UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Integrated solar energy harvesting and storage devices Mahmoudzadeh Ahmadi Nejad, Mohammad Ali 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2015_may_mahmoudzadeh_ahmadi_nejad_mohammad_ali.pdf [ 9.44MB ]
JSON: 24-1.0166258.json
JSON-LD: 24-1.0166258-ld.json
RDF/XML (Pretty): 24-1.0166258-rdf.xml
RDF/JSON: 24-1.0166258-rdf.json
Turtle: 24-1.0166258-turtle.txt
N-Triples: 24-1.0166258-rdf-ntriples.txt
Original Record: 24-1.0166258-source.json
Full Text

Full Text

Integrated Electrochemical Solar Energy Harvesting andStorage DevicesbyMohammad Ali Mahmoudzadeh Ahmadi NejadM.A.Sc., The University of British Columbia, 2009B.Sc., Sharif University of Technology, 2006A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Electrical and Computer Engineering)The University of British Columbia(Vancouver)April 2015c©Mohammad Ali Mahmoudzadeh Ahmadi Nejad, 2015AbstractLarge scale storage of electricity is a vital requirement for the realization of a carbon-neutralelectricity grid. This thesis provides a study of integrated solar energy conversion and storagesystems in order to increase the efficiency and reduce the utilization cost of solar energy. Theefficient performance of photogalvanic cells relies on high dye solubility and selective elec-trodes with fast electron transfer kinetics. A new configuration is proposed for photogalvaniccells that removes these impractical requirements. Instead of illuminating the device throughthe electrode a new vertical configuration is employed with light coming between the two elec-trodes. This way, the light absorption and hence electron generation is spread through the depthof the device which can be adjusted according to the concentration of the dyes to absorb allthe incoming photons even with low solubility dyes and slow electrode kinetics. The proposedconfiguration is mathematically studied and a numerical model is built for detailed analysis thatgives practical guidelines for working towards device parameters with high power conversionefficiency. The analysis suggests that upon the realization of highly selective electrodes and animproved dye/mediator couple, an efficiency higher than 13% should be achievable from thenew configuration compared to 3.7% at best using the conventional approach. Storage how-ever in this system will be challenging due to the characteristic recombination times of dyesand mediators in the same phase.For significant and long-lived storage we designed and demonstrated an integrated solar-battery structure based on two relatively well established technologies of the redox flow batteryand the dye-sensitized solar cell. The cell consists of a sensitized electrode in a redox flow bat-iitery structure. The design enables independent scaling of power and energy rating of the systemthus it is applicable for large scale storage purposes. An areal energy capacity of 52 µWhcm−2,charge capacity of 1.2 mAh L−1, energy efficiency of 78% and almost perfect Coulombic ef-ficiency are observed for the integrated cell. These values show a 35 times increase in chargecapacity and 13 times improvement in areal energy density compared to similar devices.iiiPrefaceThe author is the main researcher of all of the work presented in this dissertation. All the workis conducted in UBC Molecular Mechatronics Lab under the supervision of Prof. John D.W.Madden.Chapter 2. A version of this material has been published in journal of Electrochimica Acta( Mahmoudzadeh M.A., Madden J.D., ”A Vertical Architecture for Increasing PhotogalvanicSolar Cell Efficiency: Theory and Modeling”, Electrochimica Acta, Volume 143, 2014, Pages98-105). All the theoretical analysis and numerical modeling are performed by the author withsupervision of Prof. John D.W. Madden.Chapter 3. The author was the lead investigator on the concept formation with contribu-tions from Prof. John D.W. Madden and Dr. Ashwin R. Usagaocar. All the design, fabricationand characterization are the original, independent work of the author. The author also receivedvaluable feedback on this work from Professor Gordon Wallace at the University of Wollon-gong, who met approximately monthly with him by phone.The concept and supporting theory of the work in Appendix B is provided by the author.The experimental work and data analysis are performed by Dr. Ashwin R. Usagaocar and Dr.Lisheng Wang. Prof Dan Bizotto provided valuable feedback on the electrochemical analysis.The results are presented in the Electrochemical Society (ECS) Meeting. (Usgaocar A.R.,Wang L., Mahmoudzadeh M.A., Mirvakili S.M., Slota-Newson J.E., Madden J.D., Beatty J.T.,Takshi A. ”Semiconductors as Selective Electrodes for Bio-Photovoltaic Cells”, Meet Abstr.MA2013-01(4), 282., 2013)ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Energy Storage and Renewables . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Photogalvanic Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3 Redox Flow Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.4 Dye Sensitized Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.5 Research Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171.6 Thesis Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17v2 A Vertical Architecture for Increasing Photogalvanic Solar Cell Efficiency: The-ory and Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2 Analysis of the Conventional Photogalvanic Cell . . . . . . . . . . . . . . . . . 212.3 Analytical Analysis of the Vertical Photogalvanic Cell . . . . . . . . . . . . . . 242.3.1 Geometry Optimization . . . . . . . . . . . . . . . . . . . . . . . . . 282.3.2 Load Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.3.3 Estimation of Maximum Efficiency . . . . . . . . . . . . . . . . . . . 302.4 Simulation of the Photogalvanic Cell . . . . . . . . . . . . . . . . . . . . . . . 312.4.1 Equation Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.4.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383 Solar Redox Battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.5 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624 Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664.2 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684.3 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73A Publications Not Included in the Thesis . . . . . . . . . . . . . . . . . . . . . . . 81viB Semiconductors as Selective Electrodes . . . . . . . . . . . . . . . . . . . . . . . 83B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83B.2 The Gerischer Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84B.3 Experimental Measurement of the Selectivity . . . . . . . . . . . . . . . . . . 87C Solar Redox Battery Cost Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 90viiList of TablesTable 1.1 A list of commercial RFB energy storage facilities. . . . . . . . . . . . . . . 13Table 2.1 Characteristic lengths of photogalvanic devices. . . . . . . . . . . . . . . . 23Table 2.2 Albery’s recipe for the optimal cell . . . . . . . . . . . . . . . . . . . . . . 24Table 2.3 Device parameters and performance of electrode illuminated photogalvanicdevices. Efficiencies, η , are as computed using COMSOL. . . . . . . . . . . 35Table 2.4 Device parameters and performance of vertical photogalvanic devices. Effi-ciencies, η , are as computed using COMSOL. . . . . . . . . . . . . . . . . 38Table 3.1 Polarization of Ni foam electrodes in polysulfide electrolyte. . . . . . . . . . 52Table 3.2 List of materials for an efficient, high energy density solar redox battery. . . 53Table 3.3 Performance comparison of DSSC-based solar batteries. . . . . . . . . . . . 60Table B.1 Standard heterogeneous rate constants of redox couples on FTO. . . . . . . 89Table C.1 Comparison of energy conversion and storage costs for combined and sepa-rated systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91viiiList of FiguresFigure 1.1 Human Development Index (HDI) vs. per capita electricity consumption. . 2Figure 1.2 Charactristic discharge time of energy storage technologies. . . . . . . . . 4Figure 1.3 Schematic of a biological solar cell with dissolved reaction centers. . . . . 7Figure 1.4 (a)Open circuit voltage and (b)short circuit current of the reaction center-based solar cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Figure 1.5 Schematic of a traditional photogalvanic cell. . . . . . . . . . . . . . . . . 8Figure 1.6 Schematic of the vertical photogalvanic cell. . . . . . . . . . . . . . . . . . 10Figure 1.7 Schematic of a RFB system. . . . . . . . . . . . . . . . . . . . . . . . . . 12Figure 1.8 Energy diagram and the working principle of a dye sensitized solar cell. . . 15Figure 1.9 Schematic of the solar redox battery. . . . . . . . . . . . . . . . . . . . . . 16Figure 2.1 Schematic of a vertical photogalvanic cell. . . . . . . . . . . . . . . . . . . 22Figure 2.2 Load independent efficiency vs l√kr[M+]D in a vertical PGC. . . . . . . . . . 29Figure 2.3 The variation of ψ with respect to u0 in a vertical PGC. . . . . . . . . . . . 31Figure 2.4 Simulation results of the target vertical PGC. . . . . . . . . . . . . . . . . 37Figure 2.5 Effect of device length and electrode selectivity on cell efficiency for bothtop and side illuminated cells. . . . . . . . . . . . . . . . . . . . . . . . . 40Figure 3.1 Schematic of different approaches towards the integrated solar battery. . . . 44Figure 3.2 Schematic of a solar chargeable redox battery. . . . . . . . . . . . . . . . . 47Figure 3.3 Energy diagram of Redox couples for the solar redox battery. . . . . . . . . 49ixFigure 3.4 Tafel plots of several electrodes towards the polysulfide redox couples. . . . 50Figure 3.5 Tafel plots of several electrodes towards the polysulfide redox couples. . . . 51Figure 3.6 Charge/discharge characteristics of the iodide-polysulfide redox battery. . . 55Figure 3.7 (a)Complete discharge and (b) energy efficiency of the iodide-polysulfideredox battery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56Figure 3.8 IV characteristics and short circuit current of DSSC using the solar-batteryelectrolyte. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Figure 3.9 Solar redox battery sandwich structure . . . . . . . . . . . . . . . . . . . . 59Figure 3.10 (a)Photocharge/ Discharge of the solar redox battery. (b)Energy efficiencyduring charge/discharge cycles. . . . . . . . . . . . . . . . . . . . . . . . . 61Figure 3.11 Color change of sodium sulfide as it changes from colorless Na2S (a) todark red Na2S4 (e). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Figure 3.12 SEM images of the nickel foam before (a), and after treatment with acidand polysulfide (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Figure 3.13 Schematic and the image of the H-type cell used for the battery test. . . . . 65Figure 3.14 Images of the (a) light source and (b) the potentiostat setup. . . . . . . . . . 65Figure 4.1 Utilization if thermal convection for electrolyte circulation in solar redoxbatteries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71Figure B.1 Distribution of energy states in semiconductor (left) and metal (right)-electrolyteinterface. The overlap between filled and empty states on two sides deter-mines the reaction rate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84Figure B.2 Cyclic voltammetry of (a) methyl viologen (b) ferri/ferrocyanide and (c)ferric/ferrous on FTO electrode. (d) Sampled current voltammetry datameasured at 12 s after the voltage application. . . . . . . . . . . . . . . . . 88xList of AbbreviationsA-CAES Adiabatic Compressed Air Energy StorageACN AcetonitrileCAES Compressed Air Energy StorageDI DeionizedDSSC Dye Sensitized Solar CellEES Electrochemical Energy StorageFES Flywheel Energy StorageHDI Human Development IndexFTO Fluorine Doped Tin OxideHOMO Highest Occupied Molecular OrbitalLUMO Lowest Unoccupied Molecular OrbitalPB-A Lead AcidPGC Photogalvanic CellPHES Pumped Hydro Energy StoragePS PolysulfidexiPTG Power to GasRFB Redox Flow BatterySRB Solar Redox BatteryTES Thermal Energy StorageTHF TetrahydrofuranVPGC Vertical Photogalvanic CellxiiAcknowledgmentsFirst and foremost I offer my sincerest gratitude to my supervisor, Prof. John Madden, who hassupported me throughout my research with his excellent knowledge and unmatched patienceand has given me with an exceptional atmosphere for doing research. I was extremely fortunateto have him as my mentor and advisor and cannot imagine any better person for that role.I would like to express my gratitude to the members of my thesis committee who guidedme with their thoughtful feedback.I am also grateful to Prof. Curtis Berlinguette and Dr. Phil Schauer, for being generouswith their time and knowledge. I thank Prof. Jeff Young for kindly giving me access to hislaboratory.I gratefully acknowledge the financial support of this work through Discovery and StrategicGrants from the Natural Sciences and Engineering Research Council (NSERC) of Canada andThe Peter Wall Institute for Advanced Studies at UBC.I want to thank all my current and previous co-researchers during this work: Dr. AshwinUsgaocar, Dr. Arash Takshi, Dr. Rafael Saer, Dr. Jo Slota-Newson, Daniel Jun, Seyed Mo-hammad Mirvakili and Bahar Iranpour for helping me advance this work and also thank all mycolleagues at the Molecular Mechatronics lab for creating such a productive work place.Most importantly, I want to express my gratitude to my wife, my parents and my threesisters, whose love, guidance and sacrifice are always beyond what could be asked.xiiiTo my grandparents, whose love follows me to Vancouver and drives me forward in life.Chapter 1IntroductionA goal of this thesis is to investigate and propose means of harvesting solar energy, and tofind a solution to one of the challenges hindering the widespread use of clean energy - thestorage problem. We first set the stage by estimating the energy demand in the near futureand discuss the importance of energy storage. Then the candidate storage technologies will beexamined and the advantages of an electrochemical energy storage solution will be discussed.By the end of this chapter, the proposed electrochemical harvesting and storage methods andthe component technologies will be presented. The motivation for combining a redox flow cellwith a dye-sensitized solar cell is discussed, as is the justification of its feasibility for solvingthe solar energy storage challenge.1.1 Energy Storage and RenewablesAnnual global electrical energy demand surpassed 20×103 TWh in 2013 and is increasing at arate of 3 % per year. The importance of electricity in today’s life is clear. Figure 1.1 shows thehuman development index (HDI) for many countries with respect to their average electricityconsumption. In general a higher HDI is observed as a result of higher electricity use. Thisgraph can be used to estimate the global electrical energy demand in the following decades.Neglecting the off-trend values for countries such as Iceland and Norway and the hope for a10%10%20%30%40%50%60%70%80%90%100%-5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 VERY HIGHHIGHMEDIUMLOWChinaUnited StatesJapan, France, Netherlands, Italy, UK, Germany, Republic of KoreaNorwayIcelandIndiaZimbabweNigeriaSouth AfricaRussian Federation, Saudi ArabiaKuwaitHDI:CanadaAnnual Per Capita Electricity Consumption (kWh)Human Development Index4,500 kWh per person per year, threshold of developed countries.Figure 1.1: Human Development Index (HDI) vs. per capita electricity consumption.Data from United Nations Human Development Reports [2] and International En-ergy Agency [3]. 4500 kWh/year/person is the onset of acceptable standard ofliving.more efficient use of electrical energy in future, one can expect that a minimum consumptionof 4500 kWh/year/person is required for an acceptable standard of living for the 9 billionpopulation in 2050. Therefore, we will need a global electricity supply of 40×103 TWh/yearby mid century, twice the amount of today’s global generation. More detailed analysis by theInternational Energy Agency estimates the mid century electricity demand to be in the rangefrom 33 to 42× 103 TWh/year depending on energy policy choices made between now andthen [1].The realization of this amount of electrical energy with current resources will lead to anincrease of atmospheric greenhouse gas concentrations to three times the current day value [4].A perturbation on this scale has never been experienced on the planet and the outcome couldseverely endanger our ecosystem. Carbon-neutral energy sources are the logical method ofchoice. Nuclear, biomass, hydroelectric and wind power each has their own limitations thatmake them unattractive as the ultimate solution for energy generation. Nuclear power plantsneed very complicated site fabrication and given demand the installation rate needed to catch up2with the global demand is likely impossible. Biomass is not effective because of its low powerconversion efficiency and its competition with the food supply. Hydroelectricity, although cleanand inexpensive, is already utilized close to its maximum capacity and can not significantlycontribute further to global electricity generation. Wind energy capacity can supply part of theupcoming demand but is not an adequate solution for the problem [5].Solar energy is the only carbon-neutral source that can supply all the energy demand [5].Variability during the daytime and intermittency as a result of atmospheric conditions are themajor drawbacks of this energy source. An effective energy storage system is required for theglobal utilization of solar energy.Our current electric grid does not have much storage capacity. To properly function, thegrid has to always balance the energy supply and demand by tuning the power plant output.Otherwise it produces voltage fluctuations which are not acceptable, particularly for manyindustry users. Germany, with a storage capacity greater than 8% of its electrical generationcapacity, has already run into grid instability problems as the share of renewables exceeded20% [6]. A report from U.S. National Renewable Energy Laboratory (NREL) has calculatedthe maximum manageable share of renewables in a grid without storage to be 20% [7]. Theyconclude that at least 24 h of energy storage is required to balance an electricity grid with 80 %generation by variable sources.Storage solutions should have suitable capacity and response time for the specific applica-tion. Power quality applications require only sub-second storage capacity while energy man-agement applications such as load leveling and peak shaving need discharge times of 1 to 20hours. Centralized grid connected storage typically has a 100 MW power rating while a dis-tributed system can consist of several 100 kW power plants. Figure 1.2 shows the dischargetime and capacity of several energy storage technologies that have been tested for addressingthis issue. From the technologies listed in that figure, pumped hydro is the only mature tech-nology and accounts for 99% of total available storage capacity. It stores energy in the formof potential energy by using the electrical energy surplus to pump water into a reservoir of3101001k2001( 1000RFB:ARedoxAFlowABatteryPtGA:APowerAtoAGasLi-ionA:ALithiumAIonABatteryTES:AThermalAEnergyAStorageFES:FlyweelAEnergyAStoragePb-A:ALead-AcidABatteryNa-S:ASodiumASulfurABatteryCAES:ACompressedAAirAEnergyAStorageZn-air:AZinc-airAbatteryPHES:APumpedAHydroAEnergyAStorageAAA10k1(min1(hr1(day100k100 300 500400 700600 800 900TESNa-SPb-ARFBFESPHESLife(Cycle(Cost(of(Energy((W/MWh)Discharge(time(seconds)Li-ionZn-airCAESFigure 1.2: Charactristic discharge time of energy storage technologies. The horizontalaxis represents the typical life cycle cost (Levelized Cost of Energy) estimates ofeach technology. The values are adopted from [8]. The upward arrows indicatethat PHES and PtG have the potential for longer (even seasonal) energy storage.higher elevation. However, its wide use relies heavily on geographical availability of a waterreservoir in an elevated location and therefore it can not be a global solution. Furthermore, it isnot suitable for distribution storage solutions, which is what is needed in today’s Germany forexample, since hydroelectric’s overall efficiency drops to ∼ 50 % for micro (≤100 kW) appli-cations. Additionally, their slow response time makes them impractical for solar energy storagewhere seconds to minutes response time is required for unpredictable sunlight variations duringthe daytime.The second largely utilized method is compressed air energy storage (CAES) which has aglobal capacity of 440 MW [9]. Off-peak low cost electricity is used to compress air, which islater released to recover the energy. Temperature changes during compression and expansionare the main loss mechanisms, limiting widespread use. Research towards an adiabatic CAES4system is currently trying to bring cyclic efficiency of CAES up to 50 % but those systems arenot economically attractive yet [10].Another approach for energy storage is flywheel energy storage (FES), which is very at-tractive because of its scalability, fast response time and high round-trip efficiency. However,its efficiency drops significantly at low discharge rates. Also, FES systems have high self-discharge rates, which make them suitable only for short term storage.The limited energy storage capacity of supercapacitors together with their high power andcycle life make them ideal only as short term power quality applications - before the mainenergy supply kicks in [9].Power to gas (PtG) energy storage systems use electrolysis of water to produce hydrogenwhich is then stored in the form of H2 gas or is converted to synthetic methane after reactionwith CO2. Having storage capacity in the scale of terawatt-hours makes this method evensuitable for seasonal storage applications. The fact that the current natural gas network canbe employed for storage and transfer of synthetic gas together with its high energy densityand insignificant long-term storage loss make this a very attractive energy storage solution.In fact, a 6 MW (industrial scale) power-to-methane pilot plant is already demonstrated inGermany [11]. Despite theses technical advantages, this technology still suffers from relativelylow efficiency and very high cost. Therefore, significant reduction in the cost and improvementin performance are required before it can compete with conventional energy storage techniques[12].Electrochemical energy storage (EES) in several forms such as lithium-ion, sodium-sulfurand lead-acid batteries are studied for renewable energy storage. The typical main drawbackof all types is high capital cost even for the cheapest technology (lead acid battery) [13].Integration of energy conversion and storage is one method to reduce the cost of solar en-ergy systems. A method for achieving such integration is described in this thesis. Such devicescould significantly reduce the fabrication and installation cost of utilization of the renewables.Further improvement is expected from reduction in cost of balance of system and energy con-5version losses. In the next sections, we present the electrochemical energy storage systemsthat promise the possibility of such integration. The main body of this thesis describes themodification and enhancement of the devices that are now discussed.1.2 Photogalvanic Cells 1Combined solar cell- battery behavior is observed in prior work on biological solar cells inwhich the author took part. In that work, reaction center (RC) proteins from photosyntheticbacteria were used as light absorbers in a variety of solar cell architectures [14–16]. In oneconfiguration, RCs are dissolved in a solvent together with two redox couples. After the lightabsorption, a positive and a negative charge pair are separated in the RC, as indicated in thefigure by the red and blue dots. The role of the redox couples, called mediators m1 and m2,is to react with the RC and extract the charges before their recombination. The charges arethen transfered to the electrodes and external circuitry. A schematic of this device is depictedin Figure 1.3. Despite the perfect quantum efficiency of RCs and their long self-dischargetime [17], less than 0.1% photo-conversion efficiency is observed from this device. However,it showed an unexpected charge storage behavior. Figure 1.4 shows the response of a dissolvedRC-based solar cell after cell illumination ends. The open circuit voltage of the cell drops on atime scale of hours which reflects a slow self discharge and the photocurrent of the cell vanishesafter tens of minutes. This apparent storage over minutes to hours is the motivation behindstudying similar configurations in this work as the starting point for solar-battery integration.The dissolved RC solar cell belongs to a family of solar cells called photogalvanic cells (PGCs).The first architecture investigated is a modified photogalvanic cell, described in Chapter 2.Photogalvanic cells consist of an electrolyte containing a dissolved light absorbing dyeand a redox system called a mediator in contact with two charge extraction electrodes. Inaddition to solar energy conversion, PGCs naturally inherit energy storage capability from their1This section is part of a publication in a peer-reviewed journal (Reused with permission from ”Mohammad AliMahmoudzadeh, John D.W. Madden, ’A Vertical Architecture for Increasing Photogalvanic Solar Cell Efficiency:Theory and Modeling’, Electrochimica Acta, Volume 143, 2014, Pages 98-105”, Copyright 2014, Elsevier Ltd.).6Figure 1.3: Schematic of a biological solar cell with dissolved RCs. m1 and m2 are tworedox mediators. After the light absorption and the photo-excitation of the RC,positive and negative charges are separated by the mediators and then transfered tothe electrodes.0 1000 2000 3000 4000 50000. LightgON LightgOFF(a)0200400600800100005101520253035Time (s)Photocurrent (nA)Light  ONLight  OFF(b)Figure 1.4: (a) Open circuit voltage and (b) short circuit current of the reaction center-based solar cell. The RC of the photosynthetic bacterium Rhodobacter sphaeroidesis used at a 15 µM concentration in water. The mediators are 0.75 mM ferroceneand 0.75 mM methyl viologen in Tris-HCl buffer (pH 8). The light intensity is2.8 mWcm−2.battery counterpart, galvanic cells.The power generation mechanism of photogalvanic cells isshown in Figure 1.5. It starts with the photo-excitation of the dye followed by dye-mediatorreaction. The excited dye can be reduced or oxidized by the mediator depending on the dye7and mediator combination selected. The redox couple and the stabilized dye then can reacton the electrodes to generate current. In this work, the classic process of a dye-electron donorwill be explored. Chapter 2 explores the limits on power conversion efficiency, and a new cellgeometry introduced to overcome shortcomings of the traditional cell.e- e-krMM+S S*S- e-hν(2)(3) (4) (3)(1)Figure 1.5: Schematic of a traditional photogalvanic cell. The work flow is (1) light ab-sorption and dye excitation.(2) Quenching of the excited dye and production ofreduced dye and oxidized mediator.(3) Each redox couple interacts with one elec-trode and produces a current. (4) The bulk recombination tends to push the cellback to the equilibrium with rate constant kr.In the photogalvanic cell, excitation occurs almost instantly after the absorption of a photon.The relaxation process happens at a rate of kRelax and takes 10−12 ∼ 10−9s (reaction (1.1), inparentheses),S+hν → S∗( kRelax−−−→ S). (1.1)If the excited dye lives long enough to diffuse in the electrolyte and interact with a chargemediator, the excited dye will be quenched which results in two charged species;S∗+MkQ−→ S−+M+, (1.2)8where S, S∗ and S− are relaxed, excited and reduced states of the sensitizing dyes and M/M+ isthe mediating redox couple. Since the products are at high energy levels, they will recombinein the bulk with the rate constant of kr which is the main loss mechanism of the cell,S−+M+kr−→ S+M. (1.3)In order to extract the absorbed energy, the products of reaction (1.2) need to diffuse tothe electrodes before their recombination though reaction (1.3). It is also desired that eachredox couple only interact with one of the electrodes in order to avoid electrode mediatedrecombination of the species i.e. electrodes should behave selectively towards the couples. Inthis case,(Anode) S−→ S+ e− and(Cathode) M++ e−→M.(1.4)In the case of fast kinetics of these reactions, the electrode potentials will follow the po-tential of the redox couples, therefore, an open circuit potential difference of ∆E ≈ |ES−/S−EM+/M| is expected from this cell. The analytical analysis of photogalvanic cells proposed byAlbery and Archor [18] showed the possibility of high performance PGCs given certain con-ditions of device geometry and chemistry. To date the highest efficiency of a PGC is claimedby Bhimwal and Gangotri to be 1.62% with methyl orange as photosensitizer dye [19], whichexpresses how far these devices are from practical use. On the storage side, Genwa’s groupreported energy storage for up to 2 hrs from their devices [20, 21] while 1 hr storage wasobserved by Gangorti’s [22]. Selective electrodes, fast electrode kinetics and high solubilityof the dyes are the main unsatisfied properties needed to produce a good PGC. In Chapter 2a change in the configuration of the PGC is proposed that eases some of the hard to achieverequirements.The proposed new structure is justified using analytical and numerical analysis.Instead of illuminating the device through the electrode as was done in the previous work,9ltwFigure 1.6: Schematic of the vertical photogalvanic cell. The left figure shows severalaligned vertical cells to cover a large area. Each cell consists of two parallel elec-trodes, with small spacing, l. Each electrode should be selective to one of the redoxcouples.a vertical alignment of PGCs is suggested so that the light comes in from the gap betweenthe two electrodes as shown in Figure 1.6. This way, the light absorption and hence electrongeneration is spread through the depth of the device. A larger absorption length can be used. Asa result smaller current densities are expected and fast electrode kinetics are no longer required.The depth can be adjusted according to the concentration of the dyes, and thus deeper cellsenable low solubility dyes - that are often more appropriate in other respects - to be employed.Multiple devices stack next to each other to cover surfaces. The investigation of this structureis suggested due to more relaxed requirements and higher possible efficiency. However, thereare some drawbacks to this approach, including the difficulty of achieving highly selectiveelectrodes needed for efficient operation and the limited storage capability of this method. Forsignificant and relatively long-lived storage another approach is investigated that marries tworelatively well established technologies: the redox flow cell and the dye-sensitized solar cell.101.3 Redox Flow BatteriesRedox flow batteries (RFBs) are electrochemical cells with active materials stored separatelyfrom the power conversion electrodes as shown in Figure 1.7. This segmentation results in anenergy capacity independent of the power generator size, which gives more freedom with theselection of electrocatalyst and active materials. In conventional batteries, the electrodes simul-taneously act as the energy storage and power conversion media. This aggregation eliminatesany significant scale-up advantage. The dissolved redox electrochemistry additionally allowsfor higher tolerance to over-charge and over-discharge compared to conventional batteries. Un-der these conditions, the ratio of oxidized and reduced species in the electrolyte is modified tofollow the applied overpotential, meanwhile, the flow mechanism prohibits heat accumulation,whereas in Li-ion batteries as an example, an overcharge results in an internal metal plating andrapid temperature elevation which make the battery unsafe and unstable. Increasing the sizeof reservoirs relative to the electrodes can lead the device to approach the theoretical capacityof its active material as the mass and volume of other parts becomes insignificant compared tothat of active material. Because of the isolation of the two tanks, no self-discharge is expectedfrom the RFBs.Redox flow batteries have two half-cells, separated by a membrane. The active materialthat is stored separately in the form of an electrolyte, flows to these half-cells for charge anddischarge. One half-cell consists of the electrode and part of electrolyte for the negative elec-trode reactions and the other for the positive reactions. It is standard to name the electrodesfor their discharging behavior, therefore the negative electrode during discharge is called theanode and the positive electrode during discharge is named the cathode (although their reactiontypes are opposite during the charging). The electrolytes in contact with the anode and cathodeelectrodes are called negative and positive electrolytes respectively, according to the same con-vention. In order to have high power output and smooth flow of the electrolyte, a pair of porouselectrodes is required to increase the electrodes’ effective surface area and their permeabilityto the active materials.11Figure 1.7: Schematic of a RFB system. Energy is stored in active species On/Rn andOp/Rp which are the oxidized and reduces species of redox couples in negative andpositive electrolytes respectively. The electrolytes are stored in external tanks andare fed to the cell during charge and discharge.Discharging(anode) Rn→ On + e−,(cathode) Op + e−→ Rp,Charging(anode) On + e−→ Rn,(cathode) Rp→ Op + e−.(1.5)It can be seen in Figure 1.2 that the RFB has the highest discharge time among the bat-tery energy storage technologies. Current RFB systems range from 102 to 108 Wh capacity -12the latter being one order of magnitude larger than large scale electrochemical energy storagesystems [13]. Table 1.1 lists some of the commercial RFB energy storage plants across theglobe. The RFB’s capital cost is estimated to be less than 500 $/kWh for an industrial scaleTable 1.1: A list of commercial RFB energy storage facilities.2There are also severalsmaller (sub-MWh) installations in Spain and Denmark.Technology Power rating Storage capacity countryPolysulfide-Bromide 15 MW 120 MWh UKAll Vanadium 15 MW 60 MWh JapanAll Vanadium 5 MW 10 MWh ChinaAll Vanadium 4 MW 6 MWh JapanAll Vanadium 500 kW 2 MWh JapanAll Vanadium 500 MW 2 MWh USIron-Chromium 250 MW 1 MWh US1 MWh storage facility, which is less than the 800 $/kWh minimum cost of Li-ion batteriesin the same scale [23]. Further increase in size would emphasize the price gap even more.The average capital costs of RFBs and Li-ion batteries are estimated to be 300 $/kWh and 540$/kWh respectively for grid-scale storage [12]. The cycle life of RFBs are shown to be largerthan 13× 103 cycles, double the value for the best Li-ion batteries. This give RFBs furtherprice advantages in cost per cycle analysis [13].An RFB structure is used in this work to build a solar-battery by integrating it with a solarcell. Such integration requires design compatibility and availability of the materials that satisfythe requirements of both devices. Dye sensitized solar cell technology is chosen to be pairedwith RFBs as it is a relatively efficient solar cell and employs a redox process that can be sharedwith the RFB system.1.4 Dye Sensitized Solar CellsDye sensitized solar cells (DSSCs) have shown great potential for sunlight harvesting, withtheir low fabrication costs (4-10 years payback time), high efficiency even in low light condi-2From US department of energy global energy database. (http://www.energystorageexchange.org)13tions and environmentally compatible precursors [24]. In 1991, Gratzel et al. expanded the pre-vious work on photocurrent generation from sensitized semiconductors in photo-electrochemicalcells (PEC) [25] and suggested the structure of a solar cell loosely based on photosynthesis innature [26]. DSSCs are made of a nanoporous wide band gap semiconducting electrode (e.g.,TiO2) which is coated with a monolayer of light absorbing dye molecules. This electrode isimmersed in an electrolyte containing an energetically appropriate redox couple (Red/Ox) anda platinum counter electrode. Light absorption in the sensitizing dyes (S) results in an exciton(S∗) whose electron jumps rapidly to the semiconductor,S∗ −→ S++ eTiO2. (1.6)As depicted in Figure 1.8, the mediator then reduces the positively charged dye, reaction 1.7,making the dye ready to absorb another photon,S++Red −→ S+Ox. (1.7)Charge recombination is an interfacial process in DSSCs. The high surface area of thisdevice makes it critical to minimize energy waste through recombination. Separated electronsin the conduction band of TiO2 might recombine with the positive side of dye or the oxidants,eTiO2 +S+ −→ S andeTiO2 +Ox−→ Red. (1.8)Design restrictions should be applied to minimize these leakage currents. Reducing the re-sistance between the TiO2 nanoparticles normally suppresses the first reaction and coulombicscreening of photo-electrons by the surrounding electrolyte is used to limit the second lossmechanism [27].The very limited life time of photo-excited electrons in TiO2 before either flowing to the14- (V/NHE)LoadMediatorPorous TiO2 Sensitizing DyeHOMO (S)LUMO (S*)ehvOR(1)(2)(3)(2) (4)Counter ElectrodeFigure 1.8: Energy diagram and the working principle of a DSSC. Upon the absorptionof a photon, an electron is pumped to the LUMO level of the dye (1) and theninjected into the titanium oxide conduction band . The dye is then neutralized bythe electrolyte (2). The high energy electron in then travels to the external circuitryvia the conductive glass substrate(3). Finally the mediator is reduced at the counterelectrode (4). Dashed lines show the flow of electrons.external circuitry or recombining through reaction 1.8, does not allow for any storage effect inDSSCs in their current form. However, DSSC’s common nature with batteries motivated us tostore solar energy directly in the form of chemical energy without the electrical conversationmedium. This is unique compared to solar cell battery systems proposed in the literature, andhas the potential to decrease the cost and increase the efficiency of the whole system. Redoxflow batteries are used as the battery technology of choice because of cost, mobility and safety,which make them an important candidate for large scale energy storage [28].A solar chargeable redox battery is designed by adding a porous dye-sensitized electrodeto the redox battery structure. This results in a DSSC-like half cell in a battery, as depictedin Figure 1.9. During illumination and charging, the sensitized electrode is connected to thereduction electrode in the other half cell via a switch, causing the oxidation of ions in the DSSCand reduction in the second half cell. A charge balancing ion (shown as ⊕) will balance thecharge across the two half cells by transport through the membrane. The stored energy can later15PumpPumpRp/Op Rn/OnLoadPoroussensitized  electrodeOp Rp On RnChargeDischarge⊕I AnolytereservoirCatholytereservoirFigure 1.9: Schematic of the solar redox battery. The device working mode is selected byconnecting the switch to the photocharging or charge extracting electrodes. Energycapacity is defined by the size of external reservoirs.be extracted through the two non-sensitized electrodes after changing the switch to dischargemode (middle and right in Figure 1.9). A more detailed working mechanism is presented inChapter 3.The demonstration of a working device first involved the identification of proper redoxcouples that satisfy the requirements of both a DSSC and a redox battery. Secondly, highlycatalytic electrodes were selected to interact with the redox couples. These electrodes shouldminimize polarization losses in the device and should be permeable to the electrode so as notto cause significant mass transport losses. Then a membrane was selected to have maximumselectivity over the two redox pairs in order to minimize recombination losses, and have mini-mum ohmic loss in the system. Finally, the efficiency, capacity and lifetime of the device wasstudied, and feasibility of the device for the proposed applications should be analyzed.161.5 Research ObjectiveAn engineering approach towards new structures for combined energy harvesting and storage ispresented. The main objective of this work is to integrate both energy conversion and storagecapabilities in a single device. Such a device could significantly reduce the fabrication andinstallation costs of utilization of renewables by reducing the number of parts and seals inthe combined harvesting and storage system. Further improvement is expected in balance ofsystem and cuts in energy conversion losses. The candidate technologies for such a purposeare selected and design modifications are investigated to achieve the device. An appropriatematerials set (e.g. redox couples, electrodes, solvents and membrane) is