ebook img

Alkaline Quinone Flow Battery with Long Lifetime at pH 12 PDF

41 Pages·2017·2.6 MB·English
by  KwabiDavid
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Alkaline Quinone Flow Battery with Long Lifetime at pH 12

Article Alkaline Quinone Flow Battery with Long Lifetime at pH 12 DavidG.Kwabi,KaixiangLin, YunlongJi,...,Ala´n Aspuru-Guzik,RoyG.Gordon, MichaelJ.Aziz [email protected] HIGHLIGHTS Near-neutralchemistryfor rechargeable,aqueous-soluble, organicredox-flowbatteries Capacityfaderateof(cid:1)0.01%/ day,thelongestlifetimefora quinone-basedelectrolyte Meetstechnicalcriteriafor commercializationofredox-flow batteries Thisworkdemonstratesanew,organicredox-flowbattery(RFB)thatoutlivesits predecessors,offeringthelongest-livedhigh-performanceorganicflowbatteryto date.Itappearstobethefirstaqueous-solubleorganicRFBchemistrytomeetall thetechnicalcriteriaforcommercialization.Thepotentiallowreactantand membranecostsofthischemistryofferthepotentialforRFBsofthistypetobe usedcosteffectivelyatthegigawattscaleinordertoenablemassivepenetration ofintermittentrenewableelectricity. Kwabietal.,Joule2,1894–1906 September19,2018ª2018TheAuthors. PublishedbyElsevierInc. https://doi.org/10.1016/j.joule.2018.07.005 Article Alkaline Quinone Flow Battery with Long Lifetime at pH 12 David G. Kwabi,1,4 Kaixiang Lin,2,4 Yunlong Ji,2 Emily F. Kerr,2 Marc-Antoni Goulet,1 Diana De Porcellinis,1 Daniel P. Tabor,2 Daniel A. Pollack,3 Ala´n Aspuru-Guzik,2 Roy G. Gordon,1,2 and Michael J. Aziz1,5,* SUMMARY Context &Scale Wedemonstratealong-lifetime,aqueousredox-flowbatterythatcanoperate Electricitygenerationfrom atapHaslowas12whilemaintaininganopen-circuitvoltageofover1V.We renewablesourcessuchassolar functionalized2,6-dihydroxyanthraquinone(2,6-DHAQ)withhighlyalkali-solu- andwindcaneliminatefossil-fuel- ble carboxylate terminal groups. The resulting negative electrolyte material basedsystemsiftheirintermittent 4,40-((9,10-anthraquinone-2,6-diyl)dioxy)dibutyrate(2,6-DBEAQ)wassixtimes outputcanberegulatedusing moresolublethan2,6-DHAQatpH12.Symmetriccellcyclingwith2,6-DBEAQ safe,cost-effectiveenergy onbothsidesofthecelldemonstratesacapacityfaderateof<0.01%/dayand storage.Aqueous-solubleorganic <0.001%/cycle. By pairing 2,6-DBEAQ with a potassium ferri-/ferrocyanide redox-flowbatteries(RFBs)area positiveelectrolyteandutilizinganon-fluorinatedmembrane,thisnear-neutral potentiallysafer,lessexpensive flowbatteryshowsacapacityfaderatethatisthelowestofanyquinoneand alternativetolithiumionbatteries rivalsthelowesteverreportedforanyflowbatteryintheabsenceofrebalanc- andvanadiumflowbatteriesfor ingprocesses.Thisresultaddstheimportantattributeoflongcalendarlifeto long-dischargedurationstorage. quinone-based redox-flow batteries, which may enable massive penetration Wedemonstratealong-lifetime, ofintermittentrenewableelectricity. organic-molecule-basedRFBthat canoperateinweakalkaline INTRODUCTION conditionswhilemaintainingan open-circuitvoltageofover1V. Thecostofwindandphotovoltaicelectricityhasdroppedsomuchthatoneofthe Thenegativeelectrolyte greatest technical barriers to their widespread substitution for fossil electricity is comprisesananthraquinone theirintermittency. Cost-effective, safe, and scalablestationaryelectricitystorage functionalizedwithsolubilizing couldsolvethisproblem.Traditionalenclosedbatteriessuchaslithiumionbatteries carboxylategroupsandthe are common, but they cannot cost effectively store enough energy for the long positiveelectrolytecomprisesthe discharge durations at rated power that appear to be necessary for regulating foodadditivepotassium the intermittency of renewables. Redox-flow batteries (RFBs) comprise one class ferrocyanide.TheRFBcan