ContentslistsavailableatScienceDirect
Journal
of
Power
Sources
j ou rn a l h o m e pa g e :w w w . e l s e v i e r . c o m / l o c a t e / j p o w s o u r
Experimental
characterization
of
internal
currents
during
the
start-up
of
a
proton
exchange
membrane
fuel
cell
A.
Lamibrac,
G.
Maranzana
∗,
O.
Lottin,
J.
Dillet,
J.
Mainka,
S.
Didierjean,
A.
Thomas,
C.
Moyne
LEMTA-NancyUniversity-CNRS,2avenuedelaForêtdeHaye,BP160,54504VandoeuvrelesNancyCedex,France
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received2May2011
Receivedinrevisedform28June2011 Accepted5July2011
Start-up/shutdownunderinappropriateconditionsareknowntodamageprotonexchangemembrane fuelcells.Severalmeasurementtechniqueswerealreadyusedtocharacterizetheireffectsandthe degra-dationsaremainlyattributedtocarbonoxidation:adecreaseofthecathodeactiveareaiscommonly observedafterstart-up/shut-downcycles.
Inordertoachieveabetterunderstandingoftheconditionsinwhichthisphenomenonappears,a segmentedcellwasdesignedandinternalcurrentswererecordedduringstart-upunderopencircuit conditions.Theresultsofthisworkallowestimatingthecarbonlossesataglobal,aswellasatalocalscale, asafunctionofthehydrogenflowrate.Localelectrochemicalimpedancespectroscopymeasurements carriedoutondamagedMEAshowclearlythatthesegmentsthatremainforthelongesttimeinthe presenceofair(whiletheothersarefedbyhydrogen)arethemostaffectedbycarbonoxidation.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
During the start-up of a PEMFC, the anode compartment is divided momentarily into an upstream part occupied by fresh hydrogenanda downstreampartfilledwiththegaswhichwas usedtoflushtheanode: airornitrogen.Theboundarybetween thedownstreamandtheupstreampartsmovesfromtheinletto theoutletoftheanodecompartmentwithavelocitythatis gov-ernedbythehydrogenflowrate.Start-upsoccurgenerallyunder opencircuitconditionssothatpositiveandnegativeinternal cur-rentsappearbetweentheanodeandthecathode.Theseinternal currentscanbeobservedthankstodualcellconfiguration[1–3]
butalsowithinstrumentedcells[4].Theiroriginisattributedto faradicreactions[1,5],asdepictedinFig.1:hydrogenisoxidizedin theinletpartoftheanodecompartmentwhileinthepresenceof oxygen,waterandcarbonoxidationcantakeplaceatthecathode. Evidenceofcarbonoxidationwasbroughtbymeasuringinreal timetheemissionsofCO2 intheairexitingthecathodeduring start/stop cycles[6–9]. Gu etal. [7] measuredbetween60 and 100ppm(partspermillion)forstopandstart-upsequences, respec-tively,withhydrogenandairflowratessetto140sccm(standard cubiccentimetersperminute)attheanode(1000sccmforairat thecathode).
∗Correspondingauthor.Tel.:+33383595548;fax:+33383595551. E-mailaddress:[email protected](G.Maranzana).
Theaimsoftheworkpresentedinthispaperaretomeasure andanalyzetheinternalcurrentsoccurringduringPEMFCstart-up sequences,inordertobetterunderstandtheirorigin,theirnature, andtheirconsequencesintermofcathodeoxidation.These inves-tigationsarecarriedoutbymeansofasegmentedcell.Segmented cellsareincreasinglyusedforinsituinvestigationofPEMfuelcell operation[10,11],forinstancetodesignchannels[12]orto opti-mizetheirperformances[13].Althoughsuitableforthestudyof degradations[14,15],theywere rarelyemployed for character-izingstart-up andshutdownsequences[16].Theexperimental resultsareinterpretedbyreferencetoanelectricalequivalent cir-cuitproposedbyMaranzanaetal.[17],whoshowednumerically thatthecapacitivecontributiontonegativecurrentscouldbe sig-nificant,dependingontheoperatingconditions.Asaconsequence, itappearsthatcarbonoxidation,andthustheresultingfuelcell degradationandperformancelosses,dependhighlyonthe start-up/shut-downprotocol.
