Overcoming Residual Carbon-Induced Recombination in Water-Oxidation Catalysis: Combining a Superior Catalyst with Low-Carbon-Content Atomic Layer Deposition of SnO 2 for Improved Catalysis

Two routes, as well as their combination, are examined for overcoming recombination in a dye-sensitized photoelectrochemical cell for water oxidation catalysis.

For each XPS figure, the light grey spectra are raw data for each replicate and the black spectrum is an average of all replicates.Note that the peak fits shown correspond to one of the raw data (grey) spectra deemed similar to the average spectrum.Satellite peaks for iridium (at approximately 63.5 eV and 66.5 eV energy) were not fit as they proved unnecessary for a complete and accurate fit.

V vs Ag/AgCl
Current (μA) . Representative CVs at different scan rates for ALD-SnO 2 AgCl in pH 7, 0.1 M KPi buffer where the blue line is 0.1 V/s, orange is 0.05 V/s, green is 0.01 V/s and purple is 0.001 V/s.

S3. Further Discussion of Deposition of Li-IrO x
In order to better replicate the previous work using Li-IrO x , the Li-IrO x catalyst was sprayed onto the device using a nebulizer. 1A handheld nebulizer (Hudson RCI, Micro Mist) was used with air to deposit 1.0 mL a 95:5 v/v ethanol: water mixture containing 2.5 mg of Li-IrO x .
The back end of the device covered using parafilm to minimize losses of the deposition solution.
The air pressure used was the minimum value sufficient to nebulize the solution.This value was not measured, but the manufacturer reports this value to be 8 liters per minute.As the device aperture was larger than the anode, a short channel was constructed out of parafilm in the shape and size of the anode in order to ensure all of the solution was deposited onto the anode in the correct area.Additionally, it was thought that this would minimize inconsistency that could occur if manual scanning across the anode with the nebulizer was done.The anode was then heated to 120 °C under air.The precise catalyst loading was not determined.

S4. Open-Circuit Photovoltage Measurements and Recombination Rate Calculations
Under illumination, the quasi-Fermi energy (E F,n ) of electrons in the nano-SnO x C y substrate is determined by the relative rates of injection of both the injection of electrons from the photoexcited PMPDI and the sum of all recombination equations.[4] (S1) To calculate this change the constant value out front can be calculated as k B multiplied by temperature divided by e, resulting in 25.85mV.For the anodes studied herein, A (-) shift in V oc occurs due to a (+) shift in E F,n (what we were actually measuring, vs. Ag/AgCl).In the case of a change of -70 mV in the V oc as observed with Li-IrO x addition, a 15-fold decrease in recombination relative to the rate of injection can be calculated.

S5. Electrochemically-Active Surface Area Correction
In order to compare the electrochemically-active surface area of the planar and 3-dimensional SnO 2 anodes, the double-layer capacitance was determined.Use of that double-layer capacitance assumes that the two types of SnO 2 , nano and ALD SnO 2 , have the same electrochemical response.
A scalar value was determined by measuring the surface area of the planar SnO 2 anodes using photos of the anodes, and then using ImageJ to obtain the surface area, followed by plotting that surface-area value vs the current at +0.4 V vs Ag/AgCl obtained from cyclic voltammetry scans at varying scan rates, Figure 5.Using eq.S2 2 , where i c is current, A is area, C d is double layer capacitance, and v is scan rate, the slope of this line was determined to be directly proportional to the double-layer capacitance.
The double-layer capacitance determined from the slope can then be divided by the geometric surface area of the planar anodes, giving a scalar value representing the ratio of the electrochemically active surface area / the geometric surface area.
One caveat here is that due to the different thicknesses of the films (especially that of thick nano-SnO x C y ), diffusion limitations of the H 2 Q could be a factor in the relatively lower photocurrent seen in the nano-SnO x C y /PMPDI anodes.However, we think this is unlikely for two reasons.First, we do not we do not observe significant current spiking and then significant decay behavior with H 2 Q addition over time, which we would expect to see if there issues with H 2 Q diffusion in the anode.Second, using different thicknesses of nano-SnO x C y for water oxidation results in differing geometric photocurrent densities. 3One would assume that if diffusion limitations were present increasing the thickness of the film and amount of dye would not increase the photocurrent as those new surfaces would be inaccessible.
In a second correction method for comparing the anodes made with planar SnO 2 by ALD to the nano-SnO x C y anodes made in house, a correction using dye absorbance was made on photocurrent transient plots examining the sacrificial reductant hydroquinone, H 2 Q.Given that it is not known if the dye absorbance or the electrochemically-active surface area is a more accurate correction method for the present systems, the data were also corrected using the higher, 1.0:190 value as well of the planar to nanostructured, electrochemically-active surface area, Figure S3.Given that the corrected data from the planar SnO 2 anodes, using the electrochemicallyactive surface area as the correction, is significantly higher than the highest values observed in any nanostructured system, it seems likely that the true value for the planar anodes is somewhere between the two limiting, corrected values.Nevertheless, the planar SnO 2 outperforms the nano-SnO x C y no matter if the comparison is based on dye absorbance or on the electrochemically-active surface area.The background current does become exaggerated when the planar photoanode's current is scaled up for comparison to the nano-SnO x C y anodes, although it should be noted that the resultant, inflated background value is a poor representation of the true background current as well as a poor measure of the anode's photo performance in the dark.

