Increased Solar-Driven Chemical Transformations through Surface-Induced Benzoperylene Aggregation in Dye-Sensitized Photoanodes

The impact of benzo[ghi]perylenetriimide (BPTI) dye aggregation on the performance of photoelectrochemical devices was explored, through imide-substitution with either alkyl (BPTI-A, 2-ethylpropyl) or bulky aryl (BPTI-B, 2,6-diisopropylphenyl) moieties, to, respectively, enable or suppress aggregation. While both dyes demonstrated similar monomeric optoelectronic properties in solution, adsorption onto mesoporous SnO2 revealed different behavior, with BPTI-A forming aggregates via π-stacking and BPTI-B demonstrating reduced aggregation in the solid state. BPTI photoanodes were tested in dye-sensitized solar cells (DSSCs) before application to dye-sensitized photoelectrochemical cells (DSPECs) for Br2 production (a strong oxidant) coupled to H2 generation (a solar fuel). BPTI-A demonstrated a twofold higher dye loading of the SnO2 surface than BPTI-B, resulting in a fivefold enhancement to both photocurrent and Br2 production. The enhanced output of the photoelectrochemical systems (with respect to dye loading) was attributed to both J- and H- aggregation phenomena in BPTI-A photoanodes that lead to improved light harvesting. Our investigation provides a strategy to exploit self-assembly via aggregation to improve molecular light-harvesting and charge separation properties that can be directly applied to dye-sensitized photoelectrochemical devices.

Figure S23.Simple energy diagram showing the absorption, emission, and solvent reorientation of monomers and H-aggregates.S10.Solvatochromism in the BPTIs.
Table S3.Solvatochromism of BPTI-A and BPTI-B in solution (µM) for various solvents.Figure S28.τe, Rrec, Cµ, against applied voltage for DSSCs and the equivalent circuit Table S4.The findings for Rrec, Cµ, τe and ηcoll derived from impedance data analysis All reagents and solvents were obtained from Sigma-Aldrich, Fluorochem, or VWR and used without purification unless described otherwise.DCM and toluene were dried in a solvent purification system.Column Reactions were performed under atmospheric conditions unless otherwise noted.Column chromatography was performed using silica gel (SiliCycle, SiliaFlash P60, 40-63 µm, 230-400 mesh) while fractions were analyzed using TLC (TLC silica gel 60 F254, Merck KGaA) visualized with 254/350 nm light. 1 H-NMR spectra were recorded on a Bruker AV400 spectrometer at 40°C due to poor solubility of the BPTI's, with chemical shifts reported in ppm relative to tetramethylsilane by referencing the residual solvent signal.Highresolution mass spectrometry (HR-MS) was measured on an AccuTOF GC v 4g, JMS-T100GCV mass spectrometer (JEOL, Japan) equipped with field desorption (FD) emitter, Carbotec (Germany), FD 13 μm.Current rate 51.2 mA min -1 over 1.2 min machine using FD as ionization method.
S2. Analytical methods.Infrared (IR) measurements were performed on a Bruker Alpha FT-IR machine at room temperature.Ultraviolet-Visible (UV-Vis) spectra were recorded on a Shimadzu UV-2700 Spectrophotometer in a 1 cm or 0.2 cm path-length quartz cuvette.Fluorescence spectra were recorded on a Fluorolog Jobin Yvon-SPEX in a 1 cm path-length quartz cuvette.Cyclic voltammograms were recorded on a PGSTAT10 potentiostat (Autolab) with either a working electrode (WE) = glassy carbon (MetrOhm, diameter 3 mm) or FTO|SnO2|BPTI-A/B electrodes (0.19 cm 2 ) clipped with an alligator clip to a holder, reference electrode (RE) = a leakless Ag/AgCl (eDAQ, ET069) and counter electrode (CE)= Pt wire in 0.1 M TBAPF6 in MeCN (Fc 0/+ at 0.55 V vs. Ag/AgCl), Benzonitrile (Fc 0/+ at 0.55 V vs. Ag/AgCl), DMF (Fc 0/+ at 0.52 V vs. Ag/AgCl) or pH 5.4 sodium acetate (100 mM) in H2O.The solvents were degassed with N2 to suppress the overlap of the O2 reduction peak with the redox waves of the BPTIs but did not influence the redox waves themselves.Dye-sensitized solar cells (DSSCs) and dye-sensitized photoelectrochemical cells (DSPECs) were illuminated with an LED light source (Zahner, TLS3, 100 mW cm -2 ) or a Solar Simulator (Oriel, LCS-100, AM1.5G, 100 mW cm -2 ) setting the light intensity with a calibrated silicon solar cell (Newport, 91150-2000) with the spectral output given in Figure S16.IPCE was measured with a tunable optical light source (Zahner, TLS03) in continuous mode with phase at 0.1 Hz and 5 counts.For electrical impedance spectroscopic (EIS) measurements, the Zahner LSW-1 light source was controlled by Zahner PP211 potentiostat at varying light intensities (1400, 1000, 600, 200, and 100 mW cm -2 ) across frequencies ranging from 0.1 to 100 KHz, with a lower limit of 0.1 Hz and 5mV amplitude.Hydrogen gas quantification was performed on a Gas Chromatograph (Shimadzu Nexis GC-2030) with a 5Å mol.sieve column (60 m, 0.32 mm internal diameter, 25 μm film thickness) with an inlet temperature of 40°C and a flow argon rate of 8.0 mL min -1 .BET analysis was performed on a Belsorp MAXII.The height profile of SnO2 of the FTO plate was measured with a 3D laser scanning Confocal Microscope XK-X1000 Series (Keyence) at 20× magnification.

