Porphyrin Aggregation under Homogeneous Conditions Inhibits Electrocatalysis: A Case Study on CO2 Reduction

Metalloporphyrins are widely used as homogeneous electrocatalysts for transformations relevant to clean energy and sustainable organic synthesis. Metalloporphyrins are well-known to aggregate due to π–π stacking, but surprisingly, the influence of aggregation on homogeneous electrocatalytic performance has not been investigated previously. Herein, we present three structurally related iron meso-phenylporphyrins whose aggregation properties are different in commonly used N,N-dimethylformamide (DMF) electrolyte. Both spectroscopy and light scattering provide evidence of extensive porphyrin aggregation under conventional electrocatalytic conditions. Using the electrocatalytic reduction of CO2 to CO as a test reaction, cyclic voltammetry reveals an inverse dependence of the kinetics on the catalyst concentration. The inhibition extends to bulk performance, where up to 75% of the catalyst at 1 mM is inactive compared to at 0.25 mM. We additionally report how aggregation is perturbed by organic additives, axial ligands, and redox state. Periodic boundary calculations provide additional insights into aggregate stability as a function of metalloporphyrin structure. Finally, we generalize the aggregation phenomenon by surveying metalloporphyrins with different metals and substituents. This study demonstrates that homogeneous metalloporphyrins can aggregate severely in well-solubilizing organic electrolytes, that aggregation can be easily modulated through experimental conditions, and that the extent of aggregation must be considered for accurate catalytic benchmarking.


S1. General Methods
Reagents and solvents were purchased from Sigma-Aldrich or Oakwood Chemical and used without further purification unless otherwise noted.The free-base porphyrin ligands meso-5,15diphenylporphyrin (H2DiPP) and meso-tetraphenylporphyrin (H2TetraPP) were purchased from Frontier Scientific.THF was dried using a Pure Process Technology (Nashua, NH) solvent purification system. 1 H NMR spectra were recorded on Bruker Avance 300 MHz spectrometer.Chemical shifts were referenced to residual proteo-solvent signal.Electrospray ionization mass spectra (ESI-MS) were collected on a Micromass LCT time-of-flight instrument in LC-MS grade methanol.UV-Vis spectra were recorded using an Agilent Cary 60 spectrophotometer at room temperature with a 10 mm micro rectangular quartz glass cuvette for characterization and a short-path 1 mm SEC-CT thin layer quartz glass cuvette (AirekaCells) for aggregation studies and spectroelectrochemical experiments.Dynamic light scattering (DLS) experiments were performed on a NanoBrook Omni particle size analyzer (Brookhaven) with an incident light of 640 nm in a 10 mm quartz cuvette at 19.5°C.Samples were prepared in DMF electrolyte (0.1 M TBAPF6) and filtered through a 0.45 μm membrane filter.The refractive index of this solvent was measured to be 1.4306 at 19.5°C with an Abbe Refractometer (Atago).Anhydrous DMF for electrochemical experiments was purchased from Sigma-Aldrich and stored in an amber glass bottle over molecular sieves in a nitrogenfilled glovebox.Tetrabutylammonium hexafluorophosphate (TBAPF6) supporting electrolyte was purified via three subsequent recrystallizations from ethanol and stored in a sealed desiccator.High purity gas cylinders (CO2, Ar) were purchased from Linde.No unexpected or unusually high safety hazards were encountered.

S2. Synthetic Procedures
Preparation of meso-5,10,15-triphenylporphyrin (TriPP) Synthesis of TriPP was performed according to a modified literature procedure. 1Free-base meso-5,15diphenylporphyrin (100 mg, 0.22 mmol, 1 eq) was added to an oven-dried Schlenk flask containing a Teflon stir-bar under nitrogen atmosphere.The flask was charged with dry THF (70 mL) and cooled in a dry-ice acetone bath (−78 °C), after which a solution of phenyllithium (1.9 M in dibutyl ether, 1.15 mL, 10 eq) was added dropwise via syringe.The solution was stirred at −78 °C for 2 hours, after which the solution was removed from the cold bath and stirred for an additional 1.5 hours.Over the course of the reaction at room temperature, the solution changed from deep purple to green-brown in colour.The mixture was quenched with a 50:50 mixture of H2O and THF (30 mL), and stirred for 15 minutes.Subsequently, 2,3dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) (200 mg, 0.88 mmol, 4 eq) was added with another 15 minutes of stirring, then the solvent was removed under vacuum.Final purification was achieved via column chromatography (silica gel, 8:1 hexane:ethyl acetate) to afford the pure final product (84 mg, 71 % yield).

