Chemical Physics Letters 599 (2014) 23–33

Contents lists available at ScienceDirect

Chemical Physics Letters

journal homepage: www.elsevier .com/ locate /cplet t

FRONTIERS ARTICLE

Intramolecular radiationless transitions dominate exciton relaxationdynamics

http://dx.doi.org/10.1016/j.cplett.2014.03.0070009-2614/� 2014 Published by Elsevier B.V.

⇑ Corresponding author.E-mail address: greg.scholes@utoronto.ca (G.D. Scholes).

Chanelle C. Jumper a, Jessica M. Anna a, Anna Stradomska b, Juleon Schins c, Mykhaylo Myahkostupov d,Valentina Prusakova e, Daniel G. Oblinsky a, Felix N. Castellano d, Jasper Knoester b, Gregory D. Scholes a,⇑a Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario M5S 3H6, Canadab Zernike Institute for Advanced Materials, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlandsc Department of Chemical Engineering, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlandsd Department of Chemistry, North Carolina State University, Raleigh, NC 27695, USAe Department of Chemistry and Center for Photochemical Sciences, Bowling Green State University, Bowling Green, OH 43403, USA

a r t i c l e i n f o

Article history:Received 31 January 2014In final form 4 March 2014Available online 12 March 2014

a b s t r a c t

Reports of long-lived exciton coherences have lead researchers to expect that model dimer systems inev-itably generate exciton superposition states observable by two-dimensional electronic spectroscopy.Here we report a careful photophysical characterization of a model dimer system, a diacetylene-linkedperylenediimide dimer to examine that issue. The absorption spectrum of the dimer shows molecularexciton splitting, indicating that excitation is delocalized. The assignment of exciton states was supportedby other photophysical measurements as well as theoretical calculations. Ultrafast two-dimensional elec-tronic spectroscopy was employed to identify and characterize excitonic and vibrational features, as theyevolve over time. Population transfer between the two exciton states is found to happen in <50 fs, thuspreventing the sustainment of exciton coherences. We show that such fast radiationless relaxation can-not be explained by coupling to a solvent spectral density and is therefore missed by standard approachessuch as Redfield theory and the hierarchical equations of motion.

� 2014 Published by Elsevier B.V.

1. Introduction

The flow of energy through the exciton manifolds of naturallight harvesting complexes and synthetic molecular assembliesafter photoexcitation is a topic of fervent research involving avariety of state-of-the-art experimental and theoretical approaches[1–23]. One of the motivations for these studies is to pursue anunderstanding of the underlying molecular level mechanisms thatgovern the energy transfer dynamics in photosynthetic lightharvesting complexes. The initial light harvesting stage of photo-synthesis involves the transfer of excitation energy across chro-mophores before reaching a reaction center for charge separationand ultimately leading to energy storage. This process ofteninvolves transfer of excitation among excitonically coupled chro-mophores. Gaining a deeper understanding of the details of energytransfer and the role of molecular excitons supports the ultimategoal of improving the design and development of artificial solarenergy-driven technologies [24–29].

In photosynthesis, the protein structure determines the pig-ment orientations, which are often found in pairs of electronically

coupled dimers, as well as higher order assemblies. GeneralizedFörster theory [30–33] or modified Redfield theory [34–38] is thecurrently accepted framework, which describes a ‘hopping’ typeenergy transfer occurring between coherently delocalized donorand acceptor states. Recent studies have suggested that additionalcoherent superpositions between donor and acceptor excitonstates in several photosynthetic complexes [39–47] may play a rolein coherent light harvesting [7,8]. As a result, it is sometimesassumed that such coherences should be easily demonstrated inisolated dimer systems of the corresponding coupling regime in asolvent environment. However, ultrafast exciton relaxation ob-served for various systems, including the chlorosome of green sul-fur bacteria [48] and protein-supported chromophoric systems[49–54], may compete with the sustenance of exciton superposi-tion states. It is the balance of these two processes that we wishto pursue in this Letter.

To obtain a better understanding of these interdependentphotophysical processes, both theoretical [55–57] and experimen-tal studies [57–61] began to address ideal model systems. For aphotosynthetic exciton, the ideal system would comprise a pairof rigidly linked chromophores in an appropriate solvent environ-ment, where the electronic couplings between the chromophoricunits and the surrounding environment are comparable. A study

24 C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33

of such a heterodimeric system has recently been reported and evi-dence for long-lived electronic coherences was uncovered [60].However, the ultrafast relaxation among excitonic states has alsobeen observed in solvent suspended homodimers [61–63]. In thepresent contribution, we employ ultrafast two-dimensional elec-tronic spectroscopy (2DES) to explore the photophysics of a modelperylenediimide (PDI) homodimer and its corresponding monomerin order to understand the interplay between electronic coher-ences and energy equilibration.

2DES is a significant advance of pump-probe spectroscopy forelucidating ultrafast dynamics, including signatures of coherences,in complex condensed phase systems [39,64–68]. 2DES spreads theinformation contained in a pump-probe spectrum over twofrequency axes and the resulting spectra can be thought of as fre-quency-frequency correlation map. The time dependent amplitudechanges of the peaks within this map give detailed information oncoherences and energy flow through the system. Coherences areflagged by the modulation of peaks at the difference frequency ofthe states in superposition. Careful metrics are additionally appliedto characterize the coherence as either vibrational or electronic incharacter [46,69]. Population transfer between states is signifiedby a decrease in amplitude of the diagonal peak of the first stateand a corresponding increase in amplitude of the off-diagonal peaklocated at the energy coordinates of the two states in question [56].

In this Letter, our aim is to examine a simple and carefully char-acterized model dimer system in order to compare the competitionbetween excitonic coherences and population relaxation. Thus, weuse a combination of 2DES, steady-state measurements, and theo-retical modeling to carefully characterize our model dimer system,a homodimer consisting of two strongly coupled perylenediimideunits linked by a diacetylene bridge [70]. This dimer has clear exci-ton splitting determined by the angle between the two transitiondipole moments, collinear with the long C2 axis of each (unsubsti-tuted) monomeric unit [71,72]. In combination with intermediate-strength coupling to high-frequency vibrations, this PDI dimer is agood candidate for the study of exciton dynamics from coherencesto population relaxation.

The manuscript is organized as follows. In the first part, we de-scribe a systematic experimental characterization of the dimer,including the assignment of the linear spectrum and confirmationof the presence of molecular exciton states. This is further corrob-orated by calculations used to characterize ground and excitedstates of the dimer and reproduce the linear spectrum. In thesecond part, we explore the ultrafast exciton dynamics followingphotoexcitation via 2DES. These reveal ultrafast dephasing of exci-tonic superposition states dominated by an ultrafast populationrelaxation on the sub-50 fs timescale. This observation suggeststhat the dynamics are controlled by a passage through a conicalintersection, which results in the rapid deactivation of the upperexciton population.

2. Experimental

2.1. Synthesis

The synthesis of the PDI monomer and dimer has beendescribed elsewhere [70].

2.2. Sample preparation for room temperature measurements

PDI monomer or PDI dimer were individually diluted in dichlo-romethane (DCM) to an optical density (OD) at the absorptionmaximum of 0.24 for absorption experiments (/1 cm) and fornon-linear two-dimensional spectroscopy (/1 mm), and to an ODof <0.05 for fluorescence experiments (/1 cm).

2.3. Sample preparation for measurements at 10 K

Dilute solutions of PDI monomer or dimer were mixed with aDCM solution of the matrix polymer poly(ethyl methacrylate).The solution was dispensed by pipette onto microscope coverslipswith a final OD of <0.05. The thickness of the film was on the orderof �100 nm.

2.4. Steady-state spectroscopy

Room temperature absorption spectra were recorded using aCARY 100 Bio UV–visible spectrophotometer. Fluorescenceemission and excitation spectra were recorded on a Fluorolog-3(Horiba) spectrometer in a 90� mode, with a xenon arc light sourceand an R928 photomultiplier tube. To obtain room temperaturetwo-dimensional photoluminescence (2DPL), the emission wave-length was stepped by 2 nm from 550 to 700 nm while recordingexcitation scans from 400 to 600 nm by 1 nm steps. The resultingdata were combined in a two-dimensional plot. To obtain 2DPLat 10 K, prepared PEMA films containing PDI monomer or dimerwere mounted at a 45� angle and cooled to 10 K with a Cryodyne�

model 22C He cryostat (CTI-Cryogenics). Emission wavelength wasstepped by 1 nm in 500–700 nm spectral window, while scanningthe excitation as above, generating a corresponding 2D map. To ob-tain anisotropy data, excitation scans were recorded at four polar-ization combinations: VV, HH, VH, HV with respect to emission setto 600 nm, the maximum of the S1 ? S0

(0–0) transition. Anisotropyvalues were calculated as follows [73]:

hri ¼ IVV � G � IVH

IVV þ 2 � G � IVHð1Þ

where the ‘G factor’, which normalizes a polarization-dependentinstrument response, is defined as:

G ¼ IHV

IHHð2Þ

Data were recorded with excitation and emission slits set to a2.5 nm bandpass.