2. Segmentedcellprototype
The experimental results presented in the following were obtainedwiththetest benchdepictedinFig.2.Thisbenchwas developed inorder tomonitora uniquecellwith20 segments. Gasesaresuppliedthrough6massflowcontrollers,whichallow selectingabroadrangeofflowrates.Allofthemexceptnitrogen arehumidified.Hydrogenandair areintroducedinthecathode andanodecompartmentincounter-flow.Thegaslinesalsoallow 0378-7753/$–seefrontmatter© 2011 Elsevier B.V. All rights reserved.
ARTICLE IN PRESS
GModelPOWER-14621; No.ofPages8
2 A.Lamibracetal./JournalofPowerSourcesxxx (2011) xxx–xxx
Nomenclature
S fuelcellarea(cm2)
jseg localcurrentdensity(Acm−2)
t time(s)
R∞ infiniteresistanceordischargeresistance()
QPN2 passivechargeforastart-upbyinjectionof hydro-geninnitrogen(C)
QPAir passivechargeforastart-upbyinjectionof hydro-geninair(C)
QBN2 balancingchargeforastart-upbyinjectionof
hydro-geninnitrogen(C)
QBAir balancingchargeforastart-upbyinjectionof hydro-geninair(C)
QgreenareaN2 activecharge+positivebalancingforastart-up
byinjectionofhydrogeninnitrogen(C)
QgreenareaAir activecharge+positivebalancingforastart-up byinjectionofhydrogeninair(C)
QAN2 activechargeforastart-upbyinjectionofhydrogen
innitrogen(C)
QAAir activechargeforastart-upbyinjectionofhydrogen inair(C)
U fuelcellvoltage(V) RH relativehumidity
HFR highfrequencyresistance(cm2)
C capacity(Fcm−2)
flushing theanode with nitrogen or air, and the cathodewith nitrogen.
Fig.3showspicturesoftheinstrumentedcell.Theanodeflow fieldplateisin goldplatedalumina.Exceptfortheresults pre-sentedinSection3.1,theanodeplateisnotsegmentedanditis alsousedtocontrolthetemperatureinthecellthankstoacooling circuitandathermostatedbath(LaudaRE104-E100).Thecathode compartmentismachinedinapolycarbonateplate(transparent butnon-conductive)andthecurrentcollectionismadeby20 elec-tricallyinsulatedsegments(Fig.3).Thefour15mm,1mmthick goldplatedbrassstripsusedforeachsegmentdelimitfiveparallel airchannels(thedepthandwidthofthe5straightparallel chan-nelsare0.7mm×1mm,atthecathodeandattheanode).Kapton sheets(2.3mm×1mm×80m)areusedtoisolateeachstripfrom
theadjacentones.Thefourstripsofeachsegmentareelectrically connectedoutsidethecell,whichamountstoaveragingthe cur-rentoverthemembrane’swidth.Thelocalcurrentsaremeasured thanksto20shuntresistorsof5mandrecordedbyadata process-ingandacquisitionsystem(NationalInstrumentSCXImultiplexer) atafrequencyof200Hz(Fig.4).
The1cm×30cm,notsegmented,JohnsonMattheyMEA con-sists of a Nafion 212 membrane, 0.4mgPtcm−2 electrodes and 190mthickgasdiffusionlayers(Toraypaper).150mTeflon
gas-ketsareinsertedbetweentheMEAandtheflowfieldplates.The platesareheldtogetherthankstoasetof40boltstightenedtoa torqueofabout2.5Nm.
