neutron imaging, resulting in cumulative concentration maps (see the combined colour scale in Fig. 4). Although the concentration information is cumulative, by tuning experimental parameters (type of ion-exchange membrane and tank concentrations) we hypothesize that the observed changes can be attributed to the motion of certain species, which is predominantly BF4− in this configuration. This resonates with the subtle changes in the concentration profiles in Fig. 3 as a function of the applied potential, hinting that the neutron-transparent PF6− is the main charge carrier in the system. A novel approach to isolate the concentration of species in solution with neutron imaging is discussed in the energy selective imaging section.Fig. 4: Operando imaging of the active species transport in the NEUTRA beamline with the neutron attenuating BF4− supporting ion.a Schematic representation of the non-aqueous cell designs during charge and discharge mode, where the counter electrode (CE) corresponds to 0.5 M TEMPO and the working electrode (WE) to 0.5 M TEMPO+BF4−. b Electrochemical sequence over time showing the applied potential steps and measured averaged current output at two inlet flow rates of 15.1 mL min-1 and 5.6 mL min-1. c–e Cumulative active species (TEMPO/TEMPO+) and BF4− supporting ion concentration profiles over the reactor area. The averaged snapshots over the whole period of each individual potential step of the cell after image processing and the concentration profiles are shown for various applied potential steps and show the influence of various operation parameters: c applied potential magnitude (OCV, −0.3 V and −0.6 V at 15.1 mL min−1), d flow rate (−0.3 V at 15.1 mL min-1 and 5.6 mL min-1) and e polarization sign (−0.6 V and +0.6 V at 5.6 mL min-1).In the first hour of the experiment, the system is kept at OCV conditions to track diffusional crossover through the membrane (Fig. 4b). Although we track a change in OCV over time, indicating crossover and concentration equilibration, the small concentration variations in this short time period are not captured by the radiographs (OCV radiograph in Fig. 4c). In this experiment, BF4− has a strong concentration gradient towards the CE side, however the OCV profile does not show significant deviation from the initial concentrations, which is attributed to the Donnan exclusion of TEMPO+ coupled with the barrier properties of the dense anion exchange membrane59. Furthermore, the concentrations in the reactor volume are homogeneous and no local fluctuations are observed. We conclude that the timeframe of the OCV period is shorter than the time needed for diffusional crossover of species for this configuration.After the OCV period, alternating potential steps are applied to the electrochemical cell and the current response is recorded in time (Figure S4). A video sequence of the experiment can be seen in Supplementary Video 2. At negative potentials, TEMPO+ is converted to TEMPO in the WE, and BF4− migrates to the CE compartment (Fig. 4a). We can observe a local accumulation of attenuating species in the vicinity of the membrane on the CE side, while the opposite trend is observed in the WE compartment (Fig. 4c), attributed to the migration of the BF4− anion from the WE to the CE to maintain electroneutrality. This effect is more pronounced at a higher potential (−0.6 V vs. −0.3 V, Fig. 4c), which illustrates the influence of migration. A starker concentration gradient between compartments is obtained at lower flow rates (5.6 mL min-1, Fig. 4d), as the convective mass transfer is lower. At the WE, we find higher concentrations in the areas near the flow field inlets in comparison with the area under the ribs, showing an advantage of two-dimensional concentration maps obtained using neutron radiography. We hypothesize that velocity distributions within the porous electrodes, induced by the flow-by-flow field design and the relatively thick porous electrode, explain these variations in concentration through the electrode volume. This phenomenon is more visible at the highest flow rate as the convective forces pushing the electrolyte in the porous electrode are larger (Fig. 4d). In summary, at negative potentials, the profiles show an increasing species concentration in the CE occurring synchronously with a decrease in the WE (Fig. 4c).When a positive potential is applied to the WE, the reverse reactions take place, and the resulting radiographs and the concentration profiles (Fig. 4e) are closer to the initial OCV profile compared to negative applied potentials. Because of the reactor architecture used in the NEUTRA experiments (stacked paper electrodes and a flow-by-flow field) in combination with the low ionic conductivity of the electrolyte, there is only a small change in the cell capacity at negative potentials, consuming only ~8% of the total capacity (Figure S5), resulting in relatively low current densities. Therefore, at positive potentials, only a small amount of TEMPO is present in the WE compartment to be converted back to TEMPO+. As a result, large overpotentials are generated throughout the cell due to the low concentration of reactants to sustain the current. This explains the asymmetry in the current magnitudes when the polarity of the cell is reversed (i.e., +7 mA vs. −65 mA at 15.1 mL min-1 and +/−0.3 V, Fig. 4b), and the even lower capacity recovery (~1-2%) resulting in underutilized capacity over the duration of the experiment (Figure S5).When