2.5. Time-resolved fluorescence and quantum yield measurements

Photoluminescence lifetimes were measured using the gatedsecond-harmonic (4 MHz repetition rate, �1 nJ/pulse) output of atunable ultrafast Ti-sapphire oscillator (Coherent, �120 fs) as anexcitation source and a time-correlated single photon counting(TCSPC) detection system (LifeSpec II, Edinburgh Instruments,�150 ps IRF). The excited state lifetimes were obtained from thefirst-order decay fit of single wavelength emission transients usingOrigin 8.0 software. Fluorescence quantum yields were measuredusing an absolute PL quantum yield spectrometer (C11347,Hamamatsu).

2.6. Coherent two-dimensional electronic spectroscopy

The experimental set-up has been described in detail elsewhere[46]. Briefly, three incoming pulses (Supplemental Figure S2) arriveat the sample position arranged in a box geometry, leading to a sig-nal emission in the background free direction determined by thewavevectors of the incoming fields [64]. A fourth electric field,the local oscillator (ELO), is combined with the signal such that bothreal and imaginary components can be extracted through spectralinterferometry. According to the box geometry, one phase-match-ing condition leads to the generation of a non-linear spectroscopicsignal occurring in the direction of the local oscillator. Two timeorderings give rise to the rephasing and nonrephasing spectrawith wavevector combinations as follows: ks, R = �k1+k2+k3 and

C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33 25

ks, NR = +k1-k2+k3, respectively. These two pathways are achievedby simply reversing the interaction ordering of the first two elec-tric fields with the sample, each obtained independently in our set-up. There are three time intervals associated with this timeordering. The coherence period (t1) is the delay time between thefirst and second pulse, the waiting time (t2) is the delay betweenthe second and third pulse, and the signal detection time (t3) isthe delay after the third pulse. The sum of the real parts of theresulting spectra gives rise to the absorptive spectrum accordingto [74,75]:

Sðx1; t2;x3Þ ¼ ReZ 1

0dt1

Z 1

0dt3SRðt1; t2; t3Þe�ix1t1 eix3t3

�

þZ 1

0dt1

Z 1

0dt3SNRðt1; t2; t3Þeix1t1 eix3t3

�ð3Þ

where SR is the rephasing spectrum, SNR is the nonrephasing spec-trum, x1 is the excitation frequency and x3 is the detection fre-quency. The absorptive spectrum is a two-dimensional spectrumwhere the excitation and detection frequencies are coherently cor-related through the experiment. In order to obtain information fromthe real part of the spectrum, it is necessary to determine the accu-rate time delays, t1 and t3, otherwise a mixing of real and imaginarycomponents may lead to dispersive distortions in the resultingspectrum [74–77]. The removal of these phase distortions throughour precise phasing procedure is described in detail elsewhere[78]. The reported 2D spectra where obtained by stepping t1 form�45 to 45 fs in 0.15 fs steps for each t2 step where t2 was steppedfrom 0 to 400 fs in 4 fs steps and from 0 to 100 fs in 1 fs steps.The 2D data were taken in triplicate and the same spectroscopictrends where observed between data sets taken on different days.

3. Results and discussion

3.1. Part 1: characterization of excited states

3.1.1. Linear absorption, fluorescence and structuresFigure 1a compares the linear absorption spectra of the PDI

monomer and dimer in solution (DCM), and their correspondingfluorescence spectra. The visible part of the monomer absorptionspectrum derives from the p ? p⁄ electronic transition and dis-plays a well-resolved vibronic progression associated with thesymmetric ring-stretching vibrations [72]. The spectrum is wellreproduced by a single Franck–Condon progression that corre-sponds to an electronic excitation with adiabatic energy ofE = 18795 cm�1, vibrational frequency of x0 = 1395 cm�1 anddimensionless displacement parameter of g = 1.30 (equivalent toa Huang–Rhys parameter of 0.85).

The noticeably altered absorption spectrum of the PDI dimerpoints towards the substantial electronic coupling between adja-cent PDI subunits suggesting the formation of delocalized molecu-lar exciton states [79]. Combined with the considerable excitedstate displacement in the high-frequency vibration it leads tointermediate-strength vibronic coupling.

The exciton model, originally developed for molecular crystals[80,81], is also valid for dimers, trimers, and higher-order assem-blies that are bound by van der Waals, hydrogen- and even cova-lent bonds.[55,79] The resonant coupling between subunits’excited states results in splitting and shifting of the excited statesof the assembly relative to those of its building blocks, which canbe observed spectroscopically. The distribution of the oscillatorstrength between the split components (Davydov components) isgoverned by the relative orientation of transition dipole momentsof the subunits. This model has been extended to include vibroniccoupling that has relaxed symmetry restrictions compared to thesimple model [82–84]. The exciton model has been successfullyapplied to a variety of coupled chromophores [4,59,85], including

PDIs and their derivatives, to explain their photophysical behavior[71,72,86–90]. Here, we discuss the application of the excitonmodel to our model system.

The PDI dimer is composed of two identical PDI subunits bear-ing 1-hexylheptyl groups that were incorporated to enhance thesolubility (Supplemental Figure S1), held together by a covalentdiacetylene linker. The structures shown in Figure 1b and c forthe PDI monomer and dimer, respectively, were optimized usingdensity functional theory (for details see Supplementary AppendixS2). In our calculations the side chains were purposely truncated atthe imide position to minimize the computational costs, since theywere not found to significantly alter the photophysical propertiesof the monomer [91,92]. On the other hand, the presence of thelinker in the bay-area leads to the out-of-plane twisting of the per-ylene core. Our recent structural studies [70] unraveled the confor-mational dynamics related to this twisting and also suggested thatthe PDI dimer may exist as two isomeric forms in solution: the ci-soid and transoid. Experimental studies on related compounds havealso pointed to the existence of two dimeric forms in the groundstate [93,94]. However, present theoretical and experimental datapoint towards the presence of only one major isomer in solutionfor the PDI dimer (vide infra).

In our theoretical modeling of the excited states of the PDI di-mer we take into account one dipole-allowed p–p⁄ electronic exci-tation per PDI subunit, with an adiabatic excitation energy E and atransition dipole moment parallel to the long axis of the perylenecore. This excitation is linearly coupled to one symmetric vibra-tional mode of frequency x0. The harmonic nuclear potential inthe electronic excited state is shifted with respect to that of theground state by g along the (dimensionless) vibrational coordinate,resulting in the reorganization energy given by �hx0g2=2. Thus, wedescribe the excited states of the PDI dimer using the HolsteinHamiltonian:

H ¼ Eþ�hx0g2

2

� �X2

n¼1

aynan þ Jðay1a2 þ ay2a1Þ þ �hx0

X2

n¼1

bynbn

þ�hx0gffiffiffi

2p

X2

n¼1

aynanðbyn þ bnÞ þ D ð4Þ

The operators ayn (an) and byn (bn) create (annihilate) the elec-tronic excitation and a quantum of high-frequency vibration,respectively, on the PDI subunit n. J denotes the excitonic (reso-nance) coupling, while the dimerization shift D accounts for thenon-resonant interactions, which lead to lowering of the excitedstate energies in the dimer [19,79]. A contribution to D comes fromthe mixing between locally-excited and charge-transfer configura-tions [95].

We used two numerical methods to evaluate the Hamiltonian(4): the scheme of Fulton and Gouterman [38,39] (details in Sup-plementary Appendix S3) and the two-particle approach [96–98](details in Supplementary Appendix S4). For dimers composed oftwo identical subunits these two methods are numerically equiva-lent, and thus yield the same results provided that the sameparameterization is used. However, we employed two different ap-proaches: in the first case we treated the parameters of the Fultonand Gouterman model as phenomenological constants and fit theparameters to best reproduce the experimental data. In the secondcase the parameterization for the two-particle method was ob-tained from a combination of quantum-chemical calculations andexperimental data on the monomer. The two-particle approach ismore versatile because it can easily describe dimers composed ofnonequivalent molecules. Thus, in the second case it allowed usto explicitly model the effects of disorder (inhomogeneous broad-ening), by adding random uncorrelated shifts drawn from aGAUSSIAN distribution to the excitation energy of the chromophoresconstituting the dimer.

