Role of Cobalt − Iron ( Oxy ) Hydroxide ( CoFeO x ) as Oxygen Evolution Catalyst on Hematite Photoanodes

Photoelectrochemical solar water splitting into hydrogen and oxygen offers an elegant and potentially efficient way to store solar energy in the chemical bonds of hydrogen, but the oxygen evolution rate is quite limited. The deposition of an oxygen evolution catalyst on the photoanode can enhance oxygen evolution, although the precise interplay between the semiconductor and the catalyst remains poorly understood and unoptimized. In this work, we use a combination of electrochemical approaches, including photoelectrochemical impedance spectroscopy and intensity modulated photocurrent spectroscopy, to unravel the nature of the interactions between different loadings of an electrocatalyst (CoFeOx) and a hematite (α-Fe2O3) semiconductor. A thin layer of CoFeOx mainly reduces surface charge recombination, while an extremely thin layer enhances charge transfer kinetics. Moreover, an interlayer of GaOx modifies the surface state distribution and increases the charge transfer rate even further. These findings point to new opportunities for understanding and manipulating complex photoanodes for oxygen evolution.


2/15
Figure S15 (a) An energetic scheme of a hematite photoanode coated with excessive amount of CoFeO x .(b) J-V curve (under 1 sun chopped illumination, 5 mV s -1 ) of a hematite photoanode (black) coated with CoFeO x using electrodeposition at 1.8 V RHE for 25 min with same electrolyte composition for other CoFeO x coatings done in this work.Almost no net photocurrent density can be obtained at high applied potentials compared with h/cfox30 (blue).

Note S1 Justification of PEIS and Mott-Schottky analyses.
To confirm the validity of the EIS and Mott-Shottky analyses used on the porous nanostructured hematite photoanodes of this work, we need to ensure that nanostructuring does not significantly change the behavior within the space charge region and at interface.The behavior will not change if the space charge region thickness is much smaller than the nanostructure dimension.
We start with the estimation of doping density N d of hematite and flat band potential E FB by investigating a flat film using the same preparation method as reported in the Experimental but without P123 sacrificial pore templating.Capacitance is obtained by fitting impedance spectra of this film measured in the PEC setup as described in Experimental in the dark.The Mott-Schottky plot gives an intercept of 0.6 V RHE (approximated to be E FB ) and a slope of 9.13 10 9 F -2 V -1 .Hence, doping density N d is given by × where q is the electronic charge (1.6 10 -19 C); is the relative permittivity of hematite (33), 1 is the ×   0 vacuum permittivity (8.85 10 -12 F m -1 ), A is the geometric area (1.69 cm -2 ), and k is the slope of the × linear fitting.Note this value is very high due to Sn diffusion from FTO after 800 o C treatment.
Assuming a potential drop at space charge region of 0.6 V, using the abrupt approximation, 2 the ∆  differential space charge capacitance is given by where k B is the Boltzmann constant (1.38 10 -23 m 2 kg s -2 K -1 ) and T is the room temperature (298K).× It should also be noted that C SC is considerably smaller than Helmholtz capacitance (C H ), which can be estimated to be 200 µF cm -2 . 3Therefore, the effect of potential drop at Helmholtz layer is ∆  negligible.
This result for C SC is in excellent agreement with experimental value (8.06 µF cm -2 at 1.2 V RHE ).Therefore, the doping density N d of 1.64 10 20 cm -3 used in our model is a good approximation and × can be used for our nanostructured photoanodes to calculate potential drops.
The applicability of the kinetic model on nanostructures is demonstrated as follows.Since there is not an established model for mesoporous structures, we apply our nanostructure dimension (radius R=45 nm) to a model designed for nanorods recently developed by Peter et al. 3

9/15
For a rod structure, the potential drop at space charge region is now calculated as where x is the distance from the edge of the space charge region to the center of the rod.
If is again assumed to be 0.6 V, is 41.3 nm, and the space charge region width is ∆ , Here potential drop at Helmholtz layer is given by equation: and is found to be less than 0.05 V for C H =200 µF cm -2 .Again, this is negligible.
From the calculation above, we found that the depletion region is very shallow.Under illumination, strong Fermi level pinning is induced and will be much lower so the depletion width will be ∆  smaller than 3.7 nm.Therefore, our impedance as well as Mott-Schottky analyses are valid in both dark and light conditions.

