Photophysics of the sunscreen ingredient menthyl anthranilate and its precursor methyl anthranilate : a bottom up approach to photoprotection

(1)

warwick.ac.uk/lib-publications

Original citation:

Rodrigues, N. D. N., Cole-Filipiak, N. C, Horbury, M. D, Staniforth, Michael, Karsili, T. N. V.,

Peperstraete, Y. and Stavros, Vasilios G.. (2017) Photophysics of the sunscreen ingredient

menthyl anthranilate and its precursor methyl anthranilate : a bottom-up approach to

photoprotection. Journal of Photochemistry and Photobiology A: Chemistry, 353 . pp.

376-384.

Permanent WRAP URL:

http://wrap.warwick.ac.uk/96191

Copyright and reuse:

The Warwick Research Archive Portal (WRAP) makes this work of researchers of the

University of Warwick available open access under the following conditions.

This article is made available under the Creative Commons Attribution 4.0 International

license (CC BY 4.0) and may be reused according to the conditions of the license. For more

details see:

http://creativecommons.org/licenses/by/4.0/

A note on versions:

The version presented in WRAP is the published version, or, version of record, and may be

cited as it appears here.

(2)

Photophysics

of

the

sunscreen

ingredient

menthyl

anthranilate

and

its

precursor

methyl

anthranilate:

A

bottom-up

approach

to

photoprotection

N.D.N.

Rodrigues

a

,

N.C.

Cole-Filipiak

a

,

M.D.

Horbury

a

,

M. Staniforth

a,b

,

T.N.V.

Karsili

c

,

Y. Peperstraete

d

,

V.G.

Stavros

a,

*

a

UniversityofWarwick,DepartmentofChemistry,Coventry,CV47AL,UK

b

UniversityofWarwick,DepartmentofPhysics,Coventry,CV47AL,UK

c

TempleUniversity,DepartmentofChemistry,Philadelphia,PA19122,USA

d

SynchrotronSOLEIL,AILESBeamline,L’OrmedesMerisiers,SaintAubin,BP48,91192GifSurYvetteCedex,France

ARTICLE INFO

Articlehistory:

Received3October2017

Receivedinrevisedform22November2017 Accepted23November2017

Availableonlinexxx

Keywords:

Photochemistry Photophysics Photoprotection Sunscreens Time-resolved

Ultrafastlaserspectroscopy Bottom-upapproach

ABSTRACT

Theultrafastexcitedstatedynamicsofthesunscreeningredientmenthylanthranilate(MenA)andits precursormethylanthranilate(MA)werestudiedinvacuum(usingtime-resolvedionyieldspectroscopy) andinsolution(usingtransientelectronicabsorptionspectroscopy).MenAandMAbothshowlong-lived dynamics,withtheobservationofakineticisotopeeffectsuggestingthathydrogenmotionactsasthe ratedeterminingprocessintheoveralldecay.Complementarycomputationalstudiesexploringthe intuitivedecaypathwaysofMArevealedaboundS1statewithashallow‘up-hill’gradientwithrespectto

protontransfer.Fromtheseresults,itissuggestedthatphotoexcitedpopulationistrappedinthisexcited statefromwhichluminescenceoccursasaprominentdecaypathway.Thisworkhasshownthatthe photophysicsofMAandMenA–andhencetheirphotoprotectioncapabilities–arenotdrastically inﬂuencedbyaliphaticstructureorsolventenvironmentalone.Abottom-upapproach,suchastheone described herein, is essential to understand the combination of factors that afford optimum photoprotectionandtodevelopanewgenerationoftailormade,efﬁcacioussunscreens.

1.Introduction

Thedamagingeffectsofexcessiveexposuretoultraviolet(UV) radiationtolivingorganismsarewelldocumentedintheliterature [1_–3].Suchdamagesincludeerythema[4],aresultofexcessive skin irradiation with UV-A (400–315nm) and/or UV-B (315– 280nm)radiation–bothofwhicharedirectlyabsorbedbyseveral chromophores in human skin (e.g. melanins, acids and kynur-enines).UV-Aradiationisalsolinkedtoproductionoffreeradicals andskinaging[5],whilstUV-Bradiationcanbedirectlyabsorbed byDNAandinitiatethephotochemistryresponsibleformutagenic photolesions (e.g. cyclobutane pyrimidine dimers) [6]. If repair mechanismsfailtocorrectthesephotolesionsinDNA,thisdamage mayresultincarcinogenesis[7].Whilethehumanskinhasitsown naturalphotoprotectionmechanisms(providedbymelanin pig-ments),theseareofteninsufﬁcientforcontinuedandexcessivesun

exposure.Photoprotectionproducts,suchassunscreenlotions,are thereforerequiredinordertoprovideenhancedprotectionagainst UV-induced damage. Since their appearance in the early 20th century[8],sunscreenusagehasevolvedfromsimplyproviding protection against sunburn to being perceived as primary prophylaxis,actively preventing skinagingand skincancer[9]. Despiteattemptstoraiseawarenessoftherisksassociatedwith excessive sun exposure and the widespread availability of sunscreenlotions,theincidenceofskincancerhasincreasedin recentyears[10].Thereis,therefore,anobviousurgencyformore effectivesunscreens.

Anidealsunscreenmoleculeshouldabsorbacross(andhence provideprotectionagainst)theUV-AandUV-Bwavelengthrange ofthesolarspectrum[11].Onewouldexpectsunscreenmolecules toalsohavetheability todissipatetheabsorbed excessenergy rapidly –onanultrafasttimescale,i.e.femtosecond(1015s)to picosecond(1012s)–withoutdetrimenttomolecularstructure. Such detrimental photochemistry may be both intramolecular (fragmentation,isomerisation,etc.)and/orintermolecular (chem-icalreactionswithotherspeciesinthesunscreenmixtureorinthe

*Correspondingauthor.

E-mailaddress:v.stavros@warwick.ac.uk(V.G. Stavros).

https://doi.org/10.1016/j.jphotochem.2017.11.042

ContentslistsavailableatScienceDirect

Journal

of

Photochemistry

and

Photobiology

A:

Chemistry

(3)

skinitself)[12–15].Thestudyofthephotodynamicsthatfollowthe absorptionofUVradiation(photoexcitation)ofsunscreen mole-culeshasonlyrecentlygainedmomentum[14–21]andtherefore thesemechanismsremainpoorlyunderstood.

While reports on theultrafast photodynamics of commonly usedsunscreenmolecules,suchascinnamatesandsinapates[16–

18,20], alreadyexist, analogousinformation is not available for

anthranilatessuchasmethylandmenthylanthranilate(MAand MenA,Fig.1).MAisafoodgradeﬂavourandfragranceadditive usedinpersonalcareproducts[22]andisalsoaprecursortoMenA (commercialnameMeradimate),asunscreencomponentapproved by the US Food and Drug Administration[23]. In general, the anthranilatesareconsideredtobeaphotostable(non-degradable uponexposuretoUV)classofsunscreensduetotheintramolecular hydrogen bonding facilitated by the orthoposition of the NH2 group withrespect to the ester substituent [24]. Interestingly, however,underanaerobicconditions,aphotodegradation mecha-nismwasfoundtooccurforMAupon UVexposure[25].Under aerobic(morerealistic)conditions,anappreciablequantumyield ofﬂuorescence(0.376–0.549)wasobservedinvarioussolvents [26],aswellasa280

m

stripletstatelifetimeinaqueoussolution [22].Smallerquantumyieldsforsingletoxygensensitisationhave alsobeenreportedforMA[22].Suchradiativeandnon-radiative processespersistinMenA,whichpresentslargequantumyields for ﬂuorescence (0.640.06 in ethanol)[23] and intersystem crossing(0.34inethanolatroomtemperature)[27].Thepresence oftripletstatesinsolvatedMenAwasconﬁrmedbyKikuchietal. [27]byusingatripletquencher;thelifetimeofthetripletstatewas determinedtobe2.36s[27].

