An electrochemical cell for in operando 13C nuclear magnetic resonance investigations of carbon dioxide/carbonate processes in aqueous solution

Abstract

In operando nuclear magnetic resonance (NMR) spectroscopy is one method for the online investigation of electrochemical systems and reactions. It allows for real-time observations of the formation of products and intermediates, and it grants insights into the interactions of substrates and catalysts. An in operando NMR setup for the investigation of the electrolytic reduction of ${\mathrm{CO}}_2$ at silver electrodes has been developed. The electrolysis cell consists of a three-electrode setup using a working electrode of pristine silver, a chlorinated silver wire as the reference electrode, and a graphite counter electrode. The setup can be adjusted for the use of different electrode materials and fits inside a 5 mm NMR tube. Additionally, a shielding setup was employed to minimize noise caused by interference of external radio frequency (RF) waves with the conductive components of the setup. The electrochemical performance of the in operando electrolysis setup is compared with a standard ${\mathrm{CO}}_2$ electrolysis cell. The small cell geometry impedes the release of gaseous products, and thus it is primarily suited for current densities below 1 mA cm${}^{-2}$. The effect of conductive components on ${}^{13}$C NMR experiments was studied using a ${\mathrm{CO}}_2$-saturated solution of aqueous bicarbonate electrolyte. Despite the $B_0$ field distortions caused by the electrodes, a proper shimming could be attained, and line widths of ca. 1 Hz were achieved. This enables investigations in the sub-Hertz range by NMR spectroscopy. High-resolution ${}^{13}$C NMR and relaxation time measurements proved to be sensitive to changes in the sample. It was found that the dynamics of the bicarbonate electrolyte varies not only due to interactions with the silver electrode, which leads to the formation of an electrical double layer and catalyzes the exchange reaction between ${\mathrm{CO}}_2$ and ${\mathrm{HCO}}_3^{-}$, but also due to interactions with the electrochemical setup. This highlights the necessity of a step-by-step experiment design for a mechanistic understanding of processes occurring during electrochemical ${\mathrm{CO}}_2$ reduction.

Figure 1

(a) Geometry and arrangement of the three-electrode in operando NMR setup. It consists of a silver foil working electrode (WE), a graphite rod counter electrode (CE), and a micro Ag  $/$  AgCl reference electrode (RE). The reference electrode was placed on the edge of the sensitive NMR area to minimize the amount of conductive material during NMR measurements while maintaining a small ohmic potential drop between the working and reference electrodes. (b) Photograph of the electrode setup inside a 5 mm tube.

Figure 2

Cell holder consisting of the 3D-printed frame (black) and three SMA coaxial connectors (white and gold). The electrolysis cell is fixed inside the cell holder and the electrode wires are soldered to the pins of the SMA coaxial connectors.

Figure 3

Schematic drawing of the in operando NMR electrolysis setup with shielding, RF filters, and potentiostat.

Figure 4

$B_{\mathrm{1}}$ field simulation in the proximity of the metal electrode for angles of 0 ${}^{\circ }$  (a), 45 ${}^{\circ }$  (b), and 90 ${}^{\circ }$  (c) between the direction of the incoming RF field and the electrode surface, and geometry and arrangement of the metal electrode in relation to the $B_{\mathrm{1}}$ field in the simulations (d). The incoming RF field points towards the positive $x$ axis. The vectors represent the deviations in field strength and direction compared to the undistorted RF field. Deviations smaller than $B_{\mathrm{1}}/e$ are not shown in order to increase clarity. For a better visibility of phase deviations, the vectors are color coded according to their relative field strength in the $y$ direction compared to the incoming field amplitude. Note that all figures have individual color-bar ranges. No distortion is present for a parallel (0 ${}^{\circ }$ ) orientation of the RF field and electrode. The angled (45 ${}^{\circ }$ ) and perpendicular (90 ${}^{\circ }$ ) orientations cause major distortions in immediate proximity to the electrode, which diminish at a distance of 0.6–0.8 mm.

Figure 5

(a) Nutation curves of the ${}^{\mathrm{1}}$ H water resonance using the in operando cell with electrode orientations of 0 ${}^{\circ }$ (blue), 45 ${}^{\circ }$ (red), and 90 ${}^{\circ }$ (green). The nutation curve of a water sample without an electrode is shown for comparison (black dashed line). Deviations from the undistorted nutation curve are largest for the perpendicular electrode orientation and minimal for the parallel orientation. (b) Fourier transform of the nutation curves. The main component of the undistorted sample nutates at a frequency of 25.6 kHz (15  $\mathrm{\mu }$ s 90 ${}^{\circ }$ pulse length). For the samples with electrode setups, the width of the main component increases and a low-frequency component appears. (c)  ${}^{\mathrm{1}}$ H water NMR spectrum with and without electrode setup. The $B_{\mathrm{0}}$ field was not shimmed after electrode insertion. The signal shape is mainly governed by $B_{\mathrm{0}}$ field distortions and only slightly affected by deviations in the $B_{\mathrm{1}}$ field.

Figure 6

(a) Time-dependent potential curves during the chronopotentiometry measurement. Electrolytic reduction of ${\mathrm{CO}}_{\mathrm{2}}$ starts at $-$ 1.33 V vs. Ag  $/$  AgCl for the in operando cell. Compared to the bulk cell, higher overpotentials are observed. Starting at 1 mA cm ${}^{-\mathrm{2}}$ , oscillations and increased noise appear, which are caused by stuck product gas bubbles. (b) Tafel plot of the electrolytic ${\mathrm{CO}}_{\mathrm{2}}$ reduction in the in operando electrolysis cell. The Tafel slope was determined in the low current density region as 148 mV per decade, resulting in a transfer coefficient of 0.38 at 10  ${}^{\circ }$ C.

Figure 7

${}^{\mathrm{13}}$ C spectrum of the ${\mathrm{CO}}_{\mathrm{2}}$ -saturated electrolyte without (a) and with (b) electrodes. Measurements with electrodes include connection cables and a powered potentiostat but no shielding. The peak positions of bicarbonate and solvated carbon dioxide are at 160.7 and 124.7 ppm, respectively. Peak positions are shifted downfield by about 1.1 ppm when the conductive components are introduced. The spectrum in (b) suffers from increased noise as well as from external RF signals, which are comparable in intensity to the ${\mathrm{CO}}_{\mathrm{2}}$ signal. External RF signals have been marked with (*).

Figure 8

Time evolution of the ${}^{\mathrm{13}}$ C signals for ${\mathrm{HCO}}_{\mathrm{3}}^{-}$  (a) and ${\mathrm{CO}}_{\mathrm{2}}$  (b) during the OCV stage vs. electrochemical potential between the working and reference electrodes and current density between the working and counter electrodes. In each subpanel the time-dependent potential and current density are shown on the left, with the corresponding spectra given on the right. After a relaxation period the potential remains at a stable at 47 mV.

Figure 9

Time evolution of the ${\mathrm{HCO}}_{\mathrm{3}}^{-}$  (a) and ${\mathrm{CO}}_{\mathrm{2}}$  (b) signal integrals during the OCV stage. The integrals were normalized to their maximum values during the in operando experiment. Error boundaries are shown in blue. The ${\mathrm{HCO}}_{\mathrm{3}}^{-}$ signal fluctuates within the 1 % range, while the ${\mathrm{CO}}_{\mathrm{2}}$ signal decreases significantly in intensity over the 12 h period, down to 78 % of its maximum value.