Photoinduced absorption spectroscopy (PIAS) study of water and chloride oxidation by a WO 3 photoanode in acidic solution

of Cl


S1 -Schematic of electrochemical reaction cell
The reaction cell used for all photoelectrochemical measurements (except for the PIAS and IPCE experiments) is shown in Figure S1.The cell is made from PTFE (polytetrafluoroethylene), and features an opening at the bottom, sealed using the photoanode, a Pt coil as the counter electrode, an opening to accommodate a Ag/AgCl reference electrode, and an inlet and outlet for purging the electrolyte, which was 10 mL of either 1M HClO 4 (for water oxidation) or 1 M HClO 4 + 3.5 M NaCl for chloride oxidation.The photoanode was usually irradiated through the back of the photoanode using a 365 nm LED (40 mW cm -2 ).

S2 -Determination of the faradaic efficiency for O 2 generation, f O2
A schematic illustration of the setup used to determine f O2 is shown in Figure S2.1(a).In this setup, a mass flow controller (Aalborg, Orangeburg, USA) was used to regulate a constant 0.1 mL s -1 flow of argon through the PTFE vessel.Any O 2 produced in the photo-electrochemical reaction cell (see Figure S1) was carried through to the 3D printed holder (see  In this work, the WO 3 photoanode was placed in the electrochemical reactor (see Figure S1), which contained 10 mL of 1 M HClO 4 , and irradiated with 40 mW cm -2 365 nm UVA, while poised at the potential which would generate a steady current density of 0.1 mA cm -2 (ca.1.0 V vs Ag/AgCl).The reaction cell was continually flushed with Ar (0.1 mL s -1 ), and the gas outlet was monitored by the O 2 sensor, which yielded a % O 2 vs irradiation time profile illustrated in Figure S2.2, from which a plateau value of 0.0066 % (the average % O 2 value between t=100 min and t =120 min) was determined.The value of f O2 was then calculated using the following expression, where the 'actual %O 2 ' was taken as 0.0066 %, and the value of the theoretical O 2 was calculated using the following equation where i is the current (1 x10 -4 A), F is Faradays constant (96485 C mol -1 ), n is the number of electrons transferred in the oxidation of H 2 O to O 2 (4 electrons), r is the flow rate of the carrier gas (0.1 mL s -1 ), and 24200 is the volume, in mL, occupied by 1 mol of gas at room temperature and pressure.The value for the theoretical % O 2 was calculated to be 0.0063 %, and so, from equation (S1) it follows that f O2 =0.0066/0.0063 unity.Figure S3.1 -Schematic of the setup used for the determination of f Cl2 .Argon is passed through the PTFE vessel to a KI trap containing a 100 mL solution comprising of potassium iodide (0.36 M), sodium hydroxide (0.025 M), and potassium hydrogen phthalate (0.049 M). 1 Any Cl 2 flushed over to the KI trap produces I 3 -, the absorbance of which can be monitored by UV-vis spectroscopy, allowing quantification of the amount of Cl 2 produced.
The WO 3 photoanode was irradiated (40 mW cm -2 , 365 nm) and poised at 1 V vs Ag/AgCl for 30 minutes, while the sealed PTFE electrochemical reaction cell was continually purged with Ar, flushing any photogenerated Cl 2 from the reaction solution in the photoelectrochemical reactor through to the KI trap solution, where it generated I 3 -, which absorbs strongly at 353 nm (ε 353 =26400 mol -1 cm -1 ). 1,2  reaction vessel was continually flushed with Ar for 2 hours after the applied potential and irradiation had ceased, so that all the photogenerated Cl 2 was captured by the trap, and then the UVvis absorbance of the trap solution was measured using a 1 mm quartz cuvette and yielded the absorbance spectrum (solid line) shown in  The absorbance due to a chemical species, Abs, is related to the concentration, c, of the species, according to the Beer-Lambert law, where ε is the molar absorption coefficient of the species, and l is the path length the light passes through.The absorbance at 353 nm of the KI trap solution before the chloride oxidation procedure was 0.043, and after the procedure was 0.346, thus the absorbance due to I 3 -generation was (0.346-0.043) = 0.303.This absorbance was divided by the molar absorption coefficient for I 3 -at 353 nm (26400 L mol -1 cm -1 ), 1,2 and the path length (0.1 cm) to determine [I 3 -] in the KI trap to be 0.303/(26400x0.1)=1.15x10 -4 M.
Since the trapping solution was 100 mL in volume, this means the amount of I 3 -generated was 1.15 x10 -5 mol.Stoichiometrically, each molecule of I 3 -is generated by one Cl 2 molecule, so the actual yield of Cl 2 was also 1.15x10 -5 mol.
f Cl2 was determined using the expression where the theoretical yield of Cl 2 was calculated using where Q is the total charge transferred (equal to the shaded area under the chronoamperometry trace in Figure 3.2(a), multiplied by the exposed electrode surface area (ca. 1 cm 2 ), which works out as 3.58 C), F is Faraday's constant (96485 C mol -1 ) and n (=2) is the number of electrons transferred in the oxidation of chloride ions to generate 1 Cl 2 molecule (2Cl - Cl 2 + 2e -).
Thus, the theoretical yield of Cl 2 was determined to be 3.58/(2x96485) = 1.86x10 -5 mol, and so f Cl2 was determined to be 1.15x10 -5 /1.86x10 -5 = 0.62.Although this is a high yield, it is far from unity, thus indicating that either there was an oxidation product, or products, generated other than Cl 2 , or that a significant fraction of the generated Cl 2 did not make it to the trap solution, most likely by reacting with components of the cell, such as plastic tubing, the reactor cell wall or the Suba-Seals®.

