Kinetic-Model-Free Analysis of Transient Absorption Spectra Enabled by 2D Correlation Analysis - Hniopek et al. - 2021 - Unknown

《Kinetic-Model-Free Analysis of Transient Absorption Spectra Enabled by 2D Correlation Analysis - Hniopek et al. - 2021 - Unknown》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

pubs.acs.org/JPCLLetterKinetic-Model-FreeAnalysisofTransientAbsorptionSpectraEnabledby2DCorrelationAnalysis§§JulianHniopek,CarolinMüller,ThomasBocklitz,MichaelSchmitt,BenjaminDietzek,*andJürgenPopp*CiteThis:J.Phys.Chem.Lett.2021,12,4148−4153ReadOnlineACCESSMetrics&MoreArticleRecommendations*sıSupportingInformationABSTRACT:Here,wepresent,tothebestofourknowledgeforthefirsttime,asystematicstudyofutilizing2Dcorrelationanalysisinthefieldoffemtosecondtransientabsorption(fs-TA)spectroscopy.Wepresentthattheapplicationof2Dcorrelationspectroscopy(2DCOS)tofs-TAspectroscopyenablesamodel-freemeanstoanalyzeexcitedstatekinetics,whichisdemonstratedonthemodelsystem[(tbbpy)Ru(dppz)]2+indifferentsolvents.WeshowthatTA-2DCOSis2abletodeterminethenumberofprocessescontributingtothetime-resolvedspectralchangesinfs-TAdatasets,aswellasextractthespectralresponseofthesecomponents.Overall,theresultsshowthatTA-2DCOSleadstothesameresultsasobtainedwithmethodsrelyingongloballifetimeanalysisormultivariatecurveresolutionbutwithouttheneedtospecifyapredeterminedkineticmodel.TheworkpresentedthereforehighlightsthepotentialofTA-2DCOSasamodel-freeapproachforanalyzingfs-TAspectraldatasets.ime-resolvedfemtosecondtransientabsorption(fs-TA)combinationofthesecomponents.Furthermore,applyingTspectroscopyisapowerfulmethodtoinvestigatetheMCRwithoutfurtherconstraintsoftenleadstoambiguousphotoinducedprocessesinmolecularsystems,thatis,theirresults.Thatiswhymostofthetimeadditionalapriori4relaxationandreactionpathwaysuponphotoexcitationontimeinformationisusedaswellinMCRanalysis.Tosummarize,scalesbetweenafewfemtosecondsandhundredsofnano-alltheaforementionedfs-TAanalysismethodsarenotfreeofaseconds.1−3fs-TAdatasetscontaintransientspectra{λ,λ,...,prioriknowledgeandassumptions.However,sincesucha12λn}forvariousdelay-times{Δt1,Δt2,...,Δtm}betweenprioriknowledgecannotbeobtainedorisnotavailableforphotoexcitationandrecordingthetransientspectra.Multiplemanymolecularsystems,analysismethodsthatdonotrelyon6methodsexistfortheanalysisoftheinherentlymultivariatefs-modelsarerequired.TAdatasets:4ThesemethodscanbedividedintoapproachesWithinthiscontribution,wewillshowthat2DcorrelationDownloadedviaUNIVOFCONNECTICUTonMay16,2021at13:48:09(UTC).referredtoasglobalandtargetanalysis.Globalanalysisspectroscopy(2DCOS)enablestheanalysisoffs-TAdatasetsmethodsfitthefs-TAdatawithasufficientnumberNofinamodel-freemanner.2DCOS,whichwasfirstintroducedinsequential(sequentialmodel)orindependent(parallelmodel)1993byNoda,hassincethenbeenwidelyappliedasaSeehttps://pubs.acs.org/sharingguidelinesforoptionsonhowtolegitimatelysharepublishedarticles.exponentialdecays,yieldingNspeciesassociateddifferencepowerfulapproachtoemphasizespecificspectroscopicspectra(SAS)ordecayassociateddifferencespectra(DAS),informationfromadatasetbyspreadingthespectral7informationacrosstwodimensions.Itparticularlyoffersarespectively.Forapplyingtheseglobalanalysismethods,itiswaytoanalyzesystematicchangesindatasetsrecordedunderrequiredthatthespectraofthesinglecomponents(N)donotanexternalperturbationbyrecoveringthecrosscorrelationchangewithtime.Bydoingsothefs-TAdatacanbedescribedfunctionofthespectralvariablesunderthisperturbationbyacertainnumberofcompartmentsNandtheir4(mathematicaldetailsof2DCOSandageneralguidetothecorrespondingspectra(N×n)anddecaykinetics(m×N).interpretation,thatis,theso-calledNodarules,canbefoundinTargetanalysisapproachesarebasedonaprioriknowledgeoftheSupportingInformation).Bydoingso,overlappingsignalspossibledecaymechanisms(e.g.,excitedstatebranchingorcanberesolvediftheychangedifferentlyundertheexternalequilibriumbetweentwoexcitedstates).Here,thegoalistodescribetherealconcentrationsofsinglecomponentsspectrallydescribedbySAS.BesidesthesemodeldependentReceived:March15,2021methods,model-freemethods,mostprominentlymultivariateAccepted:April14,2021curveresolution(MCR)approaches,5arealsocommonlyusedPublished:April23,2021inanalyzingfs-TAdata.WhileMCRisnotdependentonaspecifickineticmodel,itstillrequiresspecifyinganumberofcomponents,N,wherethespectramustbealinear©2021TheAuthors.PublishedbyAmericanChemicalSocietyhttps://doi.org/10.1021/acs.jpclett.1c008354148J.Phys.Chem.Lett.2021,12,4148−4153

1TheJournalofPhysicalChemistryLetterspubs.acs.org/JPCLLetterperturbation.Furthermore,2DCOSallowsonetodeterminetheorderinwhichspectralchangesoccur.2DCOSshouldthereforealsobeasuitablemethodtoanalyzeTAspectroscopicdata,wheretheexternalperturbationisthetimedelaybetweenpumpandprobebeam.However,2DCOShashardlybeenusedinthefieldofTAspectrosco-8−10py.Here,2DCOShassofarbeensimplyusedtocomplementmoreconventionaldataanalysiswithoutfullyexploringthepowerofthemethodtowardestablishingamodel-freeandhenceunbiasedanalysisofTAdata.Withinthiscontribution,wedemonstratetheapplicabilityof2DCOStoanalyzingfs-TAspectroscopicdatawithoutpriorknowledgeontheexcitedstatedynamics.Indoingso,weused[(tbbpy)Ru(dppz)]2+(tbbpy=4,4′-di-tert-butyl-2,2′-bipyr-2idine;dppz=dipyrido[3,2-a:2′,3′-c]phenazine)asamodelsystemtohighlightthecapabilitiesof2DCOSinanalyzingfs-2+TAdata.[(tbbpy)2Ru(dppz)]hasgainedconsiderableFigure1.Contourplotofthefs-TAspectraof[(tbbpy)Ru(dppz)]2+2interestinthepast20yearsduetotheso-calledlight-switchinMeCNinthetime-delayrangeof500fsto1750psafter11−14effect.Here,frequency-andtime-resolvedspectroscopicphotoexcitationat403nm.studiesrevealedtheexistenceoftwolow-lyingMLCTexcitedstatesonthe1,10-phenanthroline(phen)andphenazine(phz)sphereofthedppzligand.12−20Thecomprehensivelystudied340and360nmattheexpenseoftbbpycenteredππ*ESA12,16,16,23−262+locatedataround375nm.Thislong-lived,excitedstatepropertiesmake[(tbbpy)2Ru(dppz)]anidealemissivestatedecayswithacharacteristicrateofaroundbenchmarksystemtodemonstratethepotentialof2DCOSto−113,20(180ns).SincethechangesintheTAspectraareanalyzefs-TAdata.WethereforerecordedTAspectraof2+generallysubtle,theapplicationofadvancedanalysis[(tbbpy)2Ru(dppz)]inmultiplesolvents(acetonitriletechniquesisnecessarytoidentifythoseprocessesinsidethe(MeCN),dichloromethane(DCM),andMeCN/waterdata.mixtures)overadelaytimerangefrom500fsto1750ps.TheTA-2DCOSanalysisstartsbysplittingtheTAdatasetAdditionallydifferenttimesamplingschemes(linearandintomwindowsalongthedelay-timeperturbationaxis.logarithmic)wereapplied.TheseTAdatasetswereanalyzed2122Afterward,2DCOSspectraarecalculatedforeachtimeby2DCOSbasedmethodsusingGNUR4.0.3.windowandsubsequentlyanalyzed.ThedetailedidentificationInfluenceoftimesamplingschemes.Since2DCOSisbasedonoftheunderlyingkineticprocessesintheTAdataisachievedtheapplicationofeitheradiscreteFourieroraHilbertbyusingacombinationofthefollowingtwo2DCOStransform,itisnecessarytosampleinanequidistantmannerproperties:overtheperturbationinterval.However,inTAspectroscopy,logarithmicsamplingisnotuncommontoaccountforthe1.SignificantintensityintheasynchronousTA-2DCOSnecessarytimescalesneededtorecordfastandslowkineticspectrumΨofagivendelaytimewindowtmdirectlycomponentsinasingleexperiment.SuchalogarithmicindicatesthecontributionofmorethanoneprocessinsamplingcouldposeaproblemwhenimplementingTA-thiswindow.2DCOS.Theresultsobtainedinthisstudyrevealthatthe2.IfthecorrelationpatterninthesynchronousTA-samplingregimedoesnotsignificantlyinfluencetheresulting2DCOSspectrumΦinthetimewindowtmsignificantlyTA-2DCOSspectra,thusallowing2DCOSanalysistobediffersfromthepatternintm−1,theprocessinwindowtmappliedindependentlyfromtheexperimentaltimescanisdifferentfromtheprocessinwindowstm−1parameters(formoredetails,seeSupportingInformation).ThefirstpropertyisbasedonthefactthatspectralchangesProcessidentification.Inthefollowing,wewillshowthatTA-causedbyasingleprocesswillallhavethesamerate.However,2DCOScanbeusedtoidentifythenumberofprocessessincethecorrelationintensityinΨshowsdifferentratesofnecessarytodescribethecomplexdynamicbehaviorpresentinchange,morethanoneprocessmustcontributetotheoverallafs-TAdataset.Sinceitisaninherentlymodel-freemethod,spectralchanges.ThesecondpropertyisbasedonthefactthatTA-2DCOScanserveasastartingpointformodel-dependentthespectralsignatureofasingleprocessisconstantbetweenmethodsandhelptogeneratearefinedkineticmodel.differentwindows.Therefore,correlationpatternscausedbyaThemethodisdemonstratedonafs-TAdatasetofsingleprocesswillnotchangetheirpositionalongthespectral[(tbbpy)Ru(dppz)]2+inMeCN,recordedunderlogarithmic2axis.Here,onlythecorrelationintensity,correspondingtothesamplinginthedelaytimerangebetween500fsand1750psstrengthoftheprocess,canchangebetweenthewindows.(cf.Figure1).TheobservedtimedependentexcitedstateTherefore,asignificantchangeinΦbetweenthewindowsalsoabsorptionbehaviorcanbedescribedasfollows:Uponshowsthepresenceofmorethanoneprocess.photoexcitationat403nm,anensembleofmetal-to-ligandThese2DCOSpropertieshavebeenutilizedbyustocharge-transfer(MLCT)stateslocalizedonthetbbpyidentifytheunderlyingkineticprocessesintheTAdataof(1MLCT)anddppzligands(1MLCTand1MLCT)[(tbbpy)Ru(dppz)]2+inMeCN.Figure2ashowstheTA-tbbpyphenphz212,16,19arepopulated.Inthefirstfewpicoseconds,along-lived2COSspectrumforthedelaytimewindow0.6−1.1ps.Thephen-centered3MLCTispopulatedviaintersystemcross-phenlargediagonalcorrelationintensityofΦinthesynchronousing,vibrationalcooling,andinterligandhopping(tbbpy→TA-2DCOSspectrumrevealsthatthemajorityofchangesinphen).Thisismanifestedinthebuildupofππ*excitedstatethisdelay-timewindowoccurinthewavelengthregionbelowabsorption(ESA)ofphzandphencenteredstatesbetween400nm.Furthermore,thenegativeoff-diagonalcorrelation4149https://doi.org/10.1021/acs.jpclett.1c00835J.Phys.Chem.Lett.2021,12,4148−4153

2TheJournalofPhysicalChemistryLetterspubs.acs.org/JPCLLetterFigure2.TA-2DCOSspectraof[(tbbpy)Ru(dppz)]2+inMeCNinselecteddelaytimewindows.(a)CorrelationpatternsinΨindicate2contributionofmultiplecomponents.(b)AbsenceofpatternsinΨindicateasingleprocesscontributingtothiswindow.(c)CorrelationpatternsinΨshowtwocomponentscontributeinthiswindowagain.(d)CorrelationpatternsinΦdifferdrasticallyfrompanelsa−c,whileΨisempty,showingcontributionfromasingledifferentcomponent.intensitiesindicatethatintensitychangesabove350nmdoesnotcontributetotheTA-2DCOSspectrumofthethird(excited-stateabsorption(ESA)ofphz-andphen-centeredwindow(2.6−4.7ps)anymore,indicatingalifetimesmaller16,23,24states),areoppositetothosebelow350nm(ESAofthan2.6ps.Thesecorrelationintensitypatternscanbe12,16,17,26tbbpy-centeredstates).Smallerchangesarealsoobserveduntilthesixthdelay-timewindow(17−32ps)(cf.presentabove400nm,resultinginoff-diagonalcontributionsFigure2c),whereagainoff-diagonalcorrelationintensitycorrelatingwiththestrongchangesbelow400nm.ThepatternsinΨcanbefoundbetween320−380nmand400−asynchronousTA-2DCOSspectrum(Ψ)alsorevealssignifi-460nm.Therefore,athirdprocessmustbegintocontributetocantcorrelationpatterns,forwavelengthpositionscorrespond-thespectralchangesatlargerwavelengths.However,theingtothepatternsfoundinΦ.BasedonNoda’srules(seecontributionofthisthirdprocessisnotstrongenoughto7SupportingInformation),thechangesbetween320and350significantlyalterthepatternsinΦ.Therefore,itcanbenmhappenbeforethechangesatwavelengthsbetween350concludedthatthetimeconstantforthisprocesshastobeand380nm,thatis,twoprocessesmustcontributetothesesignificantlylongerthanthedelaytimeforthiswindow.12,16,25,26observedchanges.ThecorrelationpatternsobservedNosignificantcorrelationintensitycanbefoundintheinthisdelay-timewindowcanalsobefoundinthenextseventhandeighthwindows(32−220ps).Thesefindingswindow(extendingto2.4ps,datanotshown)withoutpointtowardthefactthatthesecondprocessdoesnotchangesignificantchanges,indicatingthatbothprocessescontributethespectrumanymore,indicatingalifetimesmallerthan32ps.tobothdelay-timewindows.ThiscanserveasanestimateforTheaforementionedthirdprocessstilldoesnotcauselargethelifetimesoftheseprocesses.enoughchangestoproducesignificantcorrelationintensityinFigure2bshowstheTA-2DCOSspectrumforthethirdΦ.Beginningfromtheninthwindow(220−415ps,cf.Figurewindow(2.6−4.7ps).ThepatternsfoundinΦforthisdelay-2d),correlationintensitypatternsinΦareobservableagain.timewindow,closelyresemblethosefoundinthefirsttwoHere,thethirdprocessisnowstrongenoughtocauseawindows,albeitwithanincreaseinintensity,indicatingcorrelationintensity.Theobservedcorrelationintensitystrongerchangesinthisdelay-timewindow.Ontheotherpatternsstaysconstantuntilthelargestdelay-timewindowhand,thepatternsfoundforΨinthefirsttwodelay-time(upto1750ps),indicatingthatthetimeconstantofthethirdwindowscannotbefoundanymoreinthisdelay-timewindow,processislargerthanthemeasurementinterval(inagreementwhichdemonstratesthatallspectralchangesinthisdelay-timewiththeexperimentallyfoundexcited-statelifetimeofabout12,13,20windowhappenconcertedly.Thisfindingshowsthatthefaster180ns).processcontributingtothechangesbelow350nm(interligandAsmentionedpreviouslyanimportantcriterionforthehopping)inthepreviousdelay-timewindows(0.6−2.6ps)applicationofTA-2DCOStoidentifytheunderlyingTA4150https://doi.org/10.1021/acs.jpclett.1c00835J.Phys.Chem.Lett.2021,12,4148−4153

3TheJournalofPhysicalChemistryLetterspubs.acs.org/JPCLLetterprocessesistheappropriatechoiceofthedelay-timewindowrelationshipbetweenchangesatdifferentspectralpositions,itsize.Thewindowshavetobelargeenoughtoallowforisthereforenecessarytoseparatethelinearpart(whichstemssignificantcorrelationintensityineachdelay-timewindowbutfromchangeswiththesameconstant)fromthenonlinearpartsmallenoughsothatfastprocessdonotgetmaskedbythe(whichstemsfromchangeswithdifferentconstants).largercorrelationintensityofprocesseswithlargetimeThisseparationofΦcanbeachievedbyusingthePearson27constants.correlationcoefficientsofthespectraldatasets.IftheTohelpidentifydelay-timewindowsofinterest,thatis,changesbetweentwospectralregionshaveapurelylinearwindowsinwhichthecorrelationpatternsinΦorΨchangerelationship,thiswillresultinahighPearsoncorrelationsignificantly,theabsolutesumsofthecorrelationintensitiesofcoefficient,whileregionswithanonlinearrelationshipwilltheTA-2DCOSspectraineachtime-delaywindowgiveanresultincomparativelylowervalues.Bythismeans,Φcanbeeasywaytoassesschangesinthedataset(cf.Figure3).Largeweightedtoonlyrepresentpurelylinearrelationships:nΦ=lin(,)νμΦ(,)Corr((),νμ·|ItItνμ())|(1)Here,Φ(ν,μ)isthesynchronouscorrelationintensityand|Corr(Iν(t),Iμ(t))|istheabsolutePearsoncorrelationcoefficientforaspecificspectralposition.Theparameterntunesthesuppressionofnonlinearcomponents.Thiscanbenecessaryasacertainamountofnonlinearityduetoexperimentalinfluenceslikenoiseorintensityvariationsisexpectedevenfortheoreticallyperfectlylinearrelationships.Byusingthislinearform,Φlin,onecanachievedanextractionofthespectralresponseICompbyslicingtheTA-2DCOSspectraalongtheselectedspectralregioninwhichtheprocesscauseschanges:I=Φ()νν·sgn(ỹ())(2)ResplinProctCompFigure3.SumoftheabsolutevaluesofΦ(blackcircles)andΨ(redHereΦlin(νProc)isaslicespectrumofΦlinalongaspectralsquares)inalltimedelaywindows.Graydottedlinesindicatethepositionrepresentativefortheprocessandsgn(ỹt(νComp))ispositionsofthewindowsdisplayedinFigure2,coincidingwithlargethesignofthedynamicspectraỹtatthisspectralposition.ThischangesinthevaluesofΦorΨ.multiplicationwiththesignofthedynamicspectraisnecessary,sinceΦdoesnotcontaintheabsolutedirectionofdifferencesineitherofthosesumvaluesbetweentwowindowschange.Itonlyencodestherelationshipbetweenthedirectionindicatethatthecorrelationpatternsmightalsochangeofchangesattwospectralpositionssothatthemultiplicationisdrasticallyandtherespectivewindowshouldthereforeberequiredtoensurethatthespectralresponseshavetherightinspectedmoreclosely.Thedelay-timewindowschoseninthesign.examplediscussedpreviouslywerethefirstwindow(whichFigure4ashowsthespectralresponsesextractedfromtheshouldalwaysbeinspectedasitdefinestheinitialcorrelationTA-2DCOSspectrashowninFigure2.Forprocess1,theslicepatterns)andthenwindows3,6,and9,whichcorrespondtospectrumalong330nminthefirstdelay-timewindow(FigurestarkchangesinthevaluesoftheabsolutesumsofbothΦand2a)isused,forprocess2,theslicespectrumalong360nminΨ.Asdiscussedpreviously,thosechangeswereindeedthethirddelay-timewindow(Figure2b)isused,andforaccompaniedbystrongchangesinthecorrelationpatterns.process3,theslicespectrumalong440nmintheeighthdelay-Overall,theresultspresentedsofarnicelydemonstratethetimewindowisused(Figure2d).TheseslicespectracapabilityofTA-2DCOStoidentifythenumberofexcitedcorrespondtoregionswherethecorrelationintensitychangesstateprocessesunderlyingthespectralchangesobservedcausedbyeachprocessaremaximal.duringafs-TAexperiment.Fortheexampleof[(tbbpy)2Ru-Figure4bshowsforcomparisontheDASretrievedfromthe(dppz)]2+inMeCN,threekineticprocessescouldbedatasetusingatraditionalgloballifetimeanalysiswithamodelidentified:afastprocesswithalifetimeshorterthan2.6psofthreefirstorderprocesses.Comparingbothfiguresitcanbe(vibrationalcoolingandinterligandhopping),asecondseenthatthekineticinformationextractedviaTA-2DCOSprocesswithalifetimeshorterthan32ps(relaxationofanicelyresemblestheDASextractedviagloballifetimeanalysis.subsetof3MLCTstates),andathirdprocesswithalifetimeTheonlyslightlimitationoftheTA-2DCOSapproachphzlongerthan1750ps(formationofalong-livedphen-centeredintroducedhereisthelossofintensityinformationbetweenstate).Thisinformationcanserveasusefulaprioristartingprocesses:Whiletheintensityratiosinoneresponseareverypointforfurtherevaluationsuchasagloballifetimeanalysis.similartothosefoundintheDAS,itisnotpossibletocompareComponentextraction.Inadditiontoidentifyingkinetictheintensitiesbetweendifferentprocesses,soTA-2DCOSisprocessesdescribingthechangespresentintheTAdatasetconfinedtoamorequalitativeanalysisofthedataset.andestimatingthecorrespondinglifetimes,itisalsopossibletoHowever,thislimitationismorethancompensatedbytheextractthecorrespondingspectralresponseoftheprocessesbypossibilitytoidentifykineticprocessesandtheirspectral2DCOSanalysisofTAdata.Thiscanbeachievedbyselectingresponsewithouttheneedtorelyonanyaprioriinformation,aspectralregioninwhichagivenprocess,identifiedasthatis,subjectiveguessingofcomponentsorrelianceonadescribedpreviously,causesspectralchanges.Ifthesamespecifickineticmodel.processcauseschangesinasecondspectralregion,theTofurtherdemonstratethevalidityoftheresponsesrelationshipbetweenthesechangesmustbepurelylinear.extractedviaTA-2DCOS,anothercommonapproachtoKeepinginmindthatΦentailsthelinearaswellasnonlinearanalyzefs-TAdata,MCRusingthealternatingleastsquared4151https://doi.org/10.1021/acs.jpclett.1c00835J.Phys.Chem.Lett.2021,12,4148−4153