Metallation Procedures
A solution of free-base porphyrin (1 eq) in minimal DMF (~5 mL) was added to a solution of FeCl3•6H2O (12  eq) in a minimal amount of DMF (~1 mL).The resulting solution was stirred at reflux for about 3 hours under nitrogen.The reaction completion was confirmed via UV-Vis, at which point the cooled solution was neutralized with HCl (6 M, 10 mL).The resulting precipitate was collected via vacuum filtration and washed with HCl (3 M, 10 mL) and excess water.The precipitate was dried under reduced pressure overnight to afford the final complex.Metalations were performed on scales between 0.05-0.10mmol of free-base porphyrin.

Cyclic Voltammetry Details
Cyclic voltammograms (CVs) were performed in dry DMF containing 0.1 M TBAPF6 supporting electrolyte under Ar, CO2, or Ar/CO2 mixtures.CVs were performed with a SP-50 potentiostat (Bio-Logic) using a threeelectrode set-up: a 3.0 mm diameter glassy carbon working electrode (Bioanalytical Systems, Inc.), a platinum wire counter electrode (0.5 mm diameter), and a silver wire encased in a Vycor tip glass tube filled with 0.1 M TBAPF6 electrolyte as the pseudo-reference electrode.The working electrode was polished between each scan with a slurry of water and alumina (0.05 μm) on a felt pad, then rinsed with water followed by acetone and dried with a stream of pressurized air.An initial blank scan of the electrolyte solution was performed before each experiment.Following the experiment, the pseudo-reference electrode was referenced to the ferrocene/ferrocenium (Fc/Fc + ) redox couple.All CVs were compensated for internal resistance at 85% compensation of the uncompensated resistance (Ru).

Controlled Potential Electrolysis Details
Controlled potential electrolysis (CPE) experiments were performed using a CHI650E potentiostat (CH Instruments, Inc.) in a gas-tight custom-made PEEK cell similar to those previously reported. 2The cell consists of two compartments separated by a glass frit.The working compartment houses a glassy carbon working electrode (1 cm 2 ) and a silver wire encased in a Vycor tip glass tube filled with 0.1 M TBAPF6 electrolyte as the pseudo-reference electrode.The counter compartment houses a graphite rod counter electrode (surface area ≈ 8 cm 2 ) .The solvent (DMF) was pre-sparged with argon for 20 minutes before use to remove any residual dimethylamine impurity.The working compartment was prepared with the desired concentration of catalyst and 100 mM PhOH in 7 mL 0.1 M TBAPF6 in DMF.The counter compartment was prepared with 3 mL electrolyte solution (either 0.1 M TBAPF6 in DMF, or 0.1 M TBAOAc in DMF as a sacrificial substrate).The cell was then sparged for 30 minutes with a mixture of 95% CO2 and 5% He (internal standard), prepared with precision mass flow controllers (Alicat Scientific).The electrolysis was performed at a potential −0.3 V from the onset of the catalytic wave (~ −2.2 V vs. Fc/Fc + ), where the exact potential was determined prior to each experiment by running a CV in the cell before electrolysis and aligning based on the observed catalytic onset potential.Following a 90-minute electrolysis, the headspace of the cell was directly injected into the pre-evacuated sample loops (−20 bar passive vacuum) of a gas chromatograph (SRI Multiple Gas Analyzer #5) through a Quick-Connect valve (Swagelok).An inline thermal conductivity detector (TCD) was used to detect He and H2, while a flame ionization detector (FID) with a methanizer was used to detect CO (Figure S1).The amount of gaseous products produced was determined by comparing the ratio of product gas to internal standard peak integrals with a prepared calibration curve (Figure S2).