Inlightoftheavailableliterature,it seemsthatphotoexcited MAandMenAdissipateexcessenergyviaradiativedecay,whichis perceivablynotidealinasunscreenmolecule[14].Theradiative decayobservedinMAandMenAisatoddswithothersunscreen moleculeswhichtendtodecaytotheirgroundelectronicstatesvia non-radiativeinternal conversion(IC)onan ultrafast timescale

[16,19]. It is now well understood that ultrafast IC between

electronicstatesisfacilitatedbyconicalintersections(CIs)which arisewhencertainnuclearmotions(e.g.isomerisationand bond-stretches)drivedistinctelectronicstatestowardsdegeneraciesin conﬁgurationspace[28].Inrelatedsunscreenmolecules,manyCI geometries – ranging from trans-cis isomerisation to ring deformations, for example – have been identiﬁed as likely contributorsdrivingultrafastICfromtheexcitedtogroundstate

[16,17,29]. For some sunscreen molecules, notably oxybenzone,

isomerisation is preceded byan enol-keto type hydrogen atom motion;ketooxybenzonethenisomerisesbacktogroundstateenol oxybenzone[19].Suchtautomerisationprocessestypicallyoccur onultrafasttimescales,i.e.onmuchshortertimescalesthanthose probedintheaforementionedMAandMenAstudies.

Thenon-unityluminescencequantumyieldofMenAsuggests other,non-radiative,photophysicalprocessesarealsoinvolvedin itsrelaxationmechanisms.Ultrafastspectroscopytechniqueswere employedinthepresent worktoidentifytheseultrafast photo-physics taking place in photoexcited MenA and hence further informonitssuitabilityasasunscreeningredient.Inkeepingwith abottom-upapproach,bywhichtheeffectsofincreasingmolecular complexityareevaluated[14],bothMAandMenAwerestudied. Moreover, spectroscopic techniques were employed both in vacuum and in solution to evaluate the environmental effects ontheintrinsicphotodynamicsofthesemolecules.Computational studies were also performed to complement the experimental measurementsandprovidefurtherdetailonthetopographyofthe electronicstatesofthemoleculesstudied.Thisworkhighlightsthe importanceofinvestigativestudiestargetingtheintrinsic molec-ularandelectroniccharacteristicsthat providesunscreen mole-culeswiththeirphotoprotectioncapabilities.

2.Experimentalmethods

2.1.Absorptionspectra

TheUV/VisibleabsorptionspectraofMA(AlfaAesar,99%)and MenA (Aldrich, 98%), shown in Fig. 1, were obtained using a PerkinElmerLambda850UV/Visspectrophotometer.Thesample of each moleculewas prepared by dissolving MA or MenA in cyclohexane (99%, Fisher Scientiﬁc) with a concentration of approximately106M.

2.2.Time-resolvedionyield(TR-IY)

Thetime-resolvedionyield(TR-IY)setupusedinthisworkhas beendescribed previously[30–32]and isthereforeonly briefly described here, with furtherdetails of particular experimental conditionsprovidedwherenecessary.Acommercialfemtosecond (fs)Ti:Sapphireoscillator(Spectra-PhysicsTsunami)and regener-ativeamplifier(Spectra-PhysicsSpitfireXP)wereusedtoproduce

40fs laser pulses of 3mJ per pulse centred at 800nm. The fundamental 800nm output was subsequently split into three beams,eachwith1mJperpulse.Oneofthesebeamswasusedto pumpanopticalparametricampliﬁer(LightConversion,TOPAS-C) togenerate“pump”(

l

pu)pulsescentredeitherat300nm(4.13eV), 315nm(3.94eV), or 330nm (3.76eV). Thesewavelengths were chosentosampletheUV-AandUV-Bregionsofthesolarspectrum andthebroadabsorptionfeatureshowninFig. 1whilemaintaining adequate signaltonoise ratios.A secondlaser beampumpeda separate TOPAS-C which was used togenerate 260nm “probe” (

l

pr)pulses.Thepumpandprobepulsesweretemporallydelayed withrespecttoeachother(withadelaytimeof

D

t)byreﬂecting thepumpoff ahollowcornergoldretroreﬂector mountedona motorised delaystage allowing a maximum temporal delay of 1.2ns.

Thetwolaserbeamsintersectedamolecularbeamwhichwas producedbyseedingthetargetmolecules,heatedto50C(MA)or 90C(MenA),intohelium(3bar).Thegaseousmixturewasthen expandedintovacuum(107mbar)usinganEven-Laviepulsed solenoidvalve[33].Atthepointofintersection,

l

puexcitedthe species in the molecularbeam and

l

pr ionisedany excited (or photodissociated)species.Theresultingionswerefocusedontoa detector,consistingoftwomicrochannelplates(MCPs)coupledto

(4)

a phosphor screen, by three ion optics similar to the set up describedbyEppinkandParker[34].Thecurrentoutputfromthe frontofthephosphorscreen,gatedinionﬂighttimeoverthemass channelofeachparention(MA+_and_MenA+_),_was_measured_on_a digitaloscilloscope(LeCroyLT372Waverunner)andintegratedasa function of

D

t in order to produce TR-IY transients. These transientswerethenmodelledusingasumofexponentialdecays convolutedwitha Gaussian instrument response;more details regardingkinetic ﬁts can befoundin the Supplementarydata, sectionA(SD.A).Powerdependence studieswereperformedto ensurelinearsignalintensityvs.laserpower,i.e.thattheobserved dynamics are due to single-photon photoexcitation. Separate measurementsweretakenwiththepolarisationsofthepumpand probebeamsparallelandperpendiculartoeachotherinorderto calculatetheresultingmagicangleequivalenttransient[35].

DeuteratedMA(d1-MAandd2-MA,inwhicheitheroneortwo hydrogensonthe amine groupwere substitutedby deuterium atoms,respectively)wasproducedbystirringMAind4-methanol forapproximately72hunderanhydrousN2.Thesolventwasthen removedunderhighvacuum(0.5mbar)inanicebath(0C);the productwasimmediatelystoredunderanhydrousN2priortouse. TheH/Dexchangewasinferredbytheloss/reductionof1_H_NMR signalcorrespondingtothehydridepositionduetotheformation oftwoN–Dbonds(seeSD.B).

2.3.Transientelectronicabsorptionspectroscopy(TEAS)

The TEAS set-up usedin thepresent experiments hasbeen detailedpreviously[36,37]andisthereforeonlybriefly summar-ised.TEAS measurementswere obtainedfromseparate 1mM solutionsofMAandMenAinbothcyclohexane(VWR,>99%)and methanol(Sigma-Aldrich,99.6%).Thesesolutionswere recircu-latedthroughaflowcell(HarrickScientific)consistingoftwoCaF2 windows separated by 100

m

m thick PTFE spacers. Transient absorption spectra (TAS) were obtained by photoexciting the sampleusingthesameTOPAS-Coutputasforgas-phase experi-ments,i.e.300nm,315nmand330nmpumppulses,withﬂuences of 1–2mJcm2 per pulse. The probe pulses consisted of a broadband white light continuum (330–675nm), generated by focusingafractionofthethird800nmfundamentalbeamintoa 2mmthickCaF2window.Therelative polarisationbetweenthe pumpandprobebeamswasheldatmagicangle(54.7)byusinga

l

/2waveplate.Thetimedelaybetweenthepumpandprobepulses wascontrolledusingahollowcornergoldretroreﬂectormounted on a motorised translation stage in the probe beam path; maximum

D

t=2ns.Thesetupusedintheseexperimentsprovides an instrument response function with a full width at half maximum of 80fs. TAS for each time delay were calculated fromthe difference in probe pulse intensities passing through sequentially excited and non-excited sample prepared by a mechanicalchopperinthepumpbeampath.

