What Makes a Photobattery Light-Rechargeable?

The demand for autonomous off-grid devices has led to the development of “photobatteries”, which integrate light-energy harvesting and electrochemical energy storage in the same architecture. Despite several photobattery chemistries and designs being reported recently, there have been few insights into the physical conditions necessary for charge transfer between the photoelectrode and counter electrode. Here, we use a three-electrode photobattery with a dye-sensitized TiO2 photoelectrode, triiodide (I–/I3–) catholyte, and anodes with varying intercalation potentials to confirm that photocharging is only feasible when the conduction band quasi-Fermi level (EFc) is positioned above the anode intercalation/plating potential. We also show that parasitic reactions after the battery is fully charged can be accelerated if the voltage of the battery and solar cell are not matched. The integration of multiple anodes in the same photobattery ensures well-controlled measurement conditions, allowing us to demonstrate the physical conditions necessary for charge transfer in photobatteries, which has been a topic of controversy in the field.


Preparation of TiO2 Photoelectrode (PE)
FTO glass pieces with a thickness of 2.2 mm (Sigma Aldrich, surface resistivity ≈7 Ω/sq) and an area of 2.1cm x 1.7cm were successively sonicated in acetone (40 minutes), IPA (5 minutes) and deionized (DI) water (5 minutes).After drying, ultraviolet (UV) ozone treatment (BioForce Nanosciences) was conducted for 30 minutes to improve hydrophilicity of the surface.A compact TiO2 blocking layer was deposited by placing the FTO glass in a 40 mM solution of TiCl4 (Sigma Aldrich, 208566) in DI water for 35 minutes at 75 o C.
A mesoporous TiO2 layer was deposited by screen-printing TiO2 nanoparticles with an average size of 30 nm (30NR-D, Greatcell Solar Materials Pty.Ltd.) using a desktop screen printer (Wellcos, Korea).After drying at 120 o C for 5 minutes, a TiO2 scattering layer (particle size > 100nm) was deposited by screen printing (Ti-Nanoxide, R/SP, Greatcell).After drying at 120 o C, the electrodes were annealed in air at 500 o C for 1 hour using a tube furnace (Carbolite Gero) to form crystalline TiO2 layers.After cooling, the electrodes were immersed in 40mM TiCl4 in DI H2O at 75 o C for 35 minutes followed by a second annealing at 500 o C for 1 hour.Two 0.5 mm holes were drilled (Dremel workstation) through the photoelectrode for electrolyte injection.
Finally, the solutions were mixed in a volumetric ratio of 3:1:1.The electrodes were then washed in EtOH to remove excess dye and dried with an air gun.

Preparation of LiMn2O4 (LMO)@Gn Storage Electrode
LMO@Gn was prepared using a prior procedure [1].LMO@Gn electrode was composed of LMO@Gn as the active material, Super P (Timcal) carbon black as the conducting agent, and polyvinylidene fluoride (PVDF, Solef 5130) as the binder (weight ratio of 8:1:1), blade cast on an Al foil.Firstly, PVDF powder was completely dissolved in N-methyl pyrrolidinone (NMP).A ground mixture of LMO@Gn and Super P was finely dispersed into the PVDF/NMP solution with a mixer (ARE 310, THINKY Corp.).After casting the LMO@Gn slurry onto the Al foil using a doctor blade, it was vacuum dried at 110 °C for 10 hours, followed by natural cooling to room temperature.The loading density of LMO@Gn ranged from 3 to 3.8 mg cm -2 .

Preparation of Li4Ti5O12 (LTO) Storage Electrode
The LTO electrode was prepared as follows.Initially, carbon nanotubes (CNT) were dispersed in deionized (DI) water using carboxymethyl cellulose (CMC) as a binder which was subsequently sonicated to achieve a stabilized gel-like slurry.Next, spherical LTO powder (BTR, China) was mixed with the CNT/CMC mixture using an ARM-310 Thinky mixer for 20 minutes.This LTO/CNT/CMC slurry was then coated onto a 17 μm thick aluminium foil and dried at 60 °C on a hot plate for 1 hour, followed by overnight drying in a vacuum oven at 60 °C.The resulting dried electrode had an average thickness of 40 μm and contained 2.9 mg cm -2 of active material.

Photobattery Assembly
Pt was deposited on LICGC TM through sputtering (Quorum Technologies Q150T) in a patterned fashion using a laser cut acrylic mask, as shown in Figure S2, to serve as the counter electrode of the solar cell.The Pt sputtered LICGC TM separator (discharge electrode, DE) was sealed against the photoelectrode using two precut 120 µm thermoplastic films (Surlyn TM , Greatcell) by pressing on a hot plate at 120 o C. The storage electrode (SE) was prepared by cutting strips of the prepare anode films (active area of 0.5cm x 0.3cm) and attaching them on a predrilled piece of glass using a thermoplastic film.The SE was then sealed against the other side of the DE using two precut 120 µm thermoplastic films.An electrolyte composed of 0.1 M LiI, 0.05M I2, 0.05 M guanidine thiocyanate, 0.5 M of tert-butyl pyridine, and 0.6 M 1,2-dimethyl-3propylimidazolium iodide (DMPII) in acetonitrile was injected between the PE and DE, and the holes were then sealed using a thermoplastic film and a cover glass.An electrolyte comprising of 0.1 M LiTFSI in ethylene carbonate (EC) and dimethyl carbonate (DMC) in a 1:1 ratio by volume was injected between the DE and SE with the holes sealed by the same procedure.Electronic contacts were soldered to the PE, SE and DE using an ultrasonic soldering machine (Sunbonder) and the entire device was sealed with potting epoxy (RS Components) and left to dry for 24 h.