S3. Synthesis and characterization of BPTIs.
BPTI-A and BPTI-B were synthesized by adapting literature procedures for similar PDI and BPTI modifications. [1,2]heme S1.Chemical structures and synthesis of BPTI-A and BPTI-B derivatives through Diels-Alder reactions on the PDI scaffold followed by condensation with 4-aminobenzoic acid.BPTI-A.N,N-bis(ethylpropyl)perylene-3,4,9,10-tetracarboxylic diimide (PBI-A, 0.53 g, 1 mmol, 1 eq.), maleic anhydride (13.3 g, 0.13 mol, 130 eq.), p-chloranil (1.7 mL, 6.9 mmol, 6.9 eq.) and nitrobenzene (1.7 mL) were combined and stirred at 220°C for 24 hours.The reaction was monitored by disappearing of the characteristic peak of the starting PDI-A at 522 nm (Figure S1A).The mixture was let to cool down to room temperature, after which EtOAc:MeOH (80 mL, 1:1 v/v) was added to the mixture and sonicated for 30 minutes.The mixture was filtrated and washed with EtOAc:MeOH (1:1 v/v).The crude intermediate was not further purified and directly used in the next step.To the crude alkylbenzoperylenediimide (0.5 g, 0.8 mmol, 1 eq.) and 4-aminobenzoic acid (0.39 g, 3.2 mmol, 4 eq.) was added DMF (75 mL) and stirred at 140 °C overnight.The mixture was concentrated under reduced pressure, then MeOH was slowly added to precipitate the product.This precipitation method was performed twice, affording the product yellow/orange powder (0.53 g, 87% of the crude alkylbenzoperylenediimide).BPTI-B.N,N'-bis(2,6-diisopropylphenyl)-3,4,9,10-perylenetetracarboxylic diimide (PDI-B, 0.42 g, 0.6 mmol, 1 eq.), maleic anhydride (9 g, 0.09 mol, 150 eq.), p-chloranil (0.34 mL, 1.4 mmol, 2.3 eq.) and nitrobenzene (1.7 mL) were stirred at 220°C for 24 hours.The reaction was monitored by UV-Vis, following the disappearance of the characteristic PDI-B absorption at 522 nm (Figure S1B).The mixture was cooled to ambient temperature, after which acetone (15 mL) and HCl(aq) (40 mL, 1 M) were added to the mixture, followed by sonication for 30 minutes.The mixture was filtered, the obtained solids dissolved in chloroform and filtered again.The crude intermediate was obtained by rotary evaporation of the filtrate, which was not further purified and was directly used in the next step.The crude aryl-benzoperylenediimide (0.1 g, 0.12 mmol, 1 eq.) and 4-aminobenzoic acid (41 mg, 0.3 mmol, 2.5 eq.) were dissolved in DMF (15 mL) and stirred at 140°C overnight.The mixture was cooled to room temperature, and DMF was removed under reduced pressure.The obtained solids were dissolved in DCM and subjected to column chromatography (silica) with a 5% methanol/dichloromethane eluant.The solvents were removed by rotary evaporation and the product as a yellow-orange powder (105 mg, 95% from crude aryl-benzoperylenediimide).A clean, with one-hole predrilled (Diamond tip nr.7134, Dremel-400) FTO plate (Solaronix, 2.2 mm, 15 Ω sq -1 ) was added to a solution of concentrated 1M HCl(aq) (37% w/v) in EtOH and sonicated for 30 minutes.The electrodes were rinsed with Milli-Q, EtOH and left to dry.Electrochemical deposition of platinum was performed using an aqueous PtCl4 solution (10 mM PtCl4, 50 mM HCl, 0.47 mM 3-(2-aminoethylamino)propyldimethoxymethylsilane) in a three-electrode system, with the FTO-plate as the working electrode (WE), Ag/AgCl (leakless, eDAQ, ET069) reference electrode (RE) and a Pt-mesh counter electrode (CE).The platinum was deposited onto the FTO-electrodes by chronoamperometry (galvanostatic) employing a set current of −0.025A vs. Ag/AgCl over 30 seconds with a PGSTAT10 potentiostat (Autolab), yielding a black hue at the FTO-electrodes.Finally, the FTO|Pt electrodes were rinsed with demi water and ethanol and left to air dry before use.S4.2.SnO2 paste preparation.SnO2 paste was prepared similar to literature procedures. [3]SnO2 nanoparticles (1 g, 22-43 nm, WAKO), glacial AcOH (0.2 mL), terpineol (3 g), ethyl cellulose (0.5 g), and EtOH (30 mL) were added to a container, after which zirconium beads (100 g, 0.8-1 mm) were added to the mixture.The container was placed into a ball mill machine (Changsha Deco Laboratory Planetary Ball Mill 400-800-4256), and the mixture was left to grind at 700 rpm for 16 hours.The ball mill machine was stopped, and the mixture was put into a round bottom flask using excess EtOH (50 mL).The EtOH was removed by rotary evaporation, yielding a thick grey/white paste suitable for screen printing.

