Fluctuating exciton localization in giantp-conjugated spoked-wheel macrocyclesA. Vikas Aggarwal1, Alexander Thiessen2, Alissa Idelson1, Daniel Kalle1, Dominik Wursch3,

Thomas Stangl3, Florian Steiner3, Stefan-S. Jester1, Jan Vogelsang3, Sigurd Hoger1*and John M. Lupton2,3*

Conjugated polymers offer potential for many diverse applications, but we still lack a fundamental microscopicunderstanding of their electronic structure. Elementary photoexcitations (excitons) span only a few nanometres of amolecule, which itself can extend over microns, and how their behaviour is affected by molecular dimensions is notimmediately obvious. For example, where is the exciton formed within a conjugated segment and is it always situated onthe same repeat units? Here, we introduce structurally rigid molecular spoked wheels, 6 nm in diameter, as a model ofextended p conjugation. Single-molecule fluorescence reveals random exciton localization, which leads to temporallyvarying emission polarization. Initially, this random localization arises after every photon absorption event because oftemperature-independent spontaneous symmetry breaking. These fast fluctuations are slowed to millisecond timescalesafter prolonged illumination. Intramolecular heterogeneity is revealed in cryogenic spectroscopy by jumps in transitionenergy, but emission polarization can also switch without a spectral jump occurring, which implies long-range homogeneityin the local dielectric environment.

Cyclic structures of various levels of symmetry are ubiquitousin nature, from benzene and pyrrole to members of theporphyrin family, such as haem or chlorophylls, to photosyn-

thetic antenna complexes1, and structural rigidity is crucial to thesemolecules on different length scales. However, synthetic compoundsthat derive macroscopic functions by mimicking elementary aspectsof electron or energy transfer in organic semiconductors tend to belinear in structure2. Although such materials, most notably p-con-jugated polymers, possess a range of desirable functional character-istics, to formulate a comprehensive microscopic picture of howindividual covalently bound monomer units arrange in space toform discrete p-conjugated segments remains challenging3–6.Conjugation and shape of the molecule are fundamentally inter-linked5. On the one hand, spectroscopic techniques can, in prin-ciple, unveil information on electronic structure. On the otherhand, physical shape, which can exhibit a level of diversity reminis-cent of conformational degrees of freedom in proteins, is muchharder to assess.

A conjugated polymer consists of a chain of repeating monomerunits. p electrons delocalize between the monomers, but maybecome localized on longer scales because of the formation ofchemical or structural defects7. An individual segment of thepolymer that supports the p orbital is referred to as a chromophoreand the elementary excitation on a chromophore is an exciton(a tightly bound electron–hole pair). The optical properties ofpolymers, such as spectral shape (width of energy bands and strengthof vibronic coupling), are accounted for by excitonic couplingmodels in which intrachromophoric interactions betweenmonomers are described in the framework of J aggregates, andinteractions between chromophores are ascribed to either J (inline)or H aggregation (cofacial)8. The exciton itself is of the order of2 nm in size, which can be much smaller than the actual conjugated

segment7. Depending on the magnitude of structural relaxation inthe excited state, the exciton may be free to move within the chro-mophore, and thereby increase transition intensity because thenumber of electrons involved in the transition increases8, but itmay also become localized7. Proximal chromophores can coupleto each other, which leads to further spreading of excitationenergy in the macromolecule9,10.

Fluorescence spectroscopy is often used to infer information onelectronic structure, coupling mechanisms and conformation, butwithout definitive knowledge of molecular conformation to beginwith, the parameters remain intractable. In particular, it is not self-evident that an exciton should always form on precisely the samemonomer units of a chromophore. Also, structural relaxation in theexcited state breaks molecular symmetry and leads to exciton self-trapping11,12 (the spatial localization of excitation energy caused bythe nuclear rearrangement of the molecule after a redistribution incharge density). Does this process always follow the same pathway?

To examine this question, we designed a giant molecular spoked-wheel structure with a conjugated shape-persistent macrocyclic rimas a model of chromophore formation and interchromophoriccoupling13. Using single-molecule techniques, we uncovered twodistinct localization mechanisms: spontaneous symmetry breaking(with the exciton localizing randomly to different parts of the ringafter every photoexcitation event) and slower photoinducedsymmetry breaking, which leads to fluctuating exciton localizationon the millisecond timescale.

Results and discussionRing design. The design of conjugated macrocyclic structuresrequires careful consideration of rigidity to prevent scissions inthe p system caused by deformation of the overall ring as a resultof the limited persistence length of rigid-rod building blocks14.

1Kekule-Institut fur Organische Chemie und Biochemie der Universitat Bonn, Gerhard-Domagk-Strasse 1, 53121 Bonn, Germany, 2Department of Physics andAstronomy, University of Utah, Salt Lake City, Utah 84112, USA, 3Institut fur Experimentelle und Angewandte Physik, Universitat Regensburg,Universitatsstrasse 31, D-93040 Regensburg, Germany. *e-mail: hoeger@uni-bonn.de; john.lupton@ur.de