To extractthedynamicalinformationfromtheTAS, aglobal ﬁtting sequentialmodel(e.g.At1_Bt2_C₎_was_employed_using_the

Glotaran software package [38]. The quality of the ﬁts was evaluateduponinspectionoftheresultingresidualsshowninthe SD.C.

3.Computationalmethods

Using Molpro 2010.1 [39], relaxed potential energy curves (PECs) along the neutral singlet ground state (S0), the ﬁrst electronically excited singlet state (S1), and the cation doublet ground state (D0+₎ _were _produced _using _the _{state-averaged} complete active space self-consistent _ﬁeld(SA-CASSCF) [40,41] methodcoupled toa cc-pVDZ basis set [42]. These PECs were

obtained by ﬁxing the H-bonded amino-centred NH stretch (henceforthNHbound)atvariousvalues(RNHb)andallowingthe remaininginternaldegreesoffreedomtorelaxtotheirrespective minima.Theactivespacecomprisedtenelectronsineightorbitals (10/8), includingthree

p

, two nand three

p

*valence orbitals. Following separate S0,S1 and D0+ _relaxations _across_R_NHb, _the energiesoftheS0andS1states,theﬁrstandsecondexcitedtriplet states(T1andT2)andtheD0+_state_were_computed_within_each relaxationsubsetusing thecompleteactivespacesecond-order perturbationtheory(CASPT2),basedonanSA-CASSCFreference wavefunctionandacc-pVDZbasisset.Astandardimaginarylevel shift of 0.5 EH was used to aid convergence and mitigate the involvementofintruderstates.

TwoPECswereproducedfortheS1state:onecorrespondingto the localised excitation (S1LE) and a second corresponding to chargetransfer(S1CT).TheS1LEPECdescribesthecaseforwhich charge remains localised on the ring upon photoexcitation (

p

ring!

p

*ring).TheS1CTPECdescribesinsteadthecaseforwhich, uponphotoexcitation,chargeisdislocatedfromthering

p

orbital tothecarbonyl

p

*orbital(

p

ring!

p

*carbonyl).Similarly,twoPECs associated withthe S0 of MAwerecalculated: theS0 PECwas produced by optimising both the geometry and electronic configuration of the S0 state of MA; the S0 charge transfer (S0CT)PECwasproducedbyoptimisingtheelectronicconfi gura-tionoftheS0statewhilekeepingthegeometryfixedtothatofthe S1CTstateateachcorrespondingRNHbvalue.

FollowingcomputationsalongRNHb,analogouscalculationsof theS0andS1stateswereundertakentoassessthepotentialenergy topographybetweenRNHb=1.5Åtoalow-energyS0/S1CIusing theCASPT2/cc-pVDZleveloftheory.Theintermediategeometries betweenRNHb=1.5ÅandtheS0/S1CIwereconstructedusinga linear interpolation of internal coordinates (LIIC). The CI was optimisedusing theSA-CASSCF/6-31G(d) level of theoryin the Gaussian 09 computationalpackage [43] witha reduced active spaceofsixelectronsinsixorbitals(6/6).

4.Results

4.1.Time-resolvedionyield(TR-IY)

TR-IYmeasurementsof MAandMenAweretakenfollowing photoexcitation at 315nm and 330nm. These yielded similar mono-exponential TR-IY transients, shown in Fig.2, both with decaylifetimesconsiderablylongerthanthetemporalwindowof theseexperiments(>>1.2ns).Incontrast,photoexcitationofMA

(5)

at300nmresultsinafastermono-exponentialdecay,asshownin green in Fig. 3, with a lifetime

t

H₁₇₀₀₂₀₀_ps _(where _the superscriptHreferstothehydrogenatedMAsample).

DeuteratedMA,bothd1-MAandd2-MA,werealsophotoexcited with300nmandtheTR-IYtransientsarepresentedinFig.3,inred andblue,respectively.ItisapparentthatdeuterationofMAyields slowerphotodynamics.However,twocaveatsmustbeconsidered intheinterpretationofourTR-IYresultsfordeuteratedMA.Firstly, the lifetimes extracted from kinetic ﬁts are still outside the temporalwindowofourexperimentsandhencetheratiobetween thelifetimesofMAandd1/d2-MAisthefocusof ourdiscussion ratherthantheirabsolutevalues.Secondly,wenotethelow signal-to-noiseratioofourdata,particularlyinthed2-MAtransientthis isduetoacombinationofexcitingMAatthetailofitsabsorption( seeFig.1)andthehighlevelsofD/Hexchangeexperiencedduring thecourseofourmeasurements.Bearingthesecaveatsinmind,we neverthelessquotetheextractedlifetimesasfollows:

t

d1₂₉₀₀

500psand

t

d2₂₅₀₀₉₀_ps _(where

t

d1 _and

t

d2_refer _to _the

decay time constants of d1-MA and d2-MA, respectively). An

averagekineticisotopeeffect(KIE)of1.50.4isthusdetermined fromtheTR-IYtimeconstantsextractedfromd1-MAandd2-MA.

4.2.Transientelectronicabsorptionspectroscopy(TEAS)

AllTASdata,evolutionassociateddifferencespectra(EADS)and residualsareshownintheSD.C.Asummaryofthetimeconstants extractedforbothMAandMenAatboth photoexcitation wave-lengths(315and330nm)canbefoundinTable1.Forillustrative purposes,theTASofMAincyclohexaneafterphotoexcitationwith 330nmareshowninFig.4 alongwiththeEADSandassociated time constants resulting from the sequential global _ﬁt of the returneddata.

4.2.1.Cyclohexane

TheresultsofMAintheweaklyperturbingsolvent, cyclohex-ane, are presented ﬁrst as this should be most akin to the environment invacuo.The TAS forboth excitationwavelengths (seeFig.4(a)andSD.C)displaysimilarspectralfeaturesandtheir temporalevolutionsarealsoanalogous.Assuchtheresultsforboth wavelengthsshallbediscussedtogether.TheTASconsistoffour featuresthatareapparentaftertheinitialexcitation.Thenegative featureat385nmcorrespondstostimulatedemission(SE)from theinitialelectronicexcitedstate,whichisingoodagreementwith the measured ﬂuorescence emission wavelength of MA (in ethanol) [12]. Alongside this SE, there is a broad excited state absorption (ESA),again fromthe initialexcited state,spanning

Fig.3.TR-IYtransientsresultingfromtheexcitationofMA(greencircles),d1-MA (reddiamonds)andd2-MA(bluetriangles)at300nm.Datawerenormalisedtotheir lastdatapoint.Solidlinescorrespondtothekineticﬁtsusedtoextractlifetimes. (Forinterpretationofthereferencestocolourinthisﬁgurelegend,thereaderis referredtothewebversionofthisarticle.)

Table1

SummaryofthetimeconstantsextractedfromtheTEASresultsforMAandMenA, incyclohexane(Cy)andmethanol(MeOH),atbothexcitationwavelengthsused.In vacuotimeconstants,extractedfromourTR-IYmeasurements,arealsopresented forcomparison.