S4 -Schematic of PIAS/IPCE cell
For PIAS and IPCE experiments, the 3D printed (polylactic acid) PIAS cell illustrated below (Figure S4) was used to host the reaction components: the electrolyte (ca. 10 mL), working electrode (WO 3 ), counter electrode (Pt coil) and reference electrode (Ag/AgCl).There were quartz windows on either side of the cell, to allow transmission of excitation electromagnetic radiation, λ excit (and in the case of PIAS the monitoring beam radiation, usually set at 500 nm), through the cell.

S5 -Incident photon to current efficiency (IPCE) measurements
A schematic of the setup used to determine incident photon to current efficiencies is illustrated in   Photon flux was calculated from the irradiance measurements made at each λ excit value using the following equation where h is Planck's constant (6.63x10 -34 J s), c is the speed of light (3.0x10 8 m s -1 ), and λ is the excitation wavelength (m).To illustrate how photon flux was calculated for each irradiance spectrum recorded for a set value of λ excit , we take here as an example the irradiance profile measured at λ excit 300 nm, which is shown in  In the same manner, photon flux was calculated for each ρ λ value at each value of λ (from 280 nm to 520 nm), that forms the beam of radiation used to excite the photoanode that emerges from the 1 kW/monochromator system illustrated in  IPCE values were calculated using the following expression, 3 IPCE = 100 * photocurrent density (electrons s -1 cm -2 ) photon flux (photons s -1 cm -2 )

#(𝑆8)
As an example calculation, for the WO 3 photoanode in 1 M HClO 4 , at λ excit = 300 nm the average photocurrent density was 1.375x10 -5 A cm -2 (see Figure S5.4), which was divided by the charge of one electron (1.6x10 -19 C) to express the photocurrent density as 8.59x10 13 electrons s -1 cm -2 .Then, using equation (S8), and the value of photon flux for 300 nm emission determined previously (1.18x10 15 photons s -1 cm -2 , see  The IPCE for the WO 3 photoanode in chloride oxidation solution (1 M HClO 4 + 3.5 M NaCl) was calculated in the same way as for the IPCE in water oxidation solution, according to equation (S8).For example, at 300 nm, the average photocurrent density was 5.91x10 -5 mA, which was divided by the charge of one electron (1.6x10 -19 C), to express the photocurrent density as 3.69x10 14 electrons s -1 cm - 2 .Then, using equation (S7), and the value of photon flux for 300 nm emission determined previously