ARTICLESPUBLISHED ONLINE: 29 SEPTEMBER 2013 | DOI: 10.1038/NCHEM.1758

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 1

We employed a phenylene–ethynylene–butadiynylene-basedscaffold15, which offers an optimum in rigidity combined withdesirable optical properties in the visible spectrum, to templatecarbazole units. Previously studied carbazole-based compoundshad electronic transitions in the ultraviolet spectrum16–18, whichmakes them poorly suited to single-molecule investigations, andporphyrin-based macrocycles of comparable rigidity to our ringsshowed very low fluorescence yields19–25 with optical transitions inthe near-infrared region. Other intriguing cyclic structures were

not conjugated fully26,27. Figure 1 illustrates the design approach.Six N-phenylcarbazole units are linked to each other byphenylene–ethynylene–butadiynylene moieties and form thenominally conjugated rim (green) of the spoked wheel 1, wherethe spokes are not part of the rim conjugation. The ring issynthesized by Buchwald–Hartwig coupling of the rim segments 2(green) and spoke modules 3 (red), selective removal of thecyanopropyldimethylsilyl protecting group from 428 and couplingof the resulting acetylene to the hub 6 in a sixfold Sonogashira

NHSi Si

Si Br

OC8H17

C8H17O

NC CN

N

Si

Si

C8H17O

OC8H17

C8H17O

OC8H17

CN

CN

N

R3

R3

OR2R2O

OR2

R2O

OR2

OR2

NR3

R3

OR2

R2O

R2O

OR2

OR2

R2O

N

R3

R3R2O

OR2

R2O OR2

OR2

R2O

N

R3

R3OR2

R2O

OR2R2OR2O

OR2

N

R3

R3R2O

OR2

R2O OR2OR2

R2O

NR3 R3

OR2

R2O

R2O

OR2

OR2

R2O

N

NN

N

N

N

OR2

R2O

OR2

R2O

OR2

OR2 OR2

R2O

R2O

OR2

OR2

OR2

OR2

R2O

OR2

R2O

R2O

OR2

OR2

R2O

R2O

OR2

R2O

OR2OR2

R2O

OR2

R2O

OR2

R2O

OR2

R2O

OR2

R2O

R2O

OR2

NC

R1

R1 =

R1 = H

Si CN

+

OC8H17

C8H17O

2

3

4

5

1 R3 =

R3 = H

7

8

R2 = C8H17

iii

iii

iv

v

I

I I

I

II6

N

C8H17O

OC8H17

OC8H17

C8H17On = 1, 9; n = 2, 10; n = 6, 11Si Si

n

C8H17O

OC8H17

OC8H17

C8H17O

R2 = C8H17

Si CN

Figure 1 | Structure and synthesis of the spokes and ring 1, and of the non-cyclic analogues 9–11. (i) 2-Di-tert-butylphosphino-2′,4′,6′-triisopropylbiphenyl,

tris(dibenzylideneacetone)dipalladium(0) (Pd2dba3), NaOtBu, toluene, 80 8C, 1 h, 66%; (ii) K2CO3, THF, methanol, room temperature (r.t.), 1 h, 95%;

(iii) Pd2dba3, PtBu3, CuI, piperidine, 120 8C, 16 min microwave irradiation, 74%; (iv) tetra-n-butylammonium fluoride, THF, r.t., 3 h, 74%; (v) Pd(PPh3)2Cl2,

CuI, I2, air, THF, NHiPr2, 50 8C, 60 h, 54%.

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1758

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry2

reaction. Subsequent deprotection of 7 leads to the spoked-wheelprecursor 8, which is cyclized in a palladium-catalysed reactionunder pseudo high-dilution conditions and purified by recyclinggel permeation chromatography (GPC) to give 1 in 54% yield.The purity is confirmed by mass spectrometry, NMRspectroscopy and GPC (Supplementary Section S2). We compared

ring 1 to model oligomers 9–11. The structural rigidity isvisualized in the scanning tunnelling micrographs (STM) in Fig. 2(for details see Supplementary Section S4). Although interactionswith the hexagonal graphite substrate lattice and dense packing ofthe molecules induce slight distortions in the ring geometry, noapparent ruptures in the wheel structures are seen, whichindicates that effective tunnelling arises from conjugation in the rim.

It is not immediately obvious whether conjugation extends alongthe entire rim and whether the ring should be emissive at all. Inmolecules of six-fold symmetry, the S02S1 transition is suppressed,as, for example, in benzene. Some larger macrocycles alsoshow inhibited fundamental transitions29–33. However, slightinteractions with the environment can break molecular symmetryand so make the transition allowed. A notable example is theB850 band in the light-harvesting system LH-II; on the basis ofdipole selection rules, thermally activated emission from the B850band would be expected because the lowest-lying state is dipoleforbidden34. However, thermally activated emission is notobserved experimentally34.

1 is, indeed, highly emissive, with a quantum yield of 71+5%and a short radiative lifetime of 840+60 ps. The results ofroom-temperature absorption and emission spectroscopy(Supplementary Section S5) of the ring and model linear com-pounds in solution are summarized in Table 1. The oligomersexhibit a bathochromic shift in emission and absorption withincreasing size from the monomer to the dimer, which impliesimproved electronic delocalization35. Little difference is seenbetween dimer and hexamer, which illustrates that delocalizationdoes not extend significantly beyond two monomers. Comparableporphyrin-based conjugated ring structures21 fluoresce in thenear-infrared region with a quantum yield of 0.12%, nearly threeorders of magnitude lower than found here.