(a) (b) (c)

Figure 1. (a) Absorption (solid) and emission (dotted) spectra in DCM of PDI monomer (green) and PDI dimer (blue). Structures of PDI monomer (b) and PDI dimer (cisoidisomer) (c), shown through two perspectives. Geometries were optimized using density functional theory for the model compounds with the 1-hexylheptyl sidechainstruncated, complete line structures are available in Supplementary Figure S1.

26 C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33

The details of parameterization are described in SupplementaryAppendix S4, here we only provide a summary. The adiabatic exci-tation energy, energy of vibrational quantum and the displacementparameter were obtained from the experimental absorption spec-trum of PDI monomers in DCM. In the dimer the PDI subunits liewithin close proximity of one another precluding the use of the di-pole–dipole approximation for the Coulombic interaction betweentransition densities [99,100] often used for the calculation of exci-tonic couplings. In order to adequately capture the spatial details ofthe monomeric electronic transition and the effects of a covalentbond between the subunits we estimated the excitonic couplingfrom the TDDFT calculations on the dimer. For the cisoid isomerthe resulting electronic coupling J = 1126 cm�1, however in our fi-nal calculations we used a slightly lower value of 970 cm�1 whichyields better reproduction of the experimental spectrum and fitswell within the accuracy of the TDDFT calculations. Since ourquantum chemical calculations do not provide any estimate forthe dimerization shift, a value of �480 cm�1 was chosen to repro-duce the experimental dimer spectrum.

The above set of parameters places the PDI dimer in the compli-cated intermediate-strength vibronic coupling regime: the exci-tonic coupling, the vibrational quantum, and the relaxationenergy are all of the same order of magnitude. In the perfecthomodimer, the two PDI subunits are equivalent, thus all theeigenstates of Hamiltonian (4) are either symmetric or antisym-metric under their exchange. The transition dipole moment fromthe ground state to symmetric eigenstates is perpendicular to thatfrom the ground to antisymmetric states. Therefore, the absorptionspectrum of the PDI dimer is a superposition of two spectra, result-ing from the symmetric and antisymmetric excited states. How-ever, the vibronic progressions in both subsystems are stronglydistorted compared to the one observed for the monomer, due tothe intermediate-strength vibronic coupling. When the disorderis taken into account the symmetry is broken, and the separationinto symmetric and antisymmetric states is no longer possible.Nonetheless, the disorder in the PDI dimer is relatively weak,which allows us to separate the excitations into two groups withapproximately perpendicular transition dipoles and makes thesimple homogeneous dimer particularly useful for interpretation.

The results of our simulations are shown in Figure 2a and b, andshow reasonable agreement with the experimental data in the re-gion of the lowest energy transitions. The disagreement arises par-tially from the large value of the displacement needed for thevibrational mode in order to reproduce the intensity distributionamong the vibronic replicas in the monomer spectrum. There areevidently factors missing from the theory needed to model thespectra accurately, which will be the subject of future research.The corresponding energy level diagram is shown in Figure 2c.

We assign the two lowest energy states observed in the absorptionspectrum to zero-phonon lines of two exciton components (asym-metric and symmetric). They are both red-shifted with respect tothe monomer transition due to the large dimerization shift [19,79].

Only the cisoid isomer spectra presented here match the exper-imental dimer data, while the simulated transoid isomer spectrabear no resemblance to experimental results (see SupplementaryAppendix S4). Thus our calculations point toward one isomer(cisoid) as a dominant form in solution.

3.1.2. 2D and time-resolved photoluminescenceTwo-dimensional photoluminescence measurements were car-

ried out at 10 K and room temperature (RT) to establish a relation-ship between the excited states and the corresponding emissivestate(s) and further characterize the linear spectrum. The resultsare shown in Figure 3a and b, respectively. Since the absorptionand emission spectra are sensitive to solvent conditions, the spec-tra in a film at 10 K are slightly shifted with respect to the DCMsolution spectra acquired at RT. Both sets of data show a constantset of excited states across all emission wavelengths indicating asingle photoluminescence channel. This observation is also consis-tent with mono-exponential PL lifetime decays for both PDI mono-mer and dimer measured at RT in DCM (Table 1). We note that therelative intensities of the vibronic progression differ only in thefilm at longer emission wavelengths, which is a characteristic sig-nature of aggregation [101].

Dimerization of the PDI molecule leads to a reduction in fluores-cence quantum yield and radiative rate (Table 1), as expected inthe formation of exciton states with mostly H-character. Given thatthe radiative rate is linearly proportional to the dipole strength forthe corresponding absorption process, the approximate 10-folddecrease in radiative rate is in agreement with the division ofoscillator strength between the two exciton states, which can bequalitatively assessed from the linear absorption spectrum(Figure 1).

3.1.3. Fluorescence anisotropyTo confirm the assignment of exciton states to the two lowest

energy transitions in the absorption spectrum of the dimer, we car-ried out fluorescence anisotropy experiments on both the PDImonomer and dimer in PEMA films at 10 K. The results are shownin Figure 4a and b for the monomer and dimer, respectively. Theplots show the absorption spectra for the molecules at room tem-perature in DCM as well as the excitation scans in PEMA films at10 K, and the resulting calculated anisotropies. There is little linenarrowing at low temperatures compared to ambient temperature,showing that inhomogeneous broadening dominates over homo-geneous line broadening.

Nor

mal

ized

Abso

rban

ce

Wavelength (nm)

(a)

400 500 6000

1

2

Nor

mal

ized

Abso

rban

ce

Wavelength (nm)

(b)

(c)

400 500 6000

1

2

Figure 2. Comparison of experimental (blue) and simulated (green) absorption spectra of PDI dimer (a) using quantum-chemistry based model parameterization as describedin the text and (b) using parameterization fitted to best reproduce the experimental data (c) proposed energy level diagram of vibronic states. In panel (a) also the stickspectrum of the homogeneous dimer is presented, with the states belonging to asymmetric (�) and symmetric (+) manifolds denoted in red and dark blue respectively. Thesame color coding is used in panel (c).

480

440

400

560

520

600

Exci

tatio

n W

avel

engt

h (n

m)

550 600 650 700Emission Wavelength (nm)

1

0

500

450

400

600

550

650

Exci

tatio

n W

avel

engt

h (n

m) 1

0500 600 700

Emission Wavelength (nm)

(a)

(b)

Figure 3. Two-dimensional photoluminescence of PDI dimer at 10 K in a PEMA film(a) and at room temperature in DCM (b). Excitation scan is slightly variable withrespect to emission wavelength in the film due to aggregation of the dimer, andaffords no line narrowing. In solution, the excitation scan is consistent across allemission wavelengths.

Table 1Fluorescence lifetimes and quantum yields of the PDI monomer and dimer.

PDI kabsa kem

b sFLc UFL

d krade knonrad

e

(nm) (nm) (ns) (s�1) (s�1)

Monomer 535 545 4.9 0.883 4.9 � 109 6.5 � 108

Dimer 547, 571 600 1.3 0.097 7.5 � 108 4.7 � 108

a Peak maxima of steady-state UV–vis absorption spectra.b Peak maxima of steady-state fluorescence spectra. Excitation wavelength was

450 nm for monomer and dimer.c Fluorescence lifetimes measured by TCSPC with the photoexcitation wave-

length of 450 nm for monomer and dimer.d Fluorescence quantum yield.e Radiative and non-radiative decay rate constants, krad and knonrad respectively,

were calculated by using fluorescence quantum yieldsc and fluorescence lifetimesd;UFL = krad/(krad+knonrad), sFL = (krad+knonrad)�1.

C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33 27

As pointed out in the previous discussion, the film appears toconsist of both ‘single’ unassociated dimer molecules as well as di-mer aggregates. Thus to ensure the excitation scans are taken onlyfor the unassociated dimer molecules, we selected the emission atk = 600 nm that corresponds to the PDI dimer in the unassociatedform only. The dimer excitation scan at this wavelength (Figure 4b)is in excellent agreement with the linear absorption spectrummeasured in DCM at room temperature and can thus be reliablyused for the fluorescence anisotropy calculations.