S6. Comparison of our Previously Published Best Performing Anodes with Alumina
In another previous study of ours, we have used ultrathin AlO n prepared by ALD to improve the water oxidation performance of the SnO x C y /PMPDI anode with or without CoO x WOC. 3 Using the current density correction approach shown here, we can also examine how the choice of ultrathin ALD materials (i.e., AlO n vs. SnO 2 ) affects the water oxidation performance.
Accordingly, the photocurrent data in buffered water was corrected using the previously determined dye absorbance value in order to be comparable to the nano-SnO x C y /PMPDI system and was coplotted with the best performing version of a representative anode, nano-SnO x C y /PMPDI/PMPDI/AlO n (0.6 nm, 85 °C deposition) from the noted previous study 3 that, again, includes an overlayer of AlO n deposited by ALD, both with and without CoO x (Figure 7).The following scheme, intended as a working hypothesis for going forward, is an adaptation of our previous kinetics scheme for the nano-SnO x C y /PMPDI/CoO x system 5 .hypothesis that recombination originating from pathways attributable to carbon is minimized, while more charge transfer occurs through the optimal green pathways with the more active Li-IrO x catalyst.
What is not represented in the above kinetic scheme is the changes in the rate constant for each charge transfer pathway that occur by altering the WOCatalyst.Our data are consistent with and fully support the hypothesis that decreased recombination in the systems with Li-IrO x can be attributed to increased rate constants for charge transfer in pathways involving Li-IrO x , specifically k int and k regen .These rate constants reflect charge injection from the water oxidation reaction into Li-IrO x and charge regeneration of the dye from Li-IrO x , respectively.Finally, we emphasize once more that Scheme S1 is provided only as a working hypothesis for further investigations going forward.

Figure S1 : 2 Figure S2 : 2 QFigure S5 :Figure S1 .
Figure S1: XPS of Li-IrO x S2.Plotted Scan Rate Data for ALD-SnO 2 Figure S2: CV electrochemical data at different scan rates for ALD-SnO 2 .S3.Further Discussion of Deposition of Li-IrO x S4.Open-Circuit Photovoltage Measurements and Recombination Rate Calculations S5.Electrochemically-Active Surface Area Correction Figure S3: J−t transients of ALD-SnO 2 /PMPDI with and without CoO x in the presence of 20 mM H 2 Q sacrificial reductant corrected for electrochemically-active surface area S6.Comparison of Previous Best Performing Anodes with Alumina Figure S4: Photocurrent transients of ALD-SnO 2 /PMPDI with and without CoO x compared to nano-SnO x C y /PMPDI/AlO n (0.6 nm, 85 °C) corrected for electrochemically-active surface area.S7.Photocurrent Measurements on ALD-SnO 2 anodes with and without Li-IrO x without H 2 Q Figure S3.J−t transients in the presence of 20 mM H 2 Q sacrificial reductant, with background dark current due to dark oxidation of the hydroquinone subtracted, at +0.2 V vs. Ag/AgCl in pH 7, 0.1 M KPi buffer with 30 s light/dark transients where purple is nano-SnO x C y /PMPDI, red is nano-SnO x C y /PMPDI/CoO x , blue dashed is ALD-SnO 2 (5 nm)/PMPDI (X190), and green dashed is ALD-SnO 2 (5 nm)/PMPDI/CoO x (X190).Both planar (blue and green) anodes are corrected (X190) to match the electrochemically active surface area of the nanostructured anodes.