ALabviewprogrammakesitpossibletocontrolthemassflow andtheoperationof thevalvesasfunctionsof the experimen-taldata:temperature,voltageandlocalcurrents.Localimpedance spectroscopywas performedin galvanostaticmode witha fre-quencyrangeof0.25–800Hzandwithanamplitudeofthe(total) currentoscillationequalto5%ofthesteadystatevalue(with a NationalInstrumentUSB6221loadcontroller).
Fig.1. Reactionsoccurringduringstart-up,theanodecompartmentbeinginitially filledwithair.
3. Resultsanddiscussion
3.1. Comparisonofinternalcurrentsbetweenanodeandcathode duringstart-up
Internalcurrentsmeasuredatoneelectrodeprovideonly par-tialinformation aboutthedirectionof theflow of theprotons. Thoseproducedintheactivepartoftheanodeshouldlogicallyflow throughthemembranetowardthecathode.However,onecannot excludeapriorithepossibilitythatsomeprotonsflowalongthe membranetoreactwithoxygenpresentinthepassivepartofthe fuelcell(Fig.5).
Thisinformationcanbeobtainedbyusingacellsegmentedon bothelectrodesandbycomparingtheinternalcurrentsmeasured onbothsides.TheresultsarepresentedinFig.6.
Thecurvesin Fig.6showonlya smalldiscrepancy between theinternalcurrentsmeasuredattheanodeandatthecathode. Thisisprobablyduetothenon-uniformityofthecontact resis-tancesbetweenthecollectingstripsandthegasdiffusionlayers, whichleadstoaredistributionoftheelectronsbetweensuccessive segments:onecancheckthatthedifferencebetweentheanode andcathodecurrentsisexactlycompensatedfromonesegment toanother.Moreoverthetotalamountofchargeproducedatthe anodeisthesameastheoneconsumedatthecathode(the differ-enceislessthan3%).Consequently,theflowofprotonsalongthe membranecanbeconsideredasnegligible:usingafuelcell seg-mentedatthecathodeonlymakesitpossibletomeasureaccurately theinternalcurrents.
3.2. Equivalentelectricalcircuit
3.2.1. Start-upwiththeanodecompartmentpreviouslyfilled withnitrogen
Theequivalentelectricalcircuitthatcanbeusedtounderstand andtomodeltheoccurrenceof internalcurrentswasproposed byMaranzanaetal.[17](Fig.7).Thepurposeofthismodelisto contributetotheunderstandingofthephenomenagoverningthe internalcurrents.Masstransportisnotaccountedfor.Forthesake ofsimplicity,thefuelcellissegmentedintoonlytwosegmentsin theexampledescribedbelow.
Fig.2.Gaslinesanddataacquisitionandprocessingsystem.
increaseoftheelectrochemicalpotentialleadstothechargeofthe doublelayercapacityatbothelectrodes.
Thefuelcellbeinginopencircuit,theelectronsproducedinthe activepart(attheanode)flowtowardthepassivepartwhere
neg-Fig.3. Segmentedcellandclose-uponthecurrentcollectionstrips.
ativecurrents(byreferencetothenormalfuelcelloperation)are measured.Intheabsenceofoxygen,noelectrochemicalreaction canoccurinthepassivepart:thesenegativecurrentsarecapacitive currents.Thecorrespondingchargeofthedoublelayercapacities isofsignoppositetothatoccurringintheactivepart.
Thesecondstagecorrespondstoananodecompletelyfilledwith hydrogen(Fig.7b),thatistosaytotheinstantthesecondsegment becomesactivewhenreachedbyhydrogen.Atthistime,the poten-tialoftheelectrodes(ofthecathode)increasesuptoitssteady statevaluewhereasthedouble layercapacities remaincharged negatively(byreferencetotheregularoperation).
Asaconsequence,thefuelcellvoltageexceedstheopencircuit valueforashorttime(Fig.8);simultaneouslythenon-uniformity of thechargeof thecapacitiesover thewholemembrane elec-trodeassembly (MEA)givesrisetocapacitiveinternal currents, untilthecompleteredistributionofcharges.Then,thefuelcell volt-agedecreasesslowlydowntoitssteadystateopencircuitvalue; thisprocessisnotyetwellunderstoodbutitcanbeassumedthat thereisasmallleakagecurrentbetweenanodeandcathode,which correspondstotheresistanceR∞(actuallynotinfinite,buthigh)in parallelwiththedoublelayercapacityintheequivalentcircuit.