S4.3. Cleaning FTO electrodes.
The fluorine-doped tin oxide (FTO) electrodes (Solaronix, 2.2 mm, 15 Ω sq -1 ) were scrubbed with Deconex and rinsed with hot water, followed by wiping with an acetone-soaked tissue and air drying.The FTO electrodes were then rinsed with acetone, toluene, and ethanol and left to air dry in between.The FTO electrodes were then placed inside a glass container in a solution of a teaspoon of Deconex in Milli-Q and sonicated for 30 minutes.The sonication procedure was repeated with Milli-Q, and then ethanol and the electrodes were left to dry, after which the electrodes were treated with a UV-ozone generator (Ultra-Violet Products, PR-100) for a minimum of 30 minutes.

S4.4. Preparation of FTO|SnO2 photoanodes.
A clean FTO plate (Solaronix, 2.2 mm, 15 Ω sq -1 ) (10×5 cm) was added to a 40 mM SnCl4 solution in Milli-Q (100 mL) and placed in the oven at 70°C for 30 minutes to create a SnO2 blocking layer.The FTO plate was consecutively rinsed with Milli-Q then ethanol and air-dried.The electrodes were sintered with progressive heating (Programmer PR 5) from ambient temperature to 500°C (over 20 min, 25°C min -1 ramp, hold 500°C for 30 min).One mesoporous layer of SnO2 nanoparticles (particle size 22-43 nm, WAKO, specific surface area 36 m 2 g -1 ) was screen printed (43T screen with 0.79 cm 2 circles for DSPECs, 0.19 cm 2 circles for DSSCs) onto the cooled FTO plate.The plate was dried for 7 min on a hotplate set at 125°C and cooled to ambient temperature, before printing (43T screen) a second SnO2 layer in the same manner.The plates were sintered with progressive heating from ambient temperature to 125°C (over 5 min, 25°C min -1 ramp, hold 129°C for 5 min) then to 202°C (over 15 min, 5.1°C min -1 ramp, hold 202°C for 5 min) to 375°C (over 5 min, 34.6°C min -1 ramp, hold 375°C for 5 min) to 450°C (over 5 min, 15°C min -1 ramp, hold 450°C for 15 min) to 500°C (over 5 min, 10°C min -1 ramp, hold 500°C for 30 min) prior cooling to room temperature for sensitization.The SnO2 layer onto the plates has a thickness after annealing of 6.1 µm (Figure S11).The FTO|SnO2 electrodes were sensitized with 0.25 mM BPTI-A or BPTI-B in a 1:1 t-BuOH/THF solution in the dark overnight, rinsed afterward with a 1:1 t-BuOH/THF solution, and kept in the dark before use.