The resulting anisotropy value for the monomer across the en-tire vibronic progression is constant at 0.5 indicating that thepolarization relationship between these states is constant as ex-pected. We note that the observed anisotropy value of 0.5 is higherthan the theoretical maximum of 0.4 expected for collinear absorp-tion and emission dipole moments. However, this value (0.4) canonly be expected for a randomly oriented sample based on theprobability of projection along the excitation axis, and it is possiblethat the enhanced ordering of the molecules, and hence their

(a)

(b)

Figure 4. Fluorescence anisotropy results for the PDI monomer (a) and dimer (b) ingreen, plotted against the right hand axis. Solid blue lines are the absorptionspectrum at room temperature in DCM and dotted blue lines are the excitation scanin PEMA film at 10 K plotted against the left hand axis. Excitation scans wereselected at emission wavelengths, which isolate the species that is not aggregated,as confirmed by the agreement of the excitation scan (dimer: 600 nm, monomer:560 nm). Slight line narrowing separates the overlap in the exciton states, resultingin an apparent decrease in intensity for the lowest exciton state.

28 C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33

dipoles moments, was achieved in the film samples [73]. Anothersource of error in this value may derive from scattered light offof the surface of the film.

The dimer anisotropy value rapidly decreases when excitationis scanned across the lower exciton state to the upper exciton state.In the theoretical limit, we would expect to witness a drop to �0.2,consistent with perpendicular polarizations of exciton states.Though this limit is not perfectly reached, the large drop in anisot-ropy across these two states is consistent with the presence of twoperpendicular dipole moments considering that the spectrum islargely overlapped in this region and the theoretical maxima can-not be obtained. In agreement, there is also a small rise at the exactwavelength where the v = 1 vibrational replica of the lower excitonstate is expected to be, which would share the polarization of thelower exciton state, though it is again largely overlapped. It is rea-sonable to conclude the assignment of two exciton manifolds withperpendicular transition dipoles to the absorption spectrum.

For comparison, room temperature anisotropy experimentswere obtained for the dimer in a viscous solution of dichlorometh-ane and poly(propylene glycol). Although the anisotropies will bediminished by increased molecular motion at room temperature,these data maintain a decrease in anisotropy across the two lowestenergy transitions. The anisotropy value at the lowest energy par-allel transition is less than 0.4, consistent with some randomiza-tion of the orientations of the molecules on the time scale of themeasurement (see Supplemental Figure S3).

3.2. Part 2: ultrafast dynamics after photoexcitation

3.2.1. 2DESWe use two-dimensional electronic spectroscopy to probe the

ultrafast dynamics that occur on the 0–400 fs timescale after

photoexcitation in both the PDI monomer and dimer. Here, weuse a series of broadband laser pulses to perturb and probe theelectronic states of the monomer and dimer systems. The 2Dexperiment can be thought of as a pump-probe measurementwhere the probe transitions are directly correlated to the originat-ing pump transition, as laid out in the 2D map. A 2D map is gener-ated at different waiting times (t2), which separate the pump andprobe transitions. In this way, we map out energy transfer path-ways and the dynamics of the system. Since the same broadbandpulse is used for all light-matter interactions, we can map out var-ious pathways within the bandwidth simultaneously.

The incoming broadband laser pulse is generated with ahome-built non-collinear optical parametric amplifier [102](NOPA) and the resulting spectrum is shown in Figure 5a,together with the monomer spectrum (top panel) and dimerspectrum (bottom panel). As can be seen from Figure 5a, theincoming laser pulses are identical for both samples, and the tun-ing of the incoming laser pulse ensures that the transitions ofinterest in both the monomer and dimer systems are probed,even though the monomer and dimer spectra are not identical.This ensures that all experimental parameters, including thealignment, wedge calibration, and pulse compression remain con-sistent for comparison purposes. Compression of the spectrallybroad incoming pulse (�65 THz; intensity spectral FWHM) re-sulted in an intensity temporal FWHM of �11.8 fs with no detect-able angular dispersion (see Supplemental Figure S2), enablingthe observation of sub-50 fs dynamics. An additional advantageof employing this large spectral bandwidth to the monomer sys-tem, where there is no overlap with the linear absorption spec-trum in the lower energy region of the pulse, is that it allowsfor ground state vibrational levels to be revealed.

The resulting 2D plots are shown in Figure 5b, at 3 differentwaiting times for the monomer (top panel) and dimer (bottom pa-nel). The monomer 2D spectra show a diagonal peak correspondingto the p ? p⁄ electronic transition, as well as an off-diagonal peakdirectly below the diagonal which results from stimulated emis-sion to the 1400 cm�1 vibration in the ground electronic state.The dimer spectra at the waiting times plotted show an elongatedpeak along the diagonal which has two contributions, a strongbleach signal corresponding to the lower exciton state and a weak-er bleach signal of the upper exciton. The peak is also elongatedalong the off-diagonal axis, due to unresolvable weak cross-peaksbecause the corresponding diagonal transitions are overlapped.

Analysis of the waiting time dependent amplitudes of the peaksin the 2D spectra reveal two oscillatory components present inboth the monomeric and dimeric systems, neither of which canbe assigned to electronic coherence (see Supplementary Figure S4).The first, a characteristic high frequency oscillation at 1400 cm�1,is assigned to the ground state vibrational wavepacket of thestrongly coupled carbon–carbon stretching mode. This mode inthe electronic excited state is not coherently excited because thelaser pulse is tuned to the red side of the absorption spectrum.The second, lower frequency oscillation at �750 cm�1, is also pres-ent in both the monomer and dimer and thus cannot be assigned toelectronic coherence. Our quantum-chemical calculations of thedisplacement parameters using non-orthogonal curvilinear coordi-nates [103] (see Supplementary Appendix S2) show that in the fre-quency range from 600 to 1200 cm�1 there are no vibrationalmodes in the isolated monomer that couple strongly to the elec-tronic excitation. Careful control studies in five different solvents(dichloromethane, chlorobenzene, chloroform, carbon tetrachlo-ride, and carbon disulfide) reveal that the second oscillatory fea-ture corresponds directly to the strong Raman active mode of thesolvent, depending on the solvent (see Supplementary Figure S5).Thus, we ascertained that the long-lived oscillatory features, ob-served in both the PDI monomer and dimer, are attributed to the

(a) (b)

Figure 5. (a) Absorption spectra and NOPA spectra for PDI monomer (top panel) and PDI dimer (bottom panel). (b) Resulting two-dimensional absorptive spectra shown atthree different waiting times, where detection wavelengths are plotted against excitation wavelengths. Each plot is a snapshot that describes the energy transfer andcoherence pathways as a function of waiting time, which is the time between the excitation interaction and the detection interaction, occurring with broadband laser pulsesas exhibited in (a), red trace.

C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33 29

ground state vibrational wavepackets of the system (high fre-quency mode) and the solvent (lower frequency mode).

Based on these results, we conclude that any exciton coherenceprepared with the pulsed excitation decoheres extremely rapidlyas the dynamics after the first 50 fs do not exhibit substantial evo-lution after the removal of the oscillatory parts generated by thesolvent and PDI vibrational modes. We therefore focus our atten-tion on the first 50 fs to carefully uncover any dynamics occurringprior to the system arriving at this pseudo-steady state of popula-tion dynamics during the waiting time range 0–400 fs.

To elucidate the sub-50 fs dynamics, we performed additional2DES measurements where t2 was stepped in increments of 1 fsin order to obtain higher resolution dynamic features during thisearly time period. We note that although there are increased back-ground solvent responses due to the optical Kerr effect, the molec-ular signals can be clearly deciphered due to the large oscillatorstrength of the excitonic states of the PDI dimer.

The finely sampled 2D spectra are shown in Figure 6. Byt2 = 45 fs, the 2D spectra resemble the previously discussed 2Dspectra. The t2 = 0 fs spectrum for the dimer in DCM shows twonearly equivalent intensities for the diagonal signals belonging tothe two exciton states being probed (upper exciton labeled 2 andlower exciton labeled 1). This is what we would expect from theconvolution of the pulse spectrum with the molecular response(a preferential weighting of the lower, weak excitonic state). Asimilar pulse was also successful at generating two even signalsin previous samples of CdSe nanocrystals [104]. The initial positiveoff-diagonal signal arises from stimulated emission, i.e. thetransition from the upper exciton state to the electronic groundstate with a single vibrational excitation corresponding to the1400 cm�1 mode (peak 3). The excitation frequency associatedwith this off-diagonal peak indicates that it is coupled to the upperexciton state. We note here that the incoming laser pulse was nottuned to obtain the corresponding cross-peak coupled to the lowerexciton state.