Room-temperature single-molecule spectroscopy. What is themicroscopic nature of absorption and emission in 1? Which partof the ring absorbs light, and where is light emitted? Figure 2asketches the problem. In excitation, ring symmetry should bepreserved because incident light of any polarization can beabsorbed. Yet emission should arise from the formation of alinear transition dipole, situated anywhere on the annulus. Slightstructural relaxation in the excited state or perturbation of thestructure by the environment will break symmetry34,36–38 andgenerate distinct local potential minima11 into which the excitonrelaxes randomly. These microscopic characteristics of thep system are best resolved by single-molecule techniques thatovercome random averaging effects between molecules.

Single molecules were diluted in Zeonex or poly(methylmetha-crylate) (PMMA) matrices at picomolar concentrations andimaged at room temperature and 4 K in two separate fluorescencemicroscopes. We first addressed the nature of the absorbing tran-sition dipole by determining the distribution of excitation polariz-ation anisotropy values from molecule to molecule. Polarizationanisotropy was quantified in terms of linear dichroism. We recordedphotoluminescence (PL) intensity under alternating horizontally(H) and vertically (V) polarized excitation. Linear dichroism isdefined by the ratio of emission intensities I under the twoexcitation polarizations, Dexcitation (Dex)¼ (IV 2 IH)/(IVþ IH)(refs 4,39). A linear dipole oriented randomly in a plane will yielda distribution that peaks at +1 (ref. 40). A value of Dex¼ 0 willoriginate either from unpolarized absorption, or from a dipoleoriented at 458 with respect to both excitation planes. As moleculeand dipole orientation are distributed randomly, the statistics of Dexprovide information on whether molecular absorption is polarizedor unpolarized. Figure 2b shows histograms of linear dichroism inexcitation for the dimer, hexamer and ring. Following Table 1, thedimer constitutes the effective exciton size in the hexamer and ring.

5 nm

–100 –50 0 50 100

τ (ns)

Occ

urre

nce

10–1 100 101 102 103 104

–1

0

1

τ (μs)

Occ

urre

nce

–1.0 –0.5 0.0 0.5 1.0 –1.0 –0.5 0.0 0.5 1.0

g2 cr

oss

(τ)

Dex Dem

Pot

entia

l

Generalized coordinate

or

Excitation

Excitation Emission

IH

IV

τ

IV

IH

τ I2

I2

IH

IV

c

d

b

a

Horizontal Vertical Horizontal (Unpolarized) Vertical(Unpolarized)

Beads

2mer

6mer

Ring Beads

Figure 2 | Fluctuations in exciton localization caused by spontaneous

symmetry breaking. a, Excitation with arbitrarily polarized light leads to

exciton formation. Excitons relax to segments of length comparable to the

dimer length. Owing to bond-length changes in the excited state, the local

potential in the proximity of the exciton is modified. Such exciton localization

occurs anywhere on the ring. Linear dichroism can be measured in either

excitation (by switching laser polarization, left) or emission (by passing

fluorescence through a polarizing beam splitter, right). b, Linear dichroism

histograms in excitation and emission for 1,597 dimer (2mer), 1,273

hexamer (6mer) and 730 ring molecules. Black bars indicate instrument

response for fluorescent beads. STM images are shown with the graphite

substrate axes indicated in white. c, Temporal photon correlation (pulsed

excitation). Fluorescence passes through a beam splitter and is recorded

with two photodiodes. At delay t¼0 ns between the detectors, photon

coincidence approaches zero; photon antibunching implies the activity of

precisely one chromophore. Dashed lines show calculated thresholds for one

(lower) and two emitters (upper). The experiment was carried out

at a 20 MHz repetition rate, which gave 50 ns spacing on the t axis.

d, Cross-correlation gcross2 (t) between two detector channels of orthogonal

polarization shows no discernible timescale for polarization fluctuations,

which implies that initially ring emission appears unpolarized.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1758 ARTICLES

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 3

As expected, the dimer displays an almost linearly polarizedabsorption, the distribution matching a simple statistical simulationfor a linear dipole40. Deviations from a perfect linear dipole arisebecause the dimer is slightly bent, as seen in the STM image. Aslength increases, the conjugated system becomes more distorted,which lowers the measured Dex values and narrowsthe distribution. For 1, Dex approaches zero—effectively, themolecules are unpolarized absorbers. To assess the experimentalresolution, a histogram of 200 fluorescent beads (perfectunpolarized multichromophoric absorbers and emitters) ofcomparable spectral properties and photon count rates issuperimposed (black bars, Fig. 2b). Identical results were obtainedin wide-field microscopy, which implies that the confocal laserexcitation is free of polarization-distortion artefacts despitethe high numerical aperture (NA) objective used41 (seeSupplementary Section S9 for discussion).

Comparison of the rings to LHII is also interesting: despitesimilar dimensions and symmetry, isotropic absorption was notseen in experiments that involved single complexes36–38. Othersynthetic ring structures have been studied only by ensemblefluorescence depolarization15,21, which reveals ultrafast loss in polar-ization memory that is compatible with our single-molecule results.

