Lithium-coupled electron transfer reactions of nano-confined WOx within Zr-based metal–organic framework

Interfacial charge transfer reactions involving cations and electrons are fundamental to (photo/electro) catalysis, energy storage, and beyond. Lithium-coupled electron transfer (LCET) at the electrode-electrolyte interfaces of lithium-ion batteries (LIBs) is a preeminent example to highlight the importance of charge transfer in modern-day society. The thermodynamics of LCET reactions define the minimal energy for charge/discharge of LIBs, and yet, these parameters are rarely available in the literature. Here, we demonstrate the successful incorporation of tungsten oxides (WOx) within a chemically stable Zr-based metal−organic framework (MOF), MOF-808. Cyclic voltammograms (CVs) of the composite, WOx@MOF-808, in Li+-containing acetonitrile (MeCN)-based electrolytes showed an irreversible, cathodic Faradaic feature that shifted in a Nernstian fashion with respect to the Li+ concentration, i.e., ∼59 mV/log [(Li+)]. The Nernstian dependence established 1:1 stoichiometry of Li+ and e−. Using the standard redox potential of Li+/0, the apparent free energy of lithiation of WOx@MOF-808 (ΔGapp,Li) was calculated to be −36 ± 1 kcal mol−1. ΔGapp,Li is an intrinsic parameter of WOx@MOF-808, and thus by deriving the similar reaction free energies of other metal oxides, their direct comparisons can be achieved. Implications of the reported measurements will be further contrasted to proton-coupled electron transfer (PCET) reactions on metal oxides.


Instrumentation
N2-adsorption-desorption isotherms of Zr-MOF-808 and WOx@MOF-808 were measured using the micropore analysis port of 3Flex (Micromeritics).Prior to the isotherm measurement, both samples were dried under a dynamic vacuum at 80 °C overnight and were further activated using VacPrep (Micromeritics) at 120 °C and at <50 mTorr.All Brunauer-Emmett-Teller (BET) areas were derived using the data set between P/P0 = 0.005 -0.1.(Howarth et al., 2017) For the isotherms and the density functional theory-calculated pore size distribution derived from the isotherms using the built-in 'N2cylindrical Pores -Oxide Surface' model; see Figure 3 in the main manuscript.
All cyclic voltammograms (CVs) were measured using the CH instrument model 600D potentiostat.A WOx@MOF-808-based electrode was prepared through a simple drop-casting method applied from previous reports (see Section 3 for details).Pt wire was used as a counter electrode.Ag/Ag + pseudoreference electrode was prepared according to the reported procedure.(Wise et al., 2020) At the end of all electrochemical measurements, a small amount of ferrocene was added to the electrolyte, and glassy carbon was used as a working electrode to measure its redox for the calibration of electrochemical potential.
Grazing incidence powder X-ray diffraction (PXRD) patterns were collected using Rigaku Smartlab equipped with a Cu Kα X-ray source.2θ between 2 -60° with a step size of 0.05°/min were used as the range with a fixed ω of 0.05°.
Scanning electron microscopy images and energy-dispersive X-ray spectra (SEM-EDS) were measured using the Zeiss Neon 40 EsB field emission instrument operated at 5 kV.Prior to the measurement, a small amount of Zr-MOF-808 or WOx@MOF-808 was dispersed in acetone and drop-casted onto a polished Si wafer and was further coated with ca. 4 nm iridium using the EMS Quorum Q150 ES plus sputter coater.
1 H NMR spectra were collected using the Varian VNMR 400 MHz.

1 H NMR Spectra
1 H NMR sample preparation follows that reported previously, (Ingram et al., 2024) using ca. 1 M NaOD to digest the MOF samples and DMSO as an internal standard.The ratios of the integrations measured three times using three freshly prepared electrodes were used to determine the average loading of WOx@MOF-808.
Based on the three NMR spectra, we have determined the average loading of WOX@MOF-808 to be 0.95 ± 0.13 μmolWOX@MOF-808/cm 2 (or 5.725 ± 0.82 μmolWOx/cm 2 ).Small amounts of formate were also observed, and this is a common observation for many Zr-based MOFs.