S4. Brief Discussion of Relevant Previous Reports
Literature examples of investigations into the catalyst concentration dependence term in the rate law for metalloporphyrin electrocatalysts are limited, as most often a first-order assumption is made without experimental verification.As a result, there has been very limited discussion in the literature on the possible role of catalyst concentration-and porphyrin aggregation-on performance.
A small number of previous reports have compared metalloporphyrin catalytic currents 3,4 or rate constants 2,5-7 measured at variable catalyst concentrations to gain insight into this term in the rate law, and several have demonstrated agreement with a first-order catalyst concentration dependence.Conversely, one example observed a notable inverse relationship between the concentration of iron porphyrin catalyst and activity which was attributed to solution dimerization, however this hypothesis was not further explored or discussed in detail. 6Together, these examples provide precedent for both agreement and contradiction with a first-order assumption in catalyst; however, since many of these examples feature elaborate catalyst designs (i.e., with appended pendant groups) and variable operating conditions (i.e., solvent, electrolyte, or additives), they are unable to provide an extensive understanding of catalyst concentration dependence in the rate law or the role of catalyst aggregation.

S5. Details of Foot-of-the-Wave Analysis (FOWA) and kobs Calculations
The catalytic rate law for the electrocatalytic reduction of CO2 can be described as follows: Where the subscripts x, y, and z are the reaction orders in exogenous acid, CO2 substrate and catalyst, respectively.In the subsequent analysis procedure to calculate observed rate constants, z=1 is assumed. 8served rate constants (kobs) for the electrocatalytic reduction of CO2 were calculated from cyclic voltammograms using Foot-of-the-Wave Analysis (FOWA) as developed and described by Savéant and coworkers. 8Assuming each catalyst presented in this work follows the same mechanism as previously reported, 8 the reduction of CO2 with iron porphyrins can be described as an EC' process: The rate-determining step (RDS) includes pre-equilibrium CO2 binding to the Fe(0) active site followed by subsequent proton-coupled electron transfer.As a result, the following relationship can be derived: Where i is current,   0 is the peak current, E is the potential, v is the scan rate (in V/s), f= F/RT = 38.94V -1 , and n σ is a constant that describes the number of electrons required for catalysis (n = 2) to the power of a variable (σ) which describes the electron transfer process as being solely from the electrode (σ = 1), or resulting from disproportionation between iron species in solution (σ = 0.5).A value of σ = 1 is used as it has been proposed that this value will provide the most conservative estimate of kobs. 9,10That is, the calculated rate constants when σ = 1 is used will not be overestimated under any of the mechanistic possibilities for electron transfer, as this equation will provide the lower-limit values of kobs.
Prior to catalytic investigations, an initial CV collected under argon in the absence of PhOH was used to determine   0 , the reduction potential of the Fe I/0 couple, and the peak current (measured at the formal reduction of Fe II/I ) (  0 ).In concentration-dependence experiments,   0 was determined individually for each concentration tested.From this, a "Foot-of-the-Wave" (FOW) plot can be constructed by plotting /  0 against (1 + exp [( −   0 )]) −1 .Fitting of this plot to a linear function yields a line with slope 2.24(  )  √  / , from which kobs can be calculated.To ensure consistency between kobs values determined within this study, all linear functions were fit up to an x-axis value of 0.1, or until an R 2 of 0.98 was achieved.
We additionally note that since the equations used in FOWA assume a first-order dependence on catalyst concentration, 8 FOWA cannot be used to explicitly extract the order in catalyst concentration but can only highlight deviations from the implicitly assumed result.When this assumption is not supported experimentally, the rate constants derived from FOWA are not directly reporting inherent catalytic activity.Nevertheless, the observed rate constants extracted from FOWA can still inform agreement or disagreement with the first-order assumption and can be used to investigate trends in catalytic activity.