S4.5. Dye leaching experiments.
Dye loading was determined by adding freshly sensitized FTO|SnO2|BPTI-A or BPTI-B plates (0.19 cm 2 circles) in a solution of TBAOH in DMF (0.01 M, 1 mL, colorless) and left to soak overnight.The yellow dye-containing solutions were measured with UV-Vis to obtain absorbance spectra (Figure S12).The dye loading was determined using the molar extinction coefficient of the 0-0 vibrational band at 467 nm (BPTI-A) and 469 nm (BPTI-B).4), absorbance spectra of the reaction mixture are compared to a premade solution of Br3 -/Br2.Known amounts of Br2 have been added to the starting solution with HBr and LiBr, consecutively measured with UV-Vis spectroscopy.A molar extinction coefficient of 33469 M -1 cm -1 at 266 nm was observed for Br3 -/Br2 and concluded to be a good reference point as the starting solution HBr/LiBr solution lacked absorption at this wavelength (Figure S17).S7.Illumination setup for dye-sensitized photoelectrochemical cells.The photoreactor (Figure S18) comprises 2 Teflon compartments, which are separated by a Nafion-117 membrane (FuelCellStore).The working compartment (WEC) contained the photoanode (masked size 0.64 cm 2 ) as a WE and an Ag/AgCl (leakless, eDAQ, ET069) RE.The WEC was filled with a solution of 1.0 M LiBr, 0.1 M HBr in H2O (3 mL).The counter electrode compartment (CEC) containing a CE (Pt electrodeposited on FTO), and was filled with a solution of 1.0 M LiBr, 0.1 M HBr in H2O (3 mL).The top contains a large opening with a septum to add/remove electrolyte and/or substrate or sample aliquots of the reaction mixture.Chopped-light experiments with increasing bias potentials were performed on a P211 potentiostat (Zahner) by 5 seconds illumination and 5 seconds dark while using a LED light source (Zahner, TLS3, 100 mW cm -2 , Figure S16).Chronoamperometric photocurrent measurements at 700 mV vs. Ag/AgCl were obtained using the same setup with photoanode illumination for 16 hours with both compartments stirring.The WEC reaction mixture was sampled and measured with UV-Vis spectroscopy at the end of each long-term photoelectrochemical experiment to obtain the concentration of Br -/Br3 -produced.The Faradaic efficiencies for both BPTI-A and BPTI-B photoanodes were determined by comparing the electrons produced to the newly formed Br-Br bounds formation.The CEC was sampled to measure H2 evolution with GC (Shimadzu Nexis GC-2030).The integration of half the photocurrent (Figure S19) determines the number of electrons account for two electrons needed per oxidation reaction (i.e. 2 Br -+ 2e - Br2).Qualitative colorimetric analysis of FTO|SnO2|BPTI-A and BPTI-B photoanodes after long-term measurements (Table S2) was performed, as the disappearance of the yellow color indicates desorption or decomposition of the BPTIs due to the lingering of the dye in the oxidized state. [4]The retention of the yellow color guarantees that leaching of the BPTIs is not present in photovoltaic and photosynthetic measurements enabled by interactions mediated through aggregation phenomena and dye hydrophobicity. [5]

S8. Elaboration on H-, J-, and intermediate aggregate states.
The origin of the spectral changes emergent from the different aggregation modes found in the BPTIs can be explained using an excitonic coupling model as shown in Scheme 1C (Main Text).When two chromophores are excitonically coupled (i.e., dimer), two excitonic states arise, with one located at higher and one at lower energy compared to the transition energy of the monomer.
H-aggregates that adopt a face-to-face orientation are causing their molecular transition dipole moments to be connected perpendicular to the line to their centers (Figure S22).This electronic perturbation upon H-aggregation manifests in a hypsochromic (blue) shift of absorption maxima compared to monomer.Since the lowest splitting excited energy level is optically forbidden, this results in a quenched emission. [6]onversely, J-aggregates feature transition dipole moments parallel to the line connecting the centers of each molecule (headto-tail orientation, Figure S22), leading to a bathochromic (red) shift of absorption maxima to the monomer.J-aggregates yield higher luminescence than H-aggregates as the transition from the excited state to the ground state is allowed.

Equation S1
Aggregation modes between the "face-to-face" and "head-to-tail" also exist.These aggregation modes, denoted X-aggregates, [7] can be described by the angles (θ =slip angle, side view, and α= rotational angle, top view) [8] between the BPTIs planes of which aggregation type (J-, H-or X-) is formed, as shown in Figure S22. [9]The angle responsible for X-aggregation is also known as rotational displacement and enables allowed transitions to both exciton states, leading to the Davydov splitting (i.e., the appearance of both H-, and J-bands) in the absorbance spectrum.12][13][14][15][16][17] Equation S1 shows an estimation of the exciton coupling energy ( ) of the energy splitting of a dimer with two transition dipole moment vectors ( , and  , ).The splitting of the S1 state depends on the magnitude of the transition dipole moments and the joint arrangement of the dyes.The center-to-center distance is represented by r with r as the respective vector.