Excitation and probing of the two exciton states yields a combi-nation of stimulated emission (SE) and excited-state absorption(ESA) pathways in the nonlinear response [105,106], which givesrise to an overall negative cross-peak below the diagonal (peak21), in agreement with calculations performed on a similar dimersystem [58]. It is clear that within the first few tens of femtosec-onds, there is a marked decrease in the diagonal signal of the upperexciton state (peak 2) in concert with the off-diagonal peak

associated with the ground state vibrational mode (peak 3), whilesimultaneously the negative cross-peak increases in positiveamplitude and becomes overall net positive (peak 21).

To demonstrate these effects with more clarity, a diagonal sliceis plotted against t2 time for the first 100 fs and is shown inFigure 7a. In addition, Figure 7b presents the GAUSSIAN fits to thetwo exciton peaks in the diagonal slice and selected plots at every5 fs exhibiting an exponential decay in the upper exciton signal,which plateaus by 30 fs. It is clear that the lower exciton state re-mains relatively constant in amplitude, serving as an internal con-trol to emphasize that this is not a background-induced effect (seealso Supplementary Figure S6 for the measured background signal,which decays to a minimum by �10 fs). Figure 7c shows a trace ofthe lower cross-peak (black) which increases in positive amplitude(SE contribution) on the same timescale. The oscillations in thedata are due to the solvent, and are excluded in the exponentialfit taken of the trace (red trace).

A decay of the population on the diagonal in concert with agrowth in the corresponding cross-peak is a signature of popula-tion transfer from one state to the other [56,107]. These data indi-cate a rapid decay in the upper exciton state population andtransfer to the lower exciton state. The decay and growth are notnecessarily 1:1 in amplitude as a result of other possible contribu-tions to the observed dynamics. Firstly, the cross-peak (peak 21)has negative amplitude at t2 = 0 fs, which is only partially compen-sated by the growth of the stimulated emission due to the relaxa-tion from the upper exciton state to the lower one. Secondly, thepulses are all vertically polarized and transitions aligned with theincoming (‘pump’) fields will be photoselected. This photoselectionwill affect the amplitude of the cross-peak, which results fromtransfer between two orthogonal states [108]. Lastly, the oscillatorstrengths of the exciton states are not equal to each other, and theresulting amplitude of the cross-peak is derived from the combina-tion of these two transitions.

The rate of this population relaxation from upper to lower exci-ton is remarkably fast compared to some previous experimentaland theoretical studies [109–111]. In particular, recent calculationsof energy transfer dynamics and exciton relaxation in photosyn-thetic complexes generally assume that relaxation is governed bycoupling between exciton states and nuclear degrees of freedomof the environment. To illustrate how this approach cannot predictthe relaxation time we measure for the PDI dimer we simulate theexciton relaxation dynamics according to bath mediated dynamics

Figure 6. Two-dimensional absorptive spectra of PDI dimer at early waiting times, as indicated, for the finely sampled dataset. Detection wavelengths are plotted againstexcitation wavelengths. Peaks are labeled in the t2 = 0 fs plot for reference in the discussion.

Figure 7. (a) Surface plot of the diagonal slice along t2 from 0 to 100 fs at 1 fs intervals. A cubic spline interpolation was used to generate a smooth surface. (b) Diagonal slice(grey) shown at 5 fs intervals with GAUSSIANS fit to the upper (green) and lower (blue) exciton states, demonstrating a rapid decay in the upper exciton state. (c) Correspondinggrowth of the cross peak (black trace) taken at coordinates corresponding to the excitation of the upper exciton state and detection of the lower exciton state (560, 580 nm).The red trace represents an exponential fit to the data, excluding the oscillatory components in the rise.

30 C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33

using the hierarchical equations of motion (HEOM) [37,112]. It isthe most advanced methodology to accurately describe excitonpopulation dynamics in the presence of a bath of Brownianoscillators and has been used, for example, to model exciton relax-ation in the light harvesting complex II of Rhodopseudomonasacidophila strain 10050 [113]. More specifically, we employed thescaled HEOM approach as it has been shown to converge fasterthan the original formulation [114,115]. For these calculations,each chromophore couples to a local bath and the coupling is char-acterized by a spectral density J(x) of Ohmic form with a Drudecutoff, where J(x) = 2kcx/(x2+c2). The reorganization energyk = 410 cm�1 characterizes the system-bath coupling, determinedby half the Stokes shift, and a bath correlation time c�1 = 0.56 pstaken from experimental studies of the dynamic Stokes shift[116]. The temperature was set to 300 K, as in our experimentalconditions.

The calculated population dynamics were best fit using a biex-ponential model, yielding time constants of 2 ps and 450 ps withweightings of 81% and 19%, respectively (Supplemental Figure S7).The short time constant corresponds to upper exciton depopula-tion and the long time constant corresponds to repopulation of

the upper exciton. The simulated results are consistent withbath-mediated energy transfer timescales in the strong electroniccoupling limit, but are inconsistent with the current experimentalresult – the calculated relaxation time is too slow by two orders ofmagnitude.

Clearly, in the case of the PDI dimer (and possibly other molec-ular exciton systems in the strong coupling regime) intramoleculardegrees of freedom, that is, radiationless transitions known asinternal conversion, completely decide the rates of populationrelaxation. In the case of internal conversion in a molecule – forexample, population relaxation from the S2 to S1 state – it is wellknown that intramolecular normal modes provide the pathwayfor promoting rapid population relaxation by a breakdown of theBorn–Oppenheimer approximation [117,118]. Lin explicitly con-sidered both intramolecular and intermolecular degrees of free-dom in an early study of radiationless transition theory andconcluded that mainly as a consequence of the low frequenciesof bath modes, ‘electronic energy will first go into the internal nor-mal modes and then through the vibrational relaxation it transfersto the lattice vibration’ [119]. Our observations reveal that similarradiationless transitions govern relaxation between exciton states,

C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33 31

at least in the case of strong electronic coupling in the homodimercase.

There are a growing number of reports of ultrafast populationrelaxation among exciton states [49–54]. Deactivation of excitedstates through conical intersections, to the ground state [120–125] and to other excited states [126], is a process that is becomingincreasingly understood for many systems with predicted relaxa-tion timescales as fast as 10 fs; for example, in the case of singleDNA bases [127]. However, addressing this problem requires extre-mely high temporal resolution experimentally, along with extraor-dinary computational efforts [128]. Here, we demonstrate that2DES is a useful way to detect these ultrafast processes and suggestthat future computational efforts be made in the direction ofunderstanding the intramolecular radiationless transitionsthrough exciton states.

Non-adiabatic couplings enable transitions between electronicstates depending on their strength and their relationship withmolecular motion. Recent theoretical studies on conjugatedpolymer aggregates have suggested that torsional modes areimportant in the generation of conical intersections between exci-ton states and can drive the ultrafast relaxation in coupled systems[125,128,129]. PDI units have now been characterized to exhibit alow frequency torsional mode associated with out-of-plane twist-ing [70]. Experimentally, relaxation of exciton states formed by theinteraction between conjugated polymers has been shown toproceed with a timescale of <100 fs, although due to experimentallimitations these experiments were unable to resolve clearly thedecay constant [130]. With a pulse width of �10 fs, our experi-ments allow for the real-time direct observation of the relaxationof the upper exciton state after photoexcitation.

Understanding the dynamics occurring after photoexcitation ofexciton states is important for understanding the energy transferprocess between donor and acceptor exciton states in natural lightharvesting systems. A common picture of electronic energy trans-fer is a combination of coherent and incoherent hopping, whereexcitations are delocalized between typically 2–4 chromophores(except in the case of chlorosomes) and are subsequently trans-ferred through an incoherent Förster hopping to acceptor excitonstates. It is possible that the mechanism of energy transfer involvesan ultrafast energy equilibration as the initial step, and it would beinteresting to understand what sort of effect this process wouldhave on the energy transfer rate as well as the conditions necessaryto support it. Our current Letter indicates that exciton relaxation inour PDI model system proceeds by radiationless transitions on atimescale two orders of magnitude faster than bath-mediated pre-dictions – not only does this preclude the possibility of long-livedcoherences, but also indicates that the dynamics are dominated byintramolecular degrees of freedom. Such rapid internal conversionmay be specific to the PDI dimer, nevertheless this result suggestsit is worthwhile to examine more closely internal conversion inother excitonic systems.