ARTICLE IN PRESS
GModelPOWER-14621; No.ofPages8
4 A.Lamibracetal./JournalofPowerSourcesxxx (2011) xxx–xxx
Fig.4. Segmentedcellelectricalconnections.
Table1
Fuelcellvoltageovershootinthepresenceofnitrogenvs.hydrogenflowrate.Airandhydrogenhumidifiedto100%RH,fuelcelltemperaturesetto50◦C.
H2flowrate(slph) 0.1 0.2 0.5 1 2 5 10 20 30
Fuelcellovershoot(V) 1.06 1.09 1.10 1.09 1.08 1.06 1.03 1.01 1.00
thelastmeasurements(thefirstpointsinTable1)areaffectedby carbonoxidation.
Notethat avoltageovershoot of0.1V represents10%ofthe steadystateOCV,and15–20%ofthefuelcellvoltageinoperation.A 100cellsstackdesignedforworkingat0.6V/cell(60V/stack)could reachanOCVof110Vatstart-up,avalue80%higherthaninregular operation:onesimplewaytoreducethefuelcellvoltageovershoot duringstart-upconsistsinincreasingmonetarilythehydrogenflow rate.
3.2.2. Start-upwiththeanodecompartmentpreviouslyfilled withair
Thepresenceofoxygenintheanodecompartmententails oxi-dationreactionsatthecathodeinthepresenceofnegativecurrents (thecathodebecomingtransientlyananodeduringthefirststage ofthestart-upsequence).Theseoxidationreactions(eitheroxygen evolutionorcarboncorrosion)correspondtoafaradiccontribution tothenegativecurrentdensitymeasuredinthepassivepart;they areaccountedforintheequivalentcircuitwithachargetransfer resistance–setinparalleltothedoublelayercapacity–aswellas withanelectrodepotential(Fig.9).
Fig.5.Reactionsoccurringduringstart-upassumingpossibleflowofprotonsalong themembranefromtheactiveparttothepassivepartoftheanodecompartment (initiallyfilledwithair).
Notethatbecauseofthepresenceofairattheanode,the poten-tialdifferencebetweenthecathodeandtheelectrolytecanreach upto1.6V,correspondingtoveryoxidizingconditions[1,5–7,18].
3.3. Experimentalresults:localcurrentdensities
Fig.10ashowstheevolutionofthelocalcurrentdensities mea-suredinallthesegmentsduringastart-upwiththeanodeinitially filledwithnitrogen.Wefocusinthefollowingoncurrentdensity profilesmeasuredonthefirstsegmentsupplied withhydrogen (Fig.10b),onasegmentlocatedatanequaldistancefromtheanode inletandoutlet(Fig.10c)andonthelastsegment(Fig.10d).
Thefirstsegmentsuppliedwithhydrogenisalwaysactivewhile alltheothersbecomesuccessivelyactiveonceinthepresenceof hydrogen.Theredistributionofchargesbeginswhenthefaradic reactionsstop.Atthistime,thefirstsegmentreleasessomeofthe electronsithasaccumulatedatthecathode—whichcorresponds toanegativecurrentinFig.10bandcwhereastheinverse
phe-Fig.6. Localcurrentdensitiesmeasuredattheanodeandatthecathodesegments duringastart-upwiththeanodecompartmentinitiallyfilledwithair(H2flow rate=2slph,airflowrate=20slph).Airandhydrogenhumidifiedto100%RH,fuel celltemperaturefixedto50◦
Fig.7.(a)Equivalentelectricalcircuitcorrespondingtoastart-upwiththeanode initiallyfilledwithnitrogen—firststage,anodepartiallyfilledwithhydrogen.(b) Equivalentelectricalcircuitcorrespondingtoastart-upwithanodeinitiallyfilled withnitrogen—secondstage,anodecompletelyfilledwithhydrogen.