𝐽
, often in cm -1 , is positive for H-aggregates (blue shift, decrease in wavelengths) and negative for J-aggregates (red-shift, increase in wavelengths). [14]he X-aggregates have partially allowed transitions from both of the split excited states since the transition dipoles can never entirely cancel out each other.Therefore, both a red-and a blue-shifted band should be observed with different intensities depending on the aggregation type angle.Since our aggregated BPTI systems exhibit both red and blue-shifted features, we surmise that X-aggregates are the main species on the SnO2 surface. [9]gure S22.Illustration on the angles in H-, J-, and X-aggregates, with θ = the tilted angle between the packing direction (slip angle) and the dye plane, and α= the offset angle between the long molecular axes of the dyes (rotational angle). [8]he exciton band shapes dictate the peak height ratio of the 0-0 and 0-1 vibrational bands in the UV-Vis spectra, as shown in Figure S23.Perturbation of optical spectra arriving from the excitonic coupling between loosely stacked chromophores has been reported for various types of dyes. [18]The exciton band shapes influence the probability of the electron transition between the S0 and (the different vibrational levels) of the S1 state.If the most probable electron transition lies at the bottom of the band, the 0-0 transition will be bright, as seen in monomeric species.Conversely, H-aggregates demonstrate the bright state at the top of the band due to the displacement of the curvature giving rise to the 0-1 vibrational band. [14,17,19]The ratio between the 0-0 vibrational band 0-1 vibrational band can indicate if the BPTI is present in a more H-aggregated form. [20,21]gure S23.Simple energy diagram showing the absorption, emission, and solvent reorientation of A) monomers and B) Haggregates.The energy levels are shown in the vibronic bands for the electronic 0-0 and 0-1 transitions of S0 to S1. [21,22] The polar solvent molecules can reorientate around the excited molecule.If the excited molecule is more polar than the ground state molecule, the polar solvent molecules can stabilize the energy of the excited state more, and the energy difference decreases between the ground and excited state, resulting in a more significant Stokes shift. [23]μ= dipole moment of the ground state, μ* = dipole moment excited state.The wavelength (i.e., energy) of the 0-0 transition for BPTIs was obtained in both absorbance and fluorescence spectra at room temperature at a concentration of ~1 µM, and results are shown in Table S3 and Figure S24.A slight bathochromic (red) shift of 0-0 transition peak is visible in fluorescence spectra with an increasing dielectric constant.In general, such solvatochromism of the emission band indicates larger dipole moments of the excited state than that of the ground state.The red shift is accompanied by an increase in Stokes shift, indicating enhancement of stabilization of the excited state by solvents with increased polarity. [24]dditionally, the ratio between the maximum of the vibrational band of 0-0 and 0-1 was determined for solvents with increasing dielectric constant.The ratio between 0-0 and 0-1 absorbance maxima affords insight into H-aggregation probability (S9 and Figure S22).While the BPTIs can dissolve in low concentrations (~1-10 µM) in many different solvents, the solvation of the π-core remains poor due to the unfavorable dispersal between the π-system (high polarizability) and the solvent (low polarizability), [25][26][27] leading to higher aggregation in less polarizable solvents, i.e., lower dielectric constant. [26]Interestingly, the ratio between the maximum of the vibrational band of 0-0 and 0-1 and, therefore, the aggregation probability is very similar for both BPTI-A and BPTI-B in all solvent absorbance spectra.In contrast, in the fluorescence spectra, a smaller ratio of 0-0 and 0-1 is always visible for BPTI-A.However, a variation of this ratio related to the dielectric constant is not seen.Solventdependent aggregation can be quite complex as the hydrogen bonding ability or viscosity also plays a significant role, [21] as the difference in the ratio of 0-0 and 0-1 of BPTI-A in DMF and DCM is striking.Given that the difference peak ratios in fluorescence spectra is diagnostic of varying degrees of equilibria between monomer and aggregated BPTI, [25] we conclude that the alkyl substitution of BPTI-A affects modulating aggregation.
Table S3.Solvatochromism of the 0-0 vibronic band and ratio in 0-0 and 0-1 of the transition peak wavelength in absorbance and fluorescence spectra for BPTI-A and BPTI-B in solution (µM) for various solvents λex =420 nm.The maximal absorbance wavelength of the 0-0 transition (λabs), the maximal emission wavelength of the 0-0 transition (λem), and Stokes shift (Δλ) in nm.The ratio peak height of the 0-1 and 0-0 ( ) transition in absorbance (abs) and emission (em) is dimensionless.Dielectric constant solvents (ε) in F m -1 .DMF = N,N-dimethylformamide, t-BuOH= tert-Butyl alcohol, THF = tetrahydrofuran, DCM = dichloromethane, * = dipping solution used for MOx sensitization.Wavelength (nm S11. Elaboration on binding modes of the BPTI dyes on the SnO2 surface. Dye loading can be enhanced by aggregation in one of two proposed mechanisms.In the first mechanism, monomeric BPTIs can self-assemble in solution to yield aggregates that further chemisorb to the MOx surface. [28]In the second mechanism, a monolayer of BPTIs formed on the semiconductor surface by carboxylic-MOx sensitization, [29] after which BPTI multilayers can form through π-stacking of dissolved