4TheJournalofPhysicalChemistryLetterspubs.acs.org/JPCLLetterInconclusion,wecoulddemonstratethatTA-2DCOSallowstheextractionofthenumberofkineticprocessespresentinadatasetaswellasidentificationofthetemporalrangeinwhichtheyoccur.Thisprovidesanovelwaytoacquireaprioriinformationnecessaryfortheapplicationofadvancedmethodstoanalyzefs-TAdatasets,suchasgloballifetimeandtargetanalysis.Anobjectivewaytoobtainthisinformationposesamajorproblemintheapplicationofgloballifetimeanalysis,whichcanbeovercomebytheapplicationofthehereintroducedTA-2DCOSapproach.Furthermore,wecouldextractthespectralresponsefromtheTA-2DCOSspectrabyusingastatisticalcorrelationapproachoftheTA-2DCOSdataset.TheextractedresponsesqualitativelymatchtheresponseobtainedbyapplyinggloballifetimeanalysistothedatasetaswellascomponentspectraobtainedviaMCR-ALSanalysis,thusdemonstratingthevalidityofthisapproachonamodelsystemwithknownexcitedstatedynamics([(tbbpy)Ru(dppz)]2+).2TA-2DCOSthereforeoffersamodel-freeandaprioriknowledgefreemethodtoidentifyandqualitativelyextractthespectralresponsesofprocessespresentinaTAdataset,whichcanalsobeappliedtoprocessesthatdonotfollowtheconventionalfirstordermodels.TA-2DCOScanthereforeserveasamodel-freestartingpointforquantitativemethodslikegloballifetimeanalysisanddefinespectralresponsesthatthechosenkineticmodelshavetoreproduce.Figure4.(a),conventionalgloballifetimeanalysis(b,ComparisonofspectralresponsesextractedviaTA-2DCOScf.Supporting■*sıSupportingInformationASSOCIATEDCONTENTInformation),andMCR-ALS(c,cf.SupportingInformation).TheSupportingInformationisavailablefreeofchargeatResponsesinpanelawereextractedaccordingtoeq2usingthehttps://pubs.acs.org/doi/10.1021/acs.jpclett.1c00835.TA-2DCOSspectradepictedinFigure2.Response1,λ=330nm,Detaileddescriptionsofthemethodsusedinthisstudy,Figure2a.Response2,λ=360nm,Figure2b.Response3,λ=440nm,Figure2d.Forbettervisualization,DAS1andDAS2inpanelbashortintroductionto2Dcorrelationspectroscopy,andadditionalTA-2DCOSspectraof[(tbbpy)Ru(dppz)]2+aremultipliedby5andComp1and3inpanelcaremultipliedby3.2inDCM,H2O,andACN/H2Omixturesalongsidetherespectiveextractedresponses(PDF)4,5(ALS)approach,wasperformed.Sincethesystemstudied12−20hereiswell-known,thisanalysisrepresentsthebestcase■AUTHORINFORMATIONscenarioforMCR-ALS,asthenumberofcontributingspeciesCorrespondingAuthorsandtheirtimedependentrelativeconcentrationsarewellBenjaminDietzek−LeibnizInstituteofPhotonicTechnology,understood.Toperformthisbestcaseanalysis,thefirst,mean,07745Jena,Germany;FriedrichSchillerUniversityJena,andlastfs-TAspectrumwereusedasstartingvaluesforS(cf.InstituteofPhysicalChemistry,07743Jena,Germany;SupportingInformation,eq8).AsinitialvaluesforC(cf.FriedrichSchillerUniversityJena,AbbeCenterofPhotonics,SupportingInformation,eq8),exponentialtimeprofilesofthe07745Jena,Germany;orcid.org/0000-0002-2842-3537;forme−ktwiththelifetimes(1/k)obtainedfromgloballifetimeEmail:benjamin.dietzek@leibniz-ipht.deanalysiswereused.ThecomponentspectraextractedviathisJürgenPopp−LeibnizInstituteofPhotonicTechnology,approacharedisplayedinFigure4c.Asobservedcomparing07745Jena,Germany;FriedrichSchillerUniversityJena,theTA-2DCOSresponsestoDAS,theresponsesextractedviaInstituteofPhysicalChemistry,07743Jena,Germany;TA-2DCOSalsoresembletheMCR-ALScomponentspectraFriedrichSchillerUniversityJena,AbbeCenterofPhotonics,veryclosely,showingthatTA-2DCOSreproducesmultiple07745Jena,Germany;orcid.org/0000-0003-4257-establishedmodel-dependentapproachestoanalyzefs-TA593X;Email:juergen.popp@leibniz-ipht.despectra.Furthermore,weexploredtheuseofthenumberofAuthorsprocessesalongsidetheestimatedlifetimesandresponsesJulianHniopek−LeibnizInstituteofPhotonicTechnology,obtainedwithTA-2DCOSasstartingvaluesforanMCR-ALS07745Jena,Germany;FriedrichSchillerUniversityJena,analysis(fordetails,seeSupportingInformation).UsingthisInstituteofPhysicalChemistry,07743Jena,Germany;TA-2DCOSbasedselectionmethod,wecouldalsoreproduceFriedrichSchillerUniversityJena,AbbeCenterofPhotonics,theresultsobtainedbyMCR-ALSwiththeaprioriknowledge07745Jena,Germanyofthemodelsystem.ThisdirectlydemonstratesthepowerofCarolinMüller−LeibnizInstituteofPhotonicTechnology,TA-2DCOSasamodel-freestartingpointforquantitative07745Jena,Germany;FriedrichSchillerUniversityJena,methods.InstituteofPhysicalChemistry,07743Jena,Germany4152https://doi.org/10.1021/acs.jpclett.1c00835J.Phys.Chem.Lett.2021,12,4148−4153

5TheJournalofPhysicalChemistryLetterspubs.acs.org/JPCLLetterThomasBocklitz−LeibnizInstituteofPhotonicTechnology,orNoHydrogenBondsfromSolvent.J.Phys.Chem.A2004,108,07745Jena,Germany;FriedrichSchillerUniversityJena,4391−4398.InstituteofPhysicalChemistry,07743Jena,Germany;(12)Olson,E.J.C.;Hu,D.;Hörmann,A.;Jonkman,A.M.;Arkin,FriedrichSchillerUniversityJena,AbbeCenterofPhotonics,M.R.;Stemp,E.D.A.;Barton,J.K.;Barbara,P.F.FirstObservationoftheKeyIntermediateinthe“Light-Switch”Mechanismof07745Jena,Germany;orcid.org/0000-0003-2778-6624[Ru(phen)2dppz]2+.J.Am.Chem.Soc.1997,119,11458−11467.MichaelSchmitt−FriedrichSchillerUniversityJena,Institute(13)Brennaman,M.K.;Meyer,T.J.;Papanikolas,J.M.ofPhysicalChemistry,07743Jena,Germany;Friedrich[Ru(bpy)2dppz]2+Light-SwitchMechanisminProticSolventsasSchillerUniversityJena,AbbeCenterofPhotonics,07745StudiedthroughTemperature-DependentLifetimeMeasurements.J.Jena,Germany;orcid.org/0000-0002-3807-3630Phys.Chem.A2004,108,9938−9944.Completecontactinformationisavailableat:(14)Friedman,A.E.;Chambron,J.C.;Sauvage,J.P.;Turro,N.J.;Barton,J.K.AmolecularlightswitchforDNA:Ru(bpy)2(dppz)2+.J.https://pubs.acs.org/10.1021/acs.jpclett.1c00835Am.Chem.Soc.1990,112,4960−4962.(15)Kuhnt,C.;Tschierlei,S.;Karnahl,M.;Rau,S.;Dietzek,B.;AuthorContributionsSchmitt,M.;Popp,J.Investigationofsubstitutioneffectsonnovel§J.H.andC.M.contributedequally.Ru−dppzcomplexesbyRamanspectroscopyincombinationwithNotesDFTmethods.J.RamanSpectrosc.2010,41,922−932.Theauthorsdeclarenocompetingfinancialinterest.(16)Kuhnt,C.;Karnahl,M.;Tschierlei,S.;Griebenow,K.;Schmitt,M.;Schäfer,B.;Krieck,S.;Görls,H.;Rau,S.;Dietzek,B.;etal.■Substitution-controlledultrafastexcited-stateprocessesinRu−dppz-ACKNOWLEDGMENTSderivatives.Phys.Chem.Chem.Phys.2010,12,1357−1368.TheauthorsthanktheDeutscheForschungsgemeinschaft(17)delaCadena,A.;Davydova,D.;Tolstik,T.;Reichardt,C.;(DFG,GermanResearchFoundation)forfundingintheShukla,S.;Akimov,D.;Heintzmann,R.;Popp,J.;Dietzek,B.UltrafastframeworkofSFB/TRR234CATALIGHT(subprojectsC2andincellulophotoinduceddynamicsprocessesoftheparadigmA1,projectnumber364549901)andprojectBO4700/4-1.molecularlightswitch[Ru(bpy)2dppz]2+.Sci.Rep.2016,6,33547.TheauthorsgratefullyacknowledgeJensUhlig(Lund(18)Fees,J.;Ketterle,M.;Klein,A.;Fiedler,J.;Kaim,W.University)forprovidingthepythoncodeforgloballifetimeElectrochemical,spectroscopicandEPRstudyoftransitionmetalanalysisofthefs-TAdata.complexesofdipyrido[3,2-a:2’,3′-c]phenazine.J.Chem.Soc.,DaltonTrans.1999,2595−2600.