ofenergystoragesystemsparticularlysuitedforlong-durationdischargebecause operatewithaninexpensive energy-storingspeciesareheldinliquidforminexternaltanksthatareseparated membraneandshowsacapacity fromthepowergenerationstack.Consequently,energyandpowerratingscanbe faderatethatrivalsthelowest scaled independently, and discharge durations (energy:power ratios) of several everreportedforanyRFBinthe hourstodayscanbeachievedattheratedpower. absenceofrebalancing processes.Thisresultaddsthe The RFBs farthest along the commercial pathway are based on charge storage importantattributeoflong of vanadium ions in acidic electrolytes, but they have limited potential for calendarlifetoquinone-based widespreadadoptionduetothehighcostandlowearthabundanceofvanadium.1 RFBs. As a consequence, RFBs based on redox-active organic and organometallic reactants2–6 are receiving much research interest, as they are potentially much less expensive than their vanadium-based counterparts; in addition, organic reactants can be chemically modified through selective functionalization in order to improve aspects of performance such as voltage, rate capability, and energy density. 1894 Joule2,1894–1906,September19,2018ª2018TheAuthors.PublishedbyElsevierInc. ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/). OnedrawbackofmanyreportedorganicRFBs,however,istheirlowchemicalstabil- ity because organic molecules are susceptible to degradation reactions such as nucleophilic substitution, gem-diol formation, and self-polymerization. Several organicandorganometallicRFBshavebeenreportedinthepast5years,operating in highly acidic, highly alkaline, or neutral conditions, based on quinone,2,4,7 viologen,8–11 ferrocene,8 alloxazine,5 and nitroxide radical10,12–14 motifs. Most of them, however, experience high temporal capacity fade rates of 0.1%–3.5%/day, which limits their long-term use over many years. Such relatively short lifetimes rendermostofthesechemistriesunsuitableforcommercialization,whilemotivating further research into the development of organic molecules with high stability in bothoxidizedandreducedredoxstates. Here,wereportanewnegativeelectrolyte(negolyte)moleculesynthesizedfromour previouslyreportedDHAQchemistrythatexhibitsrecordhighchemicalstability,lead- ingtothelowesttemporalcapacityfadeamongquinone-basedflowbatteries,onthe orderof(cid:2)5%/year.ThisisamongthelowestreportedforanyRFBintheabsenceofre- balancing procedures. The approach involves the introduction of ether-linked alkyl chains with solubilizing carboxylate functional groups onto an anthraquinone core and affords highchemical stability and high solubility (0.6 M at pH 12;1.1 M at pH 14)inalkalineelectrolyte(TableS1).Wehypothesizedthatsolubilitywouldbeenhanced by frustrating crystallization ifwe attachedsolubilizing groupsvia short hydrophobic chains that were not so long that their hydrophobicity dominates. The solubility enhancement reduces the concentration of supporting electrolyte by two orders of magnitude without compromising the ionic conductivity of the system. Pairing 2,6-DBEAQwithaferro-/ferricyanide-basedpositiveelectrolyte(posolyte)resultsina battery with an open-circuit voltage of 1.05 V and theoretical volumetric energy densityof(cid:2)12Wh/LatpH14and (cid:2)17Wh/LatpH12(TableS2).Wefurthershow thatthisO-alkylatedanthraquinonecanbeusedinaflowbatteryatpH12withaninex- pensivehydrocarbon-basedmembranethatfeaturesexceptionallyhighpermselectivity and low permeability of both the alkylated anthraquinone and ferricyanide species, affording century-scale timescales for reactant crossover. These results highlight the importanceofsyntheticapproachestoincreasethechemicalstabilityofactivespecies andconstitutealeapforwardinrealizinghigh-performanceaqueousorganicRFBsthat