S6 -The PIAS/TC system
A schematic of the PIAS/TC (photoinduced absorbance spectroscopy/transient photocurrent) system used in this study is shown in Figure S6.1.In this system, a 150 W tungsten-halogen lamp with stabilized power supply (SLS301, Thorlabs, Newton, USA) was used as the monitoring light source.The WO 3 photoanode was irradiated with a 10 W, 365 nm LED (LZ4-44UV00-0000, LedEngin Inc., San Jose, USA), the irradiance of which was varied using a variable power supply unit (QL355P, Aim and Thurlby Thandar Instruments, Huntingdon, UK) for the LED.A diaphragm shutter (SHB1T, Thorlabs, Newton, USA), controlled by a data acquisition (DAQ) card (USB-6361, National Instruments, Austin, USA), was used to control the duration of this UVA irradiation.Two monochromators, positioned on either side of the sample, were used to set the wavelength of the monitoring light beam that was transmitted through the WO 3 photoanode.The intensity of the transmitted light was monitored using a Si photodiode detector (DET100A2, Thorlabs, Newton, USA) coupled to an amplifier (PDA200C, ThorLabs, Newton, USA) and recorded by the DAQ card.This setup allowed the change in absorbance of the WO 3 photoanode, ΔAbs, to be determined upon its irradiation by the 365 nm LED.In this work, the WO 3 photoanode was biased at a potential of 1.3 V vs Ag/AgCl, thereby allowing the photocurrent to be monitored as a function of irradiation time at the same time as that of ΔAbs.The absorbance of a sample is calculated as the log 10 of the ratio of the incident monitoring light intensity, I 0 , to its transmitted intensity after passing through the sample, I T : In PIAS measurements, ΔAbs, the absorbance change, is calculated similarly, but instead of I 0 and I T , the transmitted monitoring light intensity before UVA irradiation I T,0 and the transmitted intensity at every other time, t, are used, as follows: In order to determine at which monitoring wavelength the photogenerated electron holes absorb most strongly, PIAS/TC was used to measure the ΔAbs of the WO 3 photoanode in 1 M HClO 4 , upon 76 mW cm -2 365 nm UV irradiation, using 500 nm, 600 nm, 700 nm, 800 nm and 900 nm monitoring light wavelengths (λ mon values).The resulting ΔAbs vs t profiles are shown in This spectrum reveals an absorbance maximum at 500 nm, which matches well with the previously reported absorbance maximum of photogenerated holes on WO 3 at 475 nm, obtained for a WO 3 film in a solution containing the electron scavenger AgNO 3 , 4 (reproduced and shown in Figure S6.2(b)) and with other transient absorbance/diffuse reflectance studies which have reported absorbance due to photogenerated surface holes on WO 3 at 500 nm. 5,6 f the photoinduced absorbance had been due to photo-excited electrons on WO 3 , the PIAS spectrum would have been expected to show a maximum absorbance at 900-950 nm, decreasing at shorter wavelengths, in accordance with previous transient absorbance studies. 4,5 hus the absorbance at 500 nm was assigned to the photogenerated surface holes on the WO 3 photoanode, and 500 nm was used as λ mon for the remaining PIAS/TC experiments.The plateau values in Abs and J reached during irradiation are the steady state values of these parameters Abs ss and J ss , calculated as the average values between t=2.5 s and t=4.5 s.
Assuming the photoinduced absorbance is due to photogenerated surface holes, h + s , then by Beer's law, ΔAbs ss is proportional to the accumulated steady state concentration of electron holes at the photoanode surface, [h + s ].If the photooxidation reaction taking place is known to be water oxidation, because f O2 = 1, then J ss , the steady state photocurrent, is proportional to the rate of water oxidation, R. Therefore, it follows that the water oxidation rate law with respect to h + s , can be expressed in terms of the experimentally measured parameters, J ss and Abs, as so that the slope of a plot of log J ss vs log ΔAbs is equal to n, the order of water oxidation with respect to photogenerated surface electron holes.