chosen to work in thedesigned system. The results are verified through modeling and experiments.1.6 Thesis OverviewIn this chapter, the need is justified for economical large scale storage systems in a carbon neu-tral electricity grid. Potential benefits of such a storage system is explained and technologiessuitable for this purpose are presented. The working mechanism of photogalvanic cells, redoxflow batteries and dye sensitized solar cells were reviewed. These approaches are to be usedin the next chapters as the building blocks of an integrated solar energy conversion and storagedevice.In Chapter 2 an analysis of conventional photogalvanic cells is presented. The require-ments for such cells to work efficiently is derived from analytical modeling and their viabilityis discussed. Most demanding factors are identified and a novel architecture is proposed toovercome the challenging factors. The optimization of the cell is performed with both analyti-cal and numerical modeling.Steps towards the realization of a commercial photogalvanic cell are taken by looking for ameans to achieve selectivity of the electrodes. Semiconducting electrodes are investigated forthis purpose. Energy storage capability is discussed and analyzed, and the drawbacks of thissystem are discussed. In general it is found that even in the optimized configuration suggested,17efficient operation requires the discovery of new materials and techniques, and thus are still notpractical.DSSC-RFB integration is presented in Chapter 3. A solar rechargeable battery is designedand fabricated to address both storage and energy harvesting demands. The methodology isexplained and several cell parts are characterized. Experimental results for demonstration ofthe proposed cell are presented.Finally, the dissertation concludes in Chapter 4 with guides towards the next steps to befollowed for practical solar-battery systems.18Chapter 2A Vertical Architecture for IncreasingPhotogalvanic Solar Cell Efficiency:Theory and Modeling 1Photogalvanic solar cells, the original dye based solar cells, have yet to fulfill their promiseas a low fabrication cost, scalable energy conversion system. The efficient performance ofphotogalvanic cells relies on high dye solubility and selective electrodes with fast electrontransfer kinetics. A new configuration is proposed for photogalvanic cells that removes theseimpractical requirements. Instead of illuminating the device through the electrode, as is theconventional approach, a new vertical configuration is employed with light coming betweenthe two electrodes. This way, the light absorption and hence electron generation is spreadthrough the depth of the device. The depth therefore can be adjusted according to the concen-tration of the dyes to absorb all the incoming photons even with low solubility dyes. As a resultof distributed electron generation, unreasonably fast electrode kinetics are no longer required.The proposed configuration is mathematically modeled and the advantages over the conven-1A version of this chapter has been published in a peer-reviewed journal (Reused with permission from ”Mo-hammad Ali Mahmoudzadeh, John D.W. Madden, ’A Vertical Architecture for Increasing Photogalvanic SolarCell Efficiency: Theory and Modeling’, Electrochimica Acta, Volume 143, 2014, Pages 98-105”, Copyright 2014,Elsevier Ltd.).19tional cell are shown. A numerical model is built for more detailed analysis that gives practicalguidelines for working towards device parameters with high power conversion efficiency. Thereadily available thionine-iron dye-mediator couple could achieve 6 % efficiency if highly se-lective electrodes are used, compared to 0.45 % at best using the conventional approach. Theanalysis suggests that upon the realization of highly selective electrodes and an improved dye/-mediator couple, an efficiency of 13 %, and potentially higher, should be achievable from thenew configuration.2.1 IntroductionSolar energy is the most abundant of the readily available renewable energy sources. So far,the cost of conventional solar power relative to fossil fuel alternatives and the absence of sup-porting energy storage facilities in the electricity grid, have impeded its widespread use ingrid-tied locations. Many approaches are being taken to reduce the cost of the solar power.Photogalvanic cells (PGC) were studied immensely in 1980s as a cheap solar energy harvest-ing system. The photogalvanic effect was first observed in 1925 by Rideal and Williams [29],and it was Rabinowitch that initially investigated the much studied iron-thionine photogalvanicsystem [30]. Several other groups pursued the work both to understand the mechanism and tofind the optimum device configuration for PGCs [31–35]. These studies together with the workon semiconductor electrochemistry by Gerischer [25, 36], Nozik [37] and Gratzel [38] led thedesign of dye sensitized solar cells (DSSC) [26]. The analytical analysis of PGCs proposedby Albery and Archor [18] showed the possibility of high performance PGCs given certainconditions of device geometry and chemistry. Use of micelles in photogalvanic cell have beensuggested by Groenen et al. [39] in order to increase dye solubility and suppress back-reactions.Recently, a set of empirical studies examined a variety of dye/mediator couples, including thestudy of several dyes by Gangorti et al. [22, 40–42], the effect of surfactants by Genwa [21, 43]and even the use of mixed dyes by Lal et al. [44]. The highest efficiency of a PGC is claimedby Bhimwal and Gangotri to be 1.62 % with methyl orange as photosensitizer dye [19], which20expresses how far these devices are from practical use. Selective electrodes, fast electrode ki-netics and high solubility of the dyes are the main unsatisfied properties of a good PGC. Wepropose a change in the configuration of the PGC that lightens up some of the hard to achieverequirements and we justify our proposed structure by analytical and numerical analysis.Instead of illuminating the device through the electrode as was done in the previous work,we suggest vertical alignment of PGCs so that the light comes in from the gap between the twoelectrodes as shown in Figure 2.1. This way, the light absorption and hence electron generationis spread through the depth of the device. As a result of larger absorption length, smallercurrent densities are expected and fast electrode kinetics are no longer required. The depthcan be adjusted according to the concentration of the dyes, and thus deeper cells enable lowsolubility dyes to be employed. Multiple devices stack next to each other to cover surfaces. Wesuggest the investigation of this structure due to more relaxed requirements and higher possibleefficiency.In this chapter, the design guideline for an efficient cell will be presented as was publishedby Albery et al. [32]. The analysis of vertical cells is then presented and the requirements forhigh efficiencies will be derived from mathematical modeling. Several design factors will bediscussed and compared to the conventional cell and the advantages of the new configurationare shown. In the simulation sections the framework of a 2D computer model for PGCs isexplained. Both cells are then modeled and optimized, assuming in one case known propertiesof dyes and mediators, and in the second case given dyes and mediators that should be phys-ically realizable which gives a target configuration for PGCs. The expected efficiencies arecompared. The benefits of the vertical cell are demonstrated and the target device parametersare shown to make the cell more viable than the conventional cell.2.2 Analysis of the Conventional Photogalvanic CellThe working principles of PGCs are explored in section 1.2. The main reactions governingtheir performance can be summarized as,21ltwFigure 2.1: Schematic of a vertical photogalvanic cell. The left figure shows severalaligned vertical cells to cover a large area. Each cell consists of two parallel elec-trodes, with small spacing, l. Each electrode should be selective to one of the redoxcouples.(photoexcitation) S+hν → S∗( kRelax−−−→ S), (2.1)(quenching) S∗+MkQ−→ S−+M+, (2.2)(bulk recombination) S−+M+kr−→ S+M, (2.3)(Anode) S−→ S+ e− and(Cathode) M++ e−→M.(2.4)Albery examined the electrode-illuminated cell of Figure 1.5 analytically and derived somedesign criteria for photogalvanic cells [18]. The differential equation governing the photo-absorption by dye molecules, the reaction of excited dyes and mediators, and the transport ofspecies were solved simultaneously. The output of the analysis were four characteristic lengths22Table 2.1: Characteristic lengths of photogalvanic devices. ε is the extinction coefficientof the dye. [S] is the concentration of the light absorbing dye and [M+] is the concen-tration of the oxidized mediator. D is the diffusion coefficient. φ0 is the solar photonflux in units of [molcm−2 s−1].Xl l Distance between electrodesXε (ε[S])−1 Light absorption lengthXk (D/kr[M+])1/2 Typical diffusion length before recombinationXg (D/φ0ε)1/2 Typical distance dye diffuses between photon absorp-tion eventsof Table 2.1, which should be balanced according to the cell requirement to achieve a highpower efficiency.The complete absorption of the incoming light requires that the device be deep enough thatmost of the light be absorbed, or Xε << Xl . In order to extract the separated charge, generatedM+ ions need to travel to the illuminated electrode before their recombination, therefore, thedistance over which generation is occurring should be less than the length over which it willlikely diffuse before recombining through (2.3), and hence Xε << Xk. Additionally, the exciteddyes should be replaced by fresh ones before the arrival of the next photon in order to avoidsolution bleaching, and hence, Xε << Xg.The requirement of a net positive bulk generation requires that generation rate be fasterthan recombination rate therefore Xg ≤ Xk. Finally, the maximum travel length is the distancebetween the electrodes, therefore the other three length constants should be smaller than Xl .The following formula was suggested for these values:10Xε ≈ Xg ≈12Xk < Xl. (2.5)These relations define the approximate conditions for the optimized cell. Using the typicalvalues of D, ε and I0, a set of parameters were derived by Albery et al. [32] to make anefficient cell as shown in Table 2.2.One last requirement is that electrode kinetics be fast compared to the mass transport and23Table 2.2: Albery’s recipe for the optimal cell [32].Xg = 10µm D = 10−5 cm2 s−1, ε =100 mM−1 cm−1, I0 = 1.6×10−7 molcm−2 s−1Xε = 1µm ⇒ [S] =0.1 MXk = 20µm ⇒ kr[M+] =2.5 s−1Xl > 20µmrecombination rates, and thus satisfy the following conditions:k0 >>DXεand (2.6a)k0 >DXk, (2.6b)where k0 is the standard heterogeneous rate constant. The first condition ensures that the prod-uct species are generated close enough to the electrode to be able to interact with it and thesecond condition provides for a higher chance of electron extraction than bulk recombination.As explained briefly in the introduction section, the latter condition of electrode kineticsis hard to satisfy, particularly in case of selective electrodes as any surface modification im-pedes the electron transfer between ions and the electrode. Additionally, fast electrode kineticsis incompatible with slow bulk reactions according to Marcus theory [45]. In other words, nodye/mediator/electrode combination is likely to be found that offers both fast electrode kineticsand slow bulk recombination. Finally, the solubility of the dyes are much lower than the re-quirements of Table 2.2 [46], as a result, photogalvanic cells show poor efficiencies. Arrangingthe light path to be parallel to the electrode surfaces, as shown in Figure 2.1, is now shown toalleviate a number of constraints. We investigate the requirements of target vertical cell in thenext section and show that the proposed cell is less demanding in these two areas, i.e., moderateelectrode kinetics and dyes with low solubility can still be utilized in an efficient vertical cell.2.3 Analytical Analysis of the Vertical Photogalvanic CellThe vertical photogalvanic cell of Figure 1.6 with the reactions shown in Figure 1.5, is modeledin order to estimate the feasibility of the requirements for such a cell to work efficiently. First,24an analytical model is presented that allows defining guidelines for design of efficient targetvertical cells. This is followed by a numerical simulation, allowing the cell efficiency to beestimated.A comparison of maximum generation rate, which happens at the illuminated electrolytesurface, and the quenching rate constants shows that even at small concentrations of mediator,the charge separation of reaction (2.2) happens much faster than the initial photo-excitation ofthe dyes, reaction (2.1) (as has been previously assumed by Albery [18]). Therefore, the twostages of light absorption and charge separation can be simplified into a single reaction of :S+MGop+k f−−−−⇀↽ −krS−+M+, (2.7)where S/S− represent the two states of the dye and M/M+ those of the mediator. Gop, k f andkr are bulk reaction rates for optical generation, dark forward reaction and bulk recombination,respectively. M/M+ concentrations are assumed constant in the analysis by adding a conditionthat mediator concentration is much larger than that of the dye, following Albery [18], in orderto enable an analytical solution. As a result of illumination to the gap, the optical generation,Gop, varies through the depth. This generation is averaged over the cell depth in our onedimensional analysis and therefore is independent of dye concentration as long as the cell isbuilt deep enough to absorb all the incoming light. The assumptions of uniform generationand constant mediator concentration are removed in numerical analysis provided in the nextsection.A diffusion-recombination reaction mechanism is considered for the transport in the bulkat steady-state,D∂ 2[S−]∂x2 +Gop− kr[S−][M+] = 0. (2.8)In order to simplify the expressions, it is useful to rewrite the equations in dimensionless form.The length is normalized to the cell length, l, and the concentrations to the dark dye concentra-25tion, [Sd]. The unitless bulk reaction will be as follows,∂ 2u∂χ2+α2−β 2u = 0, whereχ = x/l, u = [S−]/[Sd], α2 =Gop l2D [Sd]and β 2 = l2 kr [M+]D. (2.9)Parameter α compares the cell length to generation length, the distance dye diffuses beforebeing hit by a photon. A large α guarantees dye excitation before traveling the length of thecell. β represents the cell length compared to recombination length, which is the diffusiondistance of the charged states before recombination in the bulk. A small β is desired in order toincrease the chance of charge extraction. For the analytical analysis section, completely selec-tive electrodes with fast kinetics are assumed to be employed, where each electrode interactsonly with one redox couple. The optimum cell performance conditions, which we are lookingfor, happen under this condition which guarantees minimum recombination. (The effect of thenon-perfect mediator discrimination will be explored in the numerical analysis). This assump-tion allows us to assign all the current from the left electrode to the interaction with the S/S−couple, therefore,Dd[S−]dx|x=0 =JFwhich translates to (2.10)dudχ |χ=0 = m, m =J · lFD[Sd], (2.11)where J is the cathodic current density, F is the Faraday constant, D is the diffusion coefficientand m is the normalized current density. On the right electrode, no electron transfer happenswith dyes due to the complete selectivity assumption. The boundary condition is thendudχ |χ=1 = 0. (2.12)Solving equation (2.9) with boundary conditions of (2.11) and (2.12) results in a normalized26concentration profile as followsu =α2−mβcosh(β −βχ)csch(β )β 2 . (2.13)All device characteristics can be derived from equation 2.13- most importantly, the currentdensity, m, which relates to u through (2.11). We can define the cell efficiency in terms ofconcentration to be able to calculate cell parameters, l, [Sd], [M+] and kr, for the optimizedcell. The concentration at the surface of the electrode can be written asu0 = u|χ=0 =α2−mβcoth(β )β 2 ⇒ m =α2−β 2u0βcoth(β ) . (2.14)The output voltage of the cell, the difference in electrode’s electrochemical potential, isalso normalized. The unitless potential difference, ∆P can be calculated as equation (2.17)P =FRTE (2.15)∆E = E2−E1 = E0M/M+ +RTFln[M+][M]−E0S/S−−RTFln[S][S−](2.16)[M+], [M]≈ constant⇒ ∆P = ∆P0 + ln(u−10 −1) (2.17)∆P0 =FRT(E0M/M+−E0S/S−) (2.18)The efficiency of the cell can then be calculated by dividing the product of the output currentand voltage by the incoming light power. Using equations (2.17) and (2.14), the efficiency canbe written in the following form,η = (t ·w · J) ·∆El ·w · I0×100% =FD[Sd]RTFI0tl2m ·∆P×100%=D[Sd]RT∆P0I0tl2α2β coth(β )︸ ︷︷ ︸θ , max efficiency(1−β 2u0α2)(1−ln(u−10 −1)∆P0)︸ ︷︷ ︸ψ, load dependent100%. (2.19)The two right hand side terms are functions of u0 and therefore m, the current density,27which is dependent on the load connected to cell. In order to deliver the maximum efficiency,one should maximize this part, ψ , by adjusting the load and make those terms as close tounity as possible. The part that mainly governs the magnitude of the efficiency is the leftmostproduct in (2.19), called θ , that needs to be maximized by adjusting the cell parameters. Wefirst look for the cell conditions that optimize this term, then adjust the load, and therefore u0,to maximize the two RHS terms.2.3.1 Geometry OptimizationInserting the values of α and β into θ , one can write this efficiency term in the form ofθ = D[Sd]RT∆P0I0tl2α2βcoth(β ) =D[Sd]RT∆P0I0tGtot l2/(tD[Sd])l2l√kr[M+]D coth(l√kr[M+]D)=RT∆P0GtotI01l√kr[M+]D coth(l√kr[M+]D) . (2.20)All the variable parameters of equation (2.20) are collected in the second term which rep-resents a half-bell shaped function of β whose maximum occurs at zero, as depicted in Figure2.2. Charge extraction at the electrodes always competes with bulk recombination, therefore alarger β (faster bulk recombination or wider device) consistently reduces the efficiency. Con-sequently, the cell length should be decreased to the extent that is allowed by the manufacturinglimitations to have an efficient cell. One can see that θ still has 76% of its maximum valuewhen l√kr[M+]D ' 1, which gives some room to deviate from the maximum point without ahuge efficiency loss.Assuming a slow - but feasible - bulk recombination rate of kr = 0.5× 103 M−1 s−1, atypical diffusion constant D = 10−5 cm2 s−1, and a cell length of l = 100µm, the mediatorconcentration should be smaller than 200µM in order to enable reasonable efficiency. Dyeconcentration should be at least 2 times smaller than that of the mediator so that its changedoes not disturb the mediator concentration profile, therefore [Sd]' 100µM.280 0.5 1 1.5θ/θ max(1,0.76)(0.2,0.98)Figure 2.2: Load independent efficiency vs l√kr[M+]D . θmax happens at very small devicelengths, however, the device length should be balanced between performance andfabrication limitation.The depth of the cell has no direct influence on θ output power as long as all the incominglight is absorbed in the cell, so the depth should be kept greater than 4ε [S]. The next termthat should be maximized is the ∆P0Gtot product in the numerator. ∆P0 reflects the differ-ence in the electrochemical potentials of two redox couples which should ideally match theHOMO-LUMO levels of the dye minus the required overpotential to drive the electron trans-fer. Therefore, the larger the difference is, the more significant the output voltage achievable.However, that would result in a smaller portion of incident photons to be absorbed and the Gtotto drop since lower energy photons cannot excite the dyes. Balancing the trade-off betweenGtot and ∆P0 leads to an optimum HOMO-LUMO level difference of 1.4 eV , which in turnleads to θmax = 24 %. This is the maximum efficiency regardless of the effect of the load, i.e.ISC×VOC/I0. Smaller separation between electrodes leads to higher efficiency - for example29with 20µm separation, θmax can be expected to reach 31 % as the bulk recombination loss dropssignificantly.2.3.2 Load OptimizationBoth bracketed terms in the load dependent part of equation 2.19 , ψ , should approach unityin order to achieve maximum efficiency. For this section, this condition is assumed to besatisfied and an optimum load condition is calculated. The assumption is subsequently shownto be valid. Keeping that in mind, and