Fig.8. Fuelcellvoltageduringastart-up withtheanodeinitially filledwith nitrogen—experimentalresultsmeasuredwithahydrogenflowratesetto0.2slph. Airandhydrogenhumidifiedto100%RH,fuelcelltemperaturefixedto50◦C.
Fig.9.Equivalentelectricalcircuitcorrespondingtoastart-upwiththeanode ini-tiallyfilledwithair–Firststage,anodepartiallyfilledwithhydrogen.
ARTICLE IN PRESS
GModelPOWER-14621; No.ofPages8
6 A.Lamibracetal./JournalofPowerSourcesxxx (2011) xxx–xxx
Fig.11.Totalamountofpassiveandactivechargesexchangedduringastart-up sequencevs.thehydrogenflowrate.Anodeinitiallyfilledwithnitrogen.Airand hydrogenhumidifiedto100%RH,fuelcelltemperaturefixedto50◦
C.
nomenonoccursinthelastsegment.Thebalanceofchargeshows thatthetimeandspaceintegralofthenegativecurrentsisstrictly equaltothatofthepositivecurrents.Startingfromthese consider-ations,itispossibletodefinethreetypesofchargeexchangeduring thestart-upsequence:
(i)Thepassivechargescorrespondtothenegativecurrents mea-suredinthepassivepartofthefuelcell(orangehatchedarea inFig.10candd),
(ii)Balancing charges are exchanged during the redistribution stageonceallsegmentsbecameactive;thenegative balanc-ingchargescorrespondtothebluehatchedareainFig.10b andc,
(iii)Activechargescorrespondtothepositivecurrentmeasuredin theactivepartminusthepositivebalancingcharges(equalin absolutevaluetothenegativebalancingcharges),
Qgreen areaN2 =S
IntermofMEAdegradation,carbonoxidationoccursinthe pres-enceofnegativecurrentsinthepassivepart,i.e.,itislinkedtothe amountofchargesflowingthroughthepassivesegments(aslong astheyremainpassive),whichdependsonthehydrogenflowrate andonthegasfillinginitiallytheanodecompartment,asshownin
Fig.11.
ThemostimportantresultinFig.11 concernsthesignificant impactofthegasfillinginitiallytheanodecompartmentonthe amountof charge produced duringthestart-up sequence: it is much lower in the presence of nitrogen, this difference being attributedtotheabsenceofafaradiccontributiontothenegative currentsflowingthroughthepassivesegments,i.e.,totheabsence ofcarboncorrosionatthecathodewhenthereisnooxygenatthe anode.
Onecanalsonoticethattheimpactofaironthefaradic contribu-tiontothepassivechargesdecreaseswiththehydrogenflowrates: thisisprobablyduetothelowkineticofcarbonoxidation(and possiblyofoxygenevolution)comparedtothecapacitivecurrent densitiesinthepassivepart.
3.4. Experimentalresults:carbonoxidationandperformance losses
Assumingthatthedoublelayercapacitiesareindependentof thegasintheanodecompartment,andthatoxygenevolutionis negligible compared tocarbon corrosion (which seems reason-ablebuthastobechecked),onecanestimatethemassofcarbon oxidizedduringastart-upsequencestartingfromthedifference betweentheamountofpassivechargesmeasuredwhentheanode compartmentisinitiallyfiledwithairandwithnitrogen.
ThislastresultisillustratedinFig.12a,whereonecanclearlysee thedecreaseofcarboncorrosionathighhydrogenflowrate.The estimatedweightofoxidizedcarbonperstart-upisconsistentwith theinitialamountusedascatalystsupportatthecathode(about 50mg)andwiththeamountofcarbondioxidemeasuredatthe cathodeoutletbyOfstadetal.[9].