exhibitlongcycleandcalendarlifeatlowcapitalcost. RESULTSANDDISCUSSION SynthesisandCyclicVoltammetryStudy Figure1illustratesthesyntheticroute,chemicalstructures,andcyclicvoltammograms (CVs)ofvariousisomersofDBEAQ.AllDBEAQisomerscanbesynthesizedfairlysimply byanO-alkylationreactionfollowedbyhydrolysisoftheestertoaffordthehighlyalkali- solublecarboxylicacidterminalgroups(Figure1A).Sincethesecondhydrolysisreaction isalmost100%inyieldandtheby-productisavolatileorganic,i.e.,methanol,weexpect thatfortheindustrialprocess,thenegolytecanbeprepareddirectlyafterthehydrolysis stepandtheremovalofmethanolwithouttheadditionalcostincurredbytwounneces- 1HarvardSchoolofEngineeringandApplied Sciences,Cambridge,MA02138,USA sary acid-base treatments, namely, collection of the hydrolyzed DBEAQ product via 2DepartmentofChemistryandChemicalBiology, acidification for characterization and deprotonation of DBEAQ with a designated HarvardUniversity,Cambridge,MA02138,USA amountofbaseforelectrolytepreparation. 3DepartmentofPhysics,HarvardUniversity, Cambridge,MA02138,USA Afterthefunctionalization,theresultingDBEAQisomers(Figures1BandS1–S3)showed 4Theseauthorscontributedequally asignificantimprovementinsolubilityversustheirhydroxylcounterpartsatbothpH14 5LeadContact andpH12.Forinstance,theroom-temperaturesolubilityof2,6-DBEAQexceeded1M *Correspondence:[email protected] inpH14and0.5MinpH12KOHsolution(TableS1),asopposedto0.6M4and0.1M, https://doi.org/10.1016/j.joule.2018.07.005 Joule2,1894–1906,September19,2018 1895 Figure1. SynthesisandInitialCharacterizationofDBEAQ (A)Syntheticapproachtofunctionalizehydroxyanthraquinonewithtetheredhighlyalkali-soluble carboxylicgroups. (B)Chemicalstructuresof1,2-,2,6-,and1,8-DBEAQ. (C)Cyclicvoltammogramsof1,2-(blue),2,6-(gray),and1,8-(green)DBEAQ.Theredoxpotentialvs SHEofeachisomerisindicated. SeealsoFiguresS1–S3. for2,6-DHAQundersimilarconditions,respectively.Severalavenuesexistforfurther increasingreactant solubilityand the corresponding energy density,suchasthe use of mixed cations in the electrolytes 15 or mixtures of different DBEAQ isomers or differentmoleculeswithnearlythesamereductionpotential. All the DBEAQ isomers showed very similar redox potentials between (cid:3)510 and (cid:3)540mVversusstandardhydrogenelectrode(SHE)(Figure1C),whichwouldyield batteryvoltagesabove1Vversusferri-/ferrocyanideposolyte.Despitetheincrease inreductionpotentialcomparedwith2,6-DHAQ((cid:3)680mVversusSHE),theirCVs exhibitedhighreversibilitywithredoxpeaks(cid:2)40mVapart,muchsmallerthanthe (cid:2)90mVvaluefrom2,6-DHAQ.4ALevichanalysiswasusedtoobtainthediffusion coefficientoftheoxidizedformof2,6-DBEAQ(1.58310(cid:3)6cm2/s;FigureS4),which wasthenusedinaCVsimulationofitsredoxkinetics(FigureS5).Theresultsshow,in agreementwithananalogousstudyofthe2,6-DHAQCV,thatthepeakshapeand separationareconsistentwithtwoone-electronreductionstepsatdifferentpoten- tials, E and E , whose values are modulated by the energetics of semiquinone 1 2 reduction (see Figure S6 and associated discussion in the Supplemental Informa- tion). In comparison with 2,6-DHAQ, which exhibited an E -E separation of 1 2 60 mV,4 the 2,6-DBEAQ CV exhibits a much smaller E -E separation of 6 mV, 1 2 implying that 2,6-DHAQ is more thermodynamically susceptible to the formation ofsemiquinoneradicalsthan2,6-DBEAQ. ChemicalandElectrochemicalStabilityStudy Wefirstdemonstratethat2,6-DBEAQhassuperiorchemicalstabilityoverotherqui- nonesreportedtodate,notablyincluding2,6-DHAQ.Asadirectprobeoftheeffect of O-alkylation of anthraquinone on its chemical stability, 2,6-DBEAQ was cycled potentiostatically in a volumetrically unbalanced compositionally symmetric cell 1896 Joule2,1894–1906,September19,2018 