S6.1 General Methods
UV-Vis aggregation studies were conducted in DMF containing 0.1 M TBAPF6 electrolyte in a short path (1 mm) cuvette at room temperature.Solutions were prepared first at the largest concentration, then were sequentially diluted to survey the same concentration range as in the catalytic concentration-dependence studies (as indicated; 2.0, 1.5, 1.0, 0.5, 0.25, 0.125 mM).For investigations into the Soret Band, more dilute concentrations were used (as indicated; 0.1 mM -0.0075 mM).The cuvette containing porphyrin solution was sonicated for 2 minutes prior to each measurement, and a scan was recorded at t = 0 min and at t = 20 min; no differences were observed between the spectra taken at these two time points.To normalize each spectrum to concentration, the absorbance values were divided by the concentration of porphyrin in solution, and the resulting concentration-normalized spectra were overlayed.
Previous studies have described µ-oxo dimer formation of FeTetraPP at low concentrations in DMF due to residual water. 11In order to evaluate the potential contribution of µ-oxo dimer formation in our system, the features in our porphyrin concentration-dependent spectra were compared to those of a µ-oxo dimer prepared via titration with TBAOH (Figure S3).At higher concentration regimes (Figure S3a), the low intensity porphyrin Q bands (560, 610 nm) are similar to those of the µ-oxo dimer.This comparison reveals either a small amount of µ-oxo dimer formation and/or a coincidental similarity between features associated with µ-oxo dimerization and aggregation.Dynamic light scattering (DLS) data (Figure 2b) unequivocally show that aggregates of increasing size are present at higher concentrations, thereby indicating that the UV-Vis changes correlate with aggregation at least in part.Additionally, the spectral features observed upon porphyrin dilution (Figure 2c) are also seen upon titration with pyrene as a disaggregating agent (Figure 5a) that we speculate is not capable of breaking apart covalently bound μoxo dimers.Quantification of the relative amounts of aggregated metalloporphyrins vs. μ-oxo dimers is not possible due to the similar UV-Vis characteristics of these species and the inability of other analytical methods to report on the specific chemical composition.Note that changes in the shoulder peak of the Soret band (Figure S4) are generally understood to corelate with changes in axial ligation.Thus, these changes are likely related to equilibria between chloride-bound and solvent-bound iron centers as a function of dilution.

S7. Scan Rate Dependence Experiments
Scan rate dependence experiments were performed by preparing a solution of each iron porphyrin (2.0 mM) in 0.1 M TBAPF6 in DMF.The solution was sparged thoroughly (15 min) with argon, then an initial scan was taken.Scans were then taken sequentially following dilutions of the CV solution with electrolyte solution to analyze each of the catalyst concentrations (as indicated; 2.0, 1.5, 1.0, 0.5, 0.25, 0.125 mM).At each catalyst concentration, CVs were taken at several scan rates (as indicated; 50, 100, 250, 500, 750, 1000 mV/s).The solution was kept under an argon atmosphere throughout the experiment, and the working electrode was polished before each scan.Peak currents of each iron redox couple were measured at each scan rate and plotted against the square-root of scan rate, whereby according to the Randles-Ševčík equation: a linear correlation between peak current (  0 ) and the square-root of scan rate ( 1/2 ) is characteristic of a diffusional process without the presence of electrode-bound species.A linear result was obtained for each catalyst and for each iron redox couple (Figure S5).
The slopes of each linear correlation in Figure S5 were used to calculate the diffusion coefficients (Do) of each catalyst as a function of catalyst concentration (Figure S7).We hypothesized that an increase in porphyrin concentration and aggregate size would result in a decrease to the measured diffusion coefficients.When comparing the calculated diffusion coefficients, no clear trends are observed and the values do not significantly vary with catalyst concentration at any redox couple (Figure S7).We rationalize that diffusion coefficients may not be a reliable metric for aggregation severity because the expected change in Do as a function of aggregate size is not well defined; examples in literature demonstrate that Do can increase, decrease, or remain constant with increasing aggregate size. 12,13

Phenol Concentration Dependence Experiments
A solution of iron porphyrin (0.5 mM or 1.0 mM, as indicated) was prepared in 0.1 M TBAPF6 in DMF.The solution was sparged thoroughly (15 min) with argon, then an initial scan was taken.A scan was then taken following saturation of the solution with CO2.Scans were then taken sequentially following PhOH titrations (as indicated; 50, 100, 250, 500, 750, 1000 mM), whereby a scan under both argon and CO2 were collected at each PhOH concentration.Solutions were sparged for 8 minutes between gases, and the working electrode was polished before each scan.The resulting CVs were analyzed by FOWA to calculate the observed rate constants (kobs) at each PhOH concentration (Figure S8).S1), where the peak current (  0 ) was measured individually for each catalyst concentration at the Fe II/I couple (Figure S10).We note that the peak currents are equivalent regardless of if the solutions are saturated with Ar or CO2.The order in catalyst is inverse but cannot be fit by a simple linear function across the entire range of concentrations investigated (Figure S9).That is, the order in catalyst appears to be a function of catalyst concentration as a result of the changing aggregation state influencing the order in catalyst.In general, the tangent to log(kobs) gets increasingly negative at higher catalyst concentrations (that is, the order in catalyst concentration gets more negative as catalyst concentration is increased).This is in agreement with a greater extent of aggregation and inhibition at higher catalyst loadings.