MA

lpu 315nm 330nm

Invacuo t1>1.2ns t1>1.2ns

Cy t1=10.60.3ps t1=4.60.2ps t2>2ns t2>2ns t3>2ns t3>2ns MeOH t1=3.20.1ps t1=5.90.2ps

t2>2ns t2>2ns

MenA

lpu 315nm 330nm

Invacuo t1>1.2ns t1>1.2ns

Cy t1=1.60.1ps t1=6.60.4ps t2=1.50.1ns t2=0.60.05ns

t3>2ns t3>2ns MeOH t1=2.80.1ps t1=5.40.1ps

t2>2ns t2>2ns

Fig.4. (a)TASspectraforMAincyclohexaneat330nmphotoexcitation,with(b) respectiveEADSobtainedbyﬁttingwithasequentialmodel.Theresidualsforthis

(6)

420nmto675nm,withthreedistinctpeaksat455nm,525nmand 580nm.ThedynamicsofMAincyclohexane,fromblacktoredin Fig.4(b),canbedescribed simplyasanarrowingof theexcited statespectrum(e.g.thedecreaseinintensityaround625nm)anda slightblueshiftofallfeatures(mosteasilyseeninthenegative featureat380nm)within

t

1,withexcitedstatepopulationthen persisting beyond 2ns (

t

2). A third time component is also necessarytosuccessfullyﬁttheTAS;thisisshowninFig.4(b)as theblue EADSand is associated withthe decay time constant

t

3>2ns.EquivalentspectraforMenAareverysimilar:featuresare observedwithinthesamewavelengthrangesandqualitativelythe sametime dependentbehaviour is observed, albeit withsome differencesinextractedtimeconstants,assummarisedinTable1.

4.2.2.Methanol

TEASmeasurementswerealsocarriedoutinmethanol,amore perturbingsolventthancyclohexanegivenitspolarity.TheTASfor MAandMenAatbothexcitationwavelengthsalsodisplaysimilar featuresinmethanol(seeSD.C).Thetimeconstantsobtainedfor MAandMenAinmethanolaresummarisedinTable1.Inallcases (bothmoleculesandpumpwavelengths),theTASconsistofthree mainfeatureswhich appear uponphotoexcitation.A SEfeature fromthe_ﬁrstexcitedelectronicstateisobservedat400nm,which is in good agreement with thereported ﬂuorescence emission

wavelengthofMenAinethanol.[27]ThebroadESAinmethanol extendsfrom430nmto675nm,withprominentpeaksat450nm and 600nm. Similar to the observation in cyclohexane, the dynamicsof both MAandMenA inmethanol canbedescribed as a narrowing and spectral shift of the observed features, as discussed above. However,

t

3 is not required in methanol to properlyﬁtthedata.

4.3.Computationalstudies

Therelaxedpotentialenergycurves(PECs)forseveralexcited statesof MAwerecalculated, namely:theground state(locally excited, S0LE,and charge transfer, S0CT), theﬁrstelectronically excited singlet state (locallyexcited, S1LE, and charge transfer, S1CT),theﬁrstandsecondexcitedtripletstates(T1andT2)andthe cation doublet groundstate (D0+).These PECsare presentedin Fig.5andasummaryoftheverticalexcitationenergiescalculated fortheS1,S1CT,T1andT2statesofMAisgiveninTable2.

ThePECsinFig.5(a)werecalculatedasafunctionoflengthof theintramolecularNHbond,fromRNHb=1Å(ketotautomer)to RNHb=1.5Å(enoltautomer).TheS0optimisedgeometry, corre-spondingtotheketotautomer,isalsodisplayedinFig.5(a),along withtheprotontransfer(enol)geometry.ThesePECsrevealalack ofdrivingforceforcompleteprotontransferonboththeS0and S1CT states, evidenced by theirincrease in molecularpotential energyalongtheRNHbcoordinate(1.5eVriseforS0and0.21eVfor S1CT).ThePECsforMApresentedinFig.5(a)alsosuggestastrong degeneracyoftheS1CTandT1states(nearRNHb=1.2Å).

PECsfortheS0CTandS1CTstatesofMAfromthegeometriesat RNHb=1.5ÅandthenalongtheC¼Ctwistingmotioncoordinate (i.e.rotationoftheester groupwithrespecttothephenylring) werealsoproducedandaredisplayedinFig.5(b).Whiletheupper limitbarriertothistwistingmotioninMAwascalculatedtobe 0.4eVabovetheS1CTenergyatRNHb=1.5Å,thismotionwasalso

Fig.5.(a)PECsforvariouselectronicstatesofMAalongtheRNHbcoordinate,forwhichahydrogenmigratesfromtheaminegrouptowardstheneighbouringcarbonyl oxygen.Belowtheplot,theS0optimisedgeometry(keto,left)andtheprotontransfergeometry,i.e.atRNHb=1.5Å(enol,right)arealsoshown.Presentedherearetheground state(S0,bluesquares)andthechargetransfergroundstate(S0CT,hollowlightbluesquares),thelocallyexcitedfirstsingletstate(S1LE,holloworangetriangles)andfirst excitedsingletstate,whichhaschargetransfercharacter(S1CT,redtriangles),thefirstandsecondtripletstates(T1,greendiamonds,andT2,hollowlightgreendiamonds, respectively)andthefirstcationicsurface(D0+,purplecircles).(b)PECsfortheS0CTandS1CTstatesalongthecoordinatecorrespondingtothetwistingmotionaroundthe CCbondconnectingthephenylringandtheestersubstituent.Thismotionisillustratedbytheinsetonthelowerrightcorner.(Forinterpretationofthereferencestocolour inthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

Table2

Calculatedverticalexcitationenergiesforsomeofthesingletandtripletexcited statesofMA.

ElectronicstateofMA Verticalexcitationenergy(eV)

T1(pp*) 3.12

S1CT(pp*) 3.66

T2(pp*) 3.76

(7)

foundtoleadtoaS1/S0CI.ThecalculatedgeometryatthisCIisalso presentedinFig.5(b).

Additionalcomputationalresults,includingi)PECsalong the S1LE!S1CTcoordinate;ii)PECsalongtheNHandOMebond dissociationcoordinates;iii)theoptimisedgeometryattheS1/S0 CI;andiv)theCartesiancoordinatesfortheoptimisedS0,S1andS1/ S0CIgeometries,canbefoundinSD.F.

5.Discussion

In broadterms, allthedynamics observedfor both MAand MenAat various pump energies and various environmentsare long-livedandstrikinglysimilar.Particularlyinvacuum,theresults forbothmoleculesareremarkablysimilar;theTR-IYtransients, albeitqualitativelyfasterathigherpumpenergy,allshowlifetimes >1.2ns. In solution, there are otherdifferences brought onby solvent-solute interaction. The behaviour of MA and MenA in cyclohexaneisdescribedbythreetimeconstants,ratherthanthe two requiredtoﬁtthedata in methanol(vide infrafor further discussion). Nevertheless, regardless of environment or pump wavelength,bothMAandMenAdisplaysimilarphotodynamics, alwayswithalong-livedexcitedstatewhichisobservedtopersist foratleast1–2ns.Moreover,theUV/VisiblespectraforMAand MenA (Fig.1) are, once again,very similar, indicating that the additionalmenthylunitinMenAdoesnotperturbthelowenergy electronicstatesofthesystem,ascouldbeexpectedgiventhe non-perturbativecharacterofthementhylunit(i.e.sincethementhyl unit does not provide any extra conjugation or otherwise signiﬁcantlyalter theelectronic density of MA). Therefore, the dynamicsofMAandMenAwillbediscussedjointlythroughout this section, with attention being drawn to any differences betweenthemasappropriate.

ForMA,theﬁrstbandintheUV/Visiblespectrum,rangingfrom 360nmto290nm,isassignedtotheS1(

pp

*)statebasedonthe calculatedverticalexcitationenergies(Table2).Hence,excitation within the 330

l

300nm pump wavelength range (3.76– 4.13eV)usedintheseexperimentsislikelytoexclusivelypopulate theS1state.Inthegas-phase,photoexcitationateither315nmor 330nmyieldsamono-exponentialTR-IYtransientthatdoesnot recovertothebaselinewithinthetimescaleobservableinthese experiments, i.e. 1.2ns. When MA is photoexcited at 300nm, however,thereisamarkeddifferenceintheresultingtransient,in thatasinglelifetimeof

t

H₁₇₀₀₂₀₀_ps_is_now_extracted_(with somecertainty,sincethehalf-lifeis withinourtemporalprobe window).Thelong-livedphotodynamicsobservedsuggestthatany fastrelaxationpathwaysfromtheS1statearenegligibleandthat photoexcitedpopulationistrappedintheS1stateforatleasta nanosecond.

Considering the amine group of MA, one of the intuitive ultrafast decay pathways toconsider would be NHfree (non-intramolecularlybound)bonddissociation,ashasbeenobserved inaniline [44].Experimentally, however,resonantlyprobing H-atoms (with a 243nm probe) yields no obvious H+ _signal attributabletotheformation ofsingle-photoninducedNHfree bond fission along a dissociative excited state (see SD.E) [44]. Moreover,computationalstudiesrevealalargebarriertoNHfree bondfissionrelativetotheS1minimum,furthersuggestingthat thisisnotaviablerelaxationpathwayforMA(seeSD.F).Similarly, computationalstudiespredictaboundfirstexcitedstatealongthe OMe fission coordinate; this relaxation pathwayis thus also ruledout.Prefulvenicpathwaysarealsounlikelytoberesponsible forthephotodynamicsobservedinourexperimentssincetheyare usuallyaccessedatenergiesmuchhigherthanourphotoexcitation range[44,45].

Inotherstructurallysimilarsystems[46–48],suchassalicylic

acid[49,50],aninitiallylocallyexcited1

_pp

_*_state_may_couple_to_a

low-lyingchargetransfer(CT)stateuponphotoexcitation,which can mediate proton transferalong an intramolecular hydrogen bond. This type of excited state intramolecular protontransfer (ESIPT)is well reportedin theliterature[51–54] and hasbeen observed in the sunscreen molecule oxybenzone [19]. While salicylic acid undergoes a complete hydrogen transfer [49,50], anthranilic acid (the carboxylic acid variant of MA) has been reportedtoinsteadundergoexcitedstatehydrogenatom disloca-tion, with one of the amine hydrogens migrating towards the carbonyloxygen(akintoanincompleteketo-enoltautomerisation) [55]. Parallelphotoelectron spectroscopy measurementsonMA andMenA(seeSD.G)didnotyieldtheexpecteddistinctivefeatures foracompleteprotontransfermechanism,thatis,therewereno discreteelectronkineticenergyfeaturesthatcouldbeassignedto theketo and/orenolforms ofMA[56].Hence,it isplausibleto conclude that,similar tothecase of anthranilic acid, hydrogen atom dislocationalong theintramolecularhydrogenbond takes placeinphotoexcitedMA,insteadofcompleteprotontransfer.

ThepossibilityofhydrogenatomdislocationoccurringinMA was evaluatedexperimentally byobtainingTR-IY transients for MA, d1-MA and d2-MA photoexcitedat 300nm. Fromthetime constants extracted, an average kinetic isotope effect (KIE) of 1.50.4 (approximately p2) was determined, suggesting that non-tunnellinghydrogenatommotion isindeedinvolvedinthe dynamics observed for MA. Further evidence supporting a hydrogendislocationmechanisminMAisprovidedbythePECs displayed in Fig.5(a), which reveal a lack of driving force for completeprotontransferonthesestates,aspreviouslydiscussed. However, excitation in the 330

l

300nm pumpwavelength range(3.76–4.13eV)islikelytoprovidesufﬁcientenergyforthe relativelyshallow S1CTstate tobe sampled,and hence for the aforementionedhydrogenatomdislocationtotakeplace.

Given the experimentally determinedKIE for MA and upon analysisofthePECsinFig.5,weproposethat,afterphotoexcitation with330

l

300nm,theexcitedstatepopulationofMArelaxes fromthelocallyexcited(

p

ring!

p

*ring)S1statetothemorestable CT(

p

ring!

p

*carbonyl)S1state(seeSD.F),whereitisthentrappedin a relatively shallow well. The lack of direct evidence for this S1LE!S1CT transfer in our gas-phase experiments (i.e. no correspondingdecaytimeconstant),couldbeduetothisprocess happening ona timescalefaster thanthat measurablein these

experiments. Such charge migration phenomena have been

observed to occur on sub-femtosecond timescales (see, for example, reference [57]), which would make it impossible to resolvewithourexperimentalsetup.

(8)

motion to other vibrational modes or electronic states that ultimatelyfacilitatestheﬁnaldecayofMA.

Withexcited statepopulationtrappedintheS1CTstate,and withnodirectevidenceforanyCIsoranydissociativerelaxation pathwaysthatcanbeaccessedonanultrafasttimescalefromthe S1CT state, photoexcited MA would be expected to radiatively decaytoitsgroundstate.Asmentionedintheintroduction,thereis extensivereportingofluminescenceinbothMAandMenAinthe literature. Significant fluorescence quantum yields have been reported for both molecules, [23,26,27] with the observed emissionwavelengthbeingapproximately400nm(3.1eV).We noteherethatthisemissionwavelengthof400nmmatchesbest withthecalculatedenergydifferencebetweentheS1CTandthe S0CTstatesintheregionclosertotheverticalexcitationregionof thePEC.Hence,whiletheexcitedstatepopulationmaysamplethe entiretyoftheS1CTstate,transitiontothegroundstateappearsto bemorelikelyatshortNHboundbonddistances.Thisapparent emissiondependence ongeometry, namelyonNHbound bond distances, further highlights the significance of the hydrogen motioninthephotodynamicsofMA,asdiscussedearlier.

Finally,despitetherequirementsofElSayed’srulenotbeing metforS1CT(

pp

*)!T1(

pp

*),thestrongdegeneracyofsingletand triplet states in MA (near RNHb=1.2Å, see Fig. 5(a)), could facilitateintersystemcrossing(ISC)and,consequently, phospho-rescence.WhilethereisnodirectevidenceforISCinthegas-phase experiments reported here, previous studies in solution have foundphotoexcitationofMAtoresultinanexcitedtripletstate withalifetimeof280

m

s.[22]InMenA,thereportedtripletstate lifetimeis2.360.01s,withaphosphorescencequantumyieldof

F

P=0.0920.009.[27]Theinsensitivityofourgas-phase experi-mentstoISC(S1CT!T1)maybeduetotheverysimilarsurfacesof theirrespectivePECs(seeFig.5(a)).Thetransitionbetweenthese states(S1CT!T1)maybedifﬁculttodistinguishinthese experi-mentsduetoverysimilarionisationcrosssections,theeffectlikely beingaccentuatedbytheshallowcharacterofthePECoftheﬁrst cationicstate,D0+_._Therefore,_it_is_reasonable_to_assume_that,_while ISCoccursinbothMAandMenA,itmaybeconvolutedandhence indistinguishableinourgas-phaseexperiments.Itisalsopossible thatISCoccursoutsidethetemporalwindowoftheseexperiments (asourTEASresultssuggest,videinfra).