At the excitation wavelength used (405 nm), the spokes of thewheel account for �20% of the overall absorption, because spokeand rim absorption overlap spectrally (Supplementary Section S5).It is only possible to discriminate spokes and rim in emission, asthe spectra are shifted by 43 nm. Exciton generation must lead tosymmetry breaking in the excited state because of changes inbond-length alternation and structural relaxation11. We canimage this symmetry breaking directly through the polarizationanisotropy in emission, which was again measured by the lineardichroism. Fluorescence was excited with alternating horizontaland vertical polarized light, and the emission was split into twoorthogonally polarized components from which Demission (Dem)was computed, as illustrated on the right side of Fig. 2a (seeSupplementary Section S1.2 for details). Dex and Dem could there-fore be calculated for the same molecule, which makes the Dexand Dem histograms directly comparable. Identical results werefound for excitation with circularly polarized light. Only the first100 ms of illumination are considered for reasons discussedbelow. For the dimer, the Dem distribution is virtually identical tothat in excitation because the molecule is too small for additionallocalization to arise in the excited state. For the hexamer, theeffect of exciton localization to a unit comparable to the dimer isclearly visible—the Dem distribution resembles that of the dimer.More dipole orientations are available for excitation than for emis-sion of the hexamer, which makes the Dex histogram narrower thanthat for Dem. The situation is very different for 1. Although thephotophysics is comparable to that of the dimer, the Dem histogramis much narrower than that of the dimer, which implies that singlemolecules appear to emit primarily unpolarized light. Unpolarizedreference beads show identical histograms of Dex and Dem (blackbars, Fig. 2b). However, for 1 the Dem distribution is not asnarrow as that of Dex. This broadening of the Dem histogramimplies the possibility that excited-state localization occurred on

some single rings during the experiment. In contrast, if localizationtook place on all molecules within 100 ms of illumination andalways occurred deterministically, that is at the same position, theDem histogram would match that of the dimer.

The dynamic nature of exciton–phonon coupling can give rise tolocalization fluctuations and thus unpolarized PL, but so would sim-ultaneous emission from multiple chromophores. To exclude thispossibility, we studied the statistics of single photons emitted bythe rings. Figure 2c shows the photon correlation, measured by split-ting the emission into two equal paths and recording photon arrivaltimes, as illustrated in the sketch on the left, averaged over 100 singlemolecules. The dashed horizontal lines indicate the anticipated cor-relation thresholds based on photon count rates and backgroundsignal for one and two photons (lower and upper, respectively).Only one photon is emitted at a time, which leads to the pronouncedphoton antibunching dip at zero delay between the two detectors.However, the polarization of this photon is not predetermined.Figure 2d shows the temporal cross-correlation of two orthogonallypolarized detectors averaged over 95 molecules (first 100 ms of illu-mination). The correlation is flat—there is no characteristic time-scale for polarization fluctuations. These results imply that singlephotons are emitted one at a time, and originate from different ran-domly varying segments on the ring, which leads to arbitrary fluc-tuations in emission polarization. Molecular symmetry is brokenspontaneously, and a different localization occurs after eachphotoexcitation event.

The situation becomes very different on longer timescales. Underprolonged illumination, photomodification of the molecule mayoccur, for example through the generation of a radical species10.This photomodification will lead to a quasi-static breaking of mol-ecular symmetry, sketched in Fig. 3a. Changes in photomodificationwith time can induce fluctuations in excited-state localization. Thiseffect is resolved clearly in the temporal evolution of linear dichro-ism. Figure 3b exemplifies a molecule for which linear dichroism ismeasured simultaneously in excitation and emission (furtherexamples are given in Supplementary Section S6). The total fluor-escence intensity shows discrete single-step blinking and anapproximate halving of count rate after 12 seconds of illumination.The corresponding Dex is zero for the first ten seconds, and sub-sequently rises as the overall emission intensity decreases, presum-ably because part of the ring is photobleached10. Dem sets out at zero,drifting to a value of 0.4, and subsequently shows strong discretejumps between positive and negative values. After continued illumi-nation, fluctuations in localization occur, which lead to jumps intransition dipole orientation. The temporal evolution of lineardichroism caused by photomodification can be visualized by plot-ting the Dem histogram as a function of time (Fig. 3c). We selectedmolecules that exhibit |Dem| , 0.1 within the first 100 ms of illumi-nation and did not drop by more than 30% in intensity over fiveseconds. This initial Dem value can correspond to unpolarized PLor emission from a dipole at �458 orientation with respect to thetwo detectors. Out of 2,000 single molecules, 32% fell within thisnarrow range. For a random distribution of dipoles, one wouldexpect40 only 6% of all molecules to be oriented at �458, whichimplies that the low Dem values arise primarily from unpolarized

Table 1 | Absorption, PL, PL quantum yield (PLQY) and PL lifetimes measured in toluene or chloroform solutions at r.t.

Absorptionmaximum (nm)

Molar absorptivity (cm21 M21)(at lmax (nm))*

Absorptionedge (nm)

PL maximum(nm)

PL lifetime(ns)

PLQY (%) Radiativelifetime (ns)

Monomer 9 413 61,200 (417) 431 431 0.97+0.01 84+5 1.15+0.07Dimer 10 422 200,000 (425) 459 459 0.58+0.01 69+5 0.84+0.06Hexamer 11 426 500,000 (430) 460 460 0.47+0.01 66+5 0.71+0.06Ring 1 443 420,000 (444) 462 462 0.60+0.01 71+5 0.84+0.06

*At the maximum of the lowest energy absorption band in chloroform solution.

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1758

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry4

emission. The histogram clearly broadens over time. No suchbroadening is observed for dimers (not shown). We did not find evi-dence that this broadening is reversible under interruption of theillumination (Supplementary Fig. S8). No broadening is seen inthe corresponding Dex histogram, which implies that the ringremains, to a first approximation, an isotropic absorber undercontinued illumination.