CO2 Concentration Dependence Experiments
A solution of iron porphyrin (0.25 mM or 1.0 mM, as indicated) was prepared in 0.1 M TBAPF6 in DMF.The solution was sparged thoroughly (15 min) with argon, then an initial scan was taken.PhOH (250 mM) was then introduced to the solution, and an additional argon scan was collected.Scans were then taken sequentially at different Ar/CO2 mixtures (as indicated; 10, 20, 40, 50, 60, 80, 100 % CO2) prepared with precision mass-flow controllers (Alicat Scientific).Solutions were sparged with each prepared gas mixture for 15 minutes, and the working electrode was polished before each scan.The resulting CVs were analyzed by FOWA to calculate the observed rate constants (kobs) at each CO2 concentration.The concentration of CO2 in solution was assumed to be equal to the percent of CO2 in the sparging mixture multiplied by the concentration in a CO2-saturated solution of DMF (0.23 M).

S10.1 Tabulated Faradaic Efficiencies
CPE experiments were performed in duplicate for each catalyst at both concentrations tested (0.25 mM and 1.0 mM).The average total charge passed and faradaic efficiencies are reported (Table S2).For each set of duplicate experiments, either 0.1 M TBAPF6 electrolyte or 0.1 M TBAOAc (as a sacrificial substrate) electrolyte was used in the counter compartment to determine the amount of CO produced via solvent oxidation in the counter compartment.Experiments performed with TBAOAc as a sacrificial substrate in the counter compartment had on average a 9% lower FE for CO, suggesting some amount of solvent oxidation when TBAPF6 is used.
Table S2.Results of CPE experiments, reporting charge and faradaic FE of CO for each CPE run.

S10.2 Post-CPE Catalyst Characterization
Post-electrolysis solutions were studied to investigate potential catalyst decomposition.CV's were collected pre-and post-electrolysis directly in the CPE cell immediately before and after electrolysis.For all catalysts in the series and at both catalyst concentrations, there was no evidence of significant catalyst degradation as there were only minor decreases in the catalytic currents observed in CVs of postelectrolysis solutions (Figure S17).Post-CPE solutions were also analyzed via UV-Vis by removing a 0.1 mL aliquot from the working compartment solution and diluting into 1 mL DMF containing 0.1 M TBAPF6 in a 1 mm path length quartz cuvette.Spectral measurements were taken at several time points to observe the re-oxidation of the iron porphyrin complexes following exposure to air.For each of the catalysts, the scans taken immediately upon sampling from the electrolysis cell show a sharp Soret band, characteristic of an Fe II porphyrin species.Following this initial exposure to air, the iron is re-oxidized to the Fe III species and the spectra agree with the pre-CPE spectra, showing no evidence of decomposition (Figure S18).

S11.1 Pyrene Titration Experiments
UV-Vis Pyrene Titration UV-Vis measurements were taken as described above (Section S6.1), except a constant concentration of FeTetraPP (1.0 mM) was used and spectra were recorded following titrations of a pyrene stock solution (40 mM) resulting in additions between 0 -2 molar equivalents of pyrene.No pyrene absorptions appear within the porphyrin Q band region of interest.