Fig.12bshowsthatthecarbonlossismoreimportantforthe segmentslocatednearthehydrogenoutlet,becausetheyare sub-mitted for a longer time to higher oxidizing conditions. As in
Fig.12a,thelowerthehydrogenflowrate,thehighertheamount ofcarbonoxidized.
Theperformancelossesresultingfromcarbonoxidationcanbe characterizedbytheevolutionofthefuelcellperformanceandof thelocalimpedancespectra(EIS)measuredalongthegaschannels. Thiswasdoneforasuccessionofstart/stopsequencesaccordingto threedifferentprotocols:
1.anodeflushedwithnitrogen,nitrogenflowrateatshutdown andhydrogenflowrateatstart-upsetto12slph;
2.anodeflushedwithair,airflowrateatshutdownandhydrogen flowrateatstart-upsetto12slph;
3.anodeflushedwithair,airflowrateatshutdownandhydrogen flowrateatstart-upsetto2slph.
Betweentwosuccessivesequences,thefuelcellcurrentdensity wassetto0.5Acm−2during1h,withairandhydrogenhumidified to60%RH,stoichiometricratiosequalto3and1.2,anda temper-aturefixedto50◦C.Thelocalimpedancespectraweremeasured duringthissteadystateoperation.
Theevolutionofthefuelcellvoltageasafunctionofthe num-berofstart-up/shut-downisdepictedbythecurvesofFig.13.They showadramaticvoltagedropwhentheanodeisflushedwithair, allthemore sowhenthehydrogenflow rateis low.Note that theseexperimentshavebeencarriedoutwithnewMEA,which explainsthesharpincreaseinperformanceobservedduringthe firstsequencesinthethreecases.Whentheanodeisflushedwith nitrogen,theimprovementof theperformance canbeobserved approximatelyuntilthe150thsequence;beyond,thefuelcell volt-ageat0.5Acm−2 startstodecrease slowly, probablyunderthe effectofpotentialcyclingassociatedwiththevariousstagesofthe start/stopprotocol.
Theaging testswerestoppedwhen thefuelcellvoltagefell under0.63Vat0.5Acm−2.Thisvaluewasreachedafterabout60 cycleswiththethirdprotocol(airflushat2slph)and180cycles withthesecondprotocol(airflushat12slph).Startingfromthe resultsdepictedinFig.12a,onecanestimatethatthe correspond-ingcarbonlosseswereequaltoabout0.1 and0.02mg/start-up, respectively.
Fig.12. (a)Estimationofthecarbonloss/start-upasafunctionofthehydrogenflowrate.Airandhydrogenhumidifiedto100%RH,fuelcelltemperaturefixedto50◦ C.(b) Estimationofthecarbonloss/start-upforeachsegmentasafunctionofthehydrogenflowrate.Airandhydrogenhumidifiedto100%RH,fuelcelltemperaturefixedto50◦
C.
[email protected]−2vs.numberofstart/stopsequencesfollowing threedifferentstart/stopprocedures.
us(rough)estimatesofthecarbonlossesperstart/stopsequence (0.15and 0.03mg,respectively)and thus overthewholeaging tests:about10mgwiththethirdprotocol(airflushat2slph)and 6mgwiththesecondprotocol(airflushat12slph).Thesevalues represent10–20%ofthetotal massofcarbonusedtomake the electrodes,whichisintheorderofmagnitudeofthevoltagedrop (about10%,byreferencetothemaximumvalue– 0.69V– reached withnitrogen).
Notethattheseestimationsonlytendtoconfirmour interpre-tationsofthefaradicandcapacitiveoriginsofnegativecurrents. Accuratemeasurementofcarbonoxidationandofitsconsequences startingfromtheanalysisofinternalcurrentsisanobjective diffi-culttoachievebecauseofthecomplexityofthisphenomenon:for instance,theproductionofcarbonmonoxidepoisoningtheactive sites[19–22]shouldbetakenintoaccount,aswellasthePt coars-eningresultingfromcarbonloss[23,24].