neglecting the product of the small terms, ψ can beapproximated as :ψ(u0) =(1−β 2u0α2)(1−ln(u−10 −1)∆P0)' 1−β 2u0α2 −ln(u−10 −1)∆P0. (2.21)Differentiating with respect to u0 shows a maximum at u0,m =β 2−√β 2−4α2/∆P02β 2 and ψm' 0.91.Figure 2.3 shows the variation of ψ with respect to u0 for some typical cell parameters, whereu0,m is located very close to zero. The maximum efficiency load for this case happens close tothe short circuit conditions which happens at u0 = 0 according to equation (2.14).2.3.3 Estimation of Maximum EfficiencyAltogether, the total efficiency of the vertical photogalvanic cells would appear to go as highas 28 and 22 % for 20 and 100 µm device lengths, respectively. The conditions for such per-formance are given above. In terms of device length constants, one can conclude the recom-bination length Xk should be larger than the device length Xl as shown in Figure 2.2. Thelight absorption in the vertical configuration is not limited to the electrode separation, however,since all the light needs to be captured, the cell thickness must be much larger than Xε , i.e.d >> Xε . As calculated above, the l√kr[M+]D should be smaller than unity, which puts limitson bulk recombination rate, mediator concentration and device length. In practice not all thegiven device parameters are readily achievable, but, as will be discussed later, they are morepractical than those of traditional PGCs. In the next section these parameters will be fine tuned300 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40.750.80.850.90.95u0,mψmFigure 2.3: The variation of ψ with respect to u0. l = 100µm, [Md] = 200µM, [Sd] =100µM, kr = 0.5×103 M−1 s−1 and ∆E = 1 V.in a more realistic 2D system using electrodes with less than perfect selectivity.2.4 Simulation of the Photogalvanic CellNumerical simulation enables the major assumptions of analytical analysis to be relaxed. Adepth dependent generation, variable mediator concentration and imperfectly selective elec-trodes are modeled which gives a more accurate device analysis. The photogalvanic cell wasmodeled in COMSOL Multiphysics software (COMSOL Inc., Palo Alto, CA) using a 2D ge-ometry. COMSOL Multiphysics was chosen due to its ability to simultaneously solve severaltypes of differential equations using the finite element method (FEM). The model includes tworedox couples in the bulk and two selective electrodes with different reaction rates towards re-dox species. Bulk transport, generation-recombination and interaction on the electrodes weremodeled using the Transport of Diluted Species solver in COMSOL. The Butler-Volmer equa-tion was solved on the electrodes with the built in ODE solver and light absorption was modeledusing the general PDE solver, all of which were solved self-consistently in COMSOL. Both31conventional and vertical photogalvanic cells were analyzed numerically first under optimumconditions and then with more readily available parameter values. The performance of thedevices were compared for each scenario.2.4.1 Equation SetReactions (2.1), (2.2), (2.3) and (2.4) describe the interactions happening in working photo-galvanic cells. Below, the main bulk reaction is shown using both the notations used in thesimulation and our analytical analysis,o1 + r2Gop+k f−−−−⇀↽ −krr1 +o2, or equivalentlyS+MGop+k f−−−−⇀↽ −krS−+M+.(2.22)A reaction-diffusion system of equations is set up to model the transport of species in thephotogalvanic cell as shown in equation (2.23),∂ci∂ t +∇.(−Di∇ci) = Ri i = o1,r1,o2,r2,Ro1 = Rr2 = kr[r1][o2]− k f [o1][r2]−Gop,Rr1 = Ro2 =−kr[r1][o2]+ k f [o1][r2]+Gop.(2.23)On the electrodes, reactions happen at different rates. The selectivity constraint requiresthat each electrode has fast kinetics with one couple and slow kinetics with the other oneand the dominant reaction be that of equation (2.24). The Butler-Volmer equation is used todescribe the electron transfer at the interfaces,(On electrode 1) r1→ o1 + e−,(On electrode 2) o2 + e−→ r2, (2.24)J/F = k01([o1]e−α fη1− [r1]e(1−α) fη1)+ k02([o2]e−α fη2− [r2]e(1−α) fη2), (2.25)32where η , the overpotential, is the difference between the electrode’s potential and the standardpotential of the redox couple (ηa = Eelec−E0a ). α is the transfer coefficient and chosen to be0.5 in this work which represents a symmetric energy barrier for electron transfer. k01 and k02are standard rate constants of an electrode’s interaction with redox couples [47].The light absorption follows a Beer-Lambert law behavior in which the reduction in fluxthat happens in a layer with thickness dl containing species i with concentration ci, is given byequation (2.26) where ε is the molar absorption coefficient,dφ =−(εln10)cdφdl. (2.26)The cd term of 2.26 is the dye concentration in the relaxed state i.e. [o1]. This links thetwo physics systems, light absorption and transport, together in the process of building a selfconsistent model.The other linking variable is the generation term in equation (2.23). Assuming absorptionto happen at a constant wavelength for simplicity, one can approximate the optical generationrate asGop 'φabs∆z'dφdz[molm3s], (2.27)where z can be the vertical or horizontal direction depending on the illumination direction. Ascan be seen in equations (2.26) and (2.27), the generation rate is proportional to the numberof absorbed photons at each location which is non-linearly related to the concentration of lightabsorbers in the relaxed state. The equation system is implemented in COMSOL and thetime dependent answer of the system was observed with initial conditions of [S] = [Sd] and[M] = [M+]. The steady state results were then extracted after the time dependent variablesreach a plateau. These steady state values were taken as performance figures of the cell andcompared for vertical and conventional cells in the next section.332.4.2 ResultsPhotovoltaic devices traditionally use electrodes that lie in a plane that is ideally perpendic-ular to the direction of the incident light. One can extract most of the generated charges byputting the extracting electrode close to the absorbing section (junction). For the same reason,photogalvanic cells are illuminated through one transparent electrode while the other electrodeis typically kept in the dark. This way, the peak generation happens very close to the collect-ing electrode. Despite Albery’s initial analysis that showed 18% efficiency [48], he concludedlater that some practical restrictions limit the performance to 5 % [46]. He derived conditionsfor this ’optimal’ cell as shown in Table 2.2. Simulations of the classic iron-thionine cell, andof Albery’s ’optimal’ cell integrated with a pair of selective electrodes were performed in our2D model and the device optimization resulted in a power conversion efficiency of 0.45 % and3.7 %. Thus it is not surprising that the best experimentally measured performance from thePGC is 1.62 % [19] (in which no significant selectivity is present). To understand the ultimateperformance expected from these devices, the cell parameters were investigated again in the2D model. The optimized traditional configuration cell, listed in Table 2.3, showed an effi-ciency of 2.07 %. Only the extreme case of completely selective electrodes raise the efficiencyto 3.7 %.It should be noted that even this low efficiency performance is impractical in reality. Somecharacteristics used in Table 2.3 to compute Albery’s ’optimal’ are incompatible with eachother and some are simply hard to achieve. For example, very few redox couples and elec-trodes meet the very fast electrode kinetics requirement of equation 2.6b. Fast electrode kinet-ics, needed to produce high currents, are also incompatible with slow bulk reactions (neededto reduce recombination losses between mediators) according to Marcus theory [45]. The ac-tual solubility of the dyes are much lower than those assumed here, which leads to poorerperformances in practice compared to the theory.In the suggested vertical configuration of Figure 2.1, light absorption and charge extractionlengths have been decoupled, therefore a smaller dye concentration can be utilized to reduce the34Table 2.3: Device parameters and performance of electrode illuminated photogalvanic de-vices. Efficiencies, η , are as computed using COMSOL.Albery’s Op-timal (PerfectSelectivity)Albery’s Op-timal (PartialSelectivity) [32]Iron-Thionine[46]ExplanationE0S E0S E0S 0.462 V dye standard potentialE0M E0S + 1.4 eV E0S + 1.4 eV 0.77 V M/M+ standard potentialEph 1.6 eV 1.6 eV 2.07 eV dye band gapkL,1 10−3 ms−1 10−3 ms−1 10−5 ms−1 rate constant , fast redoxkL,2 0 m/s 10−10 ms−1 10−12 ms−1 rate constant ,slow redoxkr 5×102 M−1 s−1 5×102 M−1 s−1 5×102 M−1 s−1 bulk recombination ratel 50 µm 50 µm 50 µm cell length[Sd ] 100 mM 100 mM 100 mM dye concentrationI0 1000 Wm−2 1000 Wm−2 1000 Wm−2 light intensityη 3.7 % 2.07 % 0.45 % efficiencycurrent density through the electrodes while not affecting the generated current per illuminatedsurface. Electrode kinetics need not to be particularly high if dye concentrations are low, andsimilarly diffusion lengths to electrodes can be relatively long (provided they are similar to orshorter than the recombination length). The selectivity level of each electrode - the differencein reaction rates towards the two redox couples - is investigated. The results show that a 6 to 7order of magnitude difference in rate constants is enough to achieve an efficient cell.The vertical cell is modeled in COMSOL, and shows improvement in performance. Theefficiencies of the target and iron-thionine cells were found to be 12.9% and 6%, respectivelyfor a 100 µm cell length. Because of the partial selectivity of the electrodes and the concen-tration dependent optical generation that were neglected in the theoretical model, these values,achieved with parameters of Table 2.4, are smaller than the prediction of the analytical analysis.The parameters listed in Table 2.4 are chosen with physical feasibility in mind. It is desired tohave as small a bulk recombination rate as possible, the suggested dye-mediator couple of thetarget case is assumed to have a recombination rate in the same range as the iron-thionine cou-ple, which is one of the slower known bulk reaction rates. The value for iron-thionine couple isdetermined from the literature to be 5×102 M−1 s−1 [46]. The dye concentrations are limited35to sub-millimolar range and electrode kinetic rate is on the order of 10−5 ms−1. For the targetdevice, the effective band gap is assumed to be equal to the energy difference in HOMO andLUMO of the dye. Also, a 200 mV overpotential is assumed for the bulk homogeneous chargetransfer between the dye and the mediator. This is reflected in the difference in dye band gapand M/M+ standard potential in Tables 2.3 and 2.4.Device characteristics of the target vertical cell are shown in Figures 2.4a and 2.4b. Thelow fill factor observed in the Current-Voltage characteristics is mainly due to bulk and elec-trode recombination losses. Dye concentration and cell depth were swept in value to find theoptimum concentration and size which enables both the use of the slow kinetics electrode (byreducing the generation rate) and the full absorption of the incoming light. The light intensity,I, is plotted through the depth of the cell in Figure 2.4c. The horizontal locations where I goesto zero represents full absorption by the dye. In this graph there is small part where not alllight is absorbed. The optimum dye concentration is 200 µM, which is higher than the optimaltheoretical value of the last section, because of the effect of bleaching on the surface that waspreviously ignored in the theoretical analysis. Despite the bleaching, most of the incominglight is absorbed by over a depth of 50 mm.A comparison of the cell parameters in two configurations reveals the vertical cell to beless demanding. The required electrode kinetics is reduced by two orders of magnitude in thevertical structure and partial selectivity of seven order of magnitudes proves to be sufficientfor efficient device performance. This is another advantage of the vertical cell since the 3.7%ultimate efficiency of the conventional cell was calculated based on complete selectivity andit drops to 2.07 % at a selectivity of 7 orders of magnitude. The concentration of the reduceddye, [S−], is plotted in 2.4d. It reaches zero on the dye-interacting electrode at maximumefficiency operation point, which agrees well with the theoretical prediction that the optimumload happens at very small value of u0,m, equation 2.21.As explained in the analytical section, electrode separation is inversely related to the per-formance. Figure 2.5 shows the efficiency of top and side illuminated cells for different cell360 0.2 0.4 0.6 0.8 )(a) I-V Characteristics0 0.2 0.4 0.6 0.8 102468101214V(V)Efficiency(%)(b) Cell Efficiency10008006004002000W/m2(c) Light intensity at ηmax0. 20 40 60 80 10001020304050 length (µm)Cell depth (mm)Cell(d) D− concentration at ηmaxFigure 2.4: Simulation results of the target vertical PGC. l =100 µm, [Md] =500 µM,[Sd] =200 µM, kr = 0.5×103 M−1 s−1 and ∆E = 1.4Vlengths and electrode selectivities. Device performance is more sensitive to selectivity in verti-cal cell. Therefore, a pair of selective electrodes is crucial in making a practical vertical PGC.For the target cell, this length was set to 100µm for practical reasons, however, as shown inFigure 2.5, efficiencies up to 20.2 % is achievable with thiner vertical cells. It can be seen that12.9 % efficiency of the 100µm vertical PGC is not achievable with any conventional PGCregardless of the geometry.Overall, promising device performance is expected with physically feasible parameterswhich are close to the maximum we think can be achievable. Further research is requiredto find dye/mediator couples in order to improve these parameters. The fact that each cell is37Table 2.4: Device parameters and performance of vertical photogalvanic devices. Effi-ciencies, η , are as computed using COMSOL.Target verticalcellIron-Thioninevertical cellExplanationE0S E0S 0.462 V dye standard potentialE0M E0S + 1.4 eV 0.77 V M/M+ standard potentialEph 1.6 eV 2.07 eV dye band gapkL,1 10−5 ms−1 10−5 ms−1 rate constant, fast redoxkL,2 10−10 ms−1 10−10 ms−1 rate constant, slow redoxkr 5×102 M−1 s−1 5×102 M−1 s−1 bulk recombination ratel 100 µm 100 µm cell length[Sd ] 200 µM 200 µM dye concentrationt 50 mm 50 mm cell depthI0 1000 Wm−2 1000 Wm−2 light intensityη 12.9 % 6 % efficiencyvery thin means that multiple cells must be fabricated in series to cover large areas.2.5 ConclusionVertical configuration photogalvanic cells are suggested and modeled. The analysis of an in-dividual cell shows this configuration should result in higher efficiencies than where the illu-mination is through the electrode. To be effective, sub-millimolar dye concentrations and slowbulk recombination rates on the order of 103 M−1 s−1 are required, which is not easy to achievebut seemingly achievable (the iron-thionine recombination rate is half this number). Electrodekinetics should be reasonably fast, but extending the light absorption through a deep cell makesmoderate electron transfer rate constants of 10−5 ms−1 sufficient, which is 100 times slowerthan the requirement for the traditional cell. The electrodes were assumed to be completelyselective in the analytical section, but optimizing this parameter in the numerical analysis re-vealed a need for 6 to 7 orders of magnitude difference in rate constants.Semiconductors are proposed as selective electrodes by our group [49] and so far, selec-tivity up to 3 orders was shown. The difference in kinetic rates comes from the band edgepinning that happens at a semiconductor-liquid interface. The theory and experimental work38on this method is covered in Appendix B. Given the gap between the required and the obtainedselectivity rates, further work needs to be done in this area.A further challenge is to use this approach for energy storage. With typical recombina-tion rate constants even as slow as 100 M−1 s−1, energy elevated redox couples recombine inhundreds of seconds in the bulk. Therefore, synthesis of dye/mediator pairs with slower bulkreaction rates are also required in order to achieve practical storage.All in all, we believe vertical PGCs might be used as cheap, low maintenance solar cellsassuming that an appropriate electrode-dye-mediator-electrode combination is found. Incorpo-ration of energy storage for longer than minutes will be challenging. For solar cell and storagein the scale that is demanded by the goals of this work, we investigate a less demanding ap-proach in the next chapter that is based on two relatively well established technologies: theredox flow battery and the dye-sensitized solar cell, as discussed in Chapter 3.390 1 2 3 4 5 6 7 805101520Efficiency (%)Selectivity (Log(k01/k02))20 µm50 µm100 µm200 µm(a) Top illuminated cell0 1 2 3 4 5 6 7 802468Selectivity (Log(k01/k02))Efficiency (%)20 µm50 µm100 µm200 µm(b) Side illuminated cellFigure 2.5: Effect of device length and electrode selectivity on cell efficiency for bothtop and side illuminated cells. The horizontal axis is the selectivity in orders ofmagnitudes. The effect is shown for different electrode separations as mentioned inthe figure legends.40Chapter 3Solar Redox BatteryIn this chapter, an integrated solar energy conversion and storage system is presented using adye sensitized electrode in a redox battery structure. The design enables independent scaling ofthe power and energy ratings of the system and thus it is potentially applicable for large scalegrid connected storage purposes. A 35 times increase in energy density is observed comparedto similar devices by choosing iodide/polysulfide as the pair of active materials matched withpermeable porous electrodes. An areal energy density of 52 µWhcm−2, charge capacity of1.2 Ah L−1, energy efficiency of 78% and almost perfect Coulombic efficiency are observedfor the integrated non-flow cell. The solar rechargeable battery system offers high savingson fabrication and installation costs and a higher round-trip efficiency compared to individualdevices.The chapter begins with a description of the motivation for the work and a review of relatedliterature. Cell design considerations are then discussed, followed by a summary of the results.The chapter wraps up with a detailed presentation of experimental methods.3.1 IntroductionThe global concerns over the impact of conventional energy sources on the environment haveresulted in a great interest in carbon-neutral energy generation. Solar energy is the only re-41newable source that can supply all the global energy demand projected in 2050 [5]. However,variability during daytime and intermittency as a result of atmospheric conditions are hinder-ing its widespread use. Therefore, large-scale electrical energy storage systems are requiredto balance the grid during supply shortage. Compared to solid-electrode batteries, redox flowenergy storage systems have several advantages for grid applications such as scalability andlow self discharge [50]. These features are incorporated into a design of an integrated solarenergy harvesting and storage system.A solar power rechargeable battery could lead to cost and space efficiency and reduce thecomplexity of solar energy harvesting systems. Several attempts have been made to achievesuch a device starting with back to back fabrication of the battery and solar cell [51, 52].Although this method reduces some internal losses, it still involves two individual devicesbeing fabricated and offers no material or fabrication cost benefits.Being an electrochemical cell and having structures similar to batteries, DSSCs are goodcandidates to be integrated in a solar battery. The first DSSC device capable of energy storageis introduced by Miasaka et al. [53] in 2004. Their photocapacitor is based on two membraneseparated activated carbon layers for charge storage. They used a platinum electrode on oneside and a dye sensitized electrode with a solid hole conductor on the other side. A capacitanceof 0.69 Fcm−2, discharge capacity of 0.02 mAhcm−2 and charge state voltage of 0.45 V areobserved from this device. The same group improved their work by adding a third electrodebetween the supercapacitor and the dye sensitized electrode to overcome the high internalresistance of the sensitized electrode for charge/discharge and also to prevent the contaminationof the charge storage section with redox couples from the solar cell parts [54]. The areal energydensity is improved