■(19)Coates,C.G.;Jacquet,L.;McGarvey,J.J.;Bell,S.E.J.;Al-REFERENCESObaidi,A.H.R.;Kelly,J.M.ResonanceRamanProbingofthe(1)Miao,T.J.;Tang,J.CharacterizationofchargecarrierbehaviorInteractionbetweenDipyridophenazineComplexesofRu(II)andinphotocatalysisusingtransientabsorptionspectroscopy.J.Chem.DNA.J.Am.Chem.Soc.1997,119,7130−7136.Phys.2020,152,194201.(20)Brennaman,M.K.;Alstrum-Acevedo,J.H.;Fleming,C.N.;(2)Berera,R.;vanGrondelle,R.;Kennis,J.T.M.UltrafasttransientJang,P.;Meyer,T.J.;Papanikolas,J.M.Turningthe[Ru(bpy)-absorptionspectroscopy:principlesandapplicationtophotosynthetic2dppz]2+Light-SwitchOnandOffwithTemperature.J.Am.Chem.systems.Photosynth.Res.2009,101,105−118.Soc.2002,124,15094−15098.(3)Davydova,D.;delaCadena,A.;Akimov,D.;Dietzek,B.(21)Hniopek,J.;Schmitt,M.;Popp,J.;Bocklitz,T.PC2D-COS:ATransientabsorptionmicroscopy:advancesinchemicalimagingofPrincipalComponentBaseApproachtoTwo-DimensionalCorrela-photoinduceddynamics.LaserPhotonicsRev.2016,10,62−81.tionSpectroscopy.Appl.Spectrosc.2020,74,460−472.(4)Ruckebusch,C.;Sliwa,M.;Pernot,P.;deJuan,A.;Tauler,R.(22)RCoreTeamR.ALanguageandEnvironmentforStatisticalComprehensivedataanalysisoffemtosecondtransientabsorptionComputing;RFoundationforStatisticalComputing:Vienna,Austria,spectra:Areview.J.Photochem.Photobiol.,C2012,13,1−27.2020.(5)Kvalheim,O.M.;Liang,Y.Z.Heuristicevolvinglatent(23)Stark,C.W.;Schreier,W.J.;Lucon,J.;Edwards,E.;Douglas,projections:resolvingtwo-waymulticomponentdata.1.Selectivity,T.;Kohler,B.InterligandElectronTransferinHeterolepticlatent-projectivegraph,datascope,localrank,anduniqueresolution.Ruthenium(II)ComplexesOccursonMultipleTimeScales.J.Phys.Anal.Chem.1992,64,936−946.Chem.A2015,119,4813−4824.(6)Cheshire,T.P.;Brennaman,M.K.;Giokas,P.G.;Zigler,D.F.;(24)Damrauer,N.H.;Cerullo,G.;Yeh,A.;Boussie,T.R.;Shank,C.Moran,A.M.;Papanikolas,J.M.;Meyer,G.J.;Meyer,T.J.;Houle,F.V.;McCusker,J.K.FemtosecondDynamicsofExcited-StateA.UltrafastRelaxationsinRutheniumPolypyridylChromophoresEvolutionin[Ru(bpy)3]2+.Science1997,275,54−57.DeterminedbyStochasticKineticsSimulations.J.Phys.Chem.B2020,(25)Fees,J.;Kaim,W.;Moscherosch,M.;Matheis,W.;Klima,J.;124,5971−5985.Krejcik,M.;Zalis,S.Electronicstructureofthe”molecularlight(7)Noda,I.GeneralizedTwo-DimensionalCorrelationMethodswitch”bis(bipyridine)dipyrido[3,2-a:2’,3′-c]phenazineruthenium-ApplicabletoInfrared,Raman,andotherTypesofSpectroscopy.(2+).Cyclicvoltammetric,UV/visibleandEPR/ENDORstudyofAppl.Spectrosc.1993,47,1329−1336.multiplyreducedcomplexesandligands.Inorg.Chem.1993,32,166−(8)Hung,C.-C.;Yabushita,A.;Kobayashi,T.;Chen,P.-F.;Liang,K.174.S.Ultrafastrelaxationdynamicsofnitricoxidesynthasestudiedby(26)Coates,C.G.;Callaghan,P.;McGarvey,J.J.;Kelly,J.M.;visiblebroadbandtransientabsorptionspectroscopy.Chem.Phys.Lett.Jacquet,L.;Kirsch-DeMesmaeker,A.Spectroscopicstudiesof2017,683,619−624.structurallysimilarDNA-bindingRuthenium(II)complexescontain-(9)Vdovic,S.;Wang,Y.;Li,B.;Qiu,M.;Wang,X.;Guo,Q.;Xia,A.́ingthedipyridophenazineligand.J.Mol.Struct.2001,598,15−25.Excitedstatedynamicsofβ-carotenestudiedbymeansoftransient(27)Isaksson,T.Methodstoextractexclusivelylinearrelationshipsabsorptionspectroscopyandmultivariatecurveresolutionalternatingingeneralizedtwo-dimensionalcorrelationspectroscopy(2DCOS).least-squaresanalysis.Phys.Chem.Chem.Phys.2013,15,20026−Vib.Spectrosc.2004,36,251−259.20036.(10)Ruckebusch,C.;Mouton,N.;Gladytz,T.;Rendelmann,A.;Buntinx,G.;Sliwa,M.Chemometricanalysisoffemtosecondtransientabsorptionspectroscopydata:Studyofthephotochromismofanils.Sci.China:Phys.,Mech.Astron.2010,53,1024−1035.(11)Olofsson,J.;Önfelt,B.;Lincoln,P.Three-StateLightSwitchof[Ru(phen)2dppz]2+:DistinctExcited-StateSpecieswithTwo,One,4153https://doi.org/10.1021/acs.jpclett.1c00835J.Phys.Chem.Lett.2021,12,4148−4153