Persistent and reversible solid iodine electrodeposition in nanoporous carbons

Aqueous iodine based electrochemical energy storage is considered a potential candidate to improve sustainability and performance of current battery and supercapacitor technology. It harnesses the redox activity of iodide, iodine, and polyiodide species in the confined geometry of nanoporous carbon electrodes. However, current descriptions of the electrochemical reaction mechanism to interconvert these species are elusive. Here we show that electrochemical oxidation of iodide in nanoporous carbons forms persistent solid iodine deposits. Confinement slows down dissolution into triiodide and pentaiodide, responsible for otherwise significant self-discharge via shuttling. The main tools for these insights are in situ Raman spectroscopy and in situ small and wide-angle X-ray scattering (in situ SAXS/WAXS). In situ Raman confirms the reversible formation of triiodide and pentaiodide. In situ SAXS/WAXS indicates remarkable amounts of solid iodine deposited in the carbon nanopores. Combined with stochastic modeling, in situ SAXS allows quantifying the solid iodine volume fraction and visualizing the iodine structure on 3D lattice models at the sub-nanometer scale. Based on the derived mechanism, we demonstrate strategies for improved iodine pore filling capacity and prevention of self-discharge, applicable to hybrid supercapacitors and batteries.

. Band deconvolution of the I 3and I 5bands were done using three Lorentz peaks (initial position 110 cm -1 , 143 cm -1 , and 224 cm -1 ) and one Gaussian peak (initial position 165 cm -1 ), according to Refs. 2,3. Figure 5: (a), UV-vis absorption spectra of different (poly)iodide and iodine containing solutions. The ethanol solution in contact with the charged positive electrode (red curve) shows significant amounts of I 2 indicated by the shoulder around 460 nm. The significant amounts of I 3 -(peak around 370 nm) originate from a certain fraction of (poly-)iodides left in the pores of the charged positive electrode. Note that the absorption of I 3 -is much stronger than the absorption of I 2 (see black curve, where virtually no I 3should be present); the I 3and I 2 peak height ratio is hence not proportional to the ratio of their concentrations. (b) Photograph of charged positive electrodes (after positive polarization to 1.0 V cell voltage) immersed in de-ionized water (left), in an aqueous solution with 1 M NaI (center), and in ethanol (right). Figure 6: Small and wide-angle X-ray scattering intensity (structure factors) of de-ionized H 2 O, 1 M NaI in H 2 O, 1 M NaI in H 2 O + 0.008 M I 2 and 1 M NaI + 0.08 M I 2 , recorded in quartz capillaries. Adding I 2 to the 1 M NaI electrolyte leads to the formation of polyiodides via chemical comproportionation.

Supplementary Figure 7:
Wide-angle X-ray scattering peak fit using three Gaussian peaks (black solid curves) with the parameters (x1=11.5°, sigma1=8°; x2=21.85°, sigma2=4.5°; x3=30.0°, sigma3=8°). G2 and G3 are attributed to the structure factor of solid iodine nanocrystals/nanoclusters. According to the Scherrer equation, their FWHM values correspond to a domain size (average coherence length) of 0.9 nm and 0.5 m, respectively. Figure 8: (a), Electrolyte (1 M NaI in H 2 O) and water structure factors recorded in quartz capillaries. The peak around 20 nm -1 corresponds to the H 2 O structure factor peak. Adding (uncorrelated) Na + and Iions leads to an additive constant scattering intensity contribution with respect to the pure H 2 O structure factor. The solid blue line is a polynomial fit of the third order. (b), To obtain the nanopore scattering contribution (blue data points) the electrolyte structure factor (solid grey line) and low q particle scattering contribution (solid black line) are subtracted from the scattering intensities recorded during the in situ SAXS/WAXS experiment (red data points), according to the procedure described in the experimental section. (a), The nanopore scattering intensity (blue solid line) is obtained by subtracting a constant background (determined via a standard Porod fit in the q-range 7 nm -1 < q < 9 nm -1 ) and the particle scattering contribution (determined by a power-law fit for q < 1 nm -1 ) from the SAXS intensity of the empty AC electrode (solid black line). (b), experimental nanopore scattering (blue data points) normalized by the integrated intensity (see experimental section), and modeled scattering intensity (solid red line) as a function scattering vector length q (model fit generation according to the description in the Section Methods). (c), Cross-section, and 3D cut-out of the resulting pore structure.  Supplementary Table 1.

Supplementary
Supplementary Table 1: Specific surface area (SSA) obtained from Brunauer-Emmett-Teller (BET) analysis, and quenched solid density-functional theory (QSDFT) for the AC powder. In addition, the mean pore size (d50) and the total specific pore volume determined with two different methods are given. The data are extracted from the N 2 adsorption data shown in Figure S1.
Total pore volume QSDFT (cm 3 g -1 ) Total pore volume at p/p0=0.95 (cm 3 g -1 ) AC , where the carbon skeleton density was assumed with 1.9 g cm -3 and the specific pore volume taken from gas adsorption measurements (Table S1) Within the (limited) accuracy we could not not detect severe structural differences during charge and discharge. Note the negligible differences in the SAXS intensiy shapes during charge and discharge in Supplementary Fig. 15a, b, below. Alternative correlation functions for the GRF Z(x), alternative carbons and improved data quality might help to improve the sensitivity of the plurigaussian model fit in future.