S8 -X-ray diffraction (XRD) pattern of WO 3 nanopowder
The X-ray diffraction (XRD) pattern of the WO 3 nanopowder used to prepare the photoanodes, and the XRD lines associated with monoclinic WO 3 , shown in black and red respectively, are shown in Figure S8.The lines associated with monoclinic WO 3 were obtained from an international XRD database, 13, 14 and illustrate that the WO 3 used to prepare the WO 3 photoanodes in this study was monoclinic.

Figure S8
-X-ray diffraction pattern for the WO 3 nanopowder used to prepare the WO 3 paste photoanodes (black), with the XRD lines associated with monoclinic WO 3 (red) (obtained from an international XRD database). 13, 14

S9 -Stability studies
In order to test the stability of the WO 3 photoanode, in HClO 4 and NaCl/HClO 4 electrolytes, each system was irradiated for 24 h using an irradiance of 40 mW cm -2 , with the WO 3 photoanode biased

S10 -Detection of ClO 3 -
To account for the discrepancy in the number of electrons transferred and the amount of Cl 2 generation so that f Cl2 = 0.62, see Section S3, further work was carried out to see if chlorate, ClO 3 -, was also produced, alongside Cl 2 , via the following competing oxidation reaction In this work, 1 mL solutions of NaClO 3 in 1 M HClO 4 + 3.5 M NaCl were prepared, where different chlorate concentrations, ranging from 0 -1x10 -3 M were prepared.To these 11 solutions, of different known chlorate concentration, were added 5 mL of a Lissamine green solution (1x10 -4 M, in NaCl/HClO 4 electrolyte), and its decolourisation (bleaching) by the chlorate measured.The oxidation and resulting decolourisation of Lissamine green dye solution has been exploited before as a test of the oxidizing performance of a photoanode, 15 whereas here it was used to test for the presence of oxidizing agent ClO 3 -in solution.
In the above calibration work, the resulting initially orange solutions bleached over time and at a rate that was proportional to the [ClO 3 -].Thus, after 24 h, the UV-vis spectra of the 11 solutions of known
Figure S2.1(b)), containing a fluorescence-based oxygen sensor, O 2 xyDot (O 2 xyDot®, OxySense, Devens, USA), which was maintained at 25 °C using a circulating water bath (RCB20-PLUS, Hoefer, Holliston, USA).The change in lifetime of the sensor was measured using a NEOFOX-GT probe (Ocean Insight, Orlando, USA), from which a value for the % O 2 in the argon carrier gas stream was determined, via the Stern-Volmer equation.

Figure S2. 1 -
Figure S2.1 -(a) Schematic of setup used for the determination of f O2 .Argon is passed through the PTFE vessel at 0.1 mL s -1 , carrying any O 2 produced in the PTFE vessel through to the 3D printed thermally regulated holder containing a fluorescence-based oxygen sensor, O 2 xyDot.The change in lifetime of the sensor, measured using a NeoFox-GT Probe, allows the determination of the % O 2 in the argon carrier gas stream.(b) Schematic of the custom-made, 3D-printed thermally regulated O 2 detector cell.Water was passed through the holder using a thermostatted circulating bath set at 25°C.