Solar Cell Characterization
JV curves were measured by conducting cyclic voltammetry between the PE and DE with a scan rate of 50 mVs -1 for 1 sun measurements and 5 mV s -1 for indoor light measurements using a Biologic VMP-3 potentiostat.For 1 sun measurements a solar simulator (Newport Oriel LSH-7320) was used as a light source with a 0.04 cm 2 black mask.For measurements with an LED, a white LED (Thorlabs) was used along with a 0.96 cm 2 black mask and a black box was used to block out background light.

Electrochemical Characterization
Photocharging measurements were carried out by conducting chronoamperometry (CA) at 0 V between the PE and SE using a Biologic VMP-3 potentiostat.All galvanostatic measurements were carried out by applying a current density of 50 μA cm -2 between the DE and LMO SE and a current density of 75 μA cm -2 between the DE and LTO SE.
The redox potential of I -/I 3-and LMO were recorded by the average of redox peaks in cyclic voltammogram (marked by dotted line in Figure S2 and Figure S5).For I -/I 3-, cyclic voltammetry was carried out at 100 mV s −1 scan rate in a 3-electrode system with Pt wires functioning as both working and counter electrodes, and Ag/Ag + as the reference electrode.The electrolyte was composed of 0.02M LiI, 0.01M I2, and 0.1M LiTFSI in CH3CN.For LMO, LMO cast on Al foil was used as the working electrode, and the electrolyte was 0.7M LiTFSI in acetonitrile.Ferrocene/ferrocenium pair (+0.63 V vs SHE) was used as an internal reference.
The highest occupied molecular orbital (HOMO) value of the WS72 dye was estimated from the half potential of cyclic voltammogram (Figure S4).All potential values were represented versus standard hydrogen electrode (SHE) by using an internal reference (Fc + /Fc, +0.63 V vs SHE).For the cyclic voltammetry measurement, a dye-sensitized TiO2 film on FTO glass, Pt wire, and Ag/Ag + were used as working, counter, reference electrodes, respectively.The electrolyte was prepared by dissolving 0.1M of LiTFSI in acetonitrile.To fabricate the working electrode, TiO2 solution (30NR-D:EtOH:Triton X-100 / 3:6:1) was spin-coated on FTO glass at 1500 rpm for 40 s, followed by annealing at 500 °C for 30 min in air.After cooling down to 80 °C, the electrode was immersed in a 0.0001 M WS72 dye solution (toluene:EtOH, 1:4 vol%) with 0.001 M chenodeoxycholic acid (CDCA) as the co-adsorbate at room temperature for 17h.

Optical Characterization
The dye absorption was measured by using dye-sensitized TiO2 films and the UV-vis spectra were recorded using a Perkin Elmer Lambda-750 UV-Vis-NIR Spectrometer.

Supplementary Table
Table S1.A summary of the conduction band potential of the photocathode and anode plating potentials of some two-electrode photobatteries reported in literature.

Ref
Comments         As the battery charges, the lithium-ion content in the solar cell electrolyte decreases, causing an increase in OCVss but a decrease in current [23].This sketch illustrates that the changing I-V curve of the solar cell as a function of SoC is also a contributing factor to the decrease in photocurrent and must be considered for I-V matching in photobatteries.It should be noted that this figure is a sketch for illustrative purposes and does not contain real data.

Supplementary Figures
Figure S1.The structure of the three dyes used for the co-sensitization of TiO2 photoelectrode (a) WS72, (b) MS5 and (c) XY1b.

Figure S2 .
Figure S2.CV curves of the triiodide redox couple and ferrocene redox couple in acetonitrile to obtain its redox potential.The redox potential of ferrocene (Fc + /Fc, +0.63 V vs SHE) was used to obtain the redox potential of the triiodide redox mediator vs SHE.

Figure S3 .
Figure S3.An image of the patterned Pt sputtered on LISICON TM .

Figure S4 .
Figure S4.CV curves of the WS72 dye in a solution of LiTFSI in acetonitrile.The half potential (at +1.13 V vs SHE) represents the HOMO level of the dye.All potential values are represented versus standard hydrogen electrode (SHE) by using internal reference (Fc + /Fc, +0.63 V vs SHE).

Figure S5 .
Figure S5.CV curves of the LMO anode and I3 -/I -redox couple in a solution of LiI in acetonitrile.The distance between the half potentials is about + 0.23 V, which is in good agreement with the discharge potential we obtained, indicating that the photocharging reaction is Li + intercalation into LMO.

Figure S7 .
Figure S7.(a) Long term photocharging of the I3 -/I -vs LMO battery under 1 sun conditions, with a sharp decrease in current seen during the photocharging process.(b) Discharge curves after the long-term photocharging, with the same discharge capacity as that under 200 lux and 1000 lux long-term photocharging obtained.

Figure S8 .
Figure S8.(a) Sketch of a battery voltage vs time curve with a sloping voltage profile.(b) I-V curves of the solar cell at two different SoCs, with the battery I-V curves superimposed.As the battery charges, the lithium-ion content in the solar cell electrolyte decreases, causing an increase in OCVss but a decrease in current[23].This sketch illustrates that the changing I-V curve of the solar cell as a function of SoC is also a contributing factor to the decrease in photocurrent and must be considered for I-V matching in photobatteries.It should be noted that this figure is a sketch for illustrative purposes and does not contain real data.