Figure2. SymmetricCellCyclingof2,6-DBEAQatpH14 (A)Cellschematicforunbalancedcompositionallysymmetriccellcyclingandfullcellcycling. (BandC)(B)Unbalancedcompositionallysymmetriccellcyclingof0.10M2,6-DBEAQand(C) 0.65M2,6-DBEAQ,showingcapacityasafunctionoftimeandtemporalcapacityfaderatesfrom linearfitsofthelast5daysofcycling.Thecapacity-limitingsidewas5mL2,6-DBEAQ,whilethe non-capacity-limitingsidewas10mL2,6-DBEAQ,bothatpH14.Capacitieswereobtainedbyfull potentiostaticreductionandoxidationatG0.2Vofthecapacity-limitingside;thepotentialwas switchedwhenthemagnitudeofthecurrentdensitydecayedto2mA/cm2.Notethatbothyaxis scalesrepresentabout2%ofthecapacityofthecapacity-limitingside. SeealsoFigureS7. configuration. The symmetric cell configuration, in which the two sides have the sameelectrolytecomposition,isasimpleanddirectprobeofchemicalandelectro- chemicalstabilityofRFBreactantsandisdescribedinfullinaseparatereport.16Any observedcapacityfadeisdirectlyrelatedtodeactivationofthereactantontheca- pacity-limitingsideineitheritsoxidizedorreducedstatebecause(1)thereisnegli- gible reactant crossover with symmetric compositions, (2) changes to membrane resistance do not result in temporal capacity variations in potentiostatic cycling, and(3)theunbalancedvolumespermitthecapacity-limitingsidetobetakentoits limitingstatesofcharge(SOCs)despitepotentialsidereactions. Figure2shows aschematic ofthecellsetup(Figure2A) and theresultsofunbal- anced compositionally symmetric cell cycling for 2,6-DBEAQ at 0.10 and 0.65 M for durations of 13 and 26 days, respectively (Figures 2B and 2C). In both cases, 2,6-DBEAQexhibitedtemporalcapacityfaderatesof<0.01%/dayor<3.0%/year, which suggests a loss mechanism that is first order in 2,6-DBEAQ concentration. Incontrast,2,6-DHAQcycledinasimilar cellconfigurationdemonstratedamuch higher temporal fade rate of 5%/day (Figure S7). Temporal capacity fade rates forneutralorganic/organometallicRFBshavebeensummarizedelsewhere.8Most systems show fade rates in the range of 0.1%–3.5%/day. The temporal fade rate of 2,6-DBEAQ is the lowest ever reported for a quinone-based electrolyte, with the lowest previously reported being for 9,10-anthraquinone-2,6-disulfonate (AQDS) at 0.1%–0.2%/day.17,18 It is on par with that of a recently reported Joule2,1894–1906,September19,2018 1897 viologen-basedflowbattery,8whichexhibitedthehighestcapacityretentionratefor anyflowbatteryintheabsenceofrebalancingprocesses,withatemporalcapacity faderateof<0.01%/dayinsymmetriccelltesting.16 Theoriginofthehigherchemicalstabilityof2,6-DBEAQcomparedwith2,6-DHAQ isnotcompletelyunderstood;however,wehighlighthereafewkeydistinctionsthat may be responsible. Notably, it is the reduced form of 2,6-DHAQ that has been showninapreviousstudytobeinvolvedinthelossofredoxactivity,whereasthe reducedformof2,6-DBEAQisshownheretobequitestableevenattemperatures up to 95(cid:4)C (see Supplemental Information). Our first observation is that, because the redox potential of 2,6-DBEAQ is (cid:2)200 mV higher than that of 2,6-DHAQ, 2,6-DBEAQ should be more stable thermodynamically in its reduced form. Sec- ondly,athighpH,weexpectboththereducedhydroquinonecoreandsolubilizing groups to be deprotonated and therefore negatively charged in both molecules. The closer proximity of the negatively charged deprotonated hydroxyl groups in 2,6-DHAQshouldleadtolargerintramolecularCoulombicrepulsionforces,which maycontributetothisdestabilizationofthereducedformofthemolecule.Finally, the greater susceptibility to the formation of semiquinone radicals in 2,6-DHAQ, asdiscussedabove,mayalsobeinvolvedinitsdecreasedstability. The alkyl chain functionalization, while drastically improving lifetime, does not avoid