Details on Electrochemical Measurements
WOx@MOF-808 was drop-casted onto FTO to yield the working electrode, using the modified procedure from that reported previously.(Ingram et al., 2024) Briefly, 12 mg of WOx@MOF-808 was dispersed in 1 mL of acetone and was sonicated to yield a uniform suspension.Onto a 1 × 1 cm FTO, 10 μL of the suspension was drop-casted three times.Electrodes with bare Zr-MOF-808 or bulk WO3•2H2O was synthesized analogously, but with different masses to keep the number of moles of MOF or W 6+ cation identical, respectively.
All errors presented in this section onwards are 1σ of duplicate measurements using freshly prepared electrodes.

Estimation of Electroactive Amount of WOx within WOx@MOF-808
The Faradaic feature associated with lithium-coupled electron transfer (LCET; feature highlighted as B in Figure 4A in the main manuscript) was integrated to determine the amount of electroactive WOx.By considering the average of all CVs measured in this report, and using the following equation, we have determined the average amount of electroactive WOx was determined to be 11 ± 6 nmol/cm 2 ; i.e., based on the 1 H NMR results, this indicates that ca.0.2% of all WOx within the MOF network are electroactive.In eq.S1, Q, n, F, and N refer to charge, number of electrons, the Faraday Constant (96485 C/mol), and the number of moles of electroactive WOx, respectively.

Sample-to-Sample Variation
Electrodes yielded through the above method proved successful in measuring the CVs of WOx@MOF-808 and yielding the Pourbaix diagram of the LCET reaction (Figure 4B in the main manuscript).The large sample-to-sample variation, however, precluded any detailed kinetic analysis.
As shown in the figure below, the CVs of two identically prepared electrodes in an identical electrolyte (100 mM LiClO4 and 900 mM TBAClO4 in MeCN) were quite distinct.This large sample-to-sample variation has been reported previously for MOF-based electrodes synthesized via simple drop-casting method.(Chen et al., 2021) We prefer not to use carbon black or polymeric binders, which are otherwise used commonly in the literature, as they are active towards LCET or can significantly hinder diffusion.

Peak Potentials, Current Densities, and Full-Width-Half-Maximum of the LCET Faradaic Feature
Cathodic peak potentials (Ep,c), cathodic peak current densities (jp,c), and full-width-half-maximum (FWHM) of the LCET Faradaic feature in various concentrations of Li + ion are reported in the table below.jp,c was derived after the subtraction of capacitive current; see the following reference for details.We note that these measurements were conducted at a scan rate (υ) of 100 mV/s.All errors are from two separate measurements using freshly prepared WOx@MOF-808 electrodes.

Scan Rate Dependence
CVs of WOx@MOF-808 measured in various scan rates (υ) at constant Li-ion concentrations should indicate the LCET mechanism.As shown below, the linear regression fit of the log(υ) vs. log(jp,c) resulted in a slope of 0.44 ± 0.07, indicating that the LCET reaction is diffusion-controlled.As noted in Section 3.2.1, the relatively large error bars on these figures are due to the sample-to-sample inconsistency.

Details on Thermochemical Analysis
This section describes the details on how to convert the Ep,c of LCET feature to the apparent free energy of lithiation, ΔGapp, Li.
Many standard potentials required for thermochemical conversions are reported vs. normal hydrogen electrode (NHE).Thus, we first convert Li + -ion solvated in MeCN to be that in H2O using the difference in solvation free energies between two solvents, as shown below.
As noted in the main text, we estimated the solvation free energies to be similar between different solvents, given the previous reports.(Carvalho andPliego, 2015, Itkis et al., 2021) Using Ep,c extrapolated to standard state (i.e., [Li + ] = 1 M) from Figure 4B, we calculated the ΔG of equation S5 to be 49 kcal mol -1 .
Next, the electrons involved must have the free energy references against NHE instead of Fc +/0 .The free energy of this reaction is already reported to be -15 kcal mol -1 .(Pegis et al., 2015) Finally, the redox potential of Li +/0 can be used to derive ΔGapp, Li. (Bard and Faulkner, 2001)