4. Conclusions

This Letter demonstrates that for a model homodimer system inthe strong electronic coupling regime, energy relaxation within theexcitonic manifold occurs in the very early (<50 fs) timeframe afterphotoexcitation, precluding the persistence of long-lived coher-ences between exciton states. The bath-mediated relaxation timepredicted by the hierarchical equations of motion is two ordersof magnitude longer than that observed. This indicates that intra-molecular degrees of freedom must dictate the relaxation ratethrough radiationless internal conversion pathways. In contrastto bath-mediated relaxation, electronic energy can be very rapidlyconverted to internal vibrational energy through non-adiabatic

couplings–otherwise assumed to vanish in the Born–Oppenheimerapproximation. Such couplings can enable transitions between po-tential energy surfaces through a conical intersection. In the case ofthe PDI dimer, low-frequency torsional modes, known to promotenon-adiabatic dynamics in similar excitonically coupled systems,may be involved in the observed ultrafast dynamics.

As stated in the introductory section, one of the motivations forexploring the photoinitiated dynamics in this model system was togain insights into the ultrafast mechanism(s) of photosyntheticexcitons to further understand the initial processes involved inthe efficient light harvesting processes in photosynthesis. The re-sults presented in this manuscript offer a plausible alternativemechanism for sub-100-fs energy equilibration within pigment-protein light harvesting complexes. Since the formation of stronglycoupled exciton states is a common feature among photosyntheticapparatus, it stands to question whether ultrafast internal conver-sion down the excited state manifold following photoexcitation is akey component of light harvesting efficiency or whether it is sup-pressed by chromophore-protein interactions.

Acknowledgments

This work was supported by the Natural Sciences and Engineer-ing Research Council of Canada, DARPA (QuBE) and the UnitedStates Air Force Office of Scientific Research contract FA9550-13-1-0005 to G.D.S., and by research scholarship from the Natural Sci-ences and Engineering Research Council of Canada and OntarioGraduate Scholarship to C.C.J. A.S. acknowledges the NetherlandsOrganization for Scientific Research for support through a VENIgrant. A.S. thanks prof. J. Reimers for kindly sharing his Dushin code.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.cplett.2014.03.007.

References

[1] G.D. Scholes, G.R. Fleming, A. Olaya-Castro, R. van Grondelle, Nat. Chem. 3(10) (2011) 763.

[2] O. Kühn, S. Lochbrunner, Semiconduct. Semimet. 85 (2011) 47.[3] G.D. Scholes, G. Rumbles, Nat. Mater. 5 (9) (2006) 683.[4] F.C. Spano, Annu. Rev. Phys. Chem. 57 (2006) 217.[5] J.L. Bredas, D. Beljonne, V. Coropceanu, J. Cornil, Chem. Rev. 104 (11) (2004)

4971.[6] D.J. Heijs, V.A. Malyshev, J. Knoester, Phys. Rev. Lett. 95 (17) (2005) 177402.[7] A. Ishizaki, G.R. Fleming, Ann. Rev. Condens. Matter Phys. 3 (1) (2012) 333.[8] Y.C. Cheng, G.R. Fleming, Annu. Rev. Phys. Chem. 60 (2009) 241.[9] T.G. Goodson 3rd., Acc. Chem. Res. 38 (2) (2005) 99.

[10] S.H. Yau, O. Varnavski, T. Goodson, Acc. Chem. Res. 46 (7) (2013) 1506.[11] M. Bednarz, V.A. Malyshev, J. Knoester, Phys. Rev. Lett. 91 (21) (2003) 217401.[12] D.M. Eisele, J. Knoester, S. Kirstein, J.P. Rabe, D.A. Vanden Bout, Nat.

Nanotechnol. 4 (10) (2009) 658.[13] D.M. Eisele et al., Nat. Chem. 4 (8) (2012) 655.[14] D. Beljonne, J. Cornil, R. Silbey, P. Millie, J.L. Bredas, J. Chem. Phys. 112 (10)

(2000) 4749.[15] A.V. Aggarwal et al., Nat. Chem. 5 (11) (2013) 964.[16] S. Liu, D. Schmitz, S.-S. Jester, N.J. Borys, S. Hoeger, J.M. Lupton, J. Phys. Chem.

B 117 (16) (2013) 4197–4203.[17] J.K. Sprafke et al., J. Am. Chem. Soc. 133 (43) (2011) 17262.[18] F. Paquin et al., Phys. Rev. B 88 (15) (2013).[19] F.C. Spano, Acc. Chem. Res. 43 (3) (2010) 429.[20] E.K.L. Yeow, K.P. Ghiggino, J.N.H. Reek, M.J. Crossley, A.W. Bosman, A.

Schenning, E.W. Meijer, J. Phys. Chem. B 104 (12) (2000) 2596–2606.[21] E. Schwartz et al., Chemistry 15 (11) (2009) 2536.[22] J. Wang et al., J. Phys. Chem. 117 (31) (2013) 9288.[23] C. Olbrich et al., J. Phys. Chem. B 115 (26) (2011) 8609–8621.[24] J. Barber, P.D. Tran, J. R. Soc. Interface 10 (81) (2013) 20120984.[25] G.D. Scholes, T. Mirkovic, D.B. Turner, F. Fassioli, A. Buchleitner, Energy

Environ. Sci. 5 (11) (2012) 9374.[26] R.E. Blankenship et al., Science 332 (6031) (2011) 805.[27] J. Yang, M.C. Yoon, H. Yoo, P. Kim, D. Kim, Chem. Soc. Rev. 41 (14) (2012)

4808.

32 C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33

[28] M.K. Panda, K. Ladomenou, A.G. Coutsolelos, Coord. Chem. Rev. 256 (21–22)(2012) 2601.

[29] T. Polivka, V. Sundstrom, Chem. Rev. 104 (4) (2004) 2021.[30] G.D. Scholes, G.R. Fleming, J. Phys. Chem. B 104 (8) (2000) 1854–1868.[31] S. Jang, M.D. Newton, R.J. Silbey, J. Phys. Chem. B 111 (24) (2007) 6807–6814.[32] G.D. Scholes, X.J. Jordanides, G.R. Fleming, J. Phys. Chem. B 105 (8) (2001)

1640–1651.[33] H. Sumi, J. Phys. Chem. B 103 (1) (1999) 252–260.[34] M.N. Yang, G.R. Fleming, Chem. Phys. 275 (1–3) (2002) 355.[35] W.M. Zhang, T. Meier, V. Chernyak, S. Mukamel, J. Chem. Phys. 108 (18)

(1998) 7763.[36] V.I. Novoderezhkin, A.B. Doust, C. Curutchet, G.D. Scholes, R. van Grondelle,

Biophys. J. 99 (2) (2010) 344.[37] A. Ishizaki, G.R. Fleming, J. Chem. Phys. 130 (23) (2009).[38] R. van Grondelle, V.I. Novoderezhkin, Phys. Chem. Chem. Phys. 8 (7) (2006)

793.[39] T. Brixner, J. Stenger, H.M. Vaswani, M. Cho, R.E. Blankenship, G.R. Fleming,

Nature 434 (7033) (2005) 625.[40] G.S. Engel et al., Nature 446 (7137) (2007) 782.[41] E.L. Read, G.S. Engel, T.R. Calhoun, T. Mancal, T.K. Ahn, R.E. Blankenship, G.R.

Fleming, Proc. Natl. Acad. Sci. USA 104 (36) (2007) 14203.[42] G. Panitchayangkoon et al., Proc. Natl. Acad. Sci. USA 107 (29) (2010) 12766.[43] T.R. Calhoun, N.S. Ginsberg, G.S. Schlau-Cohen, Y.C. Cheng, M. Ballottari, R.

Bassi, G.R. Fleming, J. Phys. Chem. B 113 (51) (2009) 16291–16295.[44] G.S. Schlau-Cohen, A. Ishizaki, T.R. Calhoun, N.S. Ginsberg, M. Ballottari, R.

Bassi, G.R. Fleming, Nat. Chem. 4 (5) (2012) 389.[45] E. Collini, C.Y. Wong, K.E. Wilk, P.M.G. Curmi, P. Brumer, G.D. Scholes, Nature

463 (7281) (2010) U644.[46] D.B. Turner, K.E. Wilk, P.M.G. Curmi, G.D. Scholes, J. Phys. Chem. Lett. 2 (15)

(2011) 1904.[47] R. Hildner, D. Brinks, J.B. Nieder, R.J. Cogdell, N.F. van Hulst, Science 340

(6139) (2013) 1448.[48] J. Dostál, T. Mancal, R. Augulis, F. Vácha, J. Pšencík, D. Zigmantas, J. Am. Chem.