decomposition altogether. By performing elevated temperature chemical stability studies, wehaveidentifiedthatthe cleavageofg-hydroxybutyrateisinvolvedinthe decompositionoftheoxidizedformof2,6-DBEAQ(FiguresS8andS9),andwehave characterizedthetimecourseofg-hydroxybutyratecleavageatpH12andpH14and at0.1Mand0.5Mconcentration(FiguresS10andS11).Theresultssuggestthatthe half-life of 2,6-DBEAQ atroomtemperature, pH 14,and 0.1 M concentration inthe oxidized form is on the order of 5 years, with substantially slower decomposition in thereducedformrelativetotheoxidizedformandatpH12relativetopH14;theseob- servationsareconsistentwithlongRFBlifetimeattypicaloperatingcellconditions. MembraneandFullCellStudies Ion-conductivemembranesplayacriticalroleinRFBsystemsingoverningthetrans- portofcounterionsacrossthemembranes(ionicconductivity)andthesimultaneous transportofredox-activespecies(membranecrossover).Inacompositionallyasym- metriccell,e.g.,2,6-DBEAQ-ferrocyanidesystem,capacityfaderatecanbegreatly exacerbatedbytheirrecoverablecrossoverofthereactanttotheothersideofthe electrolyte.19Tolimitcapacityfadebythismechanism,whileretaininghighpermse- lectivity,wesurveyedarangeofcation-exchangemembranes,includingtheindustry standardNafion-basedperfluorosulfonicacid(PFSA)membranes,bycharacterizing boththeirK+conductivityandpermeabilityof2,6-DBEAQandferricyanide(more permeable than ferrocyanide) (see Experimental Procedures for the customized labsetup).E-600seriesmembranes,whichcompriseanon-fluorinated,sulfonated polyaryletherketone-copolymer backbone, delivered the best performance. The membrane displayed a low area specific resistance (ASR) of (cid:2)1 U∙cm2 in 1 M K+ solution, which is comparable with Nafion 212. It also showed an extremely low 2,6-DBEAQ and ferricyanide permeability of 5.26 3 10(cid:3)13 cm2/s and 4.4 3 10(cid:3)12 cm2/s, respectively (Figure S12), which are at least an order of magnitude lowerthanNafion212systems.Tofurtherprovethelowpermeabilityofredox-active species,weconstructedalowconcentration2,6-DBEAQ-ferri-/ferrocyanidefullcell at pH 14, using a Fumasep E-620 (K) membrane. Over a period of galvanostatic cyclingtestingfor4days((cid:2)880cycles),thecellshowedimmeasurablylowcapacity fadeandacurrentefficiencyaround(cid:2)99.94%(FigureS13).Thistestisprimafacie 1898 Joule2,1894–1906,September19,2018 evidence that the low 2,6-DBEAQ permeabilities measured ex situ using the nutatingtablesetuptranslatetonegligiblecrossoverof2,6-DBEAQinanoperating cell and that additional contributions to crossover that are not present in the nutating table tests, such as electromigration and pressure from active pumping ofelectrolytes,20areunimportant.Theselowpermeabilities,togetherwiththefact that Fumasep E-600 series membranes do not contain fluorine, imply that their useinaDBEAQ-ferri-/ferrocyanidefullcellaffordsapotentiallyrobustconfiguration foralong-lastingRFBwithlowpowercost(in$/kW)forfullyinstalledsystems(seeA NoteonDBEAQ/DHAQPermeabilityinSupplementalInformation). To reduce the corrosivity of the system and the ferricyanide decomposition rate, whichisexacerbatedathighpH,21,22weperformedthefullcelltestsof2,6-DBEAQ atamoremoderateelectrolyteconditionofpH12.Symmetriccellcyclingof1,2- (Figure S14A) and 1,8-DBEAQ (Figure S14B) at pH 12 showed higher temporal faderates—greaterthan0.1%and10%/day,respectively—than2,6-DBEAQatthe samepH(<0.01%/day,FigureS15)andwerethereforenottestedinfullcellexper- iments.Theoreticalcalculationssuggestthattheseisomersaremorethermodynam- icallysusceptibletohydroxideorwater-inducedg-hydroxybutyratecleavagethan 2,6-DBEAQ(seeTableS4).Giventhe0.6Msolubilitymeasuredfor2,6-DBEAQat pH12(TableS2),subsequentcelltestswereperformedat0.5Minordertoexamine