Note S2 Demonstration of formal equivalence between the phenomenological kinetic model and EIS equivalent circuit model
The kinetic analysis of impedance data by Wijayantha et al. suggested two semicircles of high frequency (HF) and low frequency (LF). 4The intercepts of the two semicircles with Z real axis are R ser , R ser +Z 2 and R ser +Z 1 .This can be adapted to the equivalent circuit used in this paper for a bare hematite photoanode according to careful examination of Bode plots (Figure S4).It has been previously reported that charge transfer occurs on the scale of 0.1 to 1 s, 5 which matches the peaks of phase angles of lower frequency semicircles.Consequently, we attribute low frequency semicircle to interfacial charge transfer and the high frequency semicircle to surface charge recombination by trapping states.Series resistance remains the same physical meaning but is denoted as R s here.Therefore, the resistive equivalence can be described as Now Equation 4 in Wijayantha's paper 4 can be re-written as Since the rate constants of charge transfer and recombination rate constant are k ct and k rec respectively in this paper (instead of k t and k r ), we have or According to Wijayantha's model, k ct is equal to the frequency corresponding to the highest imaginary component of the low frequency semicircle, which is by definition, in the electrical circuit, the reciprocal of the time constant τ(LF) of the C ss &R ct unit (Figure 3a).This can be formulated as The surface recombination rate constant can now be expressed through Equation (10)     = 1 It can be immediately seen that the two rate constants share C ss but differ in resistance.This is understandable considering the kinetic model where charge transfer and surface recombination are two competing processes taking place at the interface that utilize charges stored in surface states (cf. Figure 1 of ref 1).
From the deductions shown above, we believe that the concepts of the phenomenological model and parameters in the equivalent circuit share the same root and our impedance data can be used for kinetic analyses.

Note S3 Brief introduction to IMPS measurement and data interpretation
IMPS serves as a tool to probe surface kinetics of photoelectrodes, which has been proved suitable for multi-step charge transfer reactions. 6It has been successfully applied to hematite photoanode systems by Peter et al. with a simple model where charge transfer only occurs at surface states while competing with surface recombination, 7 which is also used in this paper for PEIS study.In an IMPS measurement, a base light illumination intensity is employed to reach steady-state conditions while a frequencydependent sinusoidal illumination is imposed to the system and the corresponding modulated photocurrent response is measured.
The spectrum of a typical IMPS measurement for an n-type semiconductor photoanode consists of two semicircles located at first and fourth quadrant.The high frequency semicircle starts in the fourth quadrant that represents the attenuation by the total resistance of the cell and combined space charge capacitance and Helmholtz layer capacitance, giving an RC time constant of the cell.The high intercept with the real axis depicted in Figure S17 represents the modulated photocurrent that is reflective of the hole flux arriving at the surface, which happens at a certain frequency such that recombination processes are effectively "frozen out". 7By decreasing the frequency of perturbation even further surface recombination and charge transfer processes dominate, giving another semicircle in the first quadrant.The imaginary part reaches a maximum when the characteristic time constant of the system is met: 8 The system then moves toward its steady state as frequency lowers further, giving a low intercept value that corresponds to steady state photocurrent.The ratio of low frequency and high frequency intercept represents the fraction of holes that arrives at surface and is injected to the electrolyte, which is effectively charge transfer (injection) efficiency.
The two rate constants k ct and k rec can then be extracted using Equations 13 and 14.This frequency dependent measurement is particularly useful since it allows one to obtain phenomenological kinetic constants for charge transfer and recombination directly, without altering the band bending or charge distribution inside the semiconductor.
Figure S17 A representative IMPS complex plot of a hematite photoanode. 12/15

Figure
Figure S1.XRD pattern of h/cfox0 showing hematite facet indices.Stars indicate diffraction peaks from FTO

Figure S3 .
Figure S3.Current density as a function of deposition time during repeated unidirectional LSV electrodeposition of CoFeO x .Applied potential is repeatedly swept from 1.35 to 1.65 V RHE.

Figure
Figure S5 Bode plot comparison of (a) h/cfox0 and (b) h/cfox3 measured at 1 sun illumination.Lighter line colors represent lower potentials (from 1.2 to 0.6 V RHE ).The non-zero phase angles at frequencies between 10 4 and 10 5 Hz may originate from the capacitance at electrode/sample interface.

Figure S6 .
Figure S6.Bode plots of h/cfox30 measured at 1 sun illumination.Lighter line colors represent lower potentials (from 1.2 to 0.6 V RHE ).An additional peak in phase angles can be observed compared to Figure S5 above 1.0 V RHE (marked in black), which corresponds to a new semicircle at low frequencies as exemplified in (b) in the case of 1.15 V RHE .

Figure S7 .
Figure S7.Equivalent circuit used for h/cfox30 when (a) under 1 sun illumination or (b) in the dark.Elements in red box are only used when applied potential is larger than 1.0 V RHE .

Figure
Figure S8 (a) Dark J-V curve of h/cfox0 (red), h/cfox3 (green) and h/cfox30 (blue) scanned at 10 mV s -1 .(b) Nyquist plots from EIS results for h/cfox30 at various potentials in the dark.A new semicircle can be fitted from 1.00 V RHE on with strong capacitive character.

Figure
Figure S12 Calculated (a) charge transfer efficiency and (b) hole flux for h/cfox0 (red squares), h/cfox3 (green circles) and h/cfox30 (blue diamonds) for the estimation of photocurrent density shown in Figure 7c.

Figure S16 .
Figure S16.Mott-Schottky plot for a flat hematite film.

Table S1 -
S6 show original fitted EIS parameters obtained from Zview software.The working area of all electrodes is 0.25 cm 2 .