At the onset of photoexcitation, that is within the first �100 ms,single-ring fluorescence can appear unpolarized because the emis-sion jumps between equally weighted chromophores on the ringafter every absorption event. With time, random photoinducedlocalization occurs, which leads to non-zero Dem values, whichcan then switch or drift randomly over milliseconds to seconds.Although such millisecond fluctuations in linear dichroism have

been observed previously in multichromophoric macromol-ecules40,42,43, their structural or electronic origin remains unclear.The rings demonstrate that these fluctuations are a secondaryeffect and only arise as a consequence of illumination. Initially,the symmetry of the molecule is preserved, so that randomspontaneous symmetry breaking in the excited state can beobserved. Consequently, spontaneous symmetry breaking is alsovirtually independent of temperature as it constitutes a purely intra-molecular effect with no coupling to the heat bath. At 4 K, �20%of rings showed (|Dem| , 0.1) under the initial illumination,which is comparable to the room-temperature measurement(Supplementary Section S8).

Cryogenic single-molecule spectroscopy. Although they do notaffect spontaneous symmetry breaking, cryogenic temperatures dooffer the advantage of overcoming thermal broadening to reveal theenergetic heterogeneity of different chromophores within one ringmolecule. In conjugated polymers, single chromophore spectra withlinewidths orders of magnitude narrower than the ensemble havebeen identified44,45. Analogously, we can resolve typical45 single-chromophore transitions in single rings at 4 K, as described inFig. 4a (black line). The spectrum consists of a strong, asymmetricpeak at 465 nm, followed by a series of similar but weakervibronics. Whereas the single molecule exhibits a linewidth ,2 nm,the ensemble solution spectrum in Fig. 2a (green line) spans.20 nm. The histogram of 0–0 transitions of 117 different singlemolecules (grey bars) closely matches the 0–0 electronic transitionin the ensemble (green line), which implies that the ensemble ismade up of distinct single-molecule transitions that differ becauseof varying interactions with the environment. Chromophores cantherefore be distinguished by their transition energy46.

0 5 10 15 20 25 30

–0.8

0.0

0.8

Time (s)

–0.8

0.0

0.80

20

40

60

PL

inte

nsity

(kH

z)Li

near

dic

hroi

smLi

near

dic

hroi

sm

0 1 2 3 4 5

–0.8

0.0

0.8

Time (s)

–0.8

0.0

0.8

Pot

entia

l

Generalized coordinate

c

b

a

Excitation

Excitation

Emission

Emission

0 Normalized occurrence 1

Figure 3 | Photoinduced fluctuations in exciton localization apparent in the

temporal dynamics of single-ring luminescence. a, Photomodification of the

ring can distort the excited-state potential quasi-statically, which leads to a

preferred polarization in emission. Changes in photomodification result in

switching of emission polarization. b, PL intensity, excitation and emission

anisotropy of a single-ring molecule. At short times into the measurement,

the emission polarization appears isotropic. The excitation anisotropy

remains zero as the emission anisotropy initially increases from zero and

subsequently exhibits random jumps, which correspond to changes in

emissive dipole orientation. c, Evolution of the linear dichroism histogram in

excitation and emission with time for 644 single ring molecules. The

excitation remains isotropic, whereas the emission becomes anisotropic with

time because of photoinduced localization in the excited state.

460470

λ (n

m)

420 440 460 480 500 520 540 560 580

Occurrence

Wavelength (nm)

Nor

mal

ized

PL

inte

nsity

0 30 60 90

Time (s)

0 60 120 1800

1

Nor

mal

ized

PL

inte

nsity

Time (s)

b c

a

4 K 4 K

4 K300 K

IH

IV

IH

IV

Figure 4 | Low-temperature PL spectroscopy of single rings showing

switching in transition energy and polarization. a, A typical single-molecule

PL spectrum at 4 K (black line) exhibits a dominant zero-phonon line and

discrete vibronic side bands, and an inhomogeneously broadened ensemble

solution spectrum at 300 K (green line). The distribution of zero-phonon

transition wavelengths for 117 single rings is superimposed. b,c, Fluorescence

spectral trace and intensity as a function of time resolved for horizontally

(red) and vertically (blue) polarized PL. Jumps in polarization, for example at

120 s, generally coincide with a change in emission wavelength because a

different chromophore emits on the ring (b). Reversible switching of the

polarization can also occur without any change in emission energy (c),

which implies that the energy of different chromophores on the ring remains

controlled by the same dielectric environment. Arrows mark the

anticorrelation in reversible switching between polarization planes.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1758 ARTICLES

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 5

Generally, emission switching between chromophores in a singleconjugated polymer chain is accompanied not only by a change indipole orientation, but also by a modification of transitionenergy46. This situation can also be observed in 1, for which weresolved linear dichroism spectrally by splitting emission intoorthogonal polarization components. To study the fluctuations inexciton localization and their impact on the chromophore energy,we focused on molecules that show polarized emission after pro-longed illumination (see the Supplementary Information forfurther discussion). Figure 4b gives an example of a fluorescencespectral trace that resolves the two polarization components(coloured red and blue). At 4 K, fluctuations in linear dichroismare slower than at room temperature. Weak spectral jitter, character-istic of single-chromophore transitions45, is visible in the trace. Aftera blinking event (in which the molecule turned dark) at 120 secondsinto the measurement, the emission polarization jumps, as does thetransition wavelength. This situation corresponds to dipole rotationby approximately 458 because only one polarization channel isactive before the event, but both are equally strong afterwards, asseen in the integrated emission intensity (lower part of Fig. 4b).The case is very different for the molecule shown in Fig. 4c.Discrete reversible switching in dipole orientation occurs withoutdiscernible spectral change, which leads to an anticorrelation ofhorizontal and vertical components. Spectrally, it appearsthat only one chromophore is emitting, even though the emissivepart of the ring rotates by �908. This observation suggests thatthe p system experiences a homogeneous environment onlength scales that exceed the size of the dimer, the emissive unitin the ring. Even though the exciton localizes to different parts ofthe molecule, in some situations the same effective dielectricenvironment is probed so that the transition energy remainsunchanged. This phenomenon could potentially arise from long-range electronic correlations existing in the bath47. Such effectsare, however, quite rare: out of 89 single molecules at 4 K, sevenshowed jumps in polarization without a spectral shift, and fourshowed a shift.