dyes onto chemisorbed dyes.Having established the significantly higher propensity of BPTI-A to aggregate, we expected dye loading of BPTI-A on SnO2 to follow suit will also increase.A discussion on binding geometry and studies on the amount of BPTI loading are further elaborated in S13 ‡ and S4.5.The similar VOC and EIS (Table S1, Section S15) between the devices with BPTI-A or BPTI-B photoanodes suggests a negligible change in Fermi-level of the conduction band upon dye adsorption.This implies that the monolayer coverage of BPTIs is the same in all photoanodes upon chemisorption, since the H + of the carboxylic acid linker will intercalate and lower the Fermi-level, which lowers the VOC and Cµ. [30]Therefore, we surmise a monolayer of BPTIs is formed first by carboxylic-MOx sensitization, after which a second layer of BPTI-A can aggregate on top to yield a bilayer.The dye loading is relatively low compared to standard TiO2 photoanodes, known for SnO2 due to the lower isoelectric point of SnO2 (pH 4-5), compared with TiO2 (pH 6-7), which inhibit the adsorption of dye molecules with acidic carboxyl groups. [31]This suggests that the carboxylic group at the bay area of the BPTIs does play a role in anchoring the dyes on the semiconductor.To further validate this theory, efforts were made to physisorbed BPTIs (i.e., dyes without an anchoring group) on the SnO2 surface, proving unsuccessful.The necessity of the anchoring group adds to the idea of the monolayer to bilayer enhanced dye loading of BPTI-A.S12.Spectroelectrochemistry and stability studies of the BPTIs.The stability of the BPTIs on the semiconductor was investigated via spectroelectrochemistry and soaking studies.A reversible color change from yellow to green of the BPTIs on the SnO2 electrodes was seen upon reducing the first reduced species and re-oxidation to neutral species (Figure S25).The colorimetric reversibility of the BPTIs was consistent upon utilizing slower (10 mV s -1 ) scan rates in both MeCN and aqueous electrolytes.The stability of the radical anion species of the BPTIs was studied by spectroelectrochemistry by reducing the compounds in 100 mM TBAPF6 in DMF at room temperature. [1,2]In all cases, new broad bands appear during the reduction process at 650, 725, and 875 nm, as well as the disappearance of the vibronic bands at 469, 439, and 414 nm was observed.Moreover, BPTIs showed no decomposition over time, contributing to the proposal of air and water-stable radical anion. [2] The incident photon-to-current efficiency (IPCE) measures the ratio of the photocurrent produced vs. incident photons from a calibrated light source as a function of wavelength.IPCE is defined as the number of generated electrons divided by the number of incident photons, which is proportional to wavelength-depended light-harvesting efficiency and, therefore, the absorbance of the used dye.IPCE measurements of BPTI-A and BPTI-B DSSCs with the I -/I3 -and Br -/Br3 -in MeCN as well as Br -/Br3 - in 0.1 HBr in H2O are shown in Figure S26 The general formula of IPCE is shown in Equation S2 and S2, where  is the injection efficiency of the electrons in the semiconductor of the metal oxide,  is the amount of charge (electrons) collected at the back of the photoactive electrode (and can be used in the electric circuit), and the light-harvesting efficiency (LHE) is defined as the fraction of light intensity absorbed by the dye at a certain wavelength in the device.
IA is the absorbed intensity, and I0 is the incident intensity. [32,33]e normalization of the IPCE spectrum gives greater knowledge on the absorbance of each wavelength when different  and  are obtained due to different dye loading.The red-shift of BPTI-A accounts for only 16% more area in the normalized spectrum than BPTI-B calculated by integrating the normalized spectrum.We surmise that collection efficiency ( , the collection efficiency of electrons back of the photoactive electrode) is not improved in this system by using different dyes, but only by injecting more electrons.The remaining improvement in IPCE in the DSSCs using the aggregation BPTI-A can only be a result of an increased  .An interesting idea for the enhanced  by aggregation is an increase of the fluorescence lifetime.Previous studies on polymeric perylene chromophores have shown that monomeric perylenes have short-lived (3.9 ns, 0-0 at 538 nm) fluorescence lifetimes, while the self-organized, polymeric perylene chromophores show red-shifted characteristics of aggregates and have longer-lived (17 ns, red-shifted at 630-650 nm) fluorescence lifetimes. [34]Furthermore, studies have shown that a higher  might be attributed to the different binding geometry of a dye on the surface.A study on zinc porphyrins compared the  of dyes that bind planer to the surface (flat-laying on the semiconductor) to dyes that bind vertically to the surface (perpendicular to the semiconductor, increasing distance).It was shown that rigid planar systems showed an increase in  into the semiconductor thanks to flat-laying rings and minimal distance to the surface. [35]If we translate that to our study, BPTI-A has more opportunity to lay flat on the surface than the BPTI-B with increased steric bulk.This could also be an explanation of the higher  of BPTI-A.Studies on the fluorescence lifetime and binding geometry of the BPTIs are clearly of interest to reason the enhanced electron injection.Concluded, the 5-fold increase in photocurrent production is by using BPTI-A, is surmised to be a result of a higher dye loading (2-fold), and the rest is thanks to increased LHE by broader wavelength and enhanced  .