from 9.3 µWhcm−2 to 47 µWhcm−2 and the voltage is increased to 0.8 V.A pseudo-capacitor using tungsten oxide for energy storage is demonstrated by Saito etal. [55]. The work is continued by using interdigitated Nafion / polypyrrole-platinium elec-trodes [56]. The similar geometry of the device to a simple DSSC allows for its integrationinto the existing DSSC devices. A discharge capacity of 10.5 µAhcm−2 is obtained in this42cell which is almost half of the value achieved from a mesh type three electrode setup withthe same materials, mainly because of the smaller effective surface area, lower quality of PPyand leakage of I–3 through the Nafion membrane. Yang et al. developed a flexible integrateddevice using films of free-standing and aligned carbon nanotubes for energy storage, and adye sensitized electrode for charging. The photoelectric conversion efficiency and specific ca-pacitance are measured to be 6.1 % and 48 Fg−1. The areal energy density of the device ismeasured to be 54 µAhcm−2 [57]. Recently, a dye-sensitized photo-electrode was shown thatis integrated into a lithium-oxygen battery for photo-assisted charging [58]. The maximumphotocurrent density from this device is only 150 µAcm−2, however, the charging over-voltageof the battery dropped by up to 0.6 V with the assistance of the attached DSSC.These efforts in the integration of DSSCs and batteries can be classified into two groups.First, the devices where the active energy storage and energy conversion materials are sepa-rated as shown in Figure 3.1a. The three electrode setup is achieved by using a bi-functionalcenter electrode. These devices are in fact similar to back-to-back fabricated solar cell-batterystructures and their working mechanism involves undergoing similar energy conversion stepsas in the individual solar cells and batteries [53, 57, 58]. After the illumination, electrons arepumped into the TiO2 conduction band, travel through the external circuitry and reduce the ac-tive materials at the end electrode. Meanwhile in the DSSC, redox species are oxidized on thephoto-electrode and reduced at the center electrode. The latter electron transfer is balanced bythe oxidation of the active material at the other side of the bi-functional electrode. This way,solar energy is practically stored in a separate device from the DSSC. In the second group,which is the goal of this work, some active parts are shared between the energy conversion andstorage devices (Figure 3.1b) [55, 56]. The mechanism involves fewer energy conversion stepscompared to two separate devices, which could lead to improvement in solar energy storageyield and the total device cost at the expense of a more complex structure. During the photo-charging, electrons flow through the external circuitry and reduce the active materials at theend electrode. The redox species are then oxidized on the photo-electrode and the solar energy43is readily stored.LoadTCOTiO2/DyeMembraneCharge collectingelectrodesActive energy storage material(a)TCOTiO2/DyeMembraneCharge collectingelectrodesActive energy storage materialLoad(b)Figure 3.1: Schematic of different approaches towards the integrated solar battery. Theoperation mode is selected by the switch.(a) Back-to-back fabricated solar cell andbattery. The middle electrode is shared between the two devices.(b) Integrated solarbattery. Both half cells are active during energy capture and storage. Active energystorage materials are shared between the two operation modes.Although these devices might be practical to cover for short term variations in sunlight,they still do not have enough capacity for overnight energy storage even if the high energydensity of Li-Ion cells are to be achieved within an integrated solar cell. While an energydensity per unit area of at least 20 mWh cm−2 is needed to store solar energy received in fourhours of daylight even for a modest performance solar cell of 5% efficiency 1, the storagecapacity of the aforementioned devices range from 0.1 µWh cm−2 [53] to 47 µWh cm−2 [54].Considering the fundamental limitation of solid batteries, this work incorporates a technologythat allows independent scaling of energy and power ratings, i.e. redox flow batteries (RFBs),1 E4h ' 1000 W m−2 × 5 %× 4 h = 20 mWh cm−244in the proposed solar battery system.Redox flow batteries are electrochemical cells with active materials stored separately fromthe power conversion electrodes. RFBs have two half-cells, separated by a membrane as de-picted in Figure 1.7. The active material is typically a liquid containing redox active mediatorssuch as the vanadium (V2+/V3+), or bromide/polysulfide couples that can be pumped throughthe cell. This segmentation results in an energy capacity, independent of the power generatorsize, which gives more freedom in the selection of the electrocatalyst and active materials. In-creased electrode area provides more power, and increased storage volume enables a highercapacity. Porous electrodes are normally employed for charge extraction due to their perme-ability and high surface area. Increasing the size of reservoirs can allow the device to reachthe theoretical capacity of its active material as the mass and volume of other parts becomesinsignificant compared to that of active material. Currently installed RFB systems range from102 up to 107 Wh capacity which is one order of magnitude larger than any Li-ion or lead acidbattery system [13]. This size range is appropriate for industrial and grid scale applications.This work presents a scalable approach towards the integration by utilizing RFB technologyas the energy storage method while keeping the higher solar energy yield approach of sharedactive materials between solar cells and batteries (the second group explained above).A cost analysis of single-cell solar batteries based on DSSC and RFB technologies com-pared to separate solar panel/Li-ion battery systems, shows a 30 % lower cost for the integratedsystem in large scale energy conversion and storage facilities. The calculations are summarizedin Appendix C. The saving is expected to increase as the storage technology improves and itsprice drops. This way, the share of solar panels in the total cost increases and a saving in thatsegment is heightened.The only integration of a dye-sensitized electrode in a RFB is reported by Yan et al. usingaqueous Li2WO4 or quinoxaline as negative and LiI in organic electrolyte as positive half-cell electrolytes (in 2013 and 2014 [59, 60]). The schematic of such a device is depicted inFigure 3.2. This results in a DSSC-like half cell in a redox battery. During illumination and45photo-charging, Figure 3.2a, the sensitized electrode is connected to the reduction electrodein the other half cell via a switch, causing the oxidation of ions in the DSSC and reductionin the second half cell. A charge balancing ion (shown as ⊕) balances the charge across thetwo half cells by transport through the membrane. The stored energy can later be extractedthrough the two non-sensitized electrodes after changing the switch to discharge mode (Fig-ure 3.2b). While the theoretical capacity of Yan’s system is claimed to be 5.359 Ah L−1, theyonly observed a charge density of 32.9 mAh L−1 and an energy density per electrode unit areaof ∼ 3.5 µWcm−2. They attributed this marginal utilization to several factors, including partialredox reactions in the working regime of the device, large internal resistance and high polar-ization losses at the electrodes. All these, indicate that major device optimization is requiredto improve the performance of solar redox batteries.In this work, a solar redox battery with the highest reported charge density and energy ca-pacity is independently developed as a solution for the solar-battery integration challenge basedon iodide/polysulfide redox couples using electro-catalytic porous electrodes. The charge den-sity of the proposed device is measured to be 1.2 Ah L−1 which is 35 times larger than that ofYan’s work. An energy density per electrode area of 52 µWhcm−2 is achieved which is a 13times improvement over the previous work on solar redox batteries.The proposed device is demonstrated by showing the photocharge/discharge response of thecell. It involves identifying the proper redox couples and optimizing the physical parametersfor efficient performance as both a solar cell and a battery. The complete cell is assembled in asandwich structure and tested for efficiency, cycle life and energy capacity.3.2 DesignIn this section, the requirements for an efficient solar redox battery are discussed and mate-rials are chosen for each part of the cell. The theoretical capacity of the proposed device isinvestigated and compared with the experimental data.Redox couples and the electrodes. — The positive half-cell electrolyte should have max-46Solar CellPoroussensitized  electrodeOp Rp On RnCharge⊕I(a) PhotochargingLoadBatteryOp Rp On RnDischarge⊕I(b) DischargingLoadPoroussensitized  electrodeOn Rn Op RpChargeDischarge⊕I(c) Complete solar redox batteryFigure 3.2: Schematic of a solar chargeable redox battery. Solid arrows show the electronflow direction during photocharging and dashed arrows represent the discharge.The inactive parts in each scenario are faded.(a) Photocharging. The sensitizedelectrode is connected to the charge extraction electrode in the other half cell. (b)Discharging though two porous electrodes. (c) Complete schematic of the cell.47imum compatibility with the sensitized electrode for solar energy conversion purposes. A re-view of the DSSC literature shows that only very few redox couples can be effectively used asmediators in a DSSC structure. In order to achieve a highly efficient DSSC, the redox mediatorshould have an electrochemical potential very close to the HOMO of the sensitizing dye, slowkinetics of electron transfer from TiO2, a fast kinetic rate with the current collector counterelectrode, a high diffusion coefficient to minimize mass transport losses, low light absorptionin order not to compete with the dye and good stability in the cell and in contact with the metalelectrodes. Energy storage additionally requires the redox couple to be highly soluble in orderto allow for high energy capacity, and be compatible with ion selective membranes to preventleakage to the other half cell.The iodide/triiodide couple is chosen as the active material of the positive electrolyte. It hasbeen proven to be one of the very few effective mediators for DSSCs [61]. It also has highersolubility, and thus higher capacity, compared to cobalt based mediators, which are the otherpopular redox group used in DSSC fabrication.The anodic redox couple should have the last two properties of the first mediator with astandard potential close to the TiO2 conduction band. It is reasonable to look at the redox cou-ples that have been shown to be practical in RFBs as they already satisfy most of the mentionedprerequisites. Some of these redox couples with feasible redox potentials are depicted in Fig-ure 3.3. Among these, polysulfide (S2–4 /S2–2 ) is selected because of its standard potential whichis close to the conduction band edge in TiO2, its fast electron transfer rate, high solubility, itshigh stability and the fact that cross contamination can be avoided by using a cationic exchangemembrane (CEM) if coupled with a (I–3/I–) half cell. In fact, polysulfide has been widely usedtogether with bromine, an element right above iodine in the periodic table, as a high energydensity RFB [62].While Pt is the default catalyst for I–3/I– reaction, finding a current collecting electrode tointeract with PS needs further experiments and comparison of the materials activity towards thePS redox couples. The activity is measured using a potentiodynamic sweep around the redox48U0. DV/NHEyLoad MV2u/MVuiDU0.446iVyCr3u/Cr2uiDU0.407iVyV3u/V2uiDU0.26iVyFe2u/FeiDU0.3iVyiDplatingyCo3u/Co2uiD0.5iVyI3U/IUiD0.29iVyS42U/S22UiDU0.415iVyTiO2 Dye Op/Rp On/RnCurrentcollectingelectrode CurrentcollectingelectrodeEc LUMOHOMOFigure 3.3: Energy diagram of Redox couples for the solar redox battery (SRB). Theenergy levels of the TiO2 electrode, dye (HOMO and LUMO), and well establishedDSSC mediators are shown on the left hand side of the cell. On the right hand sideof the membrane (vertical dashed line), candidate redox couples for completing thecell are shown with their respective standard potentials.potential of the respective redox couple. It is assumed that in each experiment the electrode’scurrent is Faradaic and is only sourced from the reaction with the active species. A reasonabledesign goal for energy conversion and storage applications is an electrode with current densitieson the order of 10 mAcm−2 and an overpotential smaller than 100 mV [63]. The design of thiswork also requires the electrode to be permeable to the electrolyte. A literature survey showsthat polysulfide electrolytes have been a candidate for several electrochemical cells such asthe Na-S batteries [64], bromine-polysulfide redox flow batteries [62], semiconductor/liquidjunction solar cells [65] and lithium-dissolved polysulfide batteries [66]. Materials such asMoS2, Ni, Co, NiS, PbS [67], WC[68], WS2[69] and a variety of carbons [70] have beensuggested as highly catalytic electrodes for polysulfide. Based on their reported performance,it was decided to investigate Ni, NiS and Pt electrodes at the anode. The activities of theseelectrodes are investigated and depicted in Figure 3.4.The highest activity is observed from the acid treated, sulfurated nickel foam electrode. Aset of experiments is performed to maximize the electrocatalytic activity of this electrode bychanging acid and polysulfide treatment times. The Tafel parameters of these electrodes are49-0.3-0.2- (V) l og I  (A/ cm2)  Pt MeshNi MeshNi Mesh-SNiF NiF-SNiF-A/SFigure 3.4: Tafel plots of several electrodes towards the polysulfide redox couples. Theelectrolyte is 1 M Na2S4, 1 M NaClO4 in AcN/THF (2:1 v/v). The scan rate is10 mVs−1. During solar charging a current of 10 mAcm−2 must be maintainedwith minimal potential drop. Nickel Foam (NiF) is seen to enable good currentdensity with relatively small overpotential (< 200 mV).extracted by fitting the polarization data to the exponential Tafel behavior of the Butler-Volmermodel, i = i0× 10±η/b, where i is the electrode current, i0 is the exchange current density,η is the overpotential and b is the Tafel slope that could be different for anodic and cathodicbranches. Average anodic and cathodic exchange current densities for Ni foam samples withdifferent treatment times of acid and polysulfide are plotted in Figure 3.5a. The activity isincreased with polysulfide treatment time up to 120 min and reaches a plateau around thattime. It is also shown in 3.5a that long acid exposure time has an adverse effect on exchangecurrent densities, mainly due to the corrosion of the porous metal.The Tafel parameters of the Ni foam electrode with the best performance are comparedto fresh Ni foam and listed in Table 3.1. With exchange current density of ∼ 4 mAcm−2, this5015 30 60 120 7200.0001110Treatmentgtimegingpolysulfideg(min)Exchangegcurrentgdensityg(mAgcm-2 )Nogacidgtreatment4gmingacidgtreatment10gmingacidgtreatment(a)-0.2-0.15-0.1-0.0500. (V) log I (A/ cm2)  Fresh Ni Foam4 min acid/ 2 hr PS(b)Figure 3.5: (a) Average anodic and cathodic exchange current densities for Ni foam sam-ples with different treatment times of acid and polysulfide.(b) Tafel plots of freshvs. treated Ni foam electrodes. The electrolyte is 1 M Na2S4, 1 M NaClO4 inAcN/THF (2:1 v/v). The scan rate is 1 mVs−1.51electrode is sufficient to handle the photo-generated current with insignificant polarization loss.At 100 mV overpotential, the activated electrode drives 17 mAcm−2 and 14 mAcm−2 anodicand cathodic currents respectively.Table 3.1: Polarization of Ni foam electrodes in polysulfide electrolyte.Electrode Anodic Cathodici0 (mAcm−2) b(mVdecade−1) b(mVdecade−1)Ni Foam, fresh 0.62 110 188Ni Foam, acid/PS 4 155 185Electrolyte solvent. — Another component that should be mutually compatible between theDSSC and the battery is the electrolyte solvent. Aqueous electrolytes are most commonly usedfor redox flow batteries and the use of non-aqueous electrolytes is normally limited to highvoltage batteries where water electrolysis could disrupt device performance. However, waterbased DSSCs are not very promising at the moment because of dye detachment, formation ofiodate, and decrease in lifetime of the photoexcited electrons [44]. Low viscosity, polar organicsolvents such as acetonitrile (AcN), ethylene carbonate (EC), propylene carbonate (PC) andmethoxypropionitrile (MPN) are typically used for ruthenium dye based DSSCs [61]. How-ever, polysulfide solubility in PC, EC and MPN is reported to be less than 0.1 M [71] which isnot practical for a redox battery because it results in a very low energy density. First, we testedAcN as solvent of an iodine/polysulfide redox battery but it showed very poor rechargeabil-ity characteristics. The reason is found to be the limited solubility of small chain polysulfideanions in AcN. Other than water, tetrahydrofuran (THF) [66] , dimethoxy ethane (DME) and1,3-dioxolane (DOL) (1:1 v/v) [72, 73] are reported to have high solubilities of polysulfides (∼10 M). Although none of them has directly been utilized in a DSSC device, a mixture of THFin AcN is reported to have a positive effect on DSSC performance at up to a 30% volumetricratio of THF in AcN [74]. An AcN/THF (2:1 v/v) mixture is used in this work to simultane-ously address the high redox solubility need of a redox battery and low viscosity, high polarityrequirements of DSSCs.52Considering all the design requirements, a material set suitable for an integrated DSSC-RFB are selected and listed in Table 3.2. The net reaction from this battery is representedby3NaI+Na2S4charge−−−−−⇀↽ −dischargeNaI3 +2Na2S2, (3.1)which leads to a cell open circuit voltage close to the difference in electrochemical potentialsof S2–4 /S2–2 (−0.415 V vs. NHE [75]) and I–3/I–(0.29 V vs. NHE [76]), ca. 0.7 V .Theoretical charge density of a battery with these components, is estimated asQV = n×m × F ,where n is the number of electrons exchanged in the charge/discharge process, m is the molarconcentration of the redox and F is the Faraday constant which is equal to 26.8 Ahmol−1.This capacity is calculated for both half cell and the smaller value is used as the limitingcapacity. Cathodic and anodic charge densities are calculated based on 1 M solutions of iodideand polysulfide with nI = 2/3 and nPS = 2. The capacity of iodide and polysulfide half cellsare 17.8 AhL−1 and 53.6 AhL−1 respectively. Therefore, the theoretical capacity of the I-PSredox battery is ∼12.68 AhL−1 after consideration of both half cells.Performance is first investigated in an iodide-polysulfide redox battery structure without theinclusion of photocharging. Upon verification of reversible electrochemistry and the stabilityof electrolytes, the complete system is assembled and characterized. The results of these testsare presented in the next section. Details of the experimental procedures follow in section 3.5.Table 3.2: List of materials for an efficient, high energy density solar redox battery.Device part MaterialPhotoelectrode 15 µm TiO2 sensitized with N719 dyes.Cathode current collecting electrode Pt MeshPositive electrolyte 1 M I2, 0.1 M NaI, 1 M NaClO4 in AcN/THF (2:1 v/v)Anode current collecting electrode Ni Foam, acid/PS pre-treatedNegative electrolyte 1 M Na2S, 3 M S, 1 M NaClO4 in AcN/THF (2:1 v/v)Membrane CMI-7000533.3 ResultsHigh energy density and Coulombic efficiency combined with device stability and cycle lifeare the prerequisites of any battery. A solar-battery should also have high energy conversionefficiency. These characteristics are first investigated in individual battery and solar cells andthen in the solar redox battery.Iodide-Polysulfide redox battery performance. — A redox battery is fabricated and testedusing the materials listed in Table 3.2 in an H-type cell. The cell is charged/discharged (par-tially) for 10 cycles at a constant current density of 0.1 mAcm−2 to analyze the cycle life, asshown in Figures 3.6 and 3.7. A discharge voltage of 0.51 V and a charging voltage of 0.69 Vare measured for the battery. The 0.18 V voltage difference is due to ohmic losses in the mem-brane and polarization losses on the electrodes.The discharge capacity of the cell discharged to 0.1 V is 0.15 AhL−1 after the 10th cycle,which is only 1.2 % of the theoretical capacity. This low capacity can likely be attributed tothe limited contact of the electrodes with the bulk electrolyte in the non-flow structure of theexperiment, so that much of the electrolyte is left unaffected by the discharge, as ions have notreached the electrode. This will be improved in the combined solar/battery where the spacingbetween electrodes is small. Initial energy efficiency of 75% is observed during these partialcycles, which drops to 67% after the 10th cycle as depicted in Figure 3.7b. Considering thelong cycle times (2 h / cycle) and open configuration of this battery, a gradual deterioration ofthe performance is expected as the volatile solvent evaporates.DSSC performance. — The DSSC is tested using the positive electrolyte prescribed for thebattery. The electrode separation is set to 0.3 mm to mimic that of the solar-battery structure.The IV characteristics and short circuit current are depicted in Figure 3.8. The drop in shortcircuit current from 7 mAcm−2 to 4 mAcm−2 over the 1000 s of this experiment, and the re-gaining of some of the current after resting for another 1000 s, suggest a mass transfer limitedcurrent. The bulk mass transport phenomenon has been previously studied in DSSCs and a540 1 2 3 4 5 600. 