Finally,theevolutionoftheimpedancespectraprovides com-plementaryinformationaboutthelocaleffectsofcarboncorrosion. TheresultsdepictedinFig.14correspondtothesecondaging pro-tocol(airflushat12slph):acomparisonismadebetweenlocal
ARTICLE IN PRESS
GModelPOWER-14621; No.ofPages8
8 A.Lamibracetal./JournalofPowerSourcesxxx (2011) xxx–xxx
Fig.15.(a)Highfrequencyresistanceprofilesafter49and173start/stopsequencesfollowingprotocolNo.2.(b)Doublelayercapacityprofilesafter49and173start/stop sequencesfollowingprotocolNo.2.
impedancespectrameasuredafter49and173start/stopsequences, whichconfirmsthatthesegmentlocatednearthehydrogen out-let (i.e.,near theend of the anode compartment, where air is introduced)aremoreaffected,asshownbytheincreaseoftheir impedance.Nevertheless,itremainspossibletofitthese experi-mentalspectrausingtheusualRandle’sequivalentcircuitandto identifythemainimpedanceparameters[25].
Adetailedanalysisoftheirprofilewouldbeoutofthescopeof thispaper,allthemoresosincetheusualhypothesesand simpli-ficationsleadingtotheexpressionoftheWarburgimpedanceare probablynotadaptedtomodelmasstransport.However,itappears clearlythatthehighfrequencyresistanceincreasesdramatically nearthehydrogenoutlet(Fig.15a).Thishighfrequencyresistance isthesumofthecontactresistancesbetweentheMEAcomponents, oftheprotonicresistanceoftheelectrolyte,andoftheelectronic resistancesoftheelectrodes,gasdiffusionlayersandbipolarplates: inourcase,itsincreaseismostprobablyduetodamagestothe catalystlayerentailedbycarboncorrosion(i.e.,totheincreaseof thecathodeelectronicresistance).Thisisconfirmedbythe simul-taneousdecreaseofthedoublelayercapacity(Fig.15b)nearthe hydrogenoutlet,whichisinterpretedasalossofelectrochemically activearea.Notethattheimpedanceofthesegmentslocatedinthe centreofthecellorclosetotheoppositeendshowsnosignificant variation.
4. Conclusion
Amethodbasedontheanalysisoflocalcurrentdensities mea-suredinasegmentedcellisdevelopedandusedtoestimatecarbon oxidationduringfuelcellstart-up.Theresultsshowclearlythatthe rateatwhichhydrogenisintroducedintotheanodecompartment hasasignificantimpactonfuelcelldurability:itmustbeashigh aspossibleinthepresenceofairoroxygen.Similarly,ifairhasto beintroducedintotheanodecompartmentatshut-down,itsflow ratemustbeashighaspossible.
Introducinghydrogeninananodecompartmentinitially occu-piedbyaircausesprofounddamagetothecathode,andthiscan entailasteepdecreaseofthefuelcellperformance.Local electro-chemicalimpedancespectroscopymeasurementscarriedouton damagedMEAshowthatthesegmentsthatremainforthelongest timeinthepresenceofair(whiletheothersarefedbyhydrogen) arethemostaffectedbycarbonoxidation.Carbonoxidationoccurs atthecathode,inthepresenceofnegativecurrents.However,these negativecurrentshavealsoacapacitiveorigin,whichexplainsthe fuelcellvoltageovershootobservedduringstart-up.Itispossible toevaluatethefaradicandthecapacitivecontributionstothe neg-ativecurrentsbycomparingthevaluemeasuredduringstart-up,in thepresenceofairintheanodecompartment,andinthepresence
ofnitrogen.Inthislastcase,onecanreasonablyassumethatthe faradiccontributionisnegligiblesincenosignificantdegradation occursafterasuccessionofstart/stopsequences.
Theresultsofthis workshouldhelpoptimisingstart-upand shut-downprotocolsinordertoimprovefuelcellsdurability.They canalsobeusedtoimprovecarbonoxidationmodelsduring start-up[26].