Theresultsfromourexperimentsinsolutionrevealasimilar decay mechanism for both MA and MenA, with evidence for luminescence.Immediatelyobviousistheagreementbetweenthe wavelengthoftheSEcomponentobservedinourexperimentsand thepreviously measured emissionwavelengthof both MA and MenA,bothat400nm,[23,26,27]whichprovidesevidencefor ﬂuorescence.EvidencefortheexistenceoftripletstatesinbothMA andMenAand,hence,phosphorescencecanbedrawnfroma closeanalysisoftheEADSandextracteddecaytimeconstants,asis nowdiscussed.Thesequential modelﬁtof theobtainedTAS in cyclohexaneyields

t

12–11ps(depending onspeciﬁc molecule and pump wavelength). This time constant is assigned to vibrationalenergytransfer(VET),which includesboth intramo-lecularvibrationalredistribution(IVR)andintermolecularenergy transfer(IET).SinceIETisaprocessbywhichvibrationalcoolingis achievedbyenergytransfertothesolvent,thisisaprocessthat cannotoccurin thegas-phase. Thevariation in

t

1 can thenbe understood in terms of varying degrees of solvent-solute interactions which willhave an effect onthe rate of IET. [58] The TAS for both molecules in cyclohexane (at either pump wavelength)requiretwomoretimeconstantsforasuccessfulﬁt. ForMA,

t

2issimplyassignedtothedecayoftherelaxedS1state, akintoourobservationsinvacuum;

t

3isassignedtothedecayof thetripletstateof MA,sincetheEADScorrespondingto

t

3are spectrallyvery similar to the triplet state transient absorption spectrumofMenAwhichKikuchietal.obtained.[27]

There are two main differences for the analogous time constantsextractedforMenA:i)theEADSassociatedwith

t

3do notresembletheaforementionedmeasurementsbyKikuchietal., butinsteadseemtoretainsomeofthespectralfeaturesthatare assigned to thelong-lived S1 state (see SD.D) and ii)

t

2 varies between 0.6–1.5ns (c.f.>2ns for MA). As discussed in more detailintheSD.D,subtractingthenormalised

t

3EADSfromthose of

t

2 yieldsa spectrumthat isindeedsimilartomeasurements whichKikuchietal.carriedout(inethanol).Hence,wepropose that the spectral differences between our MenA results in cyclohexane and thepreviously reported tripletstate transient absorption are due to extensive convolution of the processes associated to

t

2 and the appearance of a triplet state. The shortening of

t

2 may therefore not be a reﬂection of a faster decayoftheS1state,butratheraresultofoursequentialmodelnot beingabletodeconvolutetheprocessesassociatedwith

t

2and

t

3. Indeed,thefactthatthelong-livedcomponentobservedforMenA incyclohexane(

t

3>2ns)containsspectralfeaturessimilartothe EADS corresponding to

t

2 suggests that the S1 state is still populated at 2ns (the limit of the probe window of these experiments).

Toconcludethediscussionofourresultsincyclohexane, we notethat,contrarytotheobservationinvacuum,photoexcitation ofMAat300nmdoesnotdrasticallyaccelerateitsdecay(seeSD.C). ItislikelythatIETinsolutionfacilitatesfasterrelaxationtotheS1 minimum,fromwhichthebarriertotwistingmotioncannolonger beovercome(seeabove).Moreover,wenotethatEADSforMAin cyclohexane at 300nm are similarto those of MenA (at other excitation wavelengths), which supports the theory that the spectral/temporaldiscrepanciesdiscussed earlierareduetothe sequentialmodelusedtoﬁtthesedata,ratherthananychemicalor physicaldifferencesbetweenMAorMenA.

Inmethanol,ontheotherhand,onlytwotimeconstantsare necessarytoﬁttheTASofeachmolecule;thereisnoevidencefor triplet states in this solvent for either MA or MenA. However, tripletstateswerepreviouslyobservedinMenAinethanol[27], henceitishighlylikelythattripletstatesareformedinmethanol forbothMAandMenAoutsidethetemporalwindowofourTEAS experiments.Assumingdislocationoftheintramolecularlybound hydrogenatomis asrelevantin solutionasweproposed inthe discussionofourresultsinvacuum,andhenceconsideringthat hydrogenmotionfacilitatesISC,itwouldnotbesurprisingthata polarproticsolventwouldaffecttheobservedbehaviourofMAand MenA.

In summary, regardless of pump wavelength (315nm vs. 330nm,and300nmforMA),environment(vacuumvs.solution), and/or solvent polarity (cyclohexane vs. methanol), the photo-dynamicsofMAandMenAcanessentiallybedescribedasaslow decaycorrespondingtolong-livedexcitedstateswhichwilllikely luminesce.ThedifferencesinmolecularstructurebetweenMAand MenA do not change the observed photodynamics (as is the observationforothersunscreenmoleculesandtheirprecursors)

[16,29] to such an extent that would justify the use of either

moleculeasa sunscreen,asfarasphotochemicaland/or photo-physicalfactorsareconcerned.Thelackofadrasticsolventeffect onthephotodynamicsofMAorMenA,ontheotherhand,isinstark contrastwithobservationsforothersunscreens,whichreinforces the importance of a fundamental understanding of sunscreen photophysics[16,29].

(9)

statesmeansthatphotoexcitationofMA/MenAwithUV-Aand UV-Bradiation thewavelength range of interest in a sunscreen context–providesinsufﬁcientenergyforexcitedstatepopulation toaccessanyCIsthatwouldfacilitateultrafastrelaxation.This,in turn,isanon-idealscenarioforsunscreenmoleculessincethelong lifetimeofanyexcited(potentiallyreactive)statesmayincrease the chance for undesirable intra- or intermolecular chemical reactions.Nevertheless,MenAisusedinthesunscreenindustry– albeitscarcely–whichraisesthequestion:whatisthefateofthe excess energy in MenA after absorption of UV radiation by a sunscreenlotion?Toaddressthisquestion,ourstudiesonisolated moleculeswillneedtobeexpandedtowardscomplexmixturesof sunscreen molecules, solvents, stabilisers, etc. Of particular relevancearetherecentstudiesbyMatsumotoetal.whichhave reported diffusion-controlled (highly efﬁcient) triplet–triplet

energy transfer processes from MenA to other sunscreen

molecules(octocrylene and octylmethoxycinnamate) [59]. This workbyMatsumotoetal.highlightstheimportanceofidentifying and understanding the fates of excess energy in a sunscreen context. In addition, further research into the photophysics of sunscreen molecules is necessary to identify the molecular/ electronic features that allow for ideal photoprotective characteristics.

6.Conclusion

ThesimilarityoftheresultsforMAandMenAinvacuumarein accordance with observations for other sunscreen molecules, which havesimilarphotodynamics tothose of theirprecursors [29].Nevertheless,molecularstructurealterations–particularly thoseofamoreperturbativenature–willofcoursehaveanimpact in a molecule’s photodynamics as has been shown for other systems[20].Externalenvironmentfactors,suchassolvents,also tendtohaveagreatimpactintheobservedphotodynamics[16,29]. However,interestingly,thisisnotthecaseforthesystemsunder studyinthiswork.Wehaveshownthat,photophysicallyspeaking, bothMAandMenAbehaveasnon-idealsunscreensindependently ofexternalenvironment.