Soc. 134 (28) (2012) 11611.[49] J.M. Anna, E.E. Ostroumov, K. Maghlaoui, J. Barber, G.D. Scholes, J Phys Chem

Lett 3 (24) (2012) 3677.[50] A.F. Fidler, V.P. Singh, P.D. Long, P.D. Dahlberg, G.S. Engel, J. Phys. Chem. Lett.

4 (9) (2013) 1404.[51] J.A. Myers, K.L.M. Lewis, F.D. Fuller, P.F. Tekavec, C.F. Yocum, J.P. Ogilvie, J.

Phys. Chem. Lett. 1 (19) (2010) 2774.[52] G.S. Schlau-Cohen et al., J. Phys. Chem. B 113 (46) (2009) 15352–15363.[53] J.M. Womick, A.M. Moran, J. Phys. Chem. B 113 (48) (2009) 15747.[54] D. Zigmantas, E.L. Read, T. Mancal, T. Brixner, A.T. Gardiner, R.J. Cogdell, G.R.

Fleming, Proc. Natl. Acad. Sci. USA 103 (34) (2006) 12672.[55] M. Kasha, Radiat. Res. 20 (1963) 55.[56] A.V. Pisliakov, T. Mancal, G.R. Fleming, J. Chem. Phys. 124 (23) (2006).[57] A. Halpin, P.J.M. Johnson, R. Tempelaar, R.S. Murphy, J. Knoester, T.L.C. Jansen,

R.J.D. Miller, Nat. Chem. (2014).[58] A. Perdomo-Ortiz, J.R. Widom, G.A. Lott, A. Aspuru-Guzik, A.H. Marcus, J. Phys.

Chem. B 116 (35) (2012) 10757–10770.[59] L. OddosMarcel et al., J. Phys. Chem. – US 100 (29) (1996) 11850.[60] D. Hayes, G.B. Griffin, G.S. Engel, Science 340 (6139) (2013) 1431.[61] F. Koch, M. Kullmann, U. Selig, P. Nuernberger, D.C.G. Gotz, G. Bringmann, T.

Brixner, New J. Phys. (2013) 15.[62] R. Kumble, S. Palese, V.S.Y. Lin, M.J. Therien, R.M. Hochstrasser, J. Am. Chem.

Soc. 120 (44) (1998) 11489.[63] H.S. Cho et al., J. Phys. Chem. A 104 (15) (2000) 3287.[64] S. Mukamel, Principles of Nonlinear Optical Spectroscopy, Oxford University

Press, New York, 1995.[65] D.M. Jonas, Annu. Rev. Phys. Chem. 54 (2003) 425.[66] M.H. Cho, Chem. Rev. 108 (4) (2008) 1331.[67] S.T. Cundiff, S. Mukamel, Phys. Today 66 (7) (2013) 44.[68] K.W. Stone, D.B. Turner, K. Gundogdu, S.T. Cundiff, K.A. Nelson, Acc. Chem.

Res. 42 (9) (2009) 1452.[69] D.B. Turner, R. Dinshaw, K.K. Lee, M.S. Belsley, K.E. Wilk, P.M.G. Curmi, G.D.

Scholes, Phys. Chem. Chem. Phys. 14 (14) (2012) 4857.[70] M. Myahkostupov, V. Prusakova, D.G. Oblinsky, G.D. Scholes, F.N. Castellano, J.

Org. Chem. 78 (17) (2013) 8634.[71] M.H. Hennessy, Z.G. Soos, R.A. Pascal, A. Girlando, Chem. Phys. 245 (1–3)

(1999) 199.[72] K.A. Kistler, C.M. Pochas, H. Yamagata, S. Matsika, F.C. Spano, J. Phys. Chem. B

116 (1) (2012) 77–86.[73] J.R. Lackowicz, Principles of Fluorescence Spectroscopy, third ed., Springer,

New York, 2006 (pp. 353–354, 361–364).[74] J.D. Hybl, A.A. Ferro, D.M. Jonas, J. Chem. Phys. 115 (14) (2001) 6606.[75] M. Khalil, N. Demirdoven, A. Tokmakoff, Phys. Rev. Lett. 90 (4) (2003).[76] S.M. Gallagher, A.W. Albrecht, J.D. Hybl, B.L. Landin, B. Rajaram, D.M. Jonas, J.

Opt. Soc. Am. B 15 (8) (1998) 2338–2345.[77] S. Park, K. Kwak, M.D. Fayer, Laser Phys. Lett. 4 (10) (2007) 704.[78] J.M. Anna, Y. Song, R. Dinshaw, G.D. Scholes, Pure Appl. Chem. 85 (7) (2013)

1307.[79] M. Kasha, H.R. Rawls, A.M. El-Bayoumi, Pure Appl. Chem. 11 (3–4) (1965) 371.[80] A. Davydov, Theory of Molecular Excitons, McGraw-Hill, New York, 1962.[81] J. Franck, E. Teller, J. Chem. Phys. 6 (12) (1938) 861.

[82] R.L. Fulton, M. Gouterman, J. Chem. Phys. 35 (3) (1961) 1059.[83] R.L. Fulton, M. Gouterman, J. Chem. Phys. 41 (8) (1964) 2280.[84] A. Witkowski, W. Moffitt, J. Chem. Phys. 33 (3) (1960) 872.[85] J. Roden, A. Eisfeld, M. Dvorak, O. Bunermann, F. Stienkemeier, J. Chem. Phys.

134 (5) (2011) 054907.[86] J.M. Giaimo, A.V. Gusev, M.R. Wasielewski, J. Am. Chem. Soc. 124 (29) (2002)

8530.[87] U. Selig et al., ChemPhysChem 14 (7) (2013) 1413.[88] H. Yoo et al., J. Phys. Chem. B 116 (42) (2012) 12878–12886.[89] M. Lippitz et al., Phys. Rev. Lett. 92 (10) (2004) 2565.[90] J. Hernando et al., Phys. Rev. Lett. 93 (23) (2004).[91] H. Langhals, S. Demmig, H. Huber, Spectrochim. Acta A 44 (11) (1988) 1189.[92] V. Settels, W.L. Liu, J. Pflaum, R.F. Fink, B. Engels, J. Comput. Chem. 33 (18)

(2012) 1544.[93] P. Reynders, W. Kühnle, K.A. Zachariasse, J. Am. Chem. Soc. 112 (1990) 3929.[94] P. Reynders, H. Dreeskamp, W. Kühnle, K.A. Zachariasse, J. Phys. Chem. – US

91 (1987) 3983.[95] G.D. Scholes, K.P. Ghiggino, J. Phys. Chem. – US 98 (17) (1994) 4580.[96] M.R. Philpott, J. Chem. Phys. 55 (5) (1971) 2039.[97] F.C. Spano, J. Chem. Phys. 116 (13) (2002) 5877.[98] A. Stradomska, P. Petelenz, J. Chem. Phys. 131 (4) (2009).[99] A. Olaya-Castro, G.D. Scholes, Int. Rev. Phys. Chem. 30 (1) (2011) 49.

[100] G.D. Scholes, Annu. Rev. Phys. Chem. 54 (2003) 57.[101] C. Haines, M. Chen, K.P. Ghiggino, Sol. Energy Mater. Sol. Cells 105 (2012) 287.[102] T. Wilhelm, J. Piel, E. Riedle, Opt. Lett. 22 (19) (1997) 1494.[103] J.R. Reimers, J. Chem. Phys. 115 (20) (2001) 9103.[104] D.B. Turner, Y. Hassan, G.D. Scholes, Nano Lett. 12 (2) (2012) 880.[105] D. Abramavicius, B. Palmieri, D.V. Voronine, F. Sanda, S. Mukamel, Chem. Rev.

109 (6) (2009) 2350.[106] M.H. Cho, J.Y. Yu, T.H. Joo, Y. Nagasawa, S.A. Passino, G.R. Fleming, J. Phys.