Figure S2. 2 -
Figure S2.2 -The change in % O 2 with t, recorded for the WO 3 photoanode in 1 M HClO 4 , under 40 mW cm -2 365 nm irradiation.The red dashed lines indicate the initiation and termination of applied potential (set to maintain a constant current of 0.1 mA, ca. 1 V vs Ag/AgCl), and the green dashed line indicates the actual steady state yield in % O 2 , 0.0066 %, calculated as the average % O 2 value between t =100 min and t = 120 min.

Figure S3. 2 -
Figure S3.2 -(a) The chronoamperometry profile showing the variation in current density (J) with time (t), with the shaded area representing the total charge transferred, Q, and; (b) the UV-vis absorbance spectrum of the KI trap solution before (dashed line) and after (solid line) photogenerated Cl 2 had been flushed into the trap and ΔAbs 353 stopped increasing, recorded in a 1 mm quartz cuvette.

Figure S4 -
Figure S4 -Exploded view of the PIAS cell, showing how the 3 electrodes are inserted into the 3D printed cell.The working electrode is held in place between 2 quartz windows, allowing the transmission of the excitation radiation and (for PIAS) the monitoring beam.

Figure S5. 1 -
Figure S5.1 -Schematic of the IPCE setup.Monochromatic light from 300 nm to 500 nm in 10 nm steps was achieved using a 1 kW Xe lamp connected to a monochromator.The monochromatic light was directed through the window of a PIAS cell, illuminating a 0.636 cm 2 area of the WO 3 working electrode.To calculate the IPCE value at each value of λ excit it is necessary to determine the photon flux delivered to the photoanode, and the photocurrent generated in response to this illumination.To determine the former, the irradiance vs wavelength profile of each of the apparently monochromatic light beams, set at λ excit , delivered by the 1 kW Xe-Arc lamp/monochromator, incident to the photoanode in IPCE photocurrent measurements were recorded using a calibrated spectroradiometer (OL 756, Gooch and Housego, Ilminster, UK), at 10 nm increments from 300 -500 nm (controlled by a monochromator).The irradiance vs wavelength profiles recorded for each selected value of λ excit , and the photon flux calculated from this data for each selected value of λ excit are shown in FigureS5.2.

Figure S5. 2 -
Figure S5.2 -The irradiance (ρ) of the Xe-Arc lamp, measured at different λ excit values, which range from 300 nm to 500 nm (every 10 nm).The black lines show the irradiance profiles, and the red data points show the sum photon flux output for each irradiance profile at each emission maximum, λ excit .

Figure S5. 2
and magnified in Figure S5.3.FigureS5.3shows that the spectrum for the 300 nm λ excit peak is made up of individual irradiance data points, measured for every nm in the studied range, i.e. ρ λ values, where λ ranges from 280 to 520 nm (FigureS5.3shows ρ λ values between 280 to 325 nm only, for illustrative purposes), but with a peak maximum at λ excit , which in this case is 300 nm.

Figure S5. 3 - 8 = 2 .
Figure S5.3 -The irradiance profile, shown between 280 nm -325 nm, delivered by the 1 kW Xe Arc lamp when the monochromator was used to select λ excit to be 300 nm.The diamond data points show the measured irradiance values at each wavelength, ρ λ , that comprises the beam of radiation used to excite the photoanode that emerges from the 1KW/monochromator system illustrated in FigureS5.1,when the monochromator is set at λ excit = 300 nm For each ρ λ value, the corresponding photon flux was calculated according to equation (S6).For example, in FigureS5.3, the irradiance a 291 nm, ρ 291 , was 1.6x10 -3 mW cm -2 , and so