Electrochemical Pyrene Titration Experiment
Prior to the pyrene titration experiment, CVs of pyrene were investigated to ensure no redox features would interfere with the CV analysis of FeTetraPP CO2 reduction catalysis.CVs of pyrene (0.5 mM) were taken under argon and CO2, then PhOH (10 mM) was added and again both an argon and CO2 scan were taken (Figure S19).Pyrene shows some reduction events at around −2.5 V vs. Fc/Fc + under all conditions, thus the scan window for the subsequent pyrene titrations was cut off at this potential.A solution of FeTetraPP (1.0 mM) was then prepared in 0.1 M TBAPF6 in DMF.The solution was sparged thoroughly (15 min) with argon, then an initial scan was taken.PhOH (10 mM) was added to solution, and both an argon and CO2 scan were taken.Scans were then taken sequentially following pyrene titrations (as indicated; 0.4, 0.8, 1.0, 1.4, 1.8, 2.0 mM) under CO2.The solution was kept under a CO2 atmosphere throughout the experiment, and the working electrode was polished before each scan.

Controlled Potential Electrolysis with Pyrene
A CPE with FeTetraPP (1.0 mM) and 1 molar equivalent of pyrene (1.0 mM) was performed to investigate how the presence of a disaggregating agent influences bulk CO2 reduction performance (Figure S20).The CPE was performed as described above (Section S3) except with the addition of pyrene (1.0 mM) to the working compartment.The current and total charge passed were larger than that of 1.0 mM of FeTetraPP alone (8.34 C vs 7.21 C), consistent with disaggregation of the catalyst.However, the Faradaic efficiency for CO was slightly reduced (68.0 % FE vs 77.2 % FE), which we attribute to some amount of pyrene plating on the electrode surface and/or small amounts of pyrene reduction.

S11.2 Chloride Abstraction Titration Experiments
UV-Vis Chloride Abstraction Titration UV-Vis measurements were taken as described above (Section S6.1), except a constant concentration of FeTriPP (0.5 mM) was used and spectra were recorded following titrations of a silver hexafluorophosphate (AgPF6) stock solution (4 mM) resulting in additions between 0 -0.4 molar equivalents of AgPF6.A solution of FeTriPP (0.5 mM) was then prepared in 0.1 M TBAPF6 in DMF.The solution was sparged thoroughly (15 min) with argon, then an initial scan was taken.PhOH (10 mM) was added to solution, and both an argon and CO2 scan were taken.Scans were then taken sequentially following titrations of AgPF6 under CO2.The solution was kept under a CO2 atmosphere throughout the experiment, and the working electrode was polished before each scan.

Dilution Experiment in Conditions of Excess Chloride
To further understand the role of axial ligation on the aggregation of iron phenylporphyrins, we repeated the Soret band UV-Vis dilution study (Figure S4) in the presence of an excess of chloride ions (10 mM tetrabutylammonium chloride).Under these conditions, the disaggregation effect (that is, the red-shifting of the Soret band) previously observed upon dilution is no longer observed.This suggests that equilibrium chloride ligand exchange has a significant role in the aggregation state for these complexes; potentially, a greater amount of solvent-bound species is formed following dilution and this species is less prone to aggregation.Under conditions of excess chloride, the equilibrium favours the chloride-bound species and thus there remains a more aggregated state in solution even upon dilution.These results again suggest that the chloride-bound iron porphyrin species are most prone to aggregation.

S12. Spectroelectrochemical Aggregation Studies
Spectroelectrochemical experiments were performed in an analogous manner to the previously presented concentration-normalized UV-Vis Q-Band aggregation studies, except here the reduced porphyrin species were investigated.The experimental setup consisted of a Cary 60 spectrophotometer fitted with a fiber optic coupler (Agilent Technologies, Inc.) and connected via fiber optic cables to a sample holder (OceanInsight) inside a nitrogen-filled glovebox, allowing for detection of the air-sensitive reduced porphyrin species.The electrochemical cell consisted of a 1 mm SEC-CT thin layer quartz glass cuvette, a platinum gauze flag working electrode (Bioanalytical Systems, Inc.), a platinum wire counter electrode (0.5 mm diameter), and a silver wire encased in a Vycor tip glass tube filled with 0.1 M TBAPF6 electrolyte as the pseudo-reference electrode.Electrolysis was performed with a CHI650E potentiostat (CH Instruments, Inc.), where leads were connected to electrical feedthrough cables into the glovebox.
First, a concentrated solution of catalyst (1.0 mM) was prepared and an initial CV scan was taken (Figure S24a,e) to determine the applied potentials required to target each redox state of interest.Each potential was applied at a given catalyst concentration, where the applied potential was held with UV-Vis spectra being collected every 5 minutes for about 20-30 minutes until the spectra stabilized, indicating completion of the electrolysis.The catalyst solution was then diluted and a positive potential (~ +0.5 V) was applied to return to the formal Fe III species before repeating the previous procedure at each catalyst concentration.We note that the concentration range tested in these spectroelectrochemical studies was limited due to scattering from the platinum gauze electrode resulting in a large baseline absorbance.For FeTriPP at the Fe(I) and Fe(II) redox states, (Figure 24g,h), the Q band spectra at 0.5 mM and 0.25 mM overlay, suggesting that aggregation is not significant below 0.5 mM.This is in contrast to the Fe(III) porphyrins, which show evidence of aggregation through Soret band shifts at much more dilute concentrations, suggesting less severe aggregation for reduced porphyrin species.