3.6: Charge/discharge characteristics of the iodide-polysulfide redox battery us-ing parameters of Table 3.2. The charge and discharge are at 0.1 mAcm2 constantcurrent.diffusion limited current (Ilim,b) is derived as a function of device parameters [77],Ilim,b = Ilim1+ bl εp1+3εp bl +32(bl )2, (3.2)where Ilim is the maximum photocurrent from cell, b is the bulk length, and l and εp are thethickness and the porosity of the porous TiO2 layer. To minimize the mass transfer limitations,the separation length is traditionally limited to approximately 40 µm, divided equally betweenthe nanoporous photoanode and the bulk electrolyte. The current design of this work requiresa much larger separation of electrodes to allow the storage of electrolytes and the membrane.The experiments have been performed with a bulk electrolyte that is 300 µm thick and a porouslayer that is 15 µm thick. These values predict a current that is only ∼ 18% of the maximum5501234500. (h)C el l  Potentia l  (V )(a)12345678910020406080100Cycle NumberE nergy  E f f iciency  (% )(b)Figure 3.7: (a)Complete discharge and (b) energy efficiency characteristics of the iodide-polysulfide redox battery using parameters of Table 3.2. The charge and dischargeare at 0.1 mAcm2 constant current.56value. This could be a limiting factor in the operation of the static SRB. However, after theimplementation of the flow SRB, the current will no be diffusion limited and hence the drop ofFigure 3.8a will be avoided.The efficiency of the DSSC is measured to be ∼ 1.7%. In future smaller spacings en-abled by fabrication processes operating with higher tolerances should boost the efficiency byallowing a thinner bulk thickness.Solar redox battery charge/discharge. — The solar rechargeable redox flow battery is as-sembled in a sandwich structure as shown in Figure 3.9 using the components demonstratedin the previous sections. The light window and the spacers openings are set to 1 cm2 and thethickness of the spacers is set to be 300 µm.Photocharge/discharge characteristics of the SRB are shown in Figure 3.10a. The depen-dency of the cell potential to the discharge time can be explained by Nernst equation where theequilibrium voltage changes proportional to the logarithm of the ratio between oxidized andreduced species. The open circuit potential of the cell at each charge state should follow,Eeq = ∆E0−RT2Fln[S2−4][S2−2]2[I−]3[I−3] = ∆E0− RT2Fln27(1−SOC/3)(1−SOC)32(SOC)3, (3.3)where ∆E0 is the difference in the standard potentials of the two redox couples and the log-arithmic term reflects the effect of ion concentration. SOC is the state of the charge and isdefined as the portion of iodide ions that are in the oxidized state, SOC = 3[I−3 ]/[Itotal]. For thenon-equilibrium potential of the cell, the current dependent ohmic (I Rohmic) and kinetic losses(η) should be considered as well, therefore E = Eeq−η− I(Rohmic). At the onset of discharge,minute 10, the cell voltage drops instantly for 200 mV because of ohmic losses in the mem-brane and the electrodes. Throughout the discharge, as the SOC changes from 0.75 to 0.25,the concentration dependent term of the voltage only decreases by 109 mV. This describes therelatively flat region in the cell voltage response. The fast drop at the end of the discharge oc-curs because of the large ratio of discharged species to the charged ones in the electrolyte as is57LightON LightONLightOFFIlim,b(a) Short circuit current0 0.1 0.2 0.3 0.4 0.5 0.6 0.700.511.522.533.544.555.566.57Current8density(mA/cm2)Voltage8(V) )(b) I-V CharacteristicsFigure 3.8: IV characteristics and short circuit current of DSSC using the solar-batteryelectrolyte (1 M I2, 0.1 M NaI, 1 M NaClO4 in AcN/THF (2:1 v/v)). The illumi-nation source is a 150 W Xenon lamp coupled with an AM 1.5D filter at powerintensity of 100 mWcm−2.58Figure 3.9: Solar redox battery sandwich structureobserved in equation 3.3. As the SOC approaches zero, the term inside the logarithm increasesproportional to 1/(SOC)3 which translates to a fast increase in the logarithm term.At this stage, unlike a RFB, the cell is not connected to external reservoirs. Therefore, thecapacity of the battery is limited to the amount of active material in the thin half cells. Thefull discharge revealed a 1.2 Ah L−1 charge capacity which is 35 times larger than that of theprevious work on solar redox flow battery by Yan et al. [60]. An energy density per electrodearea of 52 µWhcm−2 is achieved which is a 13 times improvement over Yan’s work. It shouldbe noted that the theoretical capacity of Yan’s solar battery is only 1 Ah L−1 based on singleelectron transfer of quinoxaline and 0.1 M concentrations of iodide and quinoxaline that havebeen used in that cell2. Although the high solubility of quinoxaline should allow a highercapacity in a more concentrated cell, even with the current target value the measured capacityis still lower than expected. This could be because of the low surface area of the electrodes (Ptcoated Ti mesh) and the incomplete electrode-electrolyte interactions.A comparison of the results of the SRB of this work with the previous DSSC-based solarbattery devices is presented in Table 3.3. It is noteworthy that even without consideration of theflow, this work presents the highest areal energy density among solar batteries. Figure 3.10bshows the cell energy efficiency, which started at more than 98% and gradually dropped to2This value is not directly reported in [60] and is calculated based on the information given in that publication.5978% after 10 cycles mainly because of solvent loss. This should be avoided in a sealed deviceusing common approaches in the redox flow industry. For separate PV panels and batteries,the electrical energy storage efficiency is ∼ 40-56 % for AC output and ∼ 49-62 % for DCoutput power [78]. Comparison of the latter value with the efficiency of this work supportsour initial expectation about reduction of energy conversion losses in an integrated solar cell-battery device.Table 3.3: Performance comparison of DSSC-based solar batteries.QV a(mAh L−1)EAb(µWhcm−2)QAc(µAhcm−2)V d(V)ηe(%)Roundtripefficiency( %)Coulombicefficiency( %)Ref9.3 76.38 0.45 80 [53]47 130.5 0.75 42 [54]31 149 0.7 36 [55]2.68 10.5 0.55 3.21 30 [56]0.1 0.3 0.68 0.1 [79]21 54 0.75 6.10 84 ∼100 [57]32.9 3.5 8.3 0.6 [60]1.2×103 52 99 0.6 1.7 78 ∼ 100 Thisworka Volumetric discharge capacity.b Areal energy density.c Areal discharge density.d Initial discharge voltage.e Photo-conversion efficiency.3.4 ConclusionA solar rechargeable redox battery is demonstrated. A set of materials is proposed for achiev-ing an efficient device and the device is demonstrated. It shows the highest performance yetachieved in any integrated DSSC-based energy storage device. Significant improvement in theareal and volumetric capacity is observed compared to the only other combined solar recharge-able redox battery demonstrated to date. The tests suggest the cell-thickness dependency ofDSSC performance as the main factor limiting performance. This effect will be greatly depre-ciated in a flow cell when the supply of iodide to the photoelectrode is provided by convection600 5 10 15 20 25 30 35 4000. LightONLightOFF LightOFFPhotochargeDischarge Discharge(a)246810020406080100Cycle NumberE nergy  E f f iciency  (% )(b)Figure 3.10: (a)Photocharge/ Discharge of the solar redox battery. (b)Energy efficiencyduring charge/discharge cycles. For charging the photoelectrode and anode aredirectly connected and the discharging is through a 1 kΩ resistor between the twodischarge electrodes.61from external sources rather than diffusion. Further improvement of the photoelectrode, mem-brane and current collecting electrodes should result in more efficient performance.Suggested future work is described in Chapter 4, following is a detailed description of theexperimental procedure.3.5 ExperimentalElectrochemical cell preparations.— Sodium sulfide nonahydrate,( Na2S ·9 H2O, 98.0 %), sul-fur, sublimed( S, 99.5.0%), iodine, resublimed (I2, 99.5.0%), sodium iodide (NaI), sodiumhydroxide pellets (NaOH, 97%) are purchased from Fisher Scientific. Sodium perchlorate(NaClO4, 98%), acetonitrile anhydrous (CH3CN, 99.8%) are supplied from Sigma Aldrich.Tetrahydrofuran (C4H8O, 99.99%) is obtained from EMD Millipore Chemicals. Nickel Foam(Ni ) is purchased readily from MTI Corporation. Ion exchange membrane of CMI-7000 issupplied by Membrane International.The conductive glass of the sensitized electrode is fluorine doped tin oxide coated withresistance of 8Ω/sq and is purchased from Solaronix. The sensitizing dye, Ruthenizer 535bis-TBA dye (N719) is purchased from Dyesol.The electrochemical glassware is washed with 2 % Extran 3000 in water solution followedsonication in ethyl alcohol and acetone. The aqueous experiments are performed using deion-ized (DI) water, purified with a Milipore water system.The DSSC/positive half-cell electrolyte is prepared by dissolving appropriate weights ofmaterials for a 1 M NaI, 0.1 M I2 and 1 M NaClO4 solution in AcN/THF (2:1 v/v). Themixture is stirred for 2 h.In order to prepare the negative electrolyte, first 3 M sodium sulfide is dissolved in THF bystirring and warming the solution to 40 ◦C. Sulfur is then added in three times the concentrationof Na2S. The mixture is heated to 50◦C and stirred at 400 RPM for 2 h before it changes colorfrom a light yellow to brownish red. The color change process of polysulfide solution is shownin Figure 3.11. After a uniform solution is achieved, additional solvent is added to maintain 362(a) (b) (c) (d) (e)Figure 3.11: Color change of sodium sulfide as it changes from colorless Na2S (a) to darkred Na2S4 (e).(a) (b)Figure 3.12: SEM images of the nickel foam before (a), and after treatment with acid andpolysulfide (b).M solution on Na2S4. Then (NaClO4) is added in 3 M concentration. Finally, the solution isdiluted with AcN to produce a 1 M solution of Na2S4 in AcN/THF (2:1 v/v).The pre-treated Ni foam electrode are prepared by first immersing a piece of Ni foam in aclean (1:1 v/v) sulfuric acid/water solution for 4 min. After rinsing with DI water, the electrodeis boiled in DI water for 1 min and then is boiled in 1 M aqueous solution of Na2S4 (preparedsimilar to the negative half-cell electrolyte) for 2 h. The electrodes are boiled in DI water foranother 1 min and then dried under vacuum at 60 ◦C. SEM images of the fresh and activatedNi foam are shown in Figure 3.12. NiS layer is shown as clusters on the smooth surface of Ni.In order to adjust the membrane for the non-aqueous solvent, it is immersed in 1 M Na2ClO4in AcN/THF (2:1 v/v). The container is then held under vacuum to enforce the substitution ofair with the solvent. The membranes is then stored in the same electrolyte for 24 h. An EISstudy of the membranes showed an ionic resistance of 240Ω/cm2. Another CEM from Fu-63matech GmbH (FKB PK130) is also tested because of previous reports on its application innon-aqueous RFBs [80]. The latter, however, showed a much higher ionic resistance in thesolvent in this work (AcN/THF (2:1 v/v)) compared to CMI-7000 which is used in these ex-periments.For the photoelectrode, the technique explained by Ito et al. [81] is followed for most ofthe electrode preparation. First, two holes are drilled in FTO coated glass electrodes usingtwo sacrificial glass layers on top and bottom to avoid chipping the glass surface during thedrilling. Then the FTO coated glass electrodes are cleaned in a sonicator bath using a 2 %Extran 3000 solution in DI water. Then the electrodes are rinsed with ethanol and DI waterand treated under a UV lamp (Mineralight UVS-54,UVP, LLC) for 18 min. A thin layer ofTiO2 is deposited on the FTO surface by immersing the electrodes in 40 mM aqueous solutionof TiCl4 kept at 70◦C and stirred on a hot plate. After 30 min, 15 µm of TiO2 paste is doctorbladed over the FTO coated glass followed by an ethanol-vapor bath treatment for 3 min. Theelectrodes are slowly heated to 400 ◦C on a hot plate at 2.5 ◦C/min until became transparent.After another TiCl4 treatment for 30 min, the electrodes are heated again at 400◦C for 30 min.The cooled electrodes are sensitized in 0.2 mM solution of N719 in ethanol for 24 h.Electrochemical and Photoelectrochemical measurements. — All the cyclic voltametriesand I-V characteristics are carried out using a PGSTAT101 Autolab potentiostat in a 2 elec-trode setup. The redox battery is tested in a H-type cell as depicted in Figure 3.13. The SRBis tested in the sandwich cell of Figure 3.9. For charging, the photoelectrode and anode aredirectly connected while the discharging happened through a 1 kΩ resistor between two dis-charge electrodes.The light source is a Newport 150 W Xenon arc lamp coupled to a filter (AM 1.5D), drivenby a Newport 69907 power supply (Newport Inc., Irvine CA). The output power is measuredusing a Newport 818-SL photodetctor and is adjusted at 100 mWcm−2. The setup for photo-electrochemical experiments is shown in Figure 3.14.64Load1 2 34 567 8(a) Schematic of the H-type cell (b) Image of the redox batteryFigure 3.13: Schematic (a) and the image (b) of the H-type cell used for the battery test.The parts are marked as: body(1), half cells (2,3), Pt mesh cathode (4), Ni foamanode (5), silicone sealant (6), ion exchange membrane (7) and the screw hous-ing(8).(a) (b)Figure 3.14: Images of the (a) light source and (b) the potentiostat setup.65Chapter 4Conclusions and Future WorkThe main conclusions and contributions derived from this work are summarized in this chapter.The approaches to explore the presented devices further are introduced next.4.1 ConclusionsWidespread application of renewable energy sources faces challenges including the high costof external storage systems needed to for grid stability. Investigation of new materials formore efficient, less expensive solar cells and batteries is an important avenue to overcomethe storage challenge. Optimization, redesign and reconfiguration of existing technologies isanother approach. This work follows the latter path, demonstrating two novel configurationsof solar energy conversion and storage devices.The vertical photogalvanic cells (VPGCs) developed in the course of this thesis work over-come challenges associated with traditional photogalvanic cells. In these devices improvementin performance is suggested by having light travel into the device parallel to electrodes ratherthan perpendicular (and through a transparent electrode). This has several advantages includ-ing higher efficiency and device compatibility with less soluble dyes and electrodes with slowerkinetics. They may be effective as cheap, low maintenance solar cells upon realization of thenow-less-demanding electrode-dye-mediator-electrode combination. It is unlikely though, that66the storage promise of photogalvanic cells will be fulfilled due to the high bulk recombinationtendency of energy elevated species.The integrated solar redox battery (SRB) of this work shows that higher solar energy storageyields are achievable with a combined device compared to a system composed of separate solarcells and batteries. Novelty is in the structure, with a redox battery combined with a dye-sensitized solar cell. Although another example of such an integrated cell was reported duringthe course of this thesis work, it’s performance is limited. Here, with careful considerationof the electrodes and redox couples, we are able to show a capacity of 1.2 Ah L−1. For thenon-flow system of this work, the integration does not severely affect the storage capacity ofthe battery but the photo conversion performance is inferior compared to a standalone dyesensitized solar cell. This difficulty is rooted in diffusion of the species in the electrolyte andmay be overcome by the convection in a flow configuration.Having a grid scale facility of solar redox battery causes important safety concerns. Thelarge reservoirs of organic solvents are health hazards since acetonitrile and tetrahydrofuranhave health hazard rating of 2 (could cause temporary incapacitation). Both have a flammabilityrating of 3 and can be ignited even at low temperatures. These issues mean the assemblyof these devices is more costly than aqueous RFBs because of more expensive fabricationprocedure and more costly packaging materials. Research in the aqueous DSSCs is constantlyimproving the performance of such devices [82–86]. A safer solar-redox battery could useaqueous solvents in both half cells - upon the realization of the water-based solar cells.Another difficulty in the development of SRBs is maintaining the flow over a large numberof solar-batteries. These devices should be installed in a solar farm and all the electronic andfluidic parts should be connected. It is unlikely that the electrolyte flow could be maintainedin hundreds of panel using a centralized pumping station. Therefore a distributed pumpingsystem should be designed which further increases the cost of SRBs compared to RFBs.The advantages of the integration, namely saving on the amount of cell material and fabri-cation steps as well as higher storage yield, should be compared to the limitations due to more67complex design of the whole system and potential safety concerns before this technology isconsidered for commercial applications.4.2 ContributionsIn Chapter 2, a vertical configuration for photogalvanic cells (PGCs) is proposed and analyzedfor the first time. In the new structure light absorption and hence electron generation is spreadthrough the depth of the device. All the incoming photons can be captured effectively even withlow solubility dyes by adjusting the depth according to the dye concentration. Because of thedistributed electron generation, unreasonably fast electrode kinetics are no longer required. Tobe efficient, the new structure requires 100 times slower electrode kinetics than conventionalPGC designs. The optimum dye concentration in conventional PGCs is ∼100 mM which isrelaxed by 3 orders of magnitudes in the proposed design. Therefore two major challenges infabrication of efficient PGCs as cheap, low maintenance solar cells are removed.The effect of device bulk recombination rate, electrode kinetic rate constant, electrode se-lectivity and device geometry are studied for the vertical photogalvanic cells and optimumvalues are listed as a guideline for future experimental work.Chapter 3 describes the work on solar cell-redox flow battery integration in a solar redoxbattery (SRB) device. The aim of the system is ultimately to offer savings on materials, fab-rication and installation costs compared to two separate devices. Most solar batteries to dateemploy solid electrodes, while in the SRB the storage capacity can be scaled by increasingthe size of the electrolyte reservoir. The integration of a combination of a dye-sensitized solarcell and a redox flow battery was recently reported for the first time by Yan et al. [59, 60],using lithium tungsten oxide (Li2WO4) and then quinoxaline electrolytes. The work of thisthesis demonstrates dramatically improved cell performance through the careful considerationand selection of electrolytes, electrodes and cell geometry. A material set for the solar batterysystem is proposed that results in the highest storage and efficiency reported for a solar-battery.The charge density of the proposed solar battery