Figure S5. 1 , 2 . 2 .
when the monochromator is set at λ excit = 300 nm.These values are then summed, to determine the total photon flux delivered by the 300 nm, λ excit , beam, which in the case of the λ excit emission peak illustrated in FigureS5.3= 18x10 15 photons s -1 cm - 2 , as shown by the appropriate red coloured data point in FigureS5.This same method was applied to determine the photon flux delivered by the 310, 320, 330 etc. nm beams of irradiation used to excite the photoanode that emerges from the 1 kW/monochromator system illustrated in Figure S5.1, when the monochromator is set at λ excit = 310, 320, 330 etc. nm, thereby generating the data necessary to construct the photon flux vs λ excit data points shown in red in Figure S5.To determine IPCE, the photocurrent generated under these same irradiation conditions was measured using the setup illustrated in FigureS5.1.Thus, the photoanode was exposed to a beam of radiation from the 1 kW/monochromator system illustrated in FigureS5.1,when the monochromator is set at λ excit , and the photocurrent measured.Photocurrent density (J) was determined by dividing the photocurrent by the irradiated photoanode surface area (ca.0.636 cm 2 ).In this work the excitation beam of radiation was mechanically chopped (5 s on 5 s of) 5 times, and the average "light on" photocurrent density (J average ) recorded, before the value of λ excit was altered by 10 nm and the process repeated as λ excit was varied from 300 to 480 nm.In this work the radiation was used to excite a WO 3 photoanode, held at 1.3 V vs. Ag/AgCl, in 1 M HClO 4 , and in 1 M HClO 4 + 3.5 M NaCl.The average photocurrent densities for each value of λ excit were calculated and are shown in Figure S5.4,for both electrolyte solutions.

Figure S5. 4 -
Figure S5.4 -The average photocurrent densities generated upon irradiation of the WO 3 photoanode, poised at 1.3 V vs Ag/AgCl, in 1 M HClO 4 (black) and 1 M HClO 4 + 3.5 M NaCl (red), using different values of excitation wavelength, λ excit .This data was combined with the photon flux for each λ excit value, to generate IPCE values, using an equation of the form of equation (S7).

Figure S6. 1 -
Figure S6.1 -(a) A schematic diagram of the setup used to obtain PIAS/TC measurements, and; (b) a simplified diagram of the PIAS cell showing the direction of the monitoring and excitation light, and the incident and transmitted light intensities, I 0 and I T , respectively.

Figure S6. 3 -
Figure S6.3 -For illustrative purposes, a typical (a) Abs vs t trace and (b) J vs t trace obtained during a PIAS experiment, in which the photoanode was irradiated between t=0 s and t=5 s.

at 1 . 3 V
photoelectrochemical systems appeared very stable, suggesting little or no photocorrosion of the photoanode.

[
ClO 3 -] (plus lissamine green) were obtained (Figure S10(a)), and for each the change in the absorbance, ΔAbs was determined (by subtracting the absorbance of the dye containing solution at 444 nm from that at 800 nm).The results of this calibration study were plotted in the form of ΔAbs vs [ClO 3 -], see Figure S10(b), to reveal a linear relationship.A used electrolyte solution was then prepared, which was the residual 1 M HClO 4 + 3.5 M NaCl electrolyte solution following a Cl 2 yield run(30 minute irradiation (40 mW cm -2 , 365 nm) and applied potential (1 V vs Ag/AgCl)).When the 'used electrolyte solution' was assessed in the same way as the chlorate calibration solutions, the measured value of ΔAbs due to the bleaching of the Lissamine green was insignificant, suggesting little or no ClO 3 -was generated.This observation suggests that the missing 38% of oxidation product could either be HClO 4 , or undetected Cl 2 , given the highly reactive nature of the latter which could lead to a significant loss in the level of Cl 2 in the Ar gas stream as it is swept from the electrochemical reaction cell to the KI trap.

F
igure S10 -(a) UV-vis absorbance spectra of the lissamine green test solutions with varying [ClO 3 -] (varying from 0 to 1x10 -3 M, shown in black) and the electrolyte solution with unknown [ClO 3 -] (red), and; (b) the calibration plot of ΔAbs (444 nm) vs [ClO 3 -] (black data points and fitted trendline), and the ΔAbs (444 nm) of the electrolyte test solution, shown as a red dashed line.