theperformanceofafullcellwithreasonableenergydensity.FromthePourbaixdi- agramof2,6-DBEAQ(FigureS16A),itsreductionpotentialbecomespHindepen- dentabovepH11.5andisnotexpectedtochangeduringcellcycling.Polarization andcapacityutilizationmeasurementswithanegolytecontaining0.5M2,6-DBEAQ areshowninFigure3.Polarizationstudies(Figure3A)atroomtemperatureshowed anear-linearrelationshipbetweencurrentdensityandvoltageatcurrentscloseto theopen-circuitvoltage(OCV),whichincreasedfrom0.97Vat10%SOCto1.12V at 100% SOC (Figure 3B). Between 80% and 90% of the polarization ASR is ac- counted for by the high-frequency resistance measured using electrochemical impedance spectroscopy (EIS), which largely reflects membrane resistance. Apeakgalvanicpowerdensityof0.24W/cm2wasrealizedat100%SOC(FigureS17). Thispowerdensityisabouthalfofthatpreviouslyreportedina2,6-DHAQ-ferrocy- anidecell,4owingtothehigherOCVofthelatter(1.20Vasopposedto1.05Vat50% SOC)andsmallerASR(0.858Ucm2asopposedto1.2Ucm2at50%SOC).When voltageandcurrenthavealinear,ohmicrelationship,thepeakgalvanicpowerden- sityisgivenbyp =V2 =r,whereV istheopen-circuitpotentialandristheASR. max OC OC The low permeability of the Fumasep membrane to the fastest-crossing species (i.e.,200yearsrequiredfor50%lossthroughcrossoverofferricyanide)permits,in principle,a4-foldreductioninthemembranethickness,whichwouldraisethepower densitysignificantly. Inordertoavoidtemporalvariationsinaccessiblecapacityduringfullcellcycling caused by changes in membrane resistance,16 each galvanostatic half cycle was finished witha potential holdatthepotentiallimit(1.4Vaftercharge,0.6Vafter discharge)untilthemagnitudeofthecurrentdensityfellbelow2mA/cm2.Overa 5-daytestperiod,acapacityfaderateof0.05%/dayor0.001%/cyclewasobserved (Figure4A).Aparallelcyclingtestwasperformedinwhichapotentiostaticcharge- discharge cycle was executed after every 20 galvanostatic cycles; a capacity fade rateof0.04%/daywasobservedinthatcase(Figure4B).Ithasbeenshownthatca- pacityretentionratesusingbothcyclingprotocolsyieldvirtuallyidenticalresults.16 Thesefullcellmeasurements,however,showedroughly4-foldincreaseincapacity fade compared with the <0.01%/day observed during symmetric cycling tests Joule2,1894–1906,September19,2018 1899 Figure3. PolarizationMeasurementsof2,6-DBEAQFullCellatpH12 (A)Cellvoltageversusdischargecurrentdensityatroomtemperatureat10%,30%,50%,70%,90%, and100%SOC.Electrolytescomprised5mLof0.5M2,6-DBEAQ(negolyte)atpH12(10mMKOH) and38mLof0.3Mpotassiumferrocyanideand0.1Mpotassiumferricyanide(posolyte)atpH12. (B)OCV,high-frequency,andpolarizationASRversusSOC. (C)Galvanostaticchargeanddischargecurvesfrom25to250mA/cm2.Theverticaldashedlines indicatethemaximumcapacityrealizedwithpotentiostaticchargeanddischargeatthevoltage cutoffs(1.4and0.6V,respectively),aswellasthetheoreticalcapacity. (D)Coulombicefficiency,round-tripenergyefficiency,andcapacityutilizationasapercentageof potentiostaticcapacityversuscurrentdensity. (Figures2Band2C),suggestingadditionalcapacityfademechanismsnotobserved duringsymmetriccellstudies.In-depthchemicalandelectrochemicalanalysis(Fig- ure S18) was performed to probe the chemical decomposition and crossover of DBEAQfromthecapacity-limitingside.Electrolytesinacellcompositionallyiden- ticaltothatinFigure4Abutcycledfor11daysweresubjectedtonuclearmagnetic resonance(NMR)(FigureS18A)andCV(FigureS18B)analysis.Fromtheseresults,no evidence of DBEAQ decomposition and crossover was found after examining the negolyte and posolyte before and after cycling. Based on the detection limit of NMR(0.1mMDBEAQ)andCVtechniquesundertheexperimentalconditionscho- sen,theupperlimitofthecapacityfaderatecausedby2,6-DBEAQdecomposition