Chem. – US 100 (29) (1996) 11944.[107] T.L.C. Jansen, J. Knoester, Acc. Chem. Res. 42 (9) (2009) 1405.[108] O. Golonzka, A. Tokmakoff, J. Chem. Phys. 115 (1) (2001) 297.[109] D. Beljonne et al., Proc. Natl. Acad. Sci. USA 99 (17) (2002) 10982.[110] R. Chang, M. Hayashi, S.H. Lin, J.H. Hsu, W.S. Fann, J. Chem. Phys. 115 (9)

(2001) 4339.[111] J. Piris et al., J. Phys. Chem. C 113 (32) (2009) 14500.[112] Y. Tanimura, R. Kubo, J. Phys. Soc. Jpn. 58 (1) (1989) 101.[113] J. Strumpfer, K. Schulten, J. Chem. Phys. 134 (9) (2011) 095102.[114] Q. Shi, L.P. Chen, G.J. Nan, R.X. Xu, Y.J. Yan, J. Chem. Phys. 130 (8) (2009).[115] J. Zhu, S. Kais, P. Rebentrost, A. Aspuru-Guzik, J. Phys. Chem. B 115 (6) (2011)

1531–1537.[116] M.L. Horng, J.A. Gardecki, A. Papazyan, M. Maroncelli, J. Phys. Chem. – US 99

(1995) 17311.[117] M. Bixon, J. Jortner, J. Chem. Phys. 48 (2) (1968) 715.[118] G.W. Robinson, R.P. Frosch, J. Chem. Phys. 38 (5) (1963) 1187.[119] S.H. Lin, J. Chem. Phys. 44 (10) (1966) 3759.[120] J.S. Lim, S.K. Kim, Nat. Chem. 2 (8) (2010) 627.[121] M.A. Robb, M. Garavelli, M. Olivucci, F. Bernardi, Rev. Comput. Chem. 15

(2000) 87.[122] A. Sanchez-Galvez, P. Hunt, M.A. Robb, M. Olivucci, T. Vreven, H.B. Schlegel, J.

Am. Chem. Soc. 122 (12) (2000) 2911.[123] A.L. Sobolewski, W. Domcke, Phys. Chem. Chem. Phys. 6 (10) (2004) 2763.[124] G. Groenhof, L.V. Schafer, M. Boggio-Pasqua, M. Goette, H. Grubmuller, M.A.

Robb, J. Am. Chem. Soc. 129 (21) (2007) 6812.[125] I. Burghardt, J.T. Hynes, E. Gindensperger, L.S. Cederbaum, Phys. Scr. 73 (1)

(2006) C42.[126] M. Schröter, O. Kühn, J. Phys. Chem. A 117 (32) (2013) 7580.[127] M. Barbatti, A.J.A. Aquino, J.J. Szymczak, D. Nachtigallova, P. Hobza, H.

Lischka, Proc. Natl. Acad. Sci. USA 107 (50) (2010) 21453.[128] T. Nelson, S. Fernandez-Alberti, V. Chernyak, A.E. Roitberg, S. Tretiak, J. Phys.

Chem. B 115 (18) (2011) 5402–5414.[129] K. Banerjee, G. Gangopadhyay, J. Phys. B – At Mol. Opt. 43 (23) (2010).[130] M.H. Chang, M.J. Frampton, H.L. Anderson, L.M. Herz, Phys. Rev. Lett. 98 (2)

(2007).

Chanelle C. Jumper received her M.Sc. in AnalyticalChemistry in 2011 from the University of Calgary underthe supervision of Prof. David Schriemer. She laterjoined Prof. Gregory Scholes at the University of Torontoas a Ph.D. student and NSERC fellow in Physical Chem-istry. Her current research concentrates on under-standing exciton dynamics in natural and synthetic lightharvesting systems, and the role that the environmentplays in these dynamics—especially the differencebetween living and non-living matter.

C.C. Jumper et al. / Chemical Physics Letters 599 (2014) 23–33 33

Jessica M. Anna obtained her Ph.D. in Physical Chem-istry from the University of Michigan in 2011. She thenjoined Prof. Gregory Scholes research group at theUniversity of Toronto as a postdoctoral research fellow.Her current research interests include using multidi-mensional spectroscopic methods to elucidate energytransfer pathways and timescales.

Anna Stradomska received her M.Sc. in 2004 and Ph.D.in 2009 (in theoretical chemistry) from the JagiellonianUniversity, Poland. She is currently a postdoctoralresearcher at the Zernike Institute for Advanced Mate-rials at the University of Groningen, The Netherlands.Her research concentrates on theoretical modelling ofexcited state dynamics in complex organic materials,charge-transfer excitons, and non-adiabatic effects.

Juleon Schins received his Ph.D. in 1992 at the Uni-versity of Amsterdam, The Netherlands. After postdoc-toral research in France (Laboratoire d’OptiqueAppliquée, ENSTA, Palaiseau) and The Netherlands(Universiteit Twente) he joined the Chemistry faculty atDelft University of Technology. His current researchinterests include charge carrier mobility and excitondynamics in photo-excited semiconductor nanostruc-tures (dots, rods, and platelets), studied by means ofpump-probe spectroscopy, probing in the visible,infrared, THz, and GHz regimes.

Mykhaylo Myahkostupov received his Master of Sci-ence in Chemical Technology and Engineering from LvivPolytechnic National University (Ukraine) in 1997. Aftercompleting his Ph.D. degree in Chemistry at RutgersUniversity (USA) in 2008, he joined the research groupof Prof. Felix N. Castellano as a Postdoctoral ResearchAssociate (Bowling Green State University, USA), fol-lowed by the appointment as a Research Scholar atNorth Carolina State University (USA) in 2013. His majorresearch interests are centered around the developmentand characterization of molecular photonic and elec-tronic materials for applications in the solar energy

conversion.

Valentina Prusakova obtained her Specialist degree atthe Department of Chemistry, Saint-Petersburg StateUniversity. After that, she received her PhD fromBowling Green State University in 2012 under thesupervision of Prof. Felix N. Castellano, followed by a 1-year postdoctoral research fellowship, focusing on thedevelopment of photoactive materials based on transi-tion metal complexes. In 2014 she joined the group ofProf. Sandra Dirè at the University of Trento as a post-doctoral research fellow in order to participate in thedevelopment of memristive devices.

Daniel G. Oblinsky is a Ph.D. student and NSERC fellowat the University of Toronto under the guidance ofProfessor Gregory D. Scholes. Some of his interestsinclude brewing craft beer, carefully perfecting the artof an espresso, sampling the wine regions of France andItaly, and golfing. His research interests are vast andrange from single molecule quantum Monte Carlo cal-culations to elucidating the dynamics of proteins andhow coherent motion may influence energy transfer.

Jasper Knoester received a PhD in Theoretical Physicsat the University of Utrecht, The Netherlands in 1987.After a postdoctoral position at the Chemistry Depart-ment of the University of Rochester (NY), in 1989 hewas awarded a Huygens Fellowship by the NetherlandsOrganization for Scientific Research (NWO) to start anew line of research at the Chemistry Department of theUniversity of Groningen, The Netherlands. He wasappointed full professor of Theory of Condensed Matterin 1993 at this university. His research interests includeexciton localization and dynamics in synthetic andnatural molecular aggregates, vibrational dynamics in

proteins and other complex molecular systems, visible and infrared nonlinear(mutidimensional) spectroscopy of complex molecular systems, and plasmonicsystems for nanophotonics.

Felix (Phil) Castellano earned a B.A. in Chemistry fromClark University in 1991 and a Ph.D. in Chemistry fromJohns Hopkins University in 1996. Following a NIHPostdoctoral Fellowship at the University of Maryland,School of Medicine, he accepted a position as AssistantProfessor at Bowling Green State University in 1998. Hewas promoted to Associate Professor in 2004, to Pro-fessor in 2006, and was appointed Director of the Centerfor Photochemical Sciences in 2011. In 2013, he movedhis research program to North Carolina State Universitywhere he is a Professor in the Department of Chemistry.His current research focuses on metal-organic chromo-

phore photophysics and energy transfer, photochemical upconversion phenomena,photocatalysis, and interfacial electron transfer processes relevant to photovolatics.

Greg D. Scholes has been a professor at the University ofToronto in the Department of Chemistry since 2000.This year he will move to Princeton University. Hisresearch group examines photophysics in systemsranging from semiconductor nanocrystals to conjugatedpolymers to photosynthetic light-harvesting proteins.He is especially interested in uncovering microscopicdetails of light-induced energy capture and conversionprocesses in photosynthesis and organic photovoltaicsusing a combination of femtosecond laser experimentsand theory. Dr. Scholes serves as an Editorial Advisor forNew Journal of Physics and Senior Editor for the Journal

of Physical Chemistry Letters.