References
[1]C.A.Reiser,L.Bregoli,T.W.Patterson,J.S.Yi,ElectrochemicalandSolid-State Letters8(6)(2005)A273–A276.
[2] H.Tang,Z.Qi,M.Ramani,J.F.Elter,JournalofPowerSources158(2006) 1306–1312.
[3]N.Takeuchi,T.F.Fuller,JournaloftheElectrochemicalSociety155(7)(2008) B770–B775.
[4] Z.Siroma,N.Fujiwara,T.Ioroi,S.Yamazaki,JournalofPowerSources172(2007) 155–162.
[5]J.P.Meyers,R.M.Darling,JournaloftheElectrochemicalSociety153(8)(2006) A1432–A1442.
[6] J.Kim,J.Lee,Y.Tak,JournalofPowerSources192(2009)674–678.
[7] W.Gu,R.N.Carter,P.T.Yu,H.U.Gasteiger,ECSTransactions11(1)(2007) 963–973.
[8]H.Chizawa,Y.Ogami,H.Naka,A.Matsunaga,N.Aoki,T.Aoki,ECSTransactions 3(1)(2006)645–655.
[9] A.B.Ofstad,J.R.Davey,S.Sunde,R.L.Borup,ECSTransactions16(2)(2008) 1301–1311.
[10]L.C.Pérez,L.Brandão,J.M.Sousa,A.Mendes,RenewableandSustainableEnergy Reviews15(2011)169–185.
[11]S.A.Freunberger,M.Reum,J.Evertz,A.Wokaun,F.N.Büchi,Journalofthe ElectrochemicalSociety153(11)(2006)A2158–A2165.
[12]M.Reum,A.Wokaun,F.N.Büchi,JournaloftheElectrochemicalSociety156 (10)(2009)B1225–B1231.
[13]M.Reum,S.A.Freunberger,A.Wokaun,F.N.Büchi,Journalofthe Electrochem-icalSociety156(3)(2009)B301–B310.
[14]F.-B.Weng,C.-Y.Hsu,C.-W.Li,InternationalJournalofHydrogenEnergy35 (2010)3664–3675.
[15]Z.Siroma,J.Takahashi,K.Yasuda,K.Tanimoto,M.Inaba,A.Tasaka,ECS Trans-actions3(1)(2006)1041–1047.
[16]D.J.L.Brett,S.Atkins,N.P.Brandon,V.Vesovic,N.Vasileiadis,A.R.Kucernak, ElectrochemistryCommunication3(2001)628–632.
[17]G.Maranzana,C.Moyne,J.Dillet,S.Didierjean,O.Lottin,JournalofPower sources195(2010)5990–5995.
[18]Q.Shen,M.Hou,D.Liang,Z.Zhou,X.Li,Z.Shao,B.Yi,JournalofPowerSources 189(2009)1114–1119.
[19]S. Gottesfeld, J. Pafford, Journal of the Electrochemical Society 135 (1988)2651.
[20]T.E.Springer,T.Rockward,T.A.Zawodzinski,S.Gottesfeld,Journalofthe Elec-trochemicalSociety148(1)(2001)A11–A23.
[21] K.K.Bhatia,C.-Y.Wang,ElectrochimicaActa49(2004)2333–2341.
[22]S.Jiménez,J.Soler,R.X.Valenzuela,L.Daza,JournalofPowerSources151(2005) 69–73.
[23]P.J. Ferreira, G.J.la O’,Y. Shao-Horn,D. Morgan, R. Makharia, S.Kocha, H.A. Gasteiger c,Journal ofthe Electrochemical Society152 (11) (2005) A2256–A2271.
[24]F.H.Garzon,J.R.Davey,R.L.Borup,ECSTransactions1(8)(2006)153–166. [25] J.Mainka,G.Maranzana,J.Dillet,S.Didierjean,O.Lottin,ProceedingsWHEC
18,EnergieAgentur,NRW,Germany,2010.