comparing the experiments with counter-ion BF4− and supporting salt KPF6, we can correlate the macroscopic performance with the concentration distributions through the reactor. For the KPF6 experiments, all active species (i.e., TEMPO, TEMPO+, PF6−) are present in both compartments. Therefore, the macroscopic performance, i.e., the current output, is symmetric when operating at negative and positive applied potentials (Fig. 3b) as the to-be-reacted species are present without the requirement of species crossover under the evaluated conditions (as the capacity change is limited to ~20%). The symmetric current output results in concentration profiles returning to the OCV profile when positive potentials are applied. Whereas for the BF4− experiment, the charged species are only present in one compartment (WE) initially and are required to cross the membrane to support the reactions, which is limited by the anion exchange membrane, resulting in asymmetric current magnitudes upon changing cell polarities (Fig. 4b). The asymmetric current can be correlated to the concentration profiles, as the concentration does not fully return to the OCV profiles for positive applied potentials (Figure S6).Using white beam neutron imaging, we have obtained cumulative concentration maps which include active species and supporting electrolytes. Using this approach, we have coupled macroscopic electrochemical cell performance with microscopic concentration distributions, revealing mass transfer modes under different cell potentials, flow rates and cell polarities. However, we are not able to isolate concentrations of active species and supporting ions with this incident beam. Acknowledging these limitations, we then utilize time-of-flight imaging to visualize reactive transport phenomena of both the active species and counter-ion, under similar experimental conditions.Energy-resolved neutron imaging (ICON)In pursuit of deconvoluting the concentrations of different species in the electrolyte, we investigate the use of energy-resolved neutron radiography at the ICON beamline. This beamline utilizes a colder neutron spectrum by secondary moderation of the neutron beam, and slower neutrons undergo inelastic scattering events with a higher probability than thermal neutrons, allowing more variations in species cross-sections to be observed. It is also possible to perform spectral neutron imaging with a time-of-flight based technique at ICON, which is currently not possible at the NEUTRA beamline due to space limitations. Since the time-of-flight of neutrons in the flight tube is inversely proportional to the square root of their energy, the ToF-NI technique can add a fourth dimension to conventional radiography. This can provide an additional mode of contrast as neutron attenuation is a function of its energy. We anticipate that if the neutron attenuation of active species and the supporting ions have distinct energy dependency profiles, we can separate the contribution of each species from the final radiograph.Correlating attenuation with neutron energyThe difference in relative neutron attenuation of materials enables tuning of the contrast between different species. To this end, we first performed calibration experiments with cuvettes, filled with 0.5 M solutions of TEMPO and TEMPO+BF4− in CD3CN. Figure 5a shows attenuation coefficients as a function of the time of flight, where the BF4− attenuation coefficient is determined by subtracting the coefficient of TEMPO from TEMPO+BF4−. Here, an increasing time-of-flight indicates a decreasing neutron energy. TEMPO+BF4− reaches nearly twice the cross-section of TEMPO at higher energies, corroborating the previous observations made at the NEUTRA beamline (Fig. 2) that TEMPO and BF4− have similar microscopic cross-sections. The linearity of the concentration with neutron attenuation was already demonstrated in the NEUTRA beamline, thus we selected only one concentration (0.5 M) corresponding to the starting concentration in the flow cell experiments.Fig. 5: Energy selective imaging at the ICON beamline.a Energy dependency of the attenuation coefficient of TEMPO, TEMPO+BF4− and BF4− obtained from the cuvette experiments (all 0.5 M), where the time-of-flight is a function of the neutron energy. b Schematic representation of the main components in the image processing sequence for the ICON beamline experiments, including the table of microscopic cross-sections, the low and high energy transmission images (grayscale) with the regions of interest and deconvoluted active species and supporting ion images of the flow cell, showing the flow fields, electrodes and membrane together with their dimensions. The transmission images are processed using the given equation to extract the concentration maps (coloured images) of TEMPO/TEMPO+ and BF4−.Using the matrix operation shown in Eq. (4) and in Fig. 5b, the respective contributions of TEMPO and BF4− from the total neutron attenuation can be separated. For this purpose, we need to define two regions within the spectrum, the high energy (HE) and low energy (LE) regions (see the Neutron Radiography section). The difference in attenuation coefficients between TEMPO and BF4− varies as a function of neutron energy, this means that the slope of the graph in Fig. 5a should be different between species, or in mathematical terms, the determinant of the microscopic cross-section matrix should not be zero. The microscopic cross-sections of the species of interest are