is measured to be 1.2 Ah L−1 which is 3568times larger than the work by Yan et al. [60]. The areal energy density per electrode area of47 µWhcm−2 is achieved which is a 13 times improvement over the previous work. The reasonfor this improvement is the combination of higher theoretical capacity of the electrolyte andhigher kinetic rates of electrodes. These properties offer hope for scalability of the technology,as discussed in future work section below.Even without the consideration of scalability offered by using flow (which still remains tobe implemented) and only comparing the capacity of the conversion cell, this work still presentsthe highest areal energy density among solar batteries, exceeding the best performance to date,reported by Murakamiet et al. [54].4.3 Future WorkThe vertical photogalvanic cell proposed in this work removed several obstacles from realiza-tion of an efficient PGC, namely, the requirement of very fast electrode kinetics and highlysoluble dyes. Work still needs to be done to satisfy other unfulfilled aspects, particularly theselective electrode. Initial research on semiconductor electrochemistry showed their poten-tial for this role as briefly explained in Appendix B [49, 87]. However, the behavior of thesemiconductor-liquid interface varies significantly with crystallinity, doping and surface finish.One aspect to expand this work in to explore a large variety of wide band gap semiconductorsfor employment in VPGCs.So far the best selectivity achieved using semiconducting electrodes is 1000 [49], andsimilar values are obtained using selective self-assembled monolayers1. This selectivity needsto be increased by a factor of 1000 to make these cells truly effective.The effect of optical properties of the electrodes on the performance of the device couldbe studied as an extension to the current model. While a reflective surface would not alterthe results significantly, the absorbing properties of the semiconducting electrodes result in amixed photoelectrochemical-photogalvanic mechanism for electricity generation.1unpublished work by Dr. Joanna Slota-Newson69Although careful considerations are taken for the design of the solar redox battery of thiswork, there still exist some aspects that can benefit from more in depth analysis. One majorlimitation of DSSC-based solar batteries is the inverse relationship between electrode spacingand photo-conversion efficiency of the cell. Solar batteries with one solid state half cell solelyrely on the capacity of redox mediators in the photosensitive half cell and therefore requirelarge spacing between electrodes for any significant storage to be achieved. This bulk size canreduce the photo-current efficiency as explained by Papageorgiou et al. [77]. The mass transferlimited current of a DSSC results from the lag in back transfer of reduced mediators from thecounter electrode. However, the adverse effect could be much weaker than the prediction ofequation (3.2) due to the thermal convection in bulk of the device.The redox flow system of this work also requires some extra spacing compared to a DSSCto make room for electrolyte flow. Therefore, analysis of this mass transfer is vital for im-provement of these devices. In fact, the photo-charging operation of the solar redox battery isslightly different from the photo conversion of DSSCs and this limitation might not be directlyapplicable. In the solar redox battery the regeneration of the dyes happens as a result of theirreduction with always-freshly-provided mediators from external reservoirs and is not limitedto the diffusion of the recycled mediators from the counter electrode as in a DSSC. Therefore,it is believed that this limitation could be alleviated by a flow of the electrolyte tuned to theconsumption rate of redox species at the photo-electrode.A numerical analysis of the solar redox battery could clarify this issue. Such model couldbe implemented in COMSOL MULTIPHYSICS software as a follow up project of this work.The flow rate of the electrolytes should be adjusted with respect to the light intensity, sothat the redox species efficiently capture the photo-generated charges. This is similar to themaximum power point tracking (MPPT) system used in separate solar panel-battery systems.The optimum flow rate at each light intensity and the control mechanism of this rate can also bestudied in the complete cell model. The numerical analysis also gives guidelines for optimumcell geometry and electrolyte parameters.70HotColdLight ConvectionDSSCFigure 4.1: Utilization if thermal convection for electrolyte circulation in solar redox bat-teries.Using solar thermosiphon effect to enhance ion circulation is one way to increase the uti-lization of solar energy by reducing the need for pumping. This way thermal convection of theelectrolyte is used to flow the ions to the tanks as is depicted in Figure 4.1. Because of the lackof control over the flow rate, a lower efficiency is expected and this method should be limitedto a trimmed-down version of the device where simplicity has priority over performance. Thefeasibility and cost effectiveness of this method can also be evaluated through the numericalmodel.Development of organic materials as active redox couples could significantly add to theflexibility if the solar battery since normally fine tuning of electrochemical potentials is possiblethough the synthesis of organic materials. Such an approach has been shown very effectivein RFBs [88]. Therefore, synthesis and incorporation of organic redox couples for SRBs isanother aspect of this work that can be expanded with collaboration of a synthetic chemistrygroup.71The round-trip energy efficiency of the device could be further enhanced by reducing poten-tial drops in the device. One major source of such loss is the potential drop over the membrane.High ionic conductivity membranes should be developed based on the charge balancing ions,active species and the electrolyte solvent used in the design of SRB.Stability of dye-based devices is currently under research by several groups. While some ofthe most commonly used dyes like N749 and N719 lack long term stability under illuminationor high temperatures, there are other stable dyes such as Z907, K19 and K77 at 80 ◦C in darkand at 60 ◦C under illumination[89]. These properties should be investigated for the utilizeddyes in the electrolyte selected for the VPG or SRB devices. Such studies are vital before anymovement towards the commercialization.72Bibliography[1] Energy Technology Perspectives 2014: Harnessing Electricity’s Potential.International Energy Agency, May 2014. → pages 2[2] “Human development index (HDI),” UNITED NATIONS DEVELOPMENTPROGRAMME, Tech. Rep., May 2014. → pages 2[3] Key World Energy Statistics. International Energy Agency, 2014. → pages 2[4] T. R. Cook, D. K. Dogutan, S. Y. Reece, Y. Surendranath, T. S. Teets, and D. G. Nocera,“Solar energy supply and storage for the legacy and nonlegacy worlds,” ChemicalReviews, vol. 110, no. 11, pp. 6474–6502, Nov. 2010. → pages 2[5] N. S. Lewis, “Powering the planet,” MRS Bulletin, vol. 32, no. 10, pp. 808–820, 2007.→ pages 3, 42[6] C. Schroder, “Instability in power grid comes at high cost for german industry,” SpiegelOnline, Aug. 2012. http://www.spiegel.de/international/germany/instability-in-power-grid-comes-at-high-cost-for-german-industry-a-850419.html[Accessed:2015-02-01] → pages 3[7] P. Denholm and M. Hand, “Grid flexibility and storage required to achieve very highpenetration of variable renewable electricity,” Energy Policy, vol. 39, no. 3, pp.1817–1830, Mar. 2011. → pages 3[8] A. A. Akhil, G. Huff, A. B. Currier, B. C. Kaun, D. M. Rastler, S. B. Chen, A. L. Cotter,D. T. Bradshaw, and W. D. Gauntlett, “DOE/EPRI 2013 electricity storage handbook incollaboration with NRECA,” Sandia National Laboratories, SANDIA REPORT, Jul.2013. → pages 4, 90[9] B. Dunn, H. Kamath, and J.-M. Tarascon, “Electrical energy storage for the grid: Abattery of choices,” Science, vol. 334, no. 6058, pp. 928–935, Nov. 2011. → pages 4, 5[10] W. F. Pickard, N. J. Hansing, and A. Q. Shen, “Can large-scale advanced-adiabaticcompressed air energy storage be justified economically in an age of sustainableenergy?” Journal of Renewable and Sustainable Energy, vol. 1, no. 3, p. 033102, May2009. → pages 573[11] “World premiere: the first industrial-scale power-to-gas plant,” Oct. 2013.http://www.audi.com/com/brand/en/vorsprung durch technik/content/2013/10/energy-turnaround-in-the-tank.html[Accessed:2015-02-01] → pages 5[12] B. Zakeri and S. Syri, “Electrical energy storage systems: A comparative life cycle costanalysis,” Renewable and Sustainable Energy Reviews, vol. 42, pp. 569–596, Feb. 2015.→ pages 5, 13[13] P. Alotto, M. Guarnieri, and F. Moro, “Redox flow batteries for the storage of renewableenergy: A review,” Renewable and Sustainable Energy Reviews, vol. 29, pp. 325–335,Apr. 2014. → pages 5, 13, 45[14] A. Takshi, J. D. Madden, A. Mahmoudzadeh, R. Saer, and J. T. Beatty, “A photovoltaicdevice using an electrolyte containing photosynthetic reaction centers,” Energies, vol. 3,no. 11, pp. 1721–1727, Oct. 2010. → pages 6[15] A. Mahmoudzadeh, R. Saer, D. Jun, S. M. Mirvakili, A. Takshi, B. Iranpour, E. Ouellet,E. T. Lagally, J. D. W. Madden, and J. T. Beatty, “Photocurrent generation by directelectron transfer using photosynthetic reaction centres,” Smart Materials and Structures,vol. 20, p. 094019, Sep. 2011. → pages[16] S. M. Mirvakili, J. E. Slota, A. R. Usgaocar, A. Mahmoudzadeh, D. Jun, M. N.Mirvakili, J. T. Beatty, and J. D. W. Madden, “Photoactive electrodes incorporatingelectrosprayed bacterial reaction centers,” Advanced Functional Materials, vol. 24,no. 30, pp. 4789–4794, Aug. 2014. → pages 6[17] G. Feher, J. P. Allen, M. Y. Okamura, and D. C. Rees, “Structure and function ofbacterial photosynthetic reaction centres,” Nature, vol. 339, no. 6220, pp. 111–116, May1989. → pages 6[18] W. J. Albery and M. D. Archer, “Photogalvanic cells: Part 3. the maximum powerobtainable from a thin layer photogalvanic concentration cell with identical electrodes,”Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 86, no. 1,pp. 1–18, Jan. 1978. → pages 9, 20, 22, 25[19] M. K. Bhimwal and K. Gangotri, “A comparative study on the performance ofphotogalvanic cells with different photosensitizers for solar energy conversion andstorage: D-xylose-NaLS systems,” Energy, vol. 36, no. 2, pp. 1324–1331, Feb. 2011. →pages 9, 20, 34[20] K. R. Genwa, A. Kumar, and A. Sonel, “Photogalvanic solar energy conversion: Studywith photosensitizers toluidine blue and malachite green in presence of NaLS,” AppliedEnergy, vol. 86, no. 9, pp. 1431–1436, Sep. 2009. → pages 9[21] K. Genwa and M. Genwa, “Photogalvanic cell: A new approach for green andsustainable chemistry,” Solar Energy Materials and Solar Cells, vol. 92, no. 5, pp.522–529, May 2008. → pages 9, 2074[22] K. Gangotri and V. Indora, “Studies in the photogalvanic effect in mixed reductantssystem for solar energy conversion and storage: Dextrose and ethylenediaminetetraaceticacidAzur a system,” Solar Energy, vol. 84, no. 2, pp. 271–276, Feb. 2010. → pages 9, 20[23] V. Viswanathan, A. Crawford, D. Stephenson, S. Kim, W. Wang, B. Li, G. Coffey,E. Thomsen, G. Graff, P. Balducci, M. Kintner-Meyer, and V. Sprenkle, “Cost andperformance model for redox flow batteries,” Journal of Power Sources, vol. 247, pp.1040–1051, Feb. 2014. → pages 13[24] M. Gratzel, “Photoelectrochemical cells,” Nature, vol. 414, no. 6861, pp. 338–344, Nov.2001. → pages 14[25] H. Gerischer, M. Michel-Beyerle, F. Rebentrost, and H. Tributsch, “Sensitization ofcharge injection into semiconductors with large band gap,” Electrochimica Acta, vol. 13,no. 6, pp. 1509–1515, Jun. 1968. → pages 14, 20[26] B. O’Regan and M. Gratzel, “A low-cost, high-efficiency solar cell based ondye-sensitized colloidal TiO2 films,” Nature, vol. 353, no. 6346, pp. 737–740, Oct.1991. → pages 14, 20[27] N. R. Neale, N. Kopidakis, J. van de Lagemaat, M. Grtzel, and A. J. Frank, “Effect of acoadsorbent on the performance of dye-sensitized TiO2 solar cells: shielding versusband-edge movement,” The Journal of Physical Chemistry B, vol. 109, no. 49, pp.23 183–23 189, Nov. 2005. → pages 14[28] A. Z. Weber, M. M. Mench, J. P. Meyers, P. N. Ross, J. T. Gostick, and Q. Liu, “Redoxflow batteries: a review,” Journal of Applied Electrochemistry, vol. 41, no. 10, pp.1137–1164, Oct. 2011. → pages 15[29] E. K. Rideal and D. C. Williams, “Photogalvanic effect,” J. Chem. Soc, vol. 127, p. 258,1925. → pages 20[30] E. Rabinowitch, “The photogalvanic effect i. the photochemical properties of thethionine-iron system,” The Journal of Chemical Physics, vol. 8, p. 551, 1940. → pages20[31] R. Gomer, “Photogalvanic cells,” Electrochimica Acta, vol. 20, no. 1, pp. 13–20, Jan.1975. → pages 20, 83[32] W. John Albery and A. W. Foulds, “Photogalvanic cells,” Journal of Photochemistry,vol. 10, no. 1, pp. 41–57, 1979. → pages 21, 23, 24, 35, 83[33] T. Sakata, Y. Suda, J. Tanaka, and H. Tsubomura, “Photogalvanic effect in thethionine-iron system,” The Journal of Physical Chemistry, vol. 81, no. 6, pp. 537–542,Mar. 1977. → pages[34] Y. Suda, Y. Shimoura, T. Sakata, and H. Tsubomura, “Photogalvanic effect in thethionine-iron system at semiconductor electrodes,” The Journal of Physical Chemistry,vol. 82, no. 3, pp. 268–271, Feb. 1978. → pages75[35] M. I. C. Ferreira and A. Harriman, “Photoredox reactions of thionine,” Journal of theChemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,vol. 73, pp. 1085–1092, Jan. 1977. → pages 20[36] H. Gerischer, “Electrochemical behavior of semiconductors under illumination,” Journalof The Electrochemical Society, vol. 113, no. 11, pp. 1174–1182, Jan. 1966. → pages 20[37] A. J. Nozik, “Photoelectrochemistry: Applications to solar energy conversion,” AnnualReview of Physical Chemistry, vol. 29, no. 1, pp. 189–222, Oct. 1978. → pages 20[38] M. Gratzel, Energy Resources through Photochemistry and Catalysis. Elsevier, Oct.1983. → pages 20[39] E. J. J. Groenen, M. S. De Groot, R. De Ruiter, and N. De Wit, “Triton x-100 micelles inthe ferrous/thionine photogalvanic cell,” The Journal of Physical Chemistry, vol. 88,no. 7, pp. 1449–1454, Mar. 1984. → pages 20[40] K. Gangotri, O. P. Regar, C. Lal, P. Kalla, K. R. Genwa, and R. Meena, “Use oftergitol-7 in photogalvanic cell for solar energy conversion and storage: Toluidineblue-glucose system,” International Journal of Energy Research, vol. 20, no. 7, pp.581–585, 1996. → pages 20[41] K. M. Gangotri and O. P. Regar, “Use of azine dye as a photosensitizer in solar cells:different reductantssafranine systems,” International Journal of Energy Research,vol. 21, no. 14, pp. 1345–1350, 1997. → pages[42] K. M. Gangotri and C. Lal, “Studies in photogalvanic effect and mixed dyes system:EDTAmethylene blue + toluidine blue system,” International Journal of EnergyResearch, vol. 24, no. 4, pp. 365–371, 2000. → pages 20[43] K. R. Genwa and N. C. Khatri, “Comparative study of photosensitizing dyes inphotogalvanic cells for solar energy conversion and storage: Brij-35Diethylenetriaminepentaacetic acid (DTPA) system,” Energy & Fuels, vol. 23, no. 2, pp. 1024–1031, Feb.2009. → pages 20[44] C. Lal, “Use of mixed dyes in a photogalvanic cell for solar energy conversion andstorage: EDTAthionineazur-B system,” Journal of Power Sources, vol. 164, no. 2, pp.926–930, Feb. 2007. → pages 20, 52[45] R. A. Marcus, “On the theory of ElectronTransfer reactions. VI. unified treatment forhomogeneous and electrode reactions,” The Journal of Chemical Physics, vol. 43, no. 2,pp. 679–701, Jul. 1965. → pages 24, 34[46] W. J. Albery, “Development of photogalvanic cells for solar energy conservation,”Accounts of Chemical Research, vol. 15, no. 5, pp. 142–148, May 1982. → pages 24,34, 35[47] A. J. Bard and L. R. Faulkner, Electrochemical methods: fundamentals and applications.New York: Wiley, 2001. → pages 3376[48] W. John Albery and M. D. Archer, “Photogalvanic cells: Part 4. the maximum powerfrom a thin layer cell with differential electrode kinetics,” Journal of ElectroanalyticalChemistry and Interfacial Electrochemistry, vol. 86, no. 1, pp. 19–34, Jan. 1978. →pages 34[49] A. R. Usgaocar, L. Wang, A. Mahmoudzadeh, S. M. Mirvakili, J. E. Slota-Newson, J. D.Madden, J. T. Beatty, and A. Takshi, “Semiconductors as selective electrodes forbio-photovoltaic cells,” Meeting Abstracts, vol. MA2013-01, no. 4, pp. 282–282, Mar.2013. → pages 38, 69, 87[50] W. Wang, Q. Luo, B. Li, X. Wei, L. Li, and Z. Yang, “Recent progress in redox flowbattery research and development,” Advanced Functional Materials, vol. 23, no. 8, pp.970–986, Feb. 2013. → pages 42[51] G. Dennler, S. Bereznev, D. Fichou, K. Holl, D. Ilic, R. Koeppe, M. Krebs, A. Labouret,C. Lungenschmied, A. Marchenko, D. Meissner, E. Mellikov, J. Mot, A. Meyer,T. Meyer, H. Neugebauer, A. pik, N. S. Sariciftci, S. Taillemite, and T. Whrle, “Aself-rechargeable and flexible polymer solar battery,” Solar Energy, vol. 81, no. 8, pp.947–957, Aug. 2007. → pages 42[52] G. Wee, T. Salim, Y. M. Lam, S. G. Mhaisalkar, and M. Srinivasan, “Printablephoto-supercapacitor using single-walled carbon nanotubes,” Energy & EnvironmentalScience, vol. 4, no. 2, pp. 413–416, Feb. 2011. → pages 42[53] T. Miyasaka and T. N. Murakami, “The photocapacitor: An efficient self-chargingcapacitor for direct storage of solar energy,” Applied Physics Letters, vol. 85, no. 17, pp.3932–3934, Oct. 2004. → pages 42, 43, 44, 60[54] T. N. Murakami, N. Kawashima, and T. Miyasaka, “A high-voltage dye-sensitizedphotocapacitor of a three-electrode system,” Chemical Communications, no. 26, pp.3346–3348, Jun. 2005. → pages 42, 44, 60, 69[55] Y. Saito, S. Uchida, T. Kubo, and H. Segawa, “Energy-storable dye-sensitized solar cellswith tungsten oxide charge-storage electrode,” ECS Transactions, vol. 16, no. 25, pp.27–34, Mar. 2009. → pages 42, 43, 60[56] Y. Saito, A. Ogawa, S. Uchida, T. Kubo, and H. Segawa, “Energy-storabledye-sensitized solar cells with interdigitated nafion/polypyrrolept comb-like electrodes,”Chemistry Letters, vol. 39, no. 5, pp. 488–489, Apr. 2010. → pages 42, 43, 60[57] Z. Yang, L. Li, Y. Luo, R. He, L. Qiu, H. Lin, and H. Peng, “An integrated device forboth photoelectric conversion and energy storage based on free-standing and alignedcarbon nanotube film,” Journal of Materials Chemistry A, vol. 1, no. 3, pp. 954–958,Dec. 2012. → pages 43, 60[58] M. Yu, X. Ren, L. Ma, and Y. Wu, “Integrating a redox-coupled dye-sensitizedphotoelectrode into a lithiumoxygen battery for photoassisted charging,” NatureCommunications, vol. 5, Oct. 2014. → pages 4377[59] N. F. Yan, G. R. Li, and X. P. Gao, “Solar rechargeable redox flow battery based onli2wo4/LiI couples in dual-phase electrolytes,” Journal of Materials Chemistry A, vol. 1,no. 24, pp. 7012–7015, May 2013. → pages 45, 68[60] N. Yan, G. R. Li, and X. Gao, “Electroactive organic compounds as anode-activematerials for solar rechargeable redox flow battery in dual-phase electrolytes,” Journalof The Electrochemical Society, vol. 161, no. 5, pp. A736–A741, Jan. 2014. → pages 45,59, 60, 68, 69[61] S. Feldt, “Alternative redox couples for dye-sensitized solar cells,” dissertation, UppsalaUniversity, 2013. → pages 48, 52[62] “Ion selective membrane separate anolyte and catholyte,” U.S. Patent US4 485 154 A,Nov., 1984. → pages 48, 49[63] P. M. Lessner, F. R. McLarnon, J. Winnick, and E. J. Cairns, “Aqueous polysulphideflow-through electrodes: Effects of electrocatalyst and electrolyte composition onperformance,” Journal of Applied Electrochemistry, vol. 22, no. 10, pp. 927–934, Oct.1992. → pages 49[64] K. M. Abraham, R. D. Rauh, and S. B. Brummer, “A low temperature Na-S batteryincorporating a soluble S cathode,” Electrochimica Acta, vol. 23, no. 6, pp. 501–507,Jun. 1978. → pages 49[65] S. Licht, R. Tenne, H. Flaisher, and J. Manassen, “Cation effects on the electrochemistryof anions in polysulfide photoelectrochemical cells,” Journal of The ElectrochemicalSociety, vol. 133, no. 1, pp. 52–59, Jan. 1986. → pages 49[66] R. D. Rauh, K. M. Abraham, G. F. Pearson, J. K. Surprenant, and S. B. Brummer, “Alithium/dissolved sulfur battery with an organic electrolyte,” Journal of TheElectrochemical Society, vol. 126, no. 4, pp. 523–527, Apr. 1979. → pages 49, 52[67] Z. Tachan, M. Shalom, I. Hod, S. Ruhle, S. Tirosh, and A. Zaban, “PbS as a highlycatalytic counter electrode for polysulfide-based quantum dot solar cells,” The Journal ofPhysical Chemistry C, vol. 115, no. 13, pp. 6162–6166, Apr. 2011. → pages 49[68] D. V. Esposito, K. D. Dobson, B. E. McCandless, R. W. Birkmire, and J. G. Chen,“Comparative study of tungsten monocarbide and platinum as counter electrodes inpolysulfide-based photoelectrochemical solar cells,” Journal of The ElectrochemicalSociety, vol. 156, no. 8, pp. B962–B969, Aug. 2009. → pages 49[69] I. E. L. Stephens, C. Ducati, and D. J. Fray, “Correlating microstructure and activity forpolysulfide reduction and oxidation at WS2 electrocatalysts,” Journal of TheElectrochemical Society, vol. 160, no. 6, pp. A757–A768, Jan. 2013. → pages 49[70] H. Zhou, H. Zhang, P. Zhao, and B. Yi, “A comparative study of carbon felt andactivated carbon based electrodes for sodium polysulfide/bromine redox flow battery,”Electrochimica Acta, vol. 51, no. 28, pp. 6304–6312, Sep. 2006. → pages 4978[71] R. D. Rauh, F. S. Shuker, J. M. Marston, and S. B. Brummer, “Formation of lithiumpolysulfides in aprotic media,” Journal of Inorganic and Nuclear Chemistry, vol. 39,no. 10, pp. 1761–1766, Oct. 1977. → pages 52[72] Y. Fu, Y. S. Su, and A. Manthiram, “Highly reversible lithium/dissolved polysulfidebatteries with carbon nanotube electrodes,” Angewandte Chemie International Edition,vol. 52, no. 27, pp. 6930–6935, Jul. 2013. → pages 52[73] Y. Yang, G. Zheng, and Y. Cui, “A membrane-free lithium/polysulfide semi-liquidbattery for large-scale energy storage,” Energy & Environmental Science, vol. 6, no. 5,pp. 1552–1558, Apr. 2013. → pages 52[74] A. Fukui, R. Komiya, R. Yamanaka, A. Islam, and L. Han, “Effect of a redox electrolytein mixed solvents on the photovoltaic performance of a dye-sensitized solar cell,” SolarEnergy Materials and Solar Cells, vol. 90, no. 5, pp. 649–658, Mar. 2006. → pages 52[75] N. S. A. Manan, L. Aldous, Y. Alias, P. Murray, L. J. Yellowlees, M. C. Lagunas, andC. Hardacre, “Electrochemistry of sulfur and polysulfides in ionic liquids,” The Journalof Physical Chemistry B, vol. 115, no. 47, pp. 13 873–13 879, Dec. 2011. → pages 53[76] G. Boschloo and A. Hagfeldt, “Characteristics of the iodide/triiodide redox mediator indye-sensitized solar cells,” Accounts of Chemical Research, vol. 42, no. 11, pp.1819–1826, Oct. 2009. → pages 53[77] N. Papageorgiou, M. Gratzel, and P. P. Infelta, “On the relevance of mass transport inthin layer nanocrystalline photoelectrochemical solar cells,” Solar Energy Materials andSolar Cells, vol. 44, no. 4, pp. 405–438, Dec. 1996. → pages 55, 70[78] V. C. Nelson, Introduction to Renewable Energy, 1st ed. Boca Raton, FL: CRC Press,Apr. 2011. → pages 60[79] P. Liu, H. Yang, X. Ai, G. Li, and X. Gao, “A solar rechargeable battery based onpolymeric charge storage electrodes,” Electrochemistry Communications, vol. 16, no. 1,pp. 69–72, Mar. 2012. → pages 60[80] S.