and/ormembranecrossoverwas(cid:2)0.01%/day,similartooursymmetriccellstudy. Wethereforehypothesizethatothercapacityfademechanisms,suchasprecipita- tion of 2,6-DBEAQ in the posolyte after crossing over, might be operative but untraceablebyNMRandCVtechniquesduetotheslowcapacityfaderate,which corresponds to a total loss of <0.5% of 2,6-DBEAQ over a 6-day testing period. Oneothersuchmechanismmightbeleakageofthenegolyteduetopooreradhe- sionbetweenthethinnerFumasepmembraneandtheVitongasketinthefullcell thanbetweenNafionN117andthegasketinthesymmetriccell.Indeed,thetotal capacityfadeinFigure4Bcorrespondstoatotallossovertheentire6-daycycling period of (cid:2)10 mL of negolyte volume, which is roughly one-fifth of a droplet. When translated to an equivalent current density (0.5 mA/cm2), this negolyte loss rateiswellwithinexpectationforseepageoftheelectrolyteintospacesbetween 1900 Joule2,1894–1906,September19,2018 Figure4. ExtendedFullCellCyclingof2,6-DBEAQat0.5MandpH12 (A)Currentefficiency(squares)andcharge(upward-pointingtriangles)anddischarge(downward- pointingtriangles)capacityversustimeandcyclenumberforanegolyte-limited2,6-DBEAQ- Fe(CN)6cell.Thecellwascycledgalvanostaticallyat100mA/cm2between1.4and0.6V,and eachhalfcycleendedwithapotentiostaticholduntilthemagnitudeofthecurrentdensityfell below2mA/cm2.Thenegolytecomprised5mLof0.5M2,6-DBEAQatpH12whiletheposolyte comprised30mLof0.3Mpotassiumferrocyanideand0.1MpotassiumferricyanideatpH12. (B)Evolutionofcharge(upward-pointingtriangles)anddischarge(downward-pointingtriangles) capacityforextendedcellcyclingat100mA/cm2.Afterevery20thgalvanostaticcycle,a potentiostaticcharge-dischargecyclewasperformed,thedischargecapacityofwhichisshown (opencircles).Becausetheconcentrationofoxidized2,6-DBEAQisatitsmaximumaftereach potentiostaticdischarge,thecapacityofthegalvanostaticchargestepimmediatelyfollowing potentiostaticcharge-dischargeishigherthaninsubsequentcycles.Thisparticularcellshowed only(cid:2)65%ofitstheoreticalcapacitybecausealargefractionofsaltwasleftoverfromthesynthesis of2,6-DBEAQ. thegasketsand/orinterfacebetweenthemembraneandgaskets,comparedwithan analogousleakrateestimatedfromapreviousstudyofcapacityfadeinananthraqui- none-basedflowbatterywiththiscellarchitecture(0.09–0.12mA/cm2).19 Quinones for aqueous flow batteries have been the subject of intensive research since their recent debut in this application. This work demonstrates that the simplefunctionalizationwithbulkychargedgroupscangreatlyextendthecalendar life and cycle life; raise the solubility, leading to a theoretical energy density of 17 Wh/L with possible routes to further increases; operate with fluoropolymer- free membranes at pH as low as 12 while maintaining the voltage against Fe(CN) 3(cid:3)/4(cid:3) above 1 V; and perform with peak power density of 0.24 W/cm2 at 6 roomtemperaturewithpossibleroutestofurtherincreases.Thesepropertiesappear tomeetallofthetechnicalrequirementscommonlyunderstoodtobenecessaryfor aqueousRFBcommercialization. At the large production volumes necessary for gigawatt-scale grid storage, it is possible that DBEAQ may approach the cost of other functionalized anthraqui- nones,suchasDHAQandAQDS,thelatterofwhichwasrecentlyestimatedtobe between$0.92/kgand$3.92/kg.23Using$2.40/kgasamid-rangecost,thisresults inacapitalcostof$17.60/kWhforthe2-electron-acceptingDBEAQinthenegolyte ina1.05Vbattery.Similarly,weestimatetheindustrial-scalecostofpotassiumferro- cyanidetobearound$2.15/kg,24whichamountstoacapitalcostofabout$32/kWh fortheposolyte(seeEstimated CostofElectrolytesintheSupplementalInforma- tion). At $50/kWh, the total capital cost of these electrolytes stands at less than one-thirdofthecostofanall-vanadiumelectrolyte,forwhichthevanadiumalone iscurrentlypricedat$160/kWh.Usingafluorine-freemembranesuchasFumasep