reported for HE and LE regions in Fig. 5b. The values reported here correspond to the microscopic cross-section averaged over LE and HE ranges, described in the Neutron Radiography section. Although maximum contrast is achieved around 8 ms ToF, the LE region was moved towards higher energies to prevent the excessive neutron edge effects/scattering at interfaces between gaskets observed at lower energies. Finally, the matrix operation is applied pixel-wise to the greyscale transmission image to calculate the contribution of species, and a colour map is applied to designate the concentrations (Fig. 5b). Achieving contrast between TEMPO and TEMPO+ is still not possible, but because the movement of TEMPO+ between compartments is mostly blocked by the anion exchange membrane, we can track the movement of BF4− and TEMPO species separately during battery operation.Deconvoluting concentrations in a flow cellTo demonstrate the potential of energy-selective and operando neutron imaging, a flow cell with asymmetric concentrations (0.5 M TEMPO on the CE and 0.5 M TEMPO+BF4− on the WE side, Fig. 6a) was imaged. The electrolyte compositions are identical to the previous experiment that utilized BF4− as the counter-ion but we increased the electrode thickness to accommodate for the lower spatial resolution. Previous experiments carried out on the NEUTRA beamline were set up with a tilted detector to increase the spatial resolution of the images60, resulting in a pixel size of ~6 µm applied to study a 630 µm thick electrode. In the ICON experiments, the ToF detection system resulted in a larger pixel size (~55 µm). Thus, to compensate for the discrepancies in spatial resolution, a thicker felt electrode (3200 µm) was employed. The cell features stacked gaskets (incompressible PTFE and compressible ePTFE) to enclose the thick felt electrode, where the interface around the ePTFE gaskets shows up in the LE image as low transmission regions (dark vertical lines in the grayscale image in Fig. 5b) due to higher neutron edge effect/scattering. Nevertheless, the central regions of the incompressible gaskets (~1 mm) were large enough to define four regions of interest (Fig. 5b) where the concentrations can be determined, and the reported concentrations are averaged over this volume for both compartments. To track the movement of species during the sequence, we opted for plotting the averaged concentrations over time (Fig. 6c).Fig. 6: Operando imaging of the active species transport in the ICON beamline with the neutron attenuating BF4− supporting ion.a Schematic representation of the non-aqueous cell designs during charge and discharge mode, where the counter electrode (CE) corresponds to 0.5 M TEMPO and the working electrode (WE) to 0.5 M TEMPO+BF4−. b Electrochemical sequence over time showing the applied potential steps and measured averaged current output at two inlet flow rates of 21.1 mL min-1 and 6.7 mL min-1. c Deconvoluted active species (TEMPO/TEMPO+) and BF4− supporting ion concentration profiles. The averaged snapshots of the cell after image processing and the concentration profiles over time are shown for various applied potential steps and flow rates: OCV, −0.6 V and +0.6 V at 21.1 mL min-1 and −0.6 V, +0.6 V, −0.3 V and +0.3 V at 6.7 mL min-1, from left to right, where the OCV images are averaged over 5 images and the applied potentials averaged over 4 images.The experiment starts with an OCV period of 10 min, after which the cell was polarized at an electrolyte flow rate of 21.1 mL min−1 followed by a reduced flow rate of 6.7 mL min-1 (Fig. 6b and S7). From the deconvoluted concentration maps (Fig. 6c), we confirm that BF4− is the main charge carrier and that the membrane blocks the transport of TEMPO/TEMPO+, as their concentration remains relatively stable in both compartments throughout the entire electrochemical sequence. When a negative potential is applied, TEMPO+ reduces to TEMPO in the WE compartment while the reverse reaction occurs in the CE (Fig. 6c). Simultaneously, the concentration maps and averaged concentrations show that BF4− moves through the membrane towards the CE to balance the positive charge of the generated TEMPO+ species. Applying a positive potential to the WE reverses the direction of the migration flux of the BF4− ions, and concentrations close to the initial state of the battery (i.e., OCV) can be recovered. These results corroborate the observations from the experiments in NEUTRA as only minor concentration fluctuations were observed with PF6− as supporting salt when an electric field is applied, whereas stark changes are detected when the supporting ion was changed to BF4− due to its strong neutron attenuation. Moreover, the concentrations of the active species within the reactor area show larger variations for the highest flow rate (Fig. 6c), induced by faster species conversion (i.e., higher current densities, Fig. 6b), and greater convective transport in the porous electrode. This brings the concentrations to extreme values due to the fast depletion of reactants in the electrolyte and promotes larger ionic currents. The difference in ionic current from the electrochemical data is correlated to the slope of the BF4− concentration variations as a function of time.From the capacity curves (Figure S8), we observe that after the first potential step (−0.6 V at 21.1 mL min-1), 60% of the total capacity is discharged. In the next step, after an applied potential of +0.6 V, only 45% of the total