S13. Details of Computational Modeling of Porphyrin Aggregates
All calculations were performed using the Fritz Haber Institute --ab initio materials simulations (FHI-aims) program. 14They used the B86bPBE density functional, 15,16 the XDM dispersion correction, 17,18 the light basis setting, dense integration grids, and the atomic Zora scalar relativity correction. 5The individual Fe complexes were assigned a high spin state with 5 unpaired electrons and their geometries fully optimized.
We then constructed periodic chains of molecules, aligned in the c lattice direction and separated from neighbouring chains in the a,b directions by vacuum, with two molecules per unit cell and a fixed spin moment of 10.The a and b lattice vectors were kept fixed, while the c lattice vector and the atomic positions were allowed to optimize, using a 1x1x2 k-point mesh.

S14.2 Survey of Various Metallo-Tetraphenylporphyrins
The UV-Vis Q band aggregation studies were repeated for tetraphenylporphyrin complexes with various divalent metals.Due to reduced solubility of some of these complexes in DMF, the range of concentrations surveyed was adjusted as necessary.The free base tetraphenyl porphyrin ligand was purchased from Frontier Scientific, and was metalated using standard procedures.The cobalt, nickel, and copper tetraphenylporphyrins show evidence of aggregation with varying levels of severity, whereas the zinc tetraphenylporphyrin does not show any concentration-dependent spectral changes.

Figure S1 .
Figure S1.Representative GC trace of post-electrolysis sample headspace.Top: FID detector showing CO peak.Bottom: TCD detector showing He internal standard peak.Time axis demarcations represent 1 minute intervals.

Figure S2 .
Figure S2.GC calibration curve for CO product formation with He as the internal standard.

Figure S3 .
Figure S3.Comparison of variable concentration UV-Vis absorption spectra in (a) the Q band and (b) the Soret band regions for FeTetraPP (solid lines) and the μ-oxo dimer of FeTetraPP (dashed lines), formed in situ by addition of 5 equivalents of tetrabutylammonium hydroxide (TBAOH).Conditions: indicated catalyst concentration, 0.1 M TBAPF6 in DMF, 1 mm path length.

Figure S5 .
Figure S5.Scan rate dependence for (a) FeDiPP (b) FeTriPP and (c) FeTetraPP.Left: peak current of the Fe III/II redox couple.Middle: peak current of the Fe II/I redox couple.Right: peak current of the Fe I/0 redox couple.Conditions: indicated catalyst concentration, 0.1 M TBAPF6 in DMF under argon.

Figure S7 .
Figure S7.Diffusion coefficients calculated from scan rate dependence experiments according to the Randles-Ševčík equation using the peak current of (a) Fe(III/II) couple, (b) the Fe(II/I) couple, and (c) the Fe(I/0) couple.Conditions: indicated catalyst concentration, 0.1 M TBAPF6 in DMF under argon.