In conclusion, we have demonstrated that exciton localization inp-conjugated macromolecules is a fundamentally non-deterministicprocess that arises randomly on different monomer units. Thephenomenon is important to microscopic modelling of energy-transfer pathways in organic electronic devices, and may contributeto the origin of intramolecular interchromophoric electroniccoherences reported in conjugated polymers2; chromophores, thepolarizable species, are not necessarily static entities.

MethodsThe synthetic methods and characterization of the materials are described in theSupplementary Information.

Single-molecule emission was studied at room temperature in air or undercryogenic conditions in vacuo. For room-temperature measurements, the analytemolecules were embedded in a PMMA (Mn¼ 46 kDa, Sigma Aldrich) host matrix.The following steps were conducted. (1) Borosilicate glass coverslips were cleaned ina 2% Hellmanex III (Hellma Analytics) solution, followed by rinsing with MilliQwater. (2) The glass coverslips were transferred into an ultraviolet–ozone cleaner(PSD Pro Series UV, Novascan) to bleach the glass coverslips from residualcontaminant fluorescent molecules. (3) The analyte was diluted in toluene to single-molecule concentration (�10212 M) and mixed with a 1% w/w PMMA/toluenesolution. (4) The analyte–PMMA–toluene solution was spin coated dynamically(Laurell, WS-400-6NPP-Lite set at 2,000 r.p.m.) onto the glass coverslips, which ledto a film thickness of about 50 nm with an average analyte density of 40 individualmolecules in a range of 50 × 50 mm2. (5) Single-molecule emission was studiedunder ambient conditions. It was found that under the latter conditions, and underexposure to dry nitrogen at room temperature, the single-molecule emission wasreduced substantially. This is most probably because of the build-up of tripletexcitons, which can be quenched by molecular oxygen.

Fluorescence transients, including the linear dichroism in excitation andemission, of single ring molecules were recorded in a confocal fluorescencemicroscope. An inverted microscope (IX71, Olympus) with a high numericalaperture objective (NA¼ 1.49, APON 60XOTIRF, Olympus) was used. Theexcitation source was a fibre-coupled diode laser (LDH-C-405, PicoQuant) with a

wavelength of 405 nm in a quasi-continuous wave mode (pulsed excitation, 20 MHzrepetition rate). The excitation light was passed through a clean-up filter (HC LaserClean-up MaxDiode 405/10, AHF Analysentechnik) and a Glan–Thompsonpolarizer to provide linearly polarized excitation light. The polarization of theexcitation light was switched by an electro-optical modulator (3079-4PW, FastPulseTechnology Inc.) and an additional l/4 waveplate between horizontal and verticalpolarization every 500 ms, as described elsewhere48. The laser beam was expandedand collimated via a lens system and coupled into the oil-immersion objectivethrough the back port of the microscope and a dichroic mirror (RDC 405 nt, AHFAnalysentechnik). A diffraction-limited spot was generated with an excitation powerof 50 nW to ensure that all measurements were performed in the linear excitationregime, far below single-molecule saturation intensity. The fluorescence signal waseither split by a polarizing beam splitter (CM1-PBS251, Thorlabs) into twoorthogonal polarizing detection channels or by a 50/50 beam splitter into twoequivalent detection channels. Avalanche photodiodes from PicoQuant (t-SPAD-20) were used as detectors and the signal was recorded with a time-correlated single-photon counting module from PicoQuant (HydraHarp 400).

For low-temperature measurements, the ring molecules were dispersed in anoptically inert polymer matrix (Zeonex 480, Zeon Corporation) at concentrations of1026 g l21 and spin coated on quartz substrates in a glove box under a nitrogenatmosphere to yield film thicknesses of 20 nm. The samples were mounted on thecold finger of a He cryostat (ST-500, Janis Research Company Inc.) and kept under avacuum of 1027 mbar during the measurement at 4 K. Fluorescence was detectedwith a long working-distance microscope objective (7.7 mm, NA 0.55, OlympusAmerica Inc.) that projects the emission onto the entrance slit of a 50 cmspectrograph (ARC-1-015-500, Princeton Instruments) with a CCD camera(CoolSnap:HQ2, Princeton Instruments).

Full details of the experimental methods are given in the SupplementaryInformation.

Received 30 April 2013; accepted 14 August 2013;published online 29 September 2013

References1. Scholes, G. D., Fleming, G. R., Olaya-Castro, A. & van Grondelle, R.

Lessons from nature about solar light harvesting. Nature Chem. 3,763–774 (2011).

2. Hwang, I. & Scholes, G. D. Electronic energy transfer and quantum-coherence inp-conjugated polymers. Chem. Mater. 23, 610–620 (2011).