capacity is recharged (due to the lower current at set time), resulting in 15% underutilized capacity after a full polarization cycle because of the starting tank solutions (no BF4− in the CE) as explained in the transport of the counter-ion section. At the lower flow rate (6.7 mL min-1), the capacity consumed at negative applied potentials is almost fully recovered at positive applied potentials, resulting in near symmetric current magnitudes. Interestingly, we find comparatively higher currents and capacity utilization with this reactor configuration in comparison with the reactor architecture used in the NEUTRA beamline (Fig. 4), which can be correlated to the significant BF4− concentration fluctuations. We attribute these differences to the use of a different electrode material (a thick felt vs. a stack of thin carbon papers) and lower compressive forces. The higher porosity, apparent permeability and internal surface area of the felt electrode can explain the higher current densities observed in this reactor configuration.Here, we demonstrate the potential of the ToF-NI spectral technique to isolate and visualize concentration distributions of active and supporting species in redox flow cells. Compared to the use of conventional neutron radiography, the ToF method requires larger acquisition times and provides lower spatial resolution but enables the detection of neutron energies necessary to deconvolute species concentrations.Practical application of the neutron imaging methodThis work demonstrates for the first time the use of neutron radiography to image concentrations of redox-active species and supporting salts in operando electrochemical flow cells. By combining macroscopic electrochemical responses with microscopic concentration distributions, neutron radiography can provide valuable insights into species motion within the reactor area, and this can be used to quantify mass transport methods (migration, diffusion, convection) and phenomena affecting the performance of the battery during operation (e.g., electrolyte depletion, precipitation, physical failure in RFB stacks). These insights can be directly used to compare and select optimal cell components and to aid computational efforts. We anticipate that the use of molecular engineering to design redox molecular probes with controlled molecular structure, diffusivity and redox potential, can enable deconvolution of different oxidation states and degradation products. Although we focus on nonaqueous redox flow cells as a case study, neutron imaging can be extended to numerous applications to quantify concentration distributions in electrochemical cells and beyond. First, the resulting concentration maps can be used as experimental data to validate computational models that describe reactive mass transport. Here we used an electrolyte composed of a solvent, supporting electrolyte, and two redox active species – hence a complex, multicomponent system close to practical devices; but model experiments can be performed (e.g. with one or two analytes) to systematically deconvolute mass transport modes (diffusion, convection and migration) and their associated transport rates. Deeper fundamental understanding of reactive mass transport in electrochemical reactors and through membranes can aid in the design of advanced electrochemical cells. Second, the methodology enables identification of local maldistributions in concentration, which can assist in designing better flow field geometries and electrodes, as well membrane crossover, Donnan exclusion and salt precipitation within the electrochemical cell, which are deleterious to performance and lifetime in several flow battery chemistries (e.g. non-aqueous, all-vanadium, all-iron). Third, we anticipate that the method – and adaptations on the detection physics – will be instrumental in advancing hybrid redox flow batteries (e.g. all-iron, zinc-bromine), where there are phase change reactions (e.g. plating and stripping or hydrogen evolution) that fundamentally limit the performance of the system. Fourth, we expect that the technique can be applied to technical systems such as electrochemical stacks, where traditional neutron imaging was instrumental in the development of fuel cell stacks through visualization of water distributions. Fifth, beyond redox flow batteries, the method can be applied to other (electro)chemical reactors where concentration profiles determines performance such as electrochemical separations, flow chemistry and chemical reactor design. To further assist the design of neutron experiments to study electrochemical systems, Tables S1 and S2 summarize the neutron attenuation coefficient of commonly used materials for reactor manufacturing and redox species/supporting salts, respectively.Finally, the use of molecularly engineered redox molecules acting as imaging probes might enable simultaneous visualization of the concentration of multiple components (> 2) when combined with energy-selective neutron imaging. For example, minimizing the neutron attenuation of the redox active molecules through a reduced hydrogen content or deuterium labelling is a viable strategy to obtain contrast. Although the technique is still in its early stages, it displays considerable potential. We anticipate that ongoing advancements in neutron detectors and choppers will enable more sophisticated analyses of complex multicomponent systems using ToF-NI, offering enhanced spatial, temporal and energy resolution.