Supplementary Note 2 | Integrated Intensity analysis
The integrated intensity or invariant f of the experimental SAXS intensity is numerically calculated by (S1) At high q the SAXS intensity requires a Porod extrapolation of the form (#) ∝ # .j (Fig. S14a,b). For a three-phase system consisting of phases A, B, C the integrated intensity can be written as 7 g 2k + = (< 5 − < l )(< 5 − < c )m4 5 − 4 5 + n + (< l − < 5 )(< l − < c )m4 l − 4 l + n Here < is the electron density, and 4 the volume fraction of phase i. Since carbon and micropore volume fractions (Supplementary Table 2) as well as electron densities are known, the iodine volume fraction can be calculated by inserting the experimental f in equation S2.
The largest error with the integrated intensity analysis is induced by the delicate background subtraction and the corresponding Porod extrapolation at high q. At high potentials the polyiodide (I 3 -) correlations cause a peak around 8 nm -1 , which is difficult to separate from the nanopore scattering (see Supplementary Fig. 14a). The Kratky plot in Supplementary Fig. 14b indicates more clearly that the porod extrapolation is not ideal at higher cell voltages. This results in too high I 2 pore occupancies at high capacities (red curve, Supplementary Fig. 14c). Besides this deviation, the integrated intensity analysis confirms the quantities derived from the SAXS model fit and the electrochemical data.
Supplementary Figure 14 | I2 pore occupancy via integrated intensity: (a), nanopore scattering (Intensity vs. scattering vector length q) with Porod extrapolation at q > 5.5 nm -1 during oxidation/ with increasingly positive cell voltage. Since the polyiodide/iodine structure factor cannot be subtracted accurately, we have only subtracted the electrolyte structure at zero cell voltage and extrapolated the SAXS intensity at q > 5.5 nm -1 with (#) ∝ # .j .
(b), Kratky plot (# + (#) op. #) of the same curves. (c), I 2 pore occupancy as a function of capacity determined by the SAXS model fit, the estimation from electrochemical data (capacity) and the analysis of the SAXS integrated intensity. The latter is calculated by solving equation S2. The SAXS integrated intensity approach deviates at high capacities due to inaccurate Porod extrapolation.

Supplementary Note 3 | In situ SAXS potential dependency during charge and discharge
Supplementary Fig. 15a, b and the surface plot in Fig. 2d show that the iodine formation is nicely reversible. The SAXS and WAXS intensity fully drops to its initial values at zero cell voltage applied.
To check for the reversibility and the time/voltage dependency we calculated the integrated intensity increase/decrease of the reduced experimental in situ SAXS data in the q-regime between q 1 = 0.8 nm -1 and q 2 = 4.0 nm -1 during two voltage cycles: g = _ # + (#) # q S q r . This implies that the absolute value of this parameter has no actual physical meaning, yet the relative change can accurately sense time and potential dependent changes of iodine formation. Supplementary Fig. 15d indicates a significant hysteresis of iodine formation/dissolution during charging/discharging. Since electrochemical charging and discharging is fast (as shown in Fig. 4), but iodide formation rather slow ( Supplementary Fig. 15), electrochemical oxidation/reduction is (to a certain extend) decoupled from solid iodine formation. We believe that during Ioxidation, I 2 is first dissolved, before it precipitates at sites where it is highly confined by the carbon. Whether the hysteresis is only a kinetic effect, (induced by fast cycling, iR drop), or incduced by an actual potential-dependency is investigated in the following. To carefully check for the potential dependency of iodine electrodeposition we carried out in situ SAXS measurements during potentiostatic charge/discharge. We mounted a silver wire as a reference electrode in the in situ SAXS cell and changed the working electrode potential step-wise (chronoamperometry)

Supplementary
Each potential was held constant for 30 min, until charging has basically stopped ( Supplementary   Fig. 15c). The SAXS/WAXS intensities increase during charge ( Supplementary Fig. 15a) and reversibly decrease during discharge ( Supplementary Fig. 15b). In Supplementary Fig. 15d, the integrated intensity value of the reduced experimental in situ SAXS data (nanopore scattering, equivalent to what is shown in Fig. 3a or Supplementary Fig. 11) at the end of each 30 min potential step is given as a function of the WE potential. The significant hysteresis points at an intrinsic overpotential necessary to dissolve iodine during discharge and the increased stability of I 2 in nanoporous confinement due to carboniodine interactions. From a physico-chemical point of view, this might be explained by interface energies between carbon, iodide and electrolyte. Given the complexity of the system, the interpretation of the hysteresis needs to be treated with caution. Ag/AgCl as a function time. (d), Integrated intensity g of reduced experimental in situ SAXS intensities (after subtraction of particle and electrolyte structure factor scattering contributions, equivalent to Fig. 3a and Supplementary Fig. 11) as a function of the WE potential. The integrated intensity is calculated for the q-regime between q 1 = 0.8 nm -1 and q 2 = 4.0 nm -1 at the end of each 30 min potential step shown in (c). The integrated intensity shows a clear hysteresis, indicating that I 2 dissolution requires some overpotential during discharge.