S14. Photocurrent dynamics studies on DSSCs
The Influence of dye aggregation can be observed through photocurrent dynamic studies on DSSCs by variation of the light intensity during chopped-light experiments. [32]The obtained Jsc is then normalized to a light intensity of 100 mW cm -2 , and the results of these experiments are shown in Figure S27.This measurement gives an idea of how efficiently light is converted to photocurrent at variable light intensities, electrolyte diffusion, and severe dye regeneration issues. [32]Disadvantages stemming from dye aggregation are visible in DSSCs when the low-light photocurrents that are normalized to 100 mW cm -2 show a substantially higher photocurrent output. [32]Even though aggregation is more present in BPTI-A DSSCs, losses in photovoltaic performance caused by aggregation in the present in the same amount in both BPTI-A and BPTI-B DSSCs combined with the I -/I3 -couple.In the case of BPTI-A and BPTI-B, DSSCs using a Br -/Br3 -couple a relatively slow light response of the DSSCs (i.e., a few seconds before reaching the maximum Jsc) as seen in Figure S27 C-F.The slow light response in DSSCs using the Br -/Br3 - couple compared to the I -/I3 --couple in DSSCs is often seen in SnO2-based DSSCs since Br − is thermodynamically more demanding to oxidize compared to I − . [36]Therefore, no conclusion on aggregation in Br -/Br3 --systems can be made as electrolyte diffusion, and severe dye regeneration issues are more pronounced then aggregation phenomena. [37]In conclusion, although some disadvantage due to aggregation of the BPTIs is seen, the amount of aggregation (i.e., comparing DSSCs with BPTI-A to BPTI-B) does not seem to influence the Jsc output of normalization of the low-light photocurrents.A) S15. Electrochemical impedance spectroscopy (EIS) Our objective was to gain insights into electron-hole recombination at the semiconductor-dye interface by performing EIS under varying light intensities.Unfortunately because of the low performance of SnO2 in DSSC application, there are very limited studies concerning the application of EIS analysis to model the electron transport behavior in the SnO2-based DSSC. [38]e aimed to determine electron lifetime (τe) as a function of VOC, which could come from a difference in recombination resistance (Rrec) or a conduction band shift causing a change in chemical capacitance (Cµ). [39]All cells were measured at open circuit voltage with an AC perturbation amplitude of 10 mV, and the AC frequency was measured from 100 mHz to 1 KHz.
To fit the measured data, we used an equivalent circuit [38,40] as shown in Figure S28D.The typical impedance spectra (Nyquist and Bode) of the DSSCs are shown in Figure S29.The Bode and Nyquist plot usually presents three semicircles that describe the electron transport processes involved in the DSSC system: Frequencies lower than 1 Hz indicate electron diffusion through photoanode and ionic diffusion in the electrolyte solution (Warburg element, W), frequencies between 1 Hz and 1 kHz was the result of the recombination resistance at the SnO2/electrolyte interface, and the capacitance of the SnO2 (Rct2 = Rrec and CPE2 = Cµ) and frequencies after 1 kHz refer to charge transfer at the counter electrode/electrolyte interfaces (Helmholtz capacitance parallel combination, Rct1, and CPE1).The initial displacement of the arcs from their origin in the semicircle observed at a high frequency was equivalent to the contribution from Ohmic serial resistance (Rh). [40]An attempt on determining the charge collection efficiency is done by using the formula ηcoll=(1+ ) -1 described by Suraya Shaban et.al. [41] Figure   .Fluorescence spectra were recorded on a Edinburgh Instruments Spectrofluorometer FS5 using a SC-05 standard cuvette holder using a cuvette of 1 cm.The cuvette was filled with electrolyte to imitate the DSPEC environment using either 1.0 M LiBr in water (pH 9) or 0.1 HBr, 1.0 M LiBr in water (pH 0.5).The spectra are normalized using the maximum intensity of fluorescence observed.Upon decreasing pH the intensity of the emission increases in the cases of both BPTI-A and -B with a slight stronger effect in in BPTI-B.Studies that use perylenes as fluorescence probes show that fluorescence decrease upon deprotonation (higher pH) compared to a protonated compound (lower pH). [42]Furthermore, the addition of protons may influence aggregation in solution-based systems, resulting in variations in emission intensity. [43]However, in our system, the addition of protons may have an influence on the SnO2 conduction band level, making the system more complicated since it may result in less favorable electron injection.Because the photovoltaic differences between BPTI-A and -B are also observed in protic solvents (acetonitrile), we do not attribute the differences to a pH effect.A) B) 1 H-NMR (THF, 400 MHz, 40°C) of BPTI-B.Figure S5.High-Resolution FD-MS of BPTI-B.Figure S6.ATR-FT-IR of BPTI-A and BPTI-B Figure S7.Absorbance spectra and normalized fluorescence spectra of BPTI-A and BPTI-B in DMF.Figure S8.Linearity of the molar extinction coefficient and UV-Vis titrations of BPTI-A BPTI-B Figure S9.Differential pulse voltammograms of BPTI-A and BPTI-B in benzonitrile and DMF Figure S10.Cyclic voltammetry measurements and ip plotted against the ν and their linear trendlines of BPTI-A and of BPTI-B S4.Counter and working electrode preparation and characterization S4.1.Preparation of platinum counter electrodes.S4.2.SnO2 paste preparation.S4.3.Cleaning FTO electrodes.S4.4.Preparation of FTO|SnO2 photoanodes.S.4.5.Dye leaching experiments.