3. Lupton, J. M. Chromophores in conjugated polymers—all straight? Chem. Phys.Chem. 13, 901–907 (2012).

4. Hu, D. H. et al. Collapse of stiff conjugated polymers with chemical defects intoordered, cylindrical conformations. Nature 405, 1030–1033 (2000).

5. Schwartz, B. J. Conjugated polymers: what makes a chromophore? Nature Mater.7, 427–428 (2008).

6. Schwartz, B. J. Conjugated polymers as molecular materials: how chainconformation and film morphology influence energy transfer and interchaininteractions. Annu. Rev. Phys. Chem. 54, 141–172 (2003).

7. Bassler, H. & Schweitzer, B. Site-selective fluorescence spectroscopy ofconjugated polymers and oligomers. Acc. Chem. Res. 32, 173–182 (1999).

8. Tempelaar, R., Stradomska, A., Knoester, J. & Spano, F. C. Anatomy of anexciton: vibrational distortion and exciton coherence in H- and J-aggregates.J. Phys. Chem. B 117, 457–466 (2013).

9. Lippitz, M. et al. Coherent electronic coupling versus localization in individualmolecular dimers. Phys. Rev. Lett. 92, 103001 (2004).

10. Hernando, J. et al. Single molecule photobleaching probes the exciton wavefunction in a multichromophoric system. Phys. Rev. Lett. 93, 236404 (2004).

11. Tretiak, S., Saxenea, A., Martin, R. L. & Bishop, A. R. Conformational dynamicsof photoexcited conjugated molecules. Phys. Rev. Lett. 89, 097402 (2002).

12. Ruseckas, A. et al. Ultrafast depolarization of the fluorescence in a conjugatedpolymer. Phys. Rev. B 72, 115214 (2005).

13. Mossinger, D., Hornung, J., Lei, S., De Feyter, S. & Hoger, S. Molecularly definedshape-persistent 2D oligomers: the covalent-template approach to molecularspoked wheels. Angew. Chem. Int. Ed. 46, 6802–6806 (2007).

14. Mayor, M. & Didschies, C. A giant conjugated molecular ring. Angew. Chem. Int.Ed. 42, 3176–3179 (2003).

15. Mossinger, D. et al. Large all-hydrocarbon spoked wheels of high symmetry:modular synthesis, photophysical properties, and surface assembly. J. Am. Chem.Soc. 132, 1410–1423 (2010).

16. Jung, S. H. et al. A conjugated polycarbazole ring around a porphyrin. Angew.Chem. Int. Ed. 45, 4685–4690 (2006).

17. Simon, S. C., Schmaltz, B., Rouhanipour, A., Rader, H. J., & Mullen, K. Amacrocyclic model dodecamer for polyfluorenes. Adv. Mater. 21, 83–85 (2009).

18. Arnold, L., Norouzi-Arasi, H., Wagner, M., Enkelmann, V. & Mullen, K. Aporphyrin-related macrocycle from carbazole and pyridine building blocks:synthesis and metal coordination. Chem. Commun. 47, 970–972 (2011).

19. Cho, H. S. et al. Excitation energy transport processes of porphyrin monomer,dimer, cyclic trimer, and hexamer probed by ultrafast fluorescence anisotropydecay. J. Am. Chem. Soc. 125, 5849–5860 (2003).

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1758

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry6

20. Hori, T. et al. Giant porphyrin wheels with large electronic coupling asmodels of light-harvesting photosynthetic antenna. Chem. Eur. J. 12,1319–1327 (2006).

21. Sprafke, J. K. et al. Belt-shaped p-systems: relating geometry to electronicstructure in a six-porphyrin nanoring. J. Am. Chem. Soc. 133,17262–17273 (2011).

22. O’Sullivan, M. C. et al. Vernier templating and synthesis of a 12-porphyrinnano-ring. Nature 469, 72–75 (2011).

23. Yang, J., Yoon, M. C., Yoo, H., Kim, P., & Kim, D. Excitation energytransfer in multiporphyrin arrays with cyclic architectures: towards artificiallight-harvesting antenna complexes. Chem. Soc. Rev. 41, 4808–4826 (2012).

24. Hoffmann, M. et al. Enhanced p conjugation around a porphyrin[6] nanoring.Angew. Chem. Int. Ed. 47, 4993–4996 (2008).

25. Iyoda, M., Yamakawa, J. & Rahman, M. J. Conjugated macrocycles: concepts andapplications. Angew. Chem. Int. Ed. 50, 10522–10553 (2011).

26. Kawase, T. & Kurata, H. Ball-, bowl-, and belt-shaped conjugated systems andtheir complexing abilities: exploration of the concave–convex p2p interaction.Chem. Rev. 106, 5250–5273 (2006).

27. Kawase, T. et al. Cyclic [5]paraphenyleneacetylene: synthesis, properties, andformation of a ring-in-ring complex showing a considerably large associationconstant and entropy effect. Angew. Chem. Int. Ed. 46, 1086–1088 (2007).

28. Hoger, S. & Bonrad, K. 3-Cyanopropyl-dimethysilyl acetylene (CPDMSacetylene), a polar analogue of trimethylsilyl acetylene: synthesis andapplications in the preparation of monoprotected bisacetylenes.J. Org. Chem. 65, 2243–2245 (2000).