E(cid:3)620(K)shouldresultinaproductofmembranecost(<$25/m2atlargeproduction Joule2,1894–1906,September19,2018 1901 volumes)andASR(<1.5Ucm2)thatisbelow$5/mU,whichaddsnegligiblecostto thesystem.25Thus,iftheDBEAQproductioncostatscaleturnsouttobenearlyas lowasforAQDSandDHAQ,23,26thenthischemistrymayhaveseriouscommercial potential, and DBEAQ-based flow batteries may be instrumental in accelerating thepenetrationofwindandphotovoltaicelectricity. EXPERIMENTALPROCEDURES SynthesisandChemicalCharacterization Synthesisof4,40-((9,10-Anthraquinone-2,6-diyl)dioxy)dibutyricAcid 2,6-DHAQ was purchased from AK Scientific. Methyl 4-bromobutyrate was pur- chased from VWR. All other chemicals were purchased from Sigma-Aldrich. All chemicalswereusedasreceivedunlessspecifiedotherwise. Dimethyl 4,40-((9,10-anthraquinone-2,6-diyl)dioxy)dibutyrate (1). 2,6-DHAQ was first converted to its dipotassium salt (2,6-DHAQK ) by adding 2,6-DHAQ (5 g, 2 20.8 mmol) to a 250 mL oven-dried flask of dimethylformamide (250 mL). Under vigorous stirring, potassium ethoxide (6.1 g, 72.9 mmol) was added. The mixture solutionwasstirredatroomtemperaturefor15min.FortheO-alkylationreaction, 2,6-DHAQK (6.5 g, 20.8 mmol) was mixed with anhydrous K CO (14.3 g, 2 2 3 104mmol)andmethyl4-bromobutyrate(12.4mL,104mmol).Thereactionmixture was then heated to 95(cid:4)C overnight. After cooling to 0(cid:4)C, deionized (DI) water (150 mL) was added to the mixture to dissolve inorganic salt and to precipitate the ester precursor of DBEAQ. The precipitate was vacuum filtered and washed thoroughlywithDIwater(50mL).Theproductwasanalyzedby1HNMRandused forthenextstepreactionwithoutfurtherpurification.Finalyield,87%. 4,40-((9,10-Anthraquinone-2,6-diyl)dioxy)dibutyricAcid(2). Theesterprecursorof DBEAQ(1g,2.23mmol)wasaddedalongwithKOH(0.52g,9mmol)toaflaskfilled withawater-isopropanolmixture(2:1v/v,60mL).Thesolutionwasstirredvigorously andheatedto60(cid:4)Cfor12hr.Duringthereaction,allsolidsdissolved,andthesolu- tionbecameadarkredcolor.Afterthereaction,thesolutionwastransferredtoa larger500mLflaskanddilutedwithDIwater(200mL).Glacialaceticacidwasadded until the pH of the solution dropped to 4. The mixture was stirred vigorously for 1 hr, followedby vacuumfiltrationand thoroughwashing withDI water (100mL). Theproductwasvacuumdried,analyzedby1HNMR,andusedforelectrochemical measurementwithoutfurtherpurification.Finalyield,99%. The 1H NMR spectrum of 4,40-((9,10-anthraquinone-2,6-diyl)dioxy)dibutyric acid is shown in Figure S1. 1,8- and 1,2-isomers were synthesized similarly, using 1,2-DHAQ and 1,8-DHAQ as precursors. Potassium tert-butoxide (8.16 g, 72.9 mmol) replaced potassium ethoxide to deprotonate the 1,2-DHAQ and 1,8- DHAQisomersbecauseofthehigherpK ofthehydroxylgroupsoftheseisomers. a 1HNMRspectraofthefinalproductsareshowninFiguresS2andS3,respectively. ElectrochemicalCharacterization CyclicVoltammetryandRotatingDiskElectrodeMeasurements Glassycarbonwasusedastheworkingelectrodeforallthree-electrodeCVtests. Rotating disk electrode experiments were conducted using a Pine Instruments ModulatedSpeedRotatorAFMSRCEequippedwitha5mmdiameterglassycarbon workingelectrode,aAg/AgClreferenceelectrode(BASi,pre-soakedin3MNaClso- lution), and a graphite counter electrode. The electrode was rotated at a specific speedwhilethevoltagewassweptlinearlyfrom(cid:3)0.4to(cid:3)0.11VversusAg/AgCl (Figure S4A). The diffusion coefficient of the oxidized form of 2,6-DBEAQ was 1902 Joule2,1894–1906,September19,2018

Description:
Alkaline Quinone Flow Battery with Long Lifetime at pH 12 . involves the introduction of ether-linked alkyl chains with solubilizing carboxylate
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.