Figure S15 .
Sandwich DSSC set-up Figure S16.Reference spectra of used LED illumination source LED and 1 Sun Table

Figure S17 .
UV-Vis titrations and linearity in molar extinction coefficient of Br2 in 1.0 M LiBr, 0.1 M HBr in H2O S7.Illumination setup for dye-sensitized photoelectrochemical cells.

Figure S18 .
Image of photoreactor used for DSPEC experiments.Table S2.Images of photoanodes of FTO|SnO2|BPTI-A and BPTI-B before and after DSPEC Figure S19.Photocurrent and amount of electrons obtained from DSPECs with FTO|SnO2|BPTI-A or BPTI-B as photoanode Figure S20.UV-Vis measurement of the reaction mixture of DSPECs Figure S21.Light-driven qualitative hydrogen gas production in the DSPEC.Additional discussions S8.Elaboration on H-, J-, and intermediate aggregate states.

Figure S24 .
Normalized absorbance and fluorescence spectra of BPTI-A and BPTI-B in various solvents.S11.Elaboration on binding modes of the BPTI dyes on the SnO2 surface.S12.Spectroelectrochemistry and stability studies of the BPTIs.

Figure S29 .
Nyquist (left )and Bode plots (right) of the DSSCs prepared with BPTI-A or BPTI-B S16.Investigation of the pH dependence of the BPTI-A and -B on the emission intensity Figure S30.Emission spectra of FTO|SnO2|BPTI-A and BPTI-B at different pH S17. References S1. Materials and methods.

Figure S1 .Figure
Figure S1.Normalized UV-Vis of A) BPTI-A, and B) BPTI-B and their respective parent PDIs.

Figure S8 .Figure S9 .Figure 1 .
Figure S8.Linearity of the molar extinction coefficient of A) BPTI-A at 467 nm (red) and B) BPTI-B at 469 nm (blue) and UV-Vis titrations of C) BPTI-A (red) and D) BPTI-B (blue) in DMF measured in a quartz cuvette with a 1 cm path length.

Figure S11 .
Figure S11.Height profile (red is vertical scan with average height = 6.22 µm, blue is vertical scan with average height = 5.89 µm) of a 1 cm circle of FTO|SnO2 plate measured with a 3D laser scanning Confocal Microscope XK-X1000 Series (Keyence) with a 20× magnification.

Figure S13 .Figure
Figure S13.Cyclic voltammetry measurements of FTO|SnO2 with BPTI-A (red) and BPTI-B (blue) A) immobilized on the SnO2 surface in 100 mM TBAPF6 in MeCN and C) 100 mM aqueous NaOAc buffer with pH 5.4.B) The dissolved BPTI-A (<0.25 mM red) and BPTI-B (1 mM, blue) in 100 mM 100 mM TBAPF6 in DMF as shown as a reference.Conditions: Working electrode = FTO|SnO2|BPTI-A or |BPTI-B or glassy carbon, reference electrode = Ag/AgCl, counter electrode = Pt-wire, scan rate = 100 mV s -1 .Measurements are scaled to visualize the variation on activity induced by the applied potential.

Figure S17 .
Figure S17.A) UV-Vis titrations of Br2 in 1.0 M LiBr, 0.1 M HBr in H2O, and B) linearity in molar extinction coefficient of Br2 in 1.0 M LiBr, 0.1 M HBr in H2O at 266 nm measured in a quartz cuvette with a 1 cm path length.

Figure S21 .
Figure S21.Light-driven qualitative hydrogen gas production in the DSPEC.The WE electrode compartment is filled with 1.0 M LiBr, 0.1 M HBr in H2O (3 mL) and consists of an FTO|SnO2|BPTI-A or BPTI-B WE (masked size = 0.64 cm 2 ), and an Ag/AgCl (leakless, eDAQ, ET069) RE.The CE compartment is separated by a Nafion-117 membrane (FuelCellStore) and is filled with 1.0 M LiBr, 0.1 M HBr in H2O (3 mL) and consists of a Pt electrodeposited on FTO CE.Chronoamperometric photocurrent measurements at 700 mV vs. Ag/AgCl were obtained using P211 potentiostat (Zahner), with both compartments stirring.The photoanode was illuminated with a LED light source (Zahner, TLS3, 100 mW cm -2 , Figure S16) for a period of 1 hour.
on peak height ratio of the 0-0 and 0-1 vibrational bands.

Figure S24 .
Figure S24.Normalized absorbance (solid lines) and fluorescence (dotted lines) spectra of BPTI-A (red) and BPTI-B (blue) in various solvents measured in a quartz cuvette with a 1 cm path length.

Figure
Figure S25.A) Photos of the color change of the first reduction of an FTO|SnO2|BPTI as a working electrode through potential cycling (both BPTI-A and BPTI-B in both MeCN and H2O show similar color changes).Spectroelectrochemistry of B) BPTI-A and C) BPTI-B at the ground state (red) and the first reduced state (purple) with the spectra in between scans (from 0 to -1.0 V vs. NHE) in a rainbow gradient.Conditions: WE: Pt grid, RE = Ag-wire, CE= Pt-wire.BPTI-A was a saturated solution that explained the different ratios because of the better solubility of the reduced state.
Table of Contents S1. Materials and methods S2.Analytical methods.S3.Synthesis and characterization of BPTIs.