29. Bednarz, M., Reineker, P., Mena-Osteritz, E. & Bauerle, P. Optical absorptionspectra of linear and cyclic thiophenes—selection rules manifestation. J. Lumin.110, 225–231 (2004).

30. Bhaskar, A. et al. Enhancement of two-photon absorption cross-section inmacrocyclic thiophenes with cavities in the nanometer regime. J. Phys.Chem. B 111, 946–954 (2007).

31. Mena-Osteritz, E., Zhang, F., Gotz, G., Reineker, P. & Bauerle, P. Opticalproperties of fully conjugated cyclo[n]thiophenes—an experimental andtheoretical approach. Beilstein J. Nanotech. 2, 720–726 (2011).

32. Varnavski, O., Bauerle, P. & Goodson, T. Strong coupling in macrocyclicthiophene investigated by time-resolved two-photon excited fluorescence.Opt. Lett. 32, 3083–3085 (2007).

33. Zhang, F., Gotz, G., Winkler, H. D. F., Schalley, C. A. & Bauerle, P. Giantcyclo[n]thiophenes with extended p conjugation. Angew. Chem. Int. Ed.48, 6632–6635 (2009).

34. Kunz, R. et al. Exciton self trapping in photosynthetic pigment–proteincomplexes studied by single-molecule spectroscopy. J. Phys. Chem. B 116,11017–11023 (2012).

35. Tretiak, S. & Mukamel, S. Density matrix analysis and simulation of electronicexcitations in conjugated and aggregated molecules. Chem. Rev. 102,3171–3212 (2002).

36. Tubasum, S., Cogdell, R. J., Scheblykin, I. G. & Pullerits, T. Excitation–emissionpolarization spectroscopy of single light harvesting complexes. J. Phys. Chem. B115, 4963–4970 (2011).

37. Tubasum, S., Thomsson, D., Cogdell, R., Scheblykin, I. & Pullerits, T.Polarization single complex imaging of circular photosynthetic antenna.Photosyn. Res. 111, 41–45 (2012).

38. Bopp, M. A., Sytnik, A., Howard, T. D., Cogdell, R. J. & Hochstrasser, R. M. Thedynamics of structural deformations of immobilized single light-harvestingcomplexes. Proc. Natl Acad. Sci. USA 96, 11271–11276 (1999).

39. Mirzov, O. et al. Polarization portraits of single multichromophoric systems:visualizing conformation and energy transfer. Small 5, 1877–1888 (2009).

40. Stangl, T. et al. Temporal switching of homo-FRET pathways in single-chromophore dimer models of p-conjugated polymers. J. Am. Chem. Soc.135, 78–81 (2013).

41. Thomasson, D., Sforazzini, G., Anderson, H. L. & Scheblykin, I. G. Excitationpolarization provides structural resolution of individual non-blinkingnano-objects. Nanoscale 5, 3070–3077 (2013).

42. Forster, M., Thomsson, D., Hania, P. R. & Scheblykin, I. G. Redistribution ofemitting state population in conjugated polymers probed by single-moleculefluorescence polarization spectroscopy. Phys. Chem. Chem. Phys. 9,761–766 (2007).

43. Flors, C. et al. Energy and electron transfer in ethynylene bridged perylenediimide multichromophores. J. Phys. Chem. C 111, 4861–4870 (2007).

44. Feist, F. A., Tommaseo, G. & Basche, T. Observation of very narrow linewidths inthe fluorescence excitation spectra of single conjugated polymer chains at1.2 K. Phys. Rev. Lett. 98, 208301 (2007).

45. Schindler, F., Lupton, J. M., Feldmann, J. & Scherf, U. A universal picture ofchromophores in p-conjugated polymers derived from single-moleculespectroscopy. Proc. Natl Acad. Sci. USA 101, 14695–14700 (2004).

46. Muller, J. G., Anni, M., Scherf, U., Lupton, J. M. & Feldmann, J. Vibrationalfluorescence spectroscopy of single conjugated polymer molecules.Phys. Rev. B 70, 035205 (2004).

47. Wu, S. L., Liu, F., Shen, Y., Cao, J. S. & Silbey, R. J. Efficient energy transfer inlight-harvesting systems, I: optimal temperature, reorganization energy andspatial–temporal correlations. New J. Phys. 12, 105012 (2010).

48. Higgins, D. A., VandenBout, D. A., Kerimo, J. & Barbara, P. F. Polarization-modulation near-field scanning optical microscopy of mesostructured materials.J. Phys. Chem. 100, 13794–13803 (1996).

AcknowledgementsThe authors are indebted to the Volkswagen Foundation for providing collaborativefunding. A.V.A. and A.T. acknowledge financial support by the Fonds der ChemischenIndustrie. J.M.L. is a David & Lucile Packard Foundation fellow and is grateful for aEuropean Research Council Starting Grant (MolMesON, #305020).

Author contributionsA.V.A., A.I., D.K. and S.H. designed and synthesized the compounds. A.T., D.W., T.S., F.S.,J.V. and J. M. L. conceived, designed and performed the spectroscopy experiments andanalysed the data. S-S.J. and S.H. performed and interpreted the STM experiments. A.T.,J.V., S.H. and J.M.L. wrote the manuscript.

Additional informationSupplementary information and chemical compound information are available in theonline version of the paper. Reprints and permissions information is available online atwww.nature.com/reprints. Correspondence and requests for materials should be addressed toS.H. and J.M.L.

Competing financial interestsThe authors declare no competing financial interests.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1758 ARTICLES

NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 7