Designingsunscreenmoleculesforoptimisedphotoprotection canonly be achievedonce the molecularcharacteristicswhich provide photoprotective capabilities are identiﬁed. To identify thesephotoprotection-affordingcharacteristics,itisnecessaryto considertheeffectsofmolecularstructureontheelectronicenergy topography of the chromophore. This should culminate in a comprehensiveunderstandingnot only of light absorption,but alsooftheensuingphotodynamicsinthesemolecules.Inthecases explored in this work, we suggest that the intramolecular hydrogen bond stabilises the excited states of MA and MenA, preventing any molecular fragmentation from occurring but ultimatelyhinderingeffectivedissipationofpotentiallyharmful excessenergy.The S1/S0CI alonga twisting motioncoordinate, common for other ESIPT molecules, is inaccessible in these anthranilates (upon photoexcitation with UV-A/B). Our work therefore highlights two aspects of the anthranilatesthat may beinterrogatedinordertodrivefasterexcitedstatedecaysinMA andMenA:i)alterationstotheintramolecularhydrogenbond,e.g. changingtheaminegroupbya differentsubstituent,changeits ring position or eliminate it altogether; and ii) changing the substituentfunctionalgroupsinanattempttolowerthebarrierto the twisting motion. In addition, these studies on isolated molecules will need to be expanded towards more realistic environmentsinthesunscreencontext,i.e.complexmixturesof sunscreenmolecules,solvents,stabilisers,etc.Workis currently underwaytoinvestigatetheseeffectsinourlaboratory.

Acknowledgements

TheauthorswouldliketothankDr.W.D.Quanforassistance with MA deuteration, and the Warwick Centre for Ultrafast Spectroscopy (WCUS; go.warwick.ac.uk/wcus) for use of the PerkinElmer Lambda 850 UV/Vis spectrophotometer. N.D.N.R. thankstheEPSRCfordoctoralfunding.N.C.C.F.andM.D.H.thank theLeverhulme Trust forpostdoctoral funding.M.S. thanksthe EPSRCforanequipmentgrant(EP/N010825).Y.P.thankstheÉcole Normale Supérieure Paris-Saclay for doctoral funding. V.G.S. thanks the EPSRC for equipment grants (EP/J007153 and EP/ N010825)andtheRoyalSocietyandLeverhulmeTrustforaRoyal SocietyLeverhulmeTrustSeniorResearchFellowship.

AppendixA.Supplementarydata

Supplementary data associated with this article can be found, in theonline version, at

https://doi.org/10.1016/j.jphoto-chem.2017.11.042.

Anyremainingdata,supportinginformationandclari_ﬁcations can be found in the supplementary data document published alongside this article. The Supplementary material provided consistsof:A.KineticFits;B.MAandd2-MA1HNMR;C.Remaining TEAS spectra for MA and MenA with respective EADs and ﬁt residuals;D.TripletstateappearanceinMenAincyclohexane;E. VelocityMapImaging(VMI)ofH-atoms.F.Additionalresultsfrom computationalstudies.G.PhotoelectronSpectroscopy.

References

[1]F.R. de Gruijl, Eur. J. Cancer 35(1999)2003–2009, doi:http://dx.doi.org/ 10.1016/s0959-8049(99)00283-x.

[2]F.R.deGruijl,MethodsEnzymol.319(2000)359–366,doi:http://dx.doi.org/ 10.1016/s0076-6879(00)19035-4.

[3]R.P.Sinha,D.P.Häder,Photochem.Photobiol.Sci.1(2002)225–236,doi:http:// dx.doi.org/10.1039/b201230h.

[4]M.Brenner,V.J.Hearing,Photochem.Photobiol.84(2008)539–549,doi:http:// dx.doi.org/10.1111/j.1751-1097.2007.00226.x.

[5]J.L.McCullough,K.M.Kelly,Ann.N.Y.Acad.Sci.1067(2006)323–331,doi: http://dx.doi.org/10.1196/annals.1354.044.

[6]D.E. Brash,Photochem. Photobiol. 91 (2015) 15–26, doi:http://dx.doi.org/ 10.1111/php.12377.

[7]F.R.DeGruijl,H.Rebel,Photochem.Photobiol.84(2008)382–387,doi:http:// dx.doi.org/10.1111/j.1751-1097.2007.00275.x.

[8]J.L.M.Hawk,EuropeanHandbookofDermatologicalTreatments,Springer, BerlinHeidelberg,2015,pp.1519–1528, doi:http://dx.doi.org/10.1007/978-3-662-45139-7_149.

[9]A.S.Farberg,A.M.Glazer,A.C.Rigel,R.White,D.S.Rigel,J.Am.Med.Assoc.: Dermatol.153(2017)99–101,doi:http://dx.doi.org/10.1001/

jamadermatol.2016.3698.

[10]N.A.Shaath,PrinciplesandPracticeofPhotoprotection,SpringerInternational Publishing,2016,pp.143–157, doi:http://dx.doi.org/10.1007/978-3-319-29382-0_9.

[11]U.Osterwalder,B.Herzog,Photochem.Photobiol.Sci.9(2010)470–481,doi: http://dx.doi.org/10.1039/b9pp00178f.

[12]A.E.Jones,DoctorofPhilosophy,DurhamUniversity,2000.

[13]R.Krishnan,N.Kollias,C.A.Elmets,B.Choi,H.Zeng,T.M.Nordlund,R.S.Malek, B.J.Wong,J.F.R.Ilgner,K.W.Gregory,etal.,SPIEProc.6842(2008)6842081– 6842088,doi:http://dx.doi.org/10.1117/12.758693.

[14]N.D. Rodrigues, M. Staniforth, V.G. Stavros,Proc.e R. Soc. A 472(2016) 20160677,doi:http://dx.doi.org/10.1098/rspa.2016.0677.

[15]L.A.Baker,V.G.Stavros,Sci.Prog.99(2016)282–311,doi:http://dx.doi.org/ 10.3184/003685016x14684992086383.

[16]E.M.Tan,M.Hilbers,W.J.Buma,J.Phys.Chem.Lett.5(2014)2464–2468,doi: http://dx.doi.org/10.1021/jz501140b.

[17]Y.Miyazaki,K.Yamamoto,J.Aoki,T.Ikeda,Y.Inokuchi,M.Ehara,T.Ebata,J. Chem.Phys.141(2014)244313,doi:http://dx.doi.org/10.1063/1.4904268. [18]J.C.Dean,R.Kusaka,P.S.Walsh,F.Allais,T.S.Zwier,J.Amer.Chem.Soc.136

(2014)14780–14795,doi:http://dx.doi.org/10.1021/ja5059026.

[19]L.A. Baker,M.D. Horbury, S.E.Greenough,P.M.Coulter, T.N. Karsili, G.M. Roberts,A.J.Orr-Ewing,M.N.Ashfold,V.G.Stavros,J.Phys.Chem.Lett.6(2015) 1363–1368,doi:http://dx.doi.org/10.1021/acs.jpclett.5b00417.

(10)

[21]M.D.Horbury,L.A.Baker,N.D.N.Rodrigues,W.-D.Quan,V.G.Stavros,Chem. Phys.Lett.673(2017)62–67,doi:http://dx.doi.org/10.1016/j.

cplett.2017.02.004.

[22]C.Gambetta,J.Natera,W.A.Massad,N.A.García,J.Photochem.Photobiol.A: Chem.269(2013)27–33,doi:http://dx.doi.org/10.1016/j.

jphotochem.2013.06.013.

[23]A.Beeby,A.E.Jones,Photochem.Photobiol.72(2000)10–15,doi:http://dx.doi. org/10.1562/0031-8655(2000)072<0010:tppoma>2.0.co;2.

[24]N.J.Lowe,N.A.Shaath,M.A.Pathak,Sunscreens:Development,Evaluation,and RegulatoryAspects,MarcelDekkerInc,NewYork,1997.