Figure S8 .
Figure S8.Observed rate constants (kobs) as a function of PhOH concentration for (a) FeDiPP (b) FeTriPP and (c) FeTetraPP at 1.0 mM and 0.5 mM.Corresponding log/log plots for (d) FeDiPP (e) FeTriPP and (f) FeTetraPP; the linear fit and slopes include only the linear regime (the first 4 points).Conditions: Indicated PhOH and catalyst concentrations, 0.1 M TBAPF6 in DMF, 100 mV/s scan rate.Catalyst Concentration Dependence ExperimentsA solution of iron porphyrin (2.0 mM) was prepared in 0.1M TBAPF6 in DMF.The solution was sparged thoroughly (15 min) with argon, then an initial scan was taken.PhOH (250 mM) was then introduced to the solution, and both an argon and CO2 scan were taken.Scans were then taken sequentially following dilutions of the CV solution (with a solution of 0.1 M TBAPF6 in DMF containing 250 mM PhOH) to each of the catalyst concentrations (as indicated; 2.0, 1.5, 1.0, 0.5, 0.25, 0.125 mM).A scan under both argon and CO2 were collected at each catalyst concentration.Solutions were sparged for 8 minutes between gases, and the working electrode was polished before each scan.The resulting CVs were analyzed by FOWA to calculate the observed rate constants (kobs) at each catalyst concentration (TableS1), where the peak current (  0 ) was measured individually for each catalyst concentration at the Fe II/I couple (FigureS10).We

Figure S10 .
Figure S10.Average of Fe II/I peak current (  0 ) values vs. catalyst concentration for (a) FeDiPP (b) FeTriPP and (c) FeTetraPP, showing a linear increase in peak current with increasing concentration as expected based on the Randles-Ševčík equation.Average   0 from triplicate experiments, error bars represent 1 standard deviation; error bars not shown are smaller than their respective data marker.Conditions: indicated catalyst concentration, 250 mM PhOH, 0.1 M TBAPF6 in CO2-saturated DMF, 100 mV/s scan rate.

Figure S11 .
Figure S11.Schematic representation of electron transfer and catalysis in iron porphyrin aggregates (a) demonstrating electron transfer is possible within the porphyrin assembly, supported by a linear increase in peak current under argon with increasing catalyst concentration and (b) demonstrating the proposed catalytic inhibition process as a result of catalyst aggregation, whereby active sites within the assembly are likely inaccessible to substrate binding.

Figure S12 .
Figure S12.Observed rate constants (kobs) as a function of CO2 concentration at (a) 0.25 mM and (b) 1.0 mM catalyst.Corresponding log/log plots at (c) 0.25 mM and (d) 1.0 mM catalyst concentrations; the linear fit and slope in (d) include only the first 4 points.Conditions: 250 mM PhOH, indicated CO2 concentration, 0.1 M TBAPF6 in DMF, 100 mV/s scan rate.

Electrochemical
Chloride Abstraction TitrationPrior to the chloride abstraction titration experiment, CVs of AgPF6 were investigated to ensure no redox features would interfere with the CV analysis of FeTriPP CO2 reduction catalysis.CVs of AgPF6 (0.1 mM) were taken under argon at two different scan windows.A larger scan window (FigureS22, black trace) shows the redox features corresponding to silver.Thus, the scan window for the subsequent AgPF6 titration was cut off at about -0.45 V vs. Fc/Fc + (FigureS22, red trace).

Figure S22 .
Figure S22.CVs of AgPF6 (0.1 mM) blank under argon.Black trace: large scan window.Red trace: scan window used in titration experiment.Conditions: 0.1 M TBAPF6 in DMF, 100 mV/s scan rate.

Figure S23 .
Figure S23.UV-Vis absorption spectra as a function of catalyst concentration under conditions of excess chloride at 0.10-0.0125mM of FeTriPP.Conditions: indicated catalyst concentration, 0.1 M TBAPF6 + 10 mM TBACl in DMF, 1 mm path length.

Figure S25 .
Figure S25.Schematic of proposed equilibria between monomeric, dimeric, and larger aggregates of iron porphyrins (horizontal equilibrium arrows) and electron transfers (vertical arrows) that highlights the solution speciation complexity.Additional speciation changes arising from axial ligand exchange or μ-oxo dimer are not depicted but likely also occur.

Figure S26 .
Figure S26.Iron(III) chloride porphyrin aggregate structures investigated, showing FeTetraPP as a representative case.Hydrogen atoms have been removed for clarity.

Table S1 .
Mean observed rate constant (kobs) values and standard deviations as a function of iron porphyrin catalyst concentration, based on three independent CV measurements for each catalyst.Data is plotted in Figure3cin the main text.