-H. Shin, S.-H. Yun, and S.-H. Moon, “A review of current developments innon-aqueous redox flow batteries: characterization of their membranes for designperspective,” RSC Adv., vol. 3, pp. 9095–9116, 2013. → pages 64[81] S. Ito, P. Chen, P. Comte, M. K. Nazeeruddin, P. Liska, P. Pchy, and M. Grtzel,“Fabrication of screen-printing pastes from TiO2 powders for dye-sensitised solar cells,”Progress in Photovoltaics: Research and Applications, vol. 15, no. 7, pp. 603–612, May2007. → pages 64[82] C. Law, S. C. Pathirana, X. Li, A. Y. Anderson, P. R. F. Barnes, A. Listorti, T. H.Ghaddar, and B. C. ORegan, “Water-based electrolytes for dye-sensitized solar cells,”Advanced Materials, vol. 22, no. 40, pp. 4505–4509, Aug. 2010. → pages 6779[83] T. Daeneke, Y. Uemura, N. W. Duffy, A. J. Mozer, N. Koumura, U. Bach, and L. Spiccia,“Aqueous dye-sensitized solar cell electrolytes based on the ferricyanideferrocyanideredox couple,” Advanced Materials, vol. 24, no. 9, pp. 1222–1225, Jan. 2012. → pages[84] W. Xiang, F. Huang, Y.-B. Cheng, U. Bach, and L. Spiccia, “Aqueous dye-sensitizedsolar cell electrolytes based on the cobalt(II)/(III) tris(bipyridine) redox couple,” Energy& Environmental Science, vol. 6, no. 1, pp. 121–127, Dec. 2012. → pages[85] H. Choi, B.-S. Jeong, K. Do, M. J. Ju, K. Song, and J. Ko, “Aqueous electrolyte baseddye-sensitized solar cells using organic sensitizers,” New Journal of Chemistry, vol. 37,no. 2, pp. 329–336, Jan. 2013. → pages[86] V. Leandri, H. Ellis, E. Gabrielsson, L. Sun, G. Boschloo, and A. Hagfeldt, “An organichydrophilic dye for water-based dye-sensitized solar cells,” Physical ChemistryChemical Physics, vol. 16, no. 37, pp. 19 964–19 971, Aug. 2014. → pages 67[87] A. Takshi, H. Yaghoubi, D. Jun, R. Saer, A. Mahmoudzadeh, J. D. Madden, and J. T.Beatty, “Application of wide band gap semiconductors to increase photocurrent in aprotein based photovoltaic device,” MRS Online Proceedings Library, vol. 1414, pp.mrsf11–1414–hh07–01, Oct. 2010. → pages 69[88] B. Huskinson, M. P. Marshak, C. Suh, S. Er, M. R. Gerhardt, C. J. Galvin, X. Chen,A. Aspuru-Guzik, R. G. Gordon, and M. J. Aziz, “A metal-free organic-inorganicaqueous flow battery,” Nature, vol. 505, no. 7482, pp. 195–198, Jan. 2014. → pages 71[89] M. I. Asghar, K. Miettunen, J. Halme, P. Vahermaa, M. Toivola, K. Aitola, and P. Lund,“Review of stability for advanced dye solar cells,” Energy & Environmental Science,vol. 3, no. 4, pp. 418–426, Mar. 2010. → pages 72[90] H. Gerischer, “The impact of semiconductors on the concepts of electrochemistry,”Electrochimica Acta, vol. 35, no. 11-12, pp. 1677–1699, Nov. 1990. → pages 84, 85[91] I. Muegge, P. X. Qi, A. J. Wand, Z. T. Chu, and A. Warshel, “The reorganization energyof cytochrome c revisited,” The Journal of Physical Chemistry B, vol. 101, no. 5, pp.825–836, Jan. 1997. → pages 85[92] “Electric power annual 2012,” U.S. Energy Information Administration (EIA), Tech.Rep., Dec. 2013. → pages 90[93] G. Hashmi, K. Miettunen, T. Peltola, J. Halme, I. Asghar, K. Aitola, M. Toivola, andP. Lund, “Review of materials and manufacturing options for large area flexible dyesolar cells,” Renewable and Sustainable Energy Reviews, vol. 15, no. 8, pp. 3717–3732,Oct. 2011. → pages 9080Appendix APublications Not Included in the ThesisJournal papers• Mirvakili S.M, Slota J., Usgaocar A., Mahmoudzadeh A., Jun D., Mirvakili M., BeattyJ.T., Madden J.D.W.(2014) ”Photoactive Electrodes Incorporating Electrosprayed Bac-terial Reaction Centers”, Advanced Functional Materials, Vol 24, Issue 30, 47894794• Mahmoudzadeh A., Saer R., Jun D, Mirvakili S.M., Takshi A., Iranpour B., OuelletE., Lagally E.T., Madden J.D.W., Beatty J.T. (2011) ”Photocurrent generation by directelectron transfer using photosynthetic reaction centres”, Smart Materials and Structures,vol 20, issue 9, 094019.• Yoo D.S., Mahmoudzadeh A., Fok E.C.W., Walus K., Madden J.D.W. (2011) ”Multi-ple time constant modelling of a printed conducting polymer electrode”, ElectrochimicaActa. vol 56 issue 13, 47116.• Takshi A., Madden J.D.W., Mahmoudzadeh A., Saer R., Beatty J.T. (2010) ”A Photo-voltaic Device Using an Electrolyte Containing Photosynthetic Reaction Centers”, En-ergies vol 3 issue 11, 1721-1727.81Presentations• Mahmoudzadeh A., Usgaocar A.R., Slota-Newson J., Wang L., Iranpour B., Jun D.,Saer R., Yaghoubi H., Mirvakili S.M., Madden J.D.W., Takshi A., Beatty J.T. (2013)”Elec-trical Energy from Photo-synthetic Proteins”, The 7th World Congress on Biomimetics,Artificial Muscles and Nano-Bio.• Slota, J.E., Christensen P., Jun D. Usgaocar, A.R., Mahmoudzadeh A., Wolf M.O.,Beatty J.T., Madden J.D.W.(2013) ”Selective self-assembled monolayer functionalisedelectrodes for photogalvanic cells”, Natl. Meet. -Am. Chem. Soc., Div. Coll. 2013,246, Coll 17.• Usgaocar A.R., Wang L., Mahmoudzadeh A., Mirvakili S.M., Slota-Newson J.E., Mad-den J.D., Beatty J.T., Takshi A. (2013) ”Semiconductors as Selective Electrodes for Bio-Photovoltaic Cells”, Meet Abstr.MA2013-01(4),282.82Appendix BSemiconductors as Selective ElectrodesElectron transfer between electrode and electrolyte is the key process in any electrochemicalcell. The vertical photogalvanic cell configuration needs selective electron transfer (ET) be-tween each redox system and the corresponding electrode. This behavior is achievable withsemiconducting electrodes. The theory behind this behavior is presented in this section fol-lowed by our experimental work to achieve such selectivity.B.1 IntroductionThe first part of this dissertation covers the guidelines for fabrication of an efficient photogal-vanic cell. The reactions on the electrodes in a PGC are summarized at:S−→ S+ e− andM++ e−→M.(B.1)If both reactions happen reversibly on the two electrodes, no photovoltage is induced in thecell [32]. However, a selective electrode could results in an electrochemical potential up to E =∆E0+ RTF ln[S−][S][M+][M] between electrodes, where ∆E0 is the difference in standard potentials ofthe two redox couples [31]. The redox electrode interaction can be explained by Gerischer83model. In this model, a reduction reaction takes place by an electron transfer from an occupiedstate in electrode to an empty state in redox ions at a constant energy (occupied and emptystates of a redox system can be interpreted as reduced and oxidized ions). As a result, theelectron transfer rate is proportional to the overlap between the occupied density of energystates in solid and empty energy states of oxidants in electrolyte. A similar discussion is validfor an oxidation reaction. This process could be quantitatively evaluated which suggests aselective behavior of semiconductors towards different redox couples. The thermal fluctuationmodel suggests a Gaussian distribution for the redox states in an electrolyte [90] as shown inFigure B.1.Figure B.1: Distribution of energy states in semiconductor (left) and metal (right)-electrolyte interface. The overlap between filled and empty states on two sidesdetermines the reaction rate.B.2 The Gerischer ModelGerischer described the electron transfer (ET) in terms of energy levels of filled and emptystates in electrode and liquid. In a redox system, occupied and empty states are interpreted asreduced and oxidized species respectively. The Gaussian distribution function of the states that84was explained above can be described byWox =W0 · exp[−(E−E0F,redox +λ )24kTλ ] (B.2)Wred =W0 · exp[−(E−E0F,redox−λ )24kTλ ] (B.3)where W is the distribution function, λ is the reorganization energy, T is the temperature, k isthe Boltzmann constant and W0 is the normalization factor and is calculated to beW0 = (4kTλ )−1/2.The half width of the distribution, ∆E1/2 is equal 0.53λ 1/2eV . A typical value for ∆E1/2 isabout 0.5 eV considering a reorganization energy of 1 eV [90, 91]. In order to get the actualdensity of states in the system, one need to include the concentration of reduced (cred) andoxidized species (cox) as well. The total distribution is given byDox(E) = coxWox and Dred(E) = credWred (B.4)The electron transfer rate is determined by the density of states on both sides e.g., forconduction band, the cathodic current, it is given byj−c = ek∫ ∞Ecf (E)ρ(E)coxWox(E)dE =ekcox(4kTλ )1/2∫ ∞Ecns exp[−(E−E0F,redox +λ )24kTλ ]dE.(B.5)in which the f (E)ρ(E) term is the distribution of occupied states in the semiconductor andis replaced by density of carriers at surface, ns(E), in the right hand side. Since having aconsiderable density of states on both sides of the interface is essential for having current,most of the electron transfer happens at the band where the overlap of two distribution occurs(depending on the band locations and redox Fermi level, ET can happen via conduction or85valence band, both or neither of them). Since the redox distribution function has a exponentialdependency on E2, most of the times the overlap between energy states of two sides of theinterface is limited to a small energy range around band edges(C.A. 1 kt). One can approximatethe integral of Equation B.5 by its linear approximation using dE = 1 kT as belowj−c ' ek0ns exp[−(Esc−E0F,redox +λ )24kTλ ], (B.6)where ns can be described by Boltzmann distribution function as ns = n0 exp(−eVSCkT ) and k0 isaccumulative of all the other terms on the left side of the integral in Equation (B.5) . Sincethe applied potential entirely drops on space charge region, cathodic current is exponentiallydependent on applied potential via ns.Electron transfer from filled states of redox system (reductants) to semiconductor’s con-duction band leads to an anodic current ofj+c ' ek0(1− f (Ec))ρ(E = Ec)cred exp[−(Esc−E0F,redox−λ )24kTλ ], (B.7)where the density of states in the semiconductorat the edge of conduction band ρ(E−Ec), isequal to Nc and since most of the states are empty in conduction band, (1− f ) ' 1. Thereforthe anodic current can be simplified toj+c ' ek0Nccred exp[−(Esc−E0F,redox−λ )24kTλ ]. (B.8)The equation can be simplified even more by collecting all the constant terms in aboveequation and the potential independent exponential term in one rate constant,j−c = ek−c nscox where k−c = kc,max− exp[−(Esc−E0F,redox +λ )24kTλ ], (B.9)j+c = ek+c Nccred where k+c = kc,max+ exp[−(Esc−E0F,redox−λ )24kTλ ]. (B.10)86Similarly, electron transfer at valence band can be described as belowj−v = ek−v Nvcox where k−v = kv,max− exp[−(Esv−E0F,redox +λ )24kTλ ] (B.11)j+v = ek+v pscred where k+v = kv,max+ exp[−(Esv−E0F,redox−λ )24kTλ ]. (B.12)Measurements of space charge capacity of several semiconductor electrodes in aqueous solu-tions indicates that position of band edges at the surface of a semiconductor are usually constanteven for samples with different doping densities. The band edges also stay pinned after addi-tion of redox system due to stronger interaction with water than redox. This pinned locationresults the potential independent rate constants of equations B.9 to B.12 be intrinsic features ofeach semiconductor-redox pair that do not change with secondary factors such as concentrationor charged state. Therefore, large and small ET rates is expected based on relative position ofband edges and redox potential. Finding semiconductors that have much larger rate constantwith only one of the redox couples in a PGC could be a solution to the selectivity challenge inPGCs.B.3 Experimental Measurement of the SelectivityThe hypothesis is tested in a series of experiments on fluorine doped tin oxide (FTO), copper(II)oxide (CuO) and nickel oxide (NiO) as wide band gap semiconducting electrodes [49]. Methylviologen (MV2+/MV+), ferrocyanide (Fe(CN)3–6 /Fe(CN)4–6 ) and ferrous (Fe3+/Fe2+) are used asthe reacting redox couples. The former has standard redox potential close to FTO conductionband edge and the latter has standard potential close to CuO valence band edge.Sampled current voltammetry is used to measure the electron transfer rate between semi-conducting electrodes and redox couples. Negative overpotentials of different values are ap-plied to the electrode and the transient current is monitored. The measure current value after12 seconds is saved as the current response to that overpotential. The reason for the delay is toavoid the double-layer capacitance charging and to only compare the Faradaic response. The87best selectivity is observed with the FTO electrode. Figure B.2 shows the cyclic voltammetryand sampled electron voltammetry results for this electrode.(a) (b)(c) (d)Figure B.2: Cyclic voltammetry of (a) methyl viologen (b) ferri/ferrocyanide and (c) fer-ric/ferrous on FTO electrode. (d) Sampled current voltammetry data measured at12 s after the voltage application. The concentrations are 10 mM in aqueous 1 MKCl. The CVs are measured at 300, 200, 100, 50 and 10 mVs−1.The cyclic voltammograms indicate that while methyl viologen has quasi-reversible kinet-ics on the FTO electrode, ferricyanide/ferrocyanide only shows onsets of cathodic peaks forextremely slow scan rates with clear anodic peaks for all scan rates. This reflects the fasterreaction rate of methyl viologen compared to ferricyanide/ferrocyanide. The absence of theanodic and cathodic peaks in the ferric/ferrous couple CVs shows a very slow kinetic rate ofthis ion on FTO.The reaction rates are derived by mapping the sampled current voltammetry data of Fig-88ure B.2d to a quasi reversible reaction model. Two orders of magnitude difference in rateconstants are observed due to the relative locations of semiconductor band edge and the redoxstandard potential as shown in Table B.1.Table B.1: Standard heterogeneous rate constants of redox couples on FTO. Eo is theredox standard potential, ko is the reaction rate constant and R2 represents the fitquality.Mediator Eo (V vs Ag/AgCl) Ec−Eo (V) ko (cms−1) R2(MV+2 /MV+) -0.62 -0.18 2.8×10−4 0.9895Fe(CN)3–6 /Fe(CN)4–6 0.25 0.65 1.4×10−4 0.9953(Fe3+/Fe2+) 0.48 1.08 5.7×10−6 0.9906It is important to note that any significant selectivity occurs only at large differences instandard potentials (∼ 1 V). This effect is predictable from Gerischer model as well sincethe reorganization energy of the species is in the same order (∼ 1 eV). Therefore, any use ofsemiconductor selectivity is only applicable for dye/mediator pairs of large standard potentialseparation. All in all, this study demonstrates the eligibility of semiconductor electrodes forselective action in photogalvanic cells.89Appendix CSolar Redox Battery Cost AnalysisThe integration of solar cell and battery has significant effects on manufacturing and operatingcost of the utilization of the solar energy. While it is understood that system level alterationsin RFBs might be required after the integration, an analysis based on modifications limitedto the energy conversion modules is presented in this chapter. The effect of system scalingis shown by calculating system cost for three system sizes. These three segments are definedas residential(10 kW/100 kWh), industrial (100 kW/1 MWh) and grid backup (1 MW/10 MWh)based on average energy and power ratings in each category [92].Based on current date expenses of Vanadium RFBs (the most expensive yet reliable RFBtechnology) and Li-ion batteries, the cost of integrated and separate solar cell batteries arecompared in Table C.1. The energy storage cost for these scenarios are extracted from a recentreport by Sandia National Laboratories based on vendors input [8].The first two rows reflects the three energy storage scale scenarios. The energy storage costis shown in row 3. RFB shows significant cost advantage over LIB in larger scales and almoston par performance in residential scale applications. Next, the cost of solar panel installationequal to the power rating of each sector is added. DSSC cost breakdown is reported by Hashmiet al. [93]. They reported that the encapsulation and counter electrode adds up to half of thesolar cell cost. These elements are shared with the RFB therefore the total cost is estimated to90Table C.1: Comparison of energy conversion and storage costs for combined and sepa-rated systems.1 Energy rating 100 kWh 1 MWh 10 MWh2 Application Residential Industrial Grid Con-nected3 Capital cost $ / kWhLi-ion Battery 1000 900 800 aRedox Flow 1000 800 6004 Average power consumption 10 kW 100 kW 1 MW5 Total Cost of solar panels ($)Separate (3,2.5,2 $/W) 30 k 250 k 2 MIntegrated (1 $/W) 10 k 100 k 1 M6Cost of solar panels ($/kWhof power plant)Separate 300 250 200Integrated 100 100 1007Cost of complete solution($/kWh)Li-ion + PV 1300 1150 1000Integrated RFB 1100 900 700a Projected value.be the cost of RFB plus the cost of the remaining elements of a DSSC. As a result the addedcost is 1 $/W as opposed to 2 $/W of a complete solar cell (row 5). The normalized cost of thesolar panels to the energy rating of the facility is shown in row 6. Overall, an integrated solarredox battery shows an economic advantage relative to Li-ion storage at all application scalesas is seen in row 7.It can be seen that because of the high cost of electrochemical energy storage in general,2/3 of the cost advantage relative to Li-ion is because of the storage unit, i.e., RFB, rather thanthe integration. In other words, 200 dollars in cost savings relative to Li-ion battery is thanksto the use of redox flow system and a further 100 dollars comes from the sharing of parts withthe solar cell bringing the total cost of an integrated 10 MWh harvesting and storage systemto 700 $/kWh. With further price drop to batteries, the share of the latter section is expectedto increase. Some further cost saving is possible because of the elimination of the DC-DCconverter and the maximum power tracker between solar cell and the battery. It is noteworthythat the design requirements of an integrated system might change these values and a moreaccurate estimate requires a finalized design and implementation of the solar redox facility.91


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items