[25]L.C.Eugeny Aronov, Pestic.Sci. 47 (1996)355–362,doi:http://dx.doi.org/ 10.1002/(SICI)1096-9063(199608)47:4<355:AID-PS429>3.0.CO;2-3. [26]W.H. Melhuish, J. Phys.Chem. 65(1961) 229–235,doi:http://dx.doi.org/

10.1021/j100820a009.

[27]A.Kikuchi,K.Shibata,R.Kumasaka,M.Yagi,Photochem.Photobiol.Sci.12 (2013)246–253,doi:http://dx.doi.org/10.1039/c2pp25190f.

[28]G.A.Worth,L.S.Cederbaum,Ann.Rev.Physi.Chem.55(2004)127–158,doi: http://dx.doi.org/10.1146/annurev.physchem.55.091602.094335.

[29]Y.Peperstraete,M.Staniforth,L.A.Baker,N.D.Rodrigues,N.C.Cole-Filipiak,W. D.Quan,V.G.Stavros,Phys.Chem.Chem.Phys.18(2016)28140–28149,doi: http://dx.doi.org/10.1039/c6cp05205c.

[30]A.Iqbal,M.S.Cheung,M.G.Nix,V.G.Stavros,J.Phys.Chem.A113(2009)8157– 8163,doi:http://dx.doi.org/10.1021/jp9031223.

[31]K.L.Wells,G.Perriam,V.G.Stavros,J.Chem.Phys.130(2009)074308,doi: http://dx.doi.org/10.1063/1.3072763.

[32]M.Staniforth,J.D.Young,D.R.Cole,T.N.Karsili,M.N.Ashfold,V.G.Stavros,J. Phys.Chem.A118(2014)10909–10918,doi:http://dx.doi.org/10.1021/ jp508919s.

[33]U.Even,J.Jortner,D.Noy,N.Lavie,C.Cossart-Magos,J.Chem.Phys.112(2000) 8068,doi:http://dx.doi.org/10.1063/1.481405.

[34]A.T.J.B.Eppink,D.H.Parker,Rev.Sci.Instrum.68(1997)3477–3484,doi:http:// dx.doi.org/10.1063/1.1148310.

[35]D.Imanbaew,M.F.Gelin,C.Riehn,Struct.Dyn.3(2016)043211,doi:http://dx. doi.org/10.1063/1.4953367.

[36]S.E.Greenough,M.D.Horbury,J.O.Thompson,G.M.Roberts,T.N.Karsili,B. Marchetti,D.Townsend,V.G.Stavros,Phys.Chem.Chem.Phys.16(2014) 16187–16195,doi:http://dx.doi.org/10.1039/c4cp02424a.

[37]S.E.Greenough,G.M.Roberts,N.A.Smith,M.D.Horbury,R.G.McKinlay,J.M. Zurek,M.J.Paterson,P.J.Sadler,V.G.Stavros,Phys.Chem.Chem.Phys. 16(2014) 19141–19155,doi:http://dx.doi.org/10.1039/c4cp02359e.

[38]J.J.Snellenburg, S.P.Laptenok, R. Seger,K.M. Mullen,I.H.M.V.Stokkum,J. StatisticalSoftw.49(2012),doi:http://dx.doi.org/10.18637/jss.v049.i03. [39]H.J. Werner, P.J. Knowles, G. Knizia, F.R. Manby, M. Schütz, Wiley

InterdisciplinaryRev.:Comput.Mol.Sci.2(2012)242–253,doi:http://dx. doi.org/10.1002/wcms.82.

[40]H.J.Werner,W.Meyer,J.Chem.Phys.74(1981)5794–5801,doi:http://dx.doi. org/10.1063/1.440892.

[41]P.G.Szalay,T.Muller,G.Gidofalvi,H.Lischka,R.Shepard,Chem.Rev.112(2012) 108–181,doi:http://dx.doi.org/10.1021/cr200137a.

[42]T.H. Dunning,J. Chem.Phys.90 (1989) 1007–1023,doi:http://dx.doi.org/ 10.1063/1.456153.

[43]M.J.Frisch,G.W.Trucks,H.B.Schlegel,G.E.Scuseria,M.A.Robb,J.R.Cheeseman, G.Scalmani,V.Barone,G.A.Petersson,H.Nakatsuji,etal.,Gaussian09(2016). [44]G.M.Roberts,C.A.Williams,J.D.Young,S.Ullrich,M.J.Paterson,V.G.Stavros,J.

Am.Chem.Soc.134(2012)12578–12589,doi:http://dx.doi.org/10.1021/ ja3029729.

[45]B.Marchetti,T.N.Karsili,M.N.Ashfold,W.Domcke,Phys.Chem.Chem.Phys. 18 (2016)20007–20027,doi:http://dx.doi.org/10.1039/c6cp00165c.

[46]T.N.Karsili,B.Marchetti,M.N.Ashfold,W.Domcke,J.Phys.Chem.A118(2014) 11999–12010,doi:http://dx.doi.org/10.1021/jp507282d.

[47]D.Tuna,N.Doslic,M.Malis,A.L.Sobolewski,W.Domcke,J.Phys.Chem.B119 (2015)2112–2124,doi:http://dx.doi.org/10.1021/jp501782v.

[48]R.Omidyan,M.Iravani,J.Phys.Chem.A120(2016)1012–1019,doi:http://dx. doi.org/10.1021/acs.jpca.5b12122.

[49]A.L.Sobolewski,W.Domcke,Phys.Chem.Chem.Phys.8(2006)3410–3417,doi: http://dx.doi.org/10.1039/b604610j.

[50]T.Raeker,B.Hartke,J.Phys.Chem.A121(2017)5967–5977,doi:http://dx.doi. org/10.1021/acs.jpca.7b03261.

[51]J.Zhao,S.Ji,Y.Chen,H.Guo,P.Yang,Phys.Chem.Chem.Phys.14(2012)8803– 8817,doi:http://dx.doi.org/10.1039/c2cp23144a.

[52]S.J.Lim,J.Seo,S.Y.Park,J.Am.Chem.Soc.128(2006)14542–14547,doi:http:// dx.doi.org/10.1021/ja0637604.

[53]J.E.Kwon,S.Y.Park,Adv.Mater.23(2011)3615–3642,doi:http://dx.doi.org/ 10.1002/adma.201102046.

[54]F.S.Rodembusch,F.P.Leusin,L.F.Campo,V.Stefani,J.Lumin.126(2007)728– 734,doi:http://dx.doi.org/10.1016/j.jlumin.2006.11.007.

[55]C.A.Southern,D.H.Levy,G.M.Florio,A.Longarte,T.S.Zwier,J.Phys.Chem.A 107(2003)4032–4040,doi:http://dx.doi.org/10.1021/jp027041x.

[56]A.Stolow,Annu.Rev.Phys.Chem.54(2003)89–119,doi:http://dx.doi.org/ 10.1146/annurev.physchem.54.011002.103809.

[57]F.Calegari,A.Trabattoni,A.Palacios,D.Ayuso,M.C.Castrovilli,J.B.Greenwood, P.Decleva,F.Martín,M.Nisoli,J.Phys.B:At.Mol.Opt.Phys.49(2016)142001, doi:http://dx.doi.org/10.1088/0953-4075/49/14/142001.

[58]H.J.Bakker,J.Chem.Phys.98(1993)8496–8506,doi:http://dx.doi.org/10.1063/ 1.464508.

[59]S.Matsumoto,R.Kumasaka,M.Yagi,A.Kikuchi,J.Photochem.Photobiol.A: Chem.